RESEARCH DOSSIER
Domestic Trade Credit Policy Loss Thresholds and Deductibles
Abstract. Domestic trade credit insurance transforms part of a seller's unpaid commercial receivables into a contingent claim against an insurer. The amount and timing of this transformation depend on the insured obligation, buyer credit limit, qualifying loss threshold, deductible, insured percentage, policy maximum, waiting period and allocation of recoveries. This monograph examines those elements as distinct contractual operations. It compares publicly accessible policy wordings from India and Australia, Canadian policy-document guidance, dated Indian and European regulatory instruments and New York insurance classification; it relates the contract evidence to economic research on monitoring, liquidity, trade finance and correlated losses. Explicit hypothetical claim ledgers, frequency-severity calculations and dependence scenarios establish how alternative retentions redistribute losses. The analysis demonstrates mathematically that a loss threshold can eliminate eligibility without eroding an annual deductible, that a per-loss deductible and an aggregate first-loss amount respond differently to fragmentation, and that the order of the insured percentage and deductible changes the indemnity. Even when the eventual nominal indemnity is identical, different waiting periods generate different liquidity exposures. The paper develops a source-grounded interpretation of established insurance mechanisms, not a proprietary theory or empirical estimate. The comparative sample identifies possible mechanisms rather than their market prevalence; all numerical examples are constructed and reproducible.
Keywords: Domestic trade credit insurance; trade receivables; nonqualifying loss; franchise; each-and-every-loss deductible; annual aggregate deductible; first loss; coinsurance; insured percentage; credit limits; protracted default; waiting period; recoveries; subrogation; liquidity; loss distribution; correlated defaults; claims administration; prudential supervision.
Introduction
A seller that delivers goods or provides services before payment acquires a financial asset and exposes its own working capital to the buyer's payment behavior. The nominal invoice amount measures the contractual payment claim, but the insured exposure is a more restricted quantity established by the sale, the insurance contract and the circumstances of the debtor's failure. A policy's advertised insured percentage therefore tells the reader relatively little about the protection available for any particular receivable. The practical subject of this monograph is the sequence of operations that separates unpaid commercial debt from collectible insurance indemnity.
The scope is domestic business-to-business trade, ordinarily involving a seller and buyer in the same national market. Currency, legal domicile, delivery locations and the place of payment can diverge, so actual domestic status must follow the contract's geographical definitions. The primary risk is commercial nonpayment. Export political risks, mortgage guarantees, consumer debtor insurance and independent documentary undertakings enter the study only where comparison illuminates the main subject. The parent Science, Credit, Surety and Financial Guarantee Insurance, provides the wider institutional classification.
Three questions organize the investigation. First, which observable events and contractual quantities determine whether a loss enters indemnity calculations? Second, how do fixed and proportional retentions operate individually and jointly over a buyer's debt and a policy period? Third, how does the retained position affect supplier liquidity, underwriting incentives, insurer accumulation and the interpretation of market statistics? The connection between all three questions is the distinction between risk transferred in principle and cash actually received under qualifying conditions.
The method is comparative documentary analysis with transparent financial arithmetic. It uses a purposive sample of accessible specimens and authoritative legal and institutional records, supplemented by the accessible abstracts of named economic studies. Document versions and access limitations are recorded in the evidence ledger. A specimen establishes the existence of a contractual mechanism; it cannot establish the prevalence of that mechanism among issued policies. Neither private schedules, insurer underwriting databases nor policyholder claim histories were available. The numerical analysis is conditional and explanatory; no sampling inference about actual premiums, claim rejection rates, waiting-period lengths or deductible distributions is attempted. Regulatory discussion is confined to the instruments and dates identified.
Conceptual and Historical Foundations
The receivable and the insurance relation
The fundamental commercial relationship is bilateral: the seller supplies and the buyer owes payment. The insurance relationship adds a third party that undertakes a conditional obligation to the insured creditor. The buyer's original obligation remains in place unless it is discharged under applicable law. The policyholder's receipt of indemnity changes who ultimately bears the economic loss and may change who has the right to pursue collection. These rights must be traced through the contract and the governing law.
A credit policy can encompass an entire portfolio or a defined subset. Whole-turnover policies can reduce the opportunity to insure only the buyers perceived as dangerous, but the scope may still be confined to specified classes of buyer or trade. Selective buyer and single-risk products address narrower exposure sets. Marsh's trade credit insurance overview (n.d.) also describes exceptional-loss protection built above ordinary bad-debt experience. These are descriptions of available market structures, not measured proportions of domestic policies.
The commercial obligation is established by the actual sale and its documentary evidence. An invoice can be valid in an accounting system while the buyer contests delivery, quantity, quality or the seller's authority. It can also be paid by an offset rather than cash. Insurance analysis therefore separates the existence of a legally enforceable debt, the buyer's inability or failure to pay it, the event covered by the issued policy, and the accounting amount remaining after adjustments.
Development of commercial risk transfer
The historical development pertinent to this investigation can be observed in the increasing differentiation of risk-transfer arrangements and supervisory responses. Short-term receivable risk exposes firms that sell on credit to events originating beyond their own balance sheets. The contractual retention becomes a means of deciding which recurrent losses the seller continues to finance and which unexpected losses enter a specialist insurance portfolio. This distinction is particularly important for a company whose receivables are replenished continuously rather than acquired as a fixed investment.
The 2020 pandemic illustrates the relevance of an older commercial mechanism to modern systemic crises. Boissay, Patel and Shin (2020), in Trade Credit, Trade Finance, and the Covid-19 Crisis, emphasize that participating firms bear much of the credit exposure associated with supply chains and that long chains are vulnerable to disruption. The central issue is the joint timing of receipts and obligations: when many buyers delay payment, a solvent supplier can face financing difficulties well before its ultimate credit losses are measured.
This historical perspective restrains a common misreading of trade credit protection. A supplier does not merely purchase compensation for a remote bankruptcy. Depending on the product, it participates in a continuing information and financial arrangement. Buyer review, shipment decisions, changes in credit limits, notice requirements and recoveries give the policy an operational dimension throughout the insured period. The deductible belongs within that continuing arrangement.
Instrument classification and jurisdictions
Insurance classification determines which risk-bearing institutions may write a product and which legal and prudential provisions are relevant. The European Union's Solvency II Annex I (Directive 2009/138/EC, EIOPA text updated 2021) places insolvency and export credit in class 14, with suretyship separately in class 15. This is a classification provision; it does not prescribe the deductible of an individual domestic trade credit policy.
New York Insurance Law section 6901 distinguishes statutory credit insurance and fidelity and surety insurance from financial guaranty insurance through specified exclusions. The archived credit insurance assignee opinion (New York State Insurance Department, 2000) addressed a particular nonrecourse purchase of commercial receivables and treated the described assignee policy as credit insurance. Its stipulated facts expressly exclude a guarantee of securitized debt instruments. The example establishes why the identity of the insured financial institution alone cannot settle the product classification.
For analytical purposes, a conventional receivable indemnity responds to a covered net loss of the creditor; a surety arrangement secures a principal's obligation to an obligee and may create recourse against the principal; a bank's independent guarantee or standby credit responds to the conditions of its own undertaking; a debt-service financial guaranty protects specified monetary obligations. Economic equivalence of two promised payment amounts is insufficient to establish equivalence of their legal triggers, defenses, collection rights or prudential treatment.
Loss Thresholds
From invoice balance to eligible loss
A scientific claim calculation begins with a reconciliation. Let G denote gross unpaid debt, A valid adjustments such as credits and offsets, R amounts recovered before assessment, and B the exposure eligible under the policy after applicable exclusions and the buyer limit. The simple expression X = max(0, G - A - R) defines a net debt in a deliberately simplified ledger. The amount B is then determined from X by the insured-debt definition, eligible invoices, security allocation, and the appropriate credit limit. In the examples below, B = min(X, C) for a debt-denominated buyer limit C; actual policies may establish a different order or a limit denominated in indemnity.
Suppose a buyer has unpaid invoices of 120,000 currency units. An agreed price adjustment reduces the balance by 8,000, and a pre-claim payment reduces it by another 12,000. The net debt is 100,000. If all remaining invoices are eligible and the buyer debt limit is 80,000, B is 80,000. A 90% insured percentage without deductible would yield 72,000. The retained loss is 28,000, including 20,000 outside the buyer limit and 8,000 within the limit. Expressed as a fraction of the net debt, the actual transfer is 72%, not 90%.
Tata AIG's trade credit policy wording (n.d., Articles I–II) illustrates how the definition of loss subtracts specified adjustments and pre-payment receipts and excludes certain invoice components. It also establishes a threshold that small losses do not exceed, and excludes such losses from contributions toward the deductible. The policy schedule supplies the relevant numerical amounts. The published wording is examined as a specimen; particular issued schedules and applicable endorsements are essential to claim interpretation.
A qualification threshold is an eligibility rule
A minimum loss threshold T determines whether a recognized loss belongs to the indemnity mechanism. If a contract excludes losses up to and including T, the qualifying amount is Q(B) = B when B > T and zero otherwise. The threshold functions as an entry gate. If the contract is a franchise under which the full otherwise eligible amount becomes indemnifiable after crossing the threshold, there is a discontinuity at T.
A deductible d, in contrast, removes a specified monetary amount from an eligible loss: the remaining base is max(0, B - d). With a 5,000 threshold and a 90% indemnity percentage, a loss of 5,001 would generate 4,500.90 under a pure franchise, assuming all other conditions hold. A 5,000 ordinary deductible on the same loss yields 0.90. These calculations differ by 4,500 even though the headline threshold is identical. The difference is the contract operation applied to the excess.
An insurer may combine both mechanisms. A loss can be excluded in its entirety when it is small, while a separate deductible applies to accepted losses, either individually or over the policy period. Under such language it is important to record excluded loss alongside eligible loss, since the insured still suffers the excluded amount even though it is absent from the insurer's accumulation ledger.
Equality requires explicit treatment. The Tata AIG specimen uses a non-exceedance test for the threshold. Generali Central's commercial policy wording (n.d., clause 14) describes exclusion of losses below the nonqualifying amount and chronological application of qualifying losses to the deductible. The present numerical examples adopt the strictly-greater-than rule for simplicity; they do not purport to resolve an actual wording ambiguity or apply identical boundary treatment to both specimens. A single monetary unit can decide attachment where a franchise is used, making exact language and rounding conventions consequential.
Buyer limits and aggregation
Three questions must be asked about any threshold: the amount measured, the entity or event at which measurement occurs, and the time window. If the minimum applies to each invoice, three unpaid invoices of 4,000 are individually below a 5,000 threshold. If the minimum applies to an admitted buyer loss aggregating all three, a 12,000 loss exceeds that threshold. The invoice count and debt amount are identical; the indemnity classification differs because the unit of aggregation differs.
Related companies present a similar issue. The legal buyer may be a particular subsidiary while credit risk is underwritten with reference to a corporate group. A buyer limit should be distinguished from any combined limit for several entities. One poorly identified corporate parent can lead to duplicate capacity, overstated portfolio diversification and inconsistent application of retained amounts. A research ledger must preserve legal entity identifiers as well as any underwriting group relationships.
Shipment dates matter when the policy attaches to risks arising from deliveries made during the policy period. A limit recorded at the date of claim may differ from the limit applicable when goods were dispatched. Debts invoiced during an earlier period may also be governed by another policy. The appropriate chronological rule derives from the issued wording. A mechanically attractive calculation that uses the wrong applicable limit or policy year is financially meaningless.
Temporal thresholds
Nonpayment is a condition observed at a date, whereas a claimable insurance event may be defined by insolvency proceedings, expiration of a specified waiting period or proof of persistent failure. Maximum payment terms govern the authorized credit extension; overdue-notification deadlines govern reporting; a waiting period governs the point at which an otherwise eligible nonpayment can become payable or mature into an insured event; and a claim submission deadline governs the filing of evidence. Each serves a separate function.
Article II.V of the Tata AIG wording cited above exempts insolvency from its stipulated waiting period. The example confirms that a waiting period need not apply uniformly to every insured cause. A legal insolvency event also does not guarantee immediate cash: the creditor may still need to establish its claim and the insurer may need to assess coverage.
For this reason the appropriate record contains at least delivery, invoice, contractual due, notification, claimable event, submission, approval, payment and recovery dates. A time series based solely on indemnity payment dates cannot recover the distribution of underlying buyer defaults. It measures a different process, filtered by contract maturity and administrative delay.
Deductible Mechanics
Loss-based and indemnity-based deductibles
Let p be the insured proportion, with 0 <= p <= 1. For an eligible loss B and a loss-based deductible d, indemnity is J = p max(0, B - d), before any policy payout cap. If the contract instead subtracts an indemnity-denominated deductible from gross otherwise payable indemnity, J = max(0, pB - d). These are ordinary alternative arithmetic operations, not descriptions of a universal legal rule.
On an eligible loss of 100,000, with p = 0.9 and d = 20,000, the first operation produces 72,000 and the second produces 70,000. Where both operations are above zero, the difference equals d(1 - p). As the insured participation becomes complete, the difference disappears; as the deductible or uninsured share rises, the difference increases. At low eligible losses, the indemnity-based amount reaches zero sooner, since the effective loss attachment is d/p whenever p is positive.
QBE Australia's SME whole-turnover wording (form QM3276-0422) expressly defines Each and Every as a scheduled deduction preceding application of the insured percentage. The same specimen includes buyer-specific definitions of insured loss and evidence for protracted default. This illustrates a verified loss-based architecture. The alternative ordering above is a mathematical comparison; no particular QBE contract is asserted to use it.
For positive loss above a loss-based deductible, the seller's ultimate retained portion within eligible debt is d + (1 - p)(B - d). The seller therefore has a fixed first component and a marginal proportional exposure. It is incorrect to calculate the retention as d plus the uninsured percentage of the entire loss: that would count part of the deductible twice.
Annual aggregate first-loss amounts
An annual aggregate deductible operates on cumulative qualifying losses rather than independently on every event. Define A as the initial loss-based aggregate deductible, Q_i as the qualifying amount of the i-th event in the contractually prescribed order, and S_i as the cumulative sum through that event. When recoveries and additional limits are temporarily ignored, incremental indemnity is
J_i = p[max(0, S_i - A) - max(0, S_(i-1) - A)].
The total over the term reduces to p max(0, sum(Q_i) - A). The equation telescopes, so the ultimate total is independent of event order under these restrictive assumptions. The timing is nevertheless dependent on order: the losses that exhaust retention arrive before indemnification. Recoveries, claim-specific limits, changing percentages and differing eligibility rules can invalidate the simplifying invariance.
An aggregate first-loss amount should also be distinguished from first-loss coverage in which an insurer pays an initial layer up to a limit. The latter places the insurance at the bottom of the loss distribution; a retention places it above an amount borne by the insured. The words first loss are insufficient on their own to identify which party bears the layer.
Reproducible claim comparison
Consider four qualifying-event candidates in chronological order, with net loss amounts of 3, 12, 30 and 60 thousand currency units. Assume full debt eligibility apart from the threshold, no recoveries, a 90% insured percentage and no binding payout limit. Contract A has a 5-thousand nonqualifying threshold and a 25-thousand aggregate loss deductible. Contract B combines the same threshold with a 10-thousand each-and-every-loss deductible. Contract C has the threshold but no deductible. Contract D uses the threshold and a 25-thousand indemnity-denominated aggregate deduction.
| Event | Net loss | Qualifying loss | Contract A payout | Contract B payout |
|---|---|---|---|---|
| First | 3 | 0 | 0 | 0 |
| Second | 12 | 12 | 0 | 1.8 |
| Third | 30 | 30 | 15.3 | 18 |
| Fourth | 60 | 60 | 54 | 45 |
| Total | 105 | 102 | 69.3 | 64.8 |
Amounts are in thousands. In A, event two uses 12 of the aggregate deductible and event three consumes its remaining 13. The third payout is 0.9 times 17. In B, the separate deductible recurs even after a previous claim has been paid. Contract C produces 91.8; D produces 66.8. Corresponding seller retentions across the entire 105 of net losses are 35.7, 40.2, 13.2 and 38.2 respectively.
The example establishes why a quoted aggregate deductible cannot be compared to an each-and-every amount without considering the count and size distribution of buyer losses. It also establishes why the underlying loss ledger must retain the losses excluded by the threshold. The insured has lost 105 in all four scenarios even though the insurer recognizes only 102 for the retention calculation.
If an annual payout cap of 50 is added to A, total indemnity becomes 50 rather than 69.3. The third event still generates 15.3 and the fourth is capped at 34.7 under the assumptions of chronological payment. The insured's total retention becomes 55. The aggregate deductible and maximum indemnity affect opposite ends of the same covered-loss layer.
Recovery and deductible reinstatement
Receivable losses are often provisional. A bankruptcy distribution, security realization, settlement or payment subsequent to indemnification can reduce the ultimate loss. The contractual allocation of recoveries is consequently as important as the initial calculation. A recovery can reduce the unpaid debt, reimburse the insurer for part of an admitted claim, restore an exhausted deductible, restore policy capacity where the contract permits, or combine several effects.
Consider a simplified portfolio with A = 25, an admitted loss of 20 and a subsequent recovery of 8 before another loss occurs. Under a rule restoring the aggregate retention for recovered retained losses, the amount used in the retention ledger declines from 20 to 12 and the amount of deductible still available increases from 5 to 13. A later 30 loss therefore generates 0.9 times 17 = 15.3, compared with 22.5 if the ledger failed to restore the deductible. Generali Central's wording, clause 16, expressly provides restoration for recoveries on deductible-retained losses. The simplified arithmetic illustrates its consequence; actual attribution of a recovery to a buyer loss and relevant period requires the policy.
QBE's comprehensive Australian wording (n.d., conditions 6 and 9) provides for retention of specified uninsured interests and a recovery allocation that takes account of insured debt, other sums owed and deductible effects. The contractual treatment of specified security proceeds is separately defined. This is evidence that a generic pro rata division of every recovery is not a sufficient claims methodology.
A useful control in a complete loss ledger is conservation of net loss across counterparties. If a debt of 100 is ultimately recovered to the extent of 20, the total economic loss attributable to unpaid principal is 80. If the original insurance payment of 72 is followed by a remittance of 14 to the insurer, final indemnity is 58 and the seller's retained loss is 22. Their sum is 80. The remittance figure is stipulated for illustration; a legally applicable allocation must be calculated independently. Collection costs, time value and premiums should be shown in separate columns rather than silently mixed into this conservation test.
Insured Retention
The marginal incentives of insurance
Retention has two financial roles: it defines the quantity of credit loss financed by the policyholder and influences the incentives of participants to control exposure. Under a constant insured percentage above a fixed deductible, a marginal increase in eligible loss is shared between seller and insurer. Below the deductible, the seller bears the marginal loss entirely. Once a payout cap has been exhausted, the seller again bears the entire marginal loss. The policy thus has different consequences for operating behavior at different points of the loss distribution.
Yang, Bakshi and Chen (2021, online 2020), in Trade Credit Insurance: Operational Value and Contract Choice, model the simultaneous roles of smoothing supplier cash flow and monitoring buyer creditworthiness. Their abstract reports that noncancelable arrangements can face a deductible trade-off, while a cancellation option changes incentives and can itself be exercised excessively. The result concerns a specified theoretical interaction. It cannot be generalized into an empirical claim that a particular deductible level or cancellation regime is optimal for every domestic supplier.
The supplier's control includes obtaining buyer information, setting and observing internal limits, considering new orders in the presence of arrears and preserving collection rights. The insurer's control may include buyer review and limit decisions. When incentives operate at both levels, retention is only one of several determinants. A large deductible can encourage scrutiny but can also make the policyholder unable to absorb ordinary adverse experience. The consequences depend on cash resources, the ability to verify information and how much each party can influence buyer selection.
Cash flow and the cost of time
An eventual loss distribution does not fully represent the financial consequences of nonpayment. A seller finances the entire delayed receipt until any buyer payment, borrowing or insurance indemnity arrives. An expected insurer payment may support borrowing under an appropriate arrangement, but that borrowing itself is conditioned on contractual and lender requirements. An indemnity payable after a waiting period is a contingent future cash inflow rather than cash already available.
Suppose a 100,000 receivable remains unpaid for 180 days rather than 60 days, and suppose the seller funds that entire balance throughout the additional 120 days at a simple annual rate of 8%. With a 365-day convention the incremental financing cost is 100,000 times 0.08 times 120/365 = 2,630.14. The example makes no assumption about the actual availability or typical cost of credit. If only part of the receivable is externally financed, the incremental cost is computed on that part. If interest compounds or rates vary, a dated cash-flow model is required.
A supplier with diversified receivables and unrestricted liquid investments may tolerate such a delay. A supplier with the same ultimate loss distribution and substantial payroll or inventory commitments may not. The cost of a high aggregate deductible therefore depends in part on when the retained losses accrue. Early exhaustion can concentrate the impact on a particular month even if the annual deductible is modest relative to yearly sales.
Claims administration and operational evidence
Coverage enforcement depends on the insured meeting the actual policy's documentary and administrative conditions. Export Development Canada's portfolio policy documents (2019) distinguish the policy text from the coverage certificate, list the deductible and insured percentage among certificate parameters, differentiate domestic-only and export-only events, and describe reporting and recovery duties. The text also distinguishes disputed debt from claims that can be clearly established.
A claim ledger ought to preserve evidence for the sale, buyer identity, shipment or service performance, contractual due date, invoices, adjustments, payments, applicable credit limit, relevant endorsement and any correspondence establishing a default or dispute. If an extension was granted, the original due date and revised date should both be retained. A disputed obligation must be separated from a purely financial default, even when the two can coexist.
A minimal reconciliation can be expressed as a table of consecutive contractual stages:
| Stage | Required observation | Principal analytical risk |
|---|---|---|
| Commercial debt | Sale, delivery and balance | Mistaking disputed or corrected debt for loss |
| Risk attachment | Policy, buyer and limit at relevant date | Using a later limit or wrong period |
| Covered event | Insolvency or contractual nonpayment trigger | Treating any arrears as payable |
| Qualification | Loss measured at required unit and threshold | Counting excluded small losses |
| Retention | Per-event or aggregate balance and percentage | Double counting or improper ordering |
| Settlement | Payout limit, evidence, waiting and decision dates | Treating approved amount as cash already received |
| Recovery | Remittance, costs and retention restoration | Overstating net indemnity or double recovery |
The table separates observations and calculations for purposes of reproducibility. It is a description of ordinary claims operations. It does not establish legal sufficiency in every jurisdiction.
International regulatory and institutional comparison
The Indian regulator's 2021 trade credit insurance guidelines came into effect on November 1, 2021. The dated text links coverage to an underlying commercial transaction, addresses buyer and policy limits, and requires risk management, claims procedures and buyer-level claims information. The comparison concerns the original 2021 framework and does not represent a certified consolidation of later Indian regulation. It shows how the supervisor can constrain the structure of insurance while detailed retained amounts remain contract-specific.
In Australia the sample includes a specifically titled SME whole-turnover specimen and a comprehensive specimen. Their different definitions of the covered loss and of retained interests are pertinent despite common jurisdictional associations. In the Canadian example, the public institutional explanation is evidence of the range of parameters in a coverage certificate, not a substitute for an issued contract. In New York, the statutory and opinion evidence addresses the classification of the protection itself. In Europe, the available prudential framework provides another comparison level rather than a schedule of standardized deductibles.
These sources are deliberately heterogeneous. They establish how contractual retentions, supervision and claims evidence intersect, but cannot be pooled into an international average insured percentage. A source describing a deductible in a certificate does not identify its amount; a regulation requiring a buyer limit does not identify how it is applied to a given loss. The absence of harmonized public schedules is itself a methodological reason to focus this research on mechanism rather than reported market averages.
Evidence Quality and Model Sensitivity
Expected loss above a deductible
For a nonnegative eligible severity X with finite expectation and no other constraints, the usual stop-loss identity is E[max(0, X - d)] = integral from d to infinity of P(X > x) dx. It follows by writing the positive part as the area of the indicator that X exceeds successive levels beyond d and exchanging a nonnegative integral with expectation. Multiplication by p yields expected payment under an uncapped loss-based deductible.
The corresponding expected retained amount is E[X] minus expected indemnity. This identity is valid only for X defined on the correct exposure basis. When thresholds remove observations, either X includes their zero indemnity effect or the separate probability of qualification must reweight the severity distribution. Treating a severity distribution conditional on paid claims as though it described all buyer defaults will bias estimates of retained loss and attachment probability.
Consider a hypothetical severity distribution per buyer-year with probabilities 0.90 for zero, 0.08 for 10, and 0.02 for 100, all amounts in thousands. Expected underlying loss is 2.8. For a fixed loss deductible of 15 and p = 0.9, expected indemnity is 0.02 times 0.9 times 85 = 1.53. Expected seller retention is 1.27. At d = 5, expected indemnity becomes 0.08 times 4.5 plus 0.02 times 85.5 = 2.07. At d = 0, it becomes 2.52. The premiums needed to support these payments are not determined by the examples; expenses, capital, timing, reinsurance and the market's pricing conditions are omitted.
Because loss severity is discrete, the marginal impact of changing d is piecewise. Where the deductible is strictly between 10 and 100, only the high-loss event responds. A premium comparison that assumes all loss severities change linearly with d would mischaracterize such a distribution. In an observed portfolio the loss distribution is more complex, but the analytical distinction remains.
Annual attachment under dependent defaults
An aggregate deductible needs a distribution of total qualifying loss. Summing each buyer's expected loss provides the mean of that total regardless of dependence, but generally cannot determine expected indemnity above the annual threshold.
Assume four independent buyers, each with a 10% probability of a loss of 40 thousand and otherwise zero. Let A = 100 thousand, p = 0.9, full eligibility and no cap. The expected total underlying loss is 16 thousand. If the number of defaults K follows a binomial distribution, only K = 3 or 4 produces a payment: respectively 18 and 54 thousand. The probabilities are 0.0036 and 0.0001. Expected indemnity is therefore 0.0702 thousand, or 70.20 currency units.
Now assume the four defaults are perfectly positively dependent, so that all four occur together with probability 0.1 and none with probability 0.9. Each individual marginal probability and severity, and the expected total credit loss, are unchanged. Indemnity is 54 thousand in the joint default state and expected indemnity is 5.4 thousand. The expected payment is much greater than under independence, while total expected credit loss remains 16 thousand. In this constructed comparison the seller's expected residual is 15.9298 under independence and 10.6 under perfect dependence.
The scenarios are endpoints for a specific hypothetical arrangement rather than calibrated bounds for all credit policies. More complex dependence, varying exposure, reinstatement, occurrence limits and insurer failure may alter outcomes. An aggregate cap, in particular, truncates the upside of insurer indemnity in severe states. The comparison nevertheless establishes the scientific necessity of accounting for dependence before interpreting the expected payout of an aggregate retention.
Concentration and supervisory stress
A seller's exposure distribution and an insurer's underwriting exposure distribution are different. A single insured buyer can be a modest portion of one supplier's sales but be present in many insured portfolios. The insurer may also have several buyers exposed to the same sector, financing source or trade chain. Retention does not eliminate these dependencies, although it changes the contractual payment conditional on them.
Article 134 of Delegated Regulation (EU) 2015/35, presented in EIOPA's credit and suretyship risk sub-module, combines separate large-default and recession capital stresses. The dated rulebook text includes identifying large credit exposures on a net loss-given-default basis after specified reinsurance and special-purpose-vehicle recoverables. These are prudential calculations at insurer level, not an insured's deductible formula or an estimate of its contractual indemnity.
The insurer's underwriting limit, expected ultimate indemnity, gross claim payment, net payment after reinsurance and regulatory capital measure consequently need separate denominators. Reinsurance can reduce the insurer's net financial exposure without increasing the direct policyholder's rights. A deductible can reduce expected claim payments without proportionately reducing the tail risk arising from correlated defaults.
Limits of empirical inference
The ESRB's 2022 macroprudential issues note discusses trade credit as a source of corporate finance and reports, based on earlier estimates, that sellers ultimately retained 85% of domestic trade credit risk. Its chart uses total domestic and international receivables as a common denominator. This historical estimate is not the uninsured percentage of domestic insurance contracts and should not be interpreted as a contemporary market share of a particular retention structure.
Van der Veer (2015), in the DNB loss shocks working paper, uses bilateral export-insurance data from 1992–2006 and reports a decline in the share of exports insured and an increase in premium levels associated with claims shocks. The related peer-reviewed loss shocks article (2019) reports evidence of transmission across markets after extreme losses. These are connected research versions rather than independent replications. Their domain is export insurance and their accessible abstracts do not identify the isolated effect of domestic deductible changes.
Van der Veer's separate private export credit insurance study (2015) reports positive effects of private export credit insurance on exports using trade models and a bilateral dataset. This supports the relevance of the insurance instrument to trade flows within the study's domain. Extrapolating the magnitude to domestic firms, or interpreting it as an estimated effect of a higher deductible, would exceed the evidence. The monograph's domestic conclusions are based on the contract operations and explicit financial calculations, not such an extrapolation.
A causal study of deductible changes would require a suitable comparison group and attention to policy selection. Suppliers may choose high retentions because they are financially stronger, or because their buyer portfolios differ. Insurers may change terms precisely when they detect worsening credit quality. Observing a higher claim rate after a lower deductible could reflect the altered eligibility and reporting threshold rather than a change in actual defaults.
Data requirements and reproducibility
To estimate an aggregate loss distribution, an investigator needs a denominator including buyers and invoices without losses, the exposure time, the covered and uncovered amounts at risk, default definitions, and recovery development. If only insurer files for paid claims are available, below-threshold losses and losses within deductible may be systematically absent. This truncation differs from right-censoring, in which an unresolved claim or later recovery remains unobserved as of the data extraction date.
At a minimum, a data model should connect policy and endorsement identifiers to buyer legal entities and corporate groups, trade transaction dates and amounts, eligible loss calculations, qualifying event types, waiting dates, deductible ledger entries, payout cap usage and final recoveries. Each monetary field requires a currency and the relevant conversion date if translation is involved. No artificial zero should replace a missing claim report or unavailable exposure amount. A zero is an observed quantity; missingness is a statement about the available information.
A useful validation approach would compare aggregate policy-period outputs with independent transaction-level and buyer-level reconciliations. An arithmetic test should exercise cases just below, at and above a threshold, zero and full insured proportions, zero deductible, exhausted payout cap and restoration following a retained-loss recovery. Time stamps should establish whether the ledger follows the required sequence. Sensitivity analysis should vary both amounts and inclusion rules: a change in the unit of aggregation may be more consequential than a change in the monetary threshold.
The main evidentiary limitation of this study is the absence of representative issued schedules and anonymized claim-level observations for domestic businesses. Some PDF visual pages could not be retrieved, although relevant extracted text was available. Some scholarly sources were available only as abstracts. The results therefore establish conditional arithmetic, documented contract examples and source-scoped institutional comparisons. They do not establish the frequency, current pricing or causal market effects of a given deductible.
Integrated Scientific Synthesis and Research Agenda
Retention is a multidimensional allocation
The preceding analysis permits an integrated interpretation of the contract. The insured's position comprises amounts outside the insured debt definition, exposure above applicable buyer limits, losses failing the qualification threshold, the deductible layer, the proportional share of the remaining covered loss, excess above maximum indemnity and the financial cost of waiting. A simple final net loss reconciliation combines the unpaid debt and net indemnity after recoveries, while a more complete economic analysis accounts separately for the funding cost, premiums and collection expenses.
Consider again the 105-thousand pattern of four losses. Under the aggregate example, the insured bears the 3 excluded loss, 25 of qualifying deductible and 7.7 of proportional retention. These sum to 35.7. When the policy payout cap falls to 50, the insured bears another 19.3 because of the cap. The policy has not changed the amount of underlying buyer loss. It has changed the ownership and timing of the economic burden. A complete evaluation identifies every component without counting the same amount twice.
This decomposition is useful for public descriptions of insurance. A stated insured percentage measures the fraction of an applicable base rather than the fraction of all commercial debt risk transferred. The credit limit, eligibility and threshold define that base; the deductible and cap shape its payoff. An illustration that begins with an insured percentage but ignores these operations is incomplete.
Frequency, severity and the supplier's financing horizon
The comparison reveals distinct consequences for three hypothetical portfolios. A portfolio with many relatively small independent losses may never exhaust a high aggregate first-loss amount, despite regular adverse experience. A concentrated portfolio with a few large buyers may attach after a single severe event. A cyclical portfolio with clustered losses may attach frequently precisely when other operating cash flows are under stress. The nominal insured percentage and annual turnover may be identical in all three cases.
This explains why the economic usefulness of a deductible requires an exposure profile. Expected annual loss alone cannot distinguish losses that occur evenly from losses concentrated into a short interval or common default cluster. Similarly, a seller's cash reserves cannot be evaluated merely as a fraction of yearly sales; required liquidity depends on the dates of outstanding obligations and receipts.
A scientifically well-defined comparison of policy structures would hold eligible exposure and policy duration constant, then calculate expected indemnity, higher loss quantiles, probability of attachment, expected time to first payment, maximum simultaneous retained exposure and the conditional distribution of recoveries. The results would be compared with premium and other contractual costs on an explicitly dated basis. Such a comparison is a conventional financial exercise, not an assertion of a new risk measure.
Institutional implications
For insurers, a recurring fixed deductible may reduce claims processing on small losses or change the mix of reported claims. An annual deductible can separate normal loss experience from a higher layer, but the calculation depends on documentary treatment of losses below the gate and on the restoration of retained amounts. For policyholders, both structures leave obligations that may require liquidity before claims are settled. Changing one element without examining others can create unintended risk concentrations or change the practical role of insurance.
For accounting and financial analysis, the existence of insurance does not by itself settle the valuation of a trade receivable or justify a uniform offset. The buyer obligation, insurer obligation and recognized accounting amounts have different bases. Recognition of insurance proceeds, treatment of credit enhancements and availability of financed cash depend on the applicable accounting rules and contractual arrangements. No accounting conclusion about any particular entity is made here.
For supervisors, the appropriate distinction is between direct contractual protection and the resilience of the institutions that provide it. An individual policy might compensate a creditor in a severe loss state, yet a collection of correlated policies can expose the insurer to simultaneous claims. Oversight of aggregation and solvency therefore complements, rather than substitutes for, precise policy wording. The historical export studies and the ESRB note make this question relevant, but they do not replace domestic claims evidence.
Research agenda
There is a concrete need for comparative datasets that preserve the distinction between nonqualifying losses, retained eligible losses and payable losses. Research could begin with a public, versioned corpus of policy wordings and endorsements, coding the unit of aggregation, threshold equality, deductible order, insured percentage base, waiting period and recovery rules. This corpus would establish structural possibilities, not market weights.
A second project could collect anonymized domestic buyer-policy-period observations under agreements protecting commercial confidentiality. It would require actual issued schedule terms, insured exposures, claims and nonclaims, reporting and payment dates, recoveries and explicit flags for disputed debt. Such a dataset would permit evaluation of the association between policy form and both underlying nonpayment and insurer payout. It would also allow identification of differences in data quality across product types.
A third investigation could exploit externally imposed or predetermined term changes, where genuinely comparable exposures are observable before and after the change. Selection into insurance, changed credit limits, changes in premium and common sector shocks would need to be controlled. Without a credible identification strategy, the proper conclusion would remain associational. Separate analysis of supplier liquidity should retain the time to payment rather than treating all monetary indemnities as instantaneous.
The research agenda is a consequence of the evidence boundaries established here. The required observations and potential comparisons are specified to make future work falsifiable and reproducible. No unsupported scientific priority, proprietary doctrine, market-wide optimal retention or empirical discovery is claimed.
Conclusion
Domestic trade credit loss thresholds and deductibles determine the allocation of net commercial credit loss through a sequence of eligibility, attachment and payment operations. A minimum-loss condition can exclude a loss without reducing the deductible balance. An each-and-every-loss deductible attaches to a specified event or loss unit. An annual aggregate deductible accumulates eligible amounts under a policy-period rule. The insured percentage and maximum payout further transform the loss into a contractual indemnity. Waiting periods and claim conditions determine when the financial consequence of the transfer is realized; recoveries can alter ultimate retained loss and, under particular language, restore the first-loss layer.
The comparative documents establish that these operations have concrete institutional and contractual counterparts. The arithmetic demonstrates that contracts with the same insured percentage can produce materially different net indemnities and that the dependence structure of buyer defaults can alter expected aggregate payments while individual expected losses remain unchanged. The timing examples show that nominal loss compensation and liquidity protection are distinct. The scientific result is an integrated explanation of established contractual mechanisms grounded in transparent assumptions and dated evidence.
This investigation supports precise comparison of domestic receivable protections at the level of actual policy language, buyer group and loss ledger. It also establishes limits to conclusions drawn from aggregate trade finance estimates, accessible abstracts and policy specimens. Issued schedules and representative claims data are needed to estimate empirical effects or prevalence; legal enforceability requires the applicable jurisdiction and actual contract. The appropriate object of scientific measurement is therefore the complete allocation of loss and timing of cash flows, with explicit conditions and denominators.
References
Generali Central Insurance (n.d.). Trade Credit Insurance: Commercial Policy Wording. Product identifier IRDAN132CPTC0020V01201920. https://www.generalicentralinsurance.com/downloads/commercial-insurance/trade-credit-insurance/policy-wording/trade-credit-insurance-policy-commercial-policy-wordings.pdf.
Tata AIG General Insurance Company Limited (n.d.). Trade Credit Insurance Policy Wording. https://tata-cms.s3.ap-south-1.amazonaws.com/Policy_Wordings_trade_credit.pdf_894c45e0b2.pdf.
QBE Insurance (Australia) Limited (2022). SME Trade Credit Insurance Policy: Whole Turnover Australia. Form QM3276-0422. https://nci.com.au/app/uploads/2025/06/QM3276-0422-SME-Trade-Credit-Policy-Whole-Turnover-Australia.pdf.
QBE Insurance (Australia) Limited (n.d.). Comprehensive Trade Credit Insurance Policy Wording, Australia/Asia-Pacific. https://www.nci.com.au/app/uploads/2022/01/QBE-Comprehensive-AP-Policy-Wording-AUS.pdf.
Export Development Canada (2019). Policy Documents: Portfolio Credit Insurance. Last modified March 12, 2019. https://www.edc.ca/en/support/topics/credit-insurance/managing-your-policy/documents.html.
Insurance Regulatory and Development Authority of India (2021). IRDAI (Trade Credit Insurance) Guidelines, 2021. IRDAI/NL/GDL/MISC/244/09/2021, September 8; effective November 1, 2021. https://www.actuariesindia.org/sites/default/files/inline-files/11.%20IRDAI%20%28Trade%20Credit%20Insurance%29%20Guidelines%2C%2020212021-09-08.pdf.
European Parliament and Council (2009). Directive 2009/138/EC (Solvency II), Annex I, Classes of Non-Life Insurance. EIOPA Single Rulebook text, updated May 3, 2021. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2408_en.
European Commission (2015). Commission Delegated Regulation (EU) 2015/35, Article 134, Credit and Suretyship Risk Sub-Module. EIOPA Single Rulebook text, updated March 26, 2024. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5754_en.
New York State (n.d.). Insurance Law § 6901: Definitions. New York State Senate, Open Legislation. https://www.nysenate.gov/legislation/laws/ISC/6901.
New York State Insurance Department (2000). OGC Opinion No. 00-06-08: Credit Insurance Policy Issued to Financial Institution. New York Department of Financial Services archive. https://www.dfs.ny.gov/insurance/ogco2000/rg006161.htm.
Yang, S. Alex, Nitin Bakshi, and Christopher J. Chen (2021; online July 1, 2020). Trade Credit Insurance: Operational Value and Contract Choice. Management Science 67(2): 875–891. DOI: 10.1287/mnsc.2019.3571. https://pubsonline.informs.org/doi/10.1287/mnsc.2019.3571.
European Systemic Risk Board (2022). Issues Note on Macroprudential Aspects of Trade Credit Insurance. August 2022. https://www.esrb.europa.eu/pub/pdf/reports/esrb.issuesnoteonmacroprudentialaspectstradecreditinsurance202208~eaa8c9c764.en.pdf.
Boissay, Frederic, Nikhil Patel, and Hyun Song Shin (2020). Trade Credit, Trade Finance, and the Covid-19 Crisis. BIS Bulletin No. 24, June 19. Bank for International Settlements. https://www.bis.org/publ/bisbull24.htm.
Van der Veer, Koen J. M. (2015). Loss Shocks and the Quantity and Price of Private Export Credit Insurance: Evidence from a Global Insurer. DNB Working Paper No. 462, February 26. https://www.dnb.nl/en/publications/research-publications/working-paper-2015/462-loss-shocks-and-the-quantity-and-price-of-private-export-credit-insurance-evidence-from-a-global-insurer/.
Van der Veer, Koen J. M. (2019). Loss Shocks in Export Credit Insurance Markets: Evidence From a Global Insurance Group. Journal of Risk and Insurance 86(1): 73–102. DOI: 10.1111/jori.12197. https://ideas.repec.org/a/bla/jrinsu/v86y2019i1p73-102.html.
Van der Veer, Koen J. M. (2015). The Private Export Credit Insurance Effect on Trade. Journal of Risk and Insurance 82(3): 601–624. DOI: 10.1111/jori.12034. https://ideas.repec.org/a/bla/jrinsu/v82y2015i3p601-624.html.
Marsh (n.d.). Trade Credit Insurance. UK market practice overview. https://www.marsh.com/en-gb/services/trade-credit/expertise/trade-credit-insurance.html.