RESEARCH DOSSIER
Underwriting Delegation Data Reporting and Bordereaux Governance
Abstract. Delegated underwriting disperses insurance binding decisions across legally and operationally distinct organizations while the risk-bearing insurer retains responsibility for the resulting portfolio. Bordereaux governance determines whether the insurer can reconstruct these decisions, identify unreported exposures, reconcile financial transactions, estimate emerging liabilities, and intervene before risk appetite is exceeded. This monograph develops a comparative documentary analysis of delegated-market reporting, supervisory expectations, statutory managing general agent obligations, data quality scholarship, and actuarial practice. It distinguishes risk, premium, and claims records; evaluates granular and aggregate reporting, contract-version semantics, effective-time and receipt-time controls, independently anchored reconciliation, subscription allocations, correction provenance, and claims development. Synthetic worked examples demonstrate compensating accounting errors, delayed concentration recognition, and the consequences of missing exposure denominators. Comparative findings concern Lloyd's, international supervisory standards, the European Union, Virginia, and Australia. No insurer-level proprietary data, newly conducted survey, or causal estimate is claimed. The contribution consists of a critical synthesis of existing methods and their implications for accountable portfolio oversight.
Keywords: delegated underwriting; bordereaux; coverholder; managing general agent; data governance; data quality; risk accumulation; premium reconciliation; claims development; reporting latency; underwriting oversight; insurance supervision.
Introduction and disciplinary foundation
Delegation as an information problem
A bordereau is a structured report of insurance business or transactions supplied by one participant to another, commonly under a binding authority, facility, or reinsurance arrangement. Bordereaux constitute the plural; individual reports may describe risks, premium movements, claims, or combined records. Their institutional significance resides in the relationship between the documented obligations and the insurer's independent knowledge of its liabilities. The delegated underwriter can create a valid insurance obligation before the insurer's central information systems record it. Consequently, the timing of information delivery and the completeness of the reporting population affect the insurer's practical control over its balance sheet.
The principal problem is threefold. First, which observations must an insurer receive to understand its exposure and claims obligations? Second, against what independent records can reported observations be tested? Third, how can data correction and escalation preserve the insurer's ability to respond to a changing portfolio? The scope is primary insurance delegated underwriting and its interfaces with premium collection, claims administration, financial reporting, and portfolio accumulation. Treaty reinsurance bordereaux, the legal validity of individual binding decisions, and the full methodology of claims settlement are adjacent subjects with their own research objects.
Delegated authority can give the risk bearer access to specialist expertise, customer segments, and distribution that would be expensive to develop internally. The arrangements create a division between decision location and risk-bearing capacity. Lloyd's (n.d.-a), in its delegated underwriting guidance, explicitly connects delegation with retained responsibility for underwriting results and with the need to agree information requirements in advance. The practical implication is that the reporting contract is part of the underwriting arrangement itself. The amount and quality of retained knowledge determine the range of decisions the carrier can responsibly delegate.
The economic interpretation draws on Jensen and Meckling's (1976) analysis of agency costs. Their original paper addresses ownership and agency relationships rather than insurance bordereaux; its abstract and journal identity establish the general analytical problem of separation and control. Applying that perspective to insurance suggests a conflict between the benefits of local discretion and the costs of monitoring decisions taken with another institution's capital. Even perfectly aligned participants face monitoring costs because they possess different records, systems, time horizons, and operational capacities. The incentive problem intensifies where remuneration depends on short-term premium volume or provisional underwriting profit and the risk bearer experiences losses over a longer period.
A principal can design onerous reporting rules and still fail to observe meaningful deviations if the reported variables bear little relation to the delegated decisions. Conversely, inexpensive, independently verifiable information may provide substantial control. The insurer should therefore evaluate reporting requirements in terms of information value, verification cost, and the consequences of error. An economically efficient arrangement retains the information necessary for intervention while allowing the intermediary to exercise the authorized commercial discretion for which it was selected.
Method and evidentiary boundaries
The research combines comparative documentary interpretation, a structured semantic examination of reporting artifacts, and deterministic synthetic examples. The documentary population consists of selected accessible official market and supervisory pages, a state statute, an actuarial standard, and scholarly abstracts. The UK Financial Conduct Authority's 2015 thematic review is used as historical observational evidence, within its stated sample. The legal comparison concerns specifically identified instruments and jurisdictions; it does not attempt a complete jurisdictional census or establish legal advice. The relevant institutional documents were inspected in the research session; where full scholarly texts and ancillary templates were unavailable, the discussion is confined to accessible abstracts or guidance summaries. No institutional interview, external peer review, or newly collected commercial dataset is represented.
Three kinds of propositions are separated. Documentary findings identify requirements or statements made by a named authority. Analytical inferences derive consequences for the information architecture of insurance. Numerical examples calculate results from explicit hypothetical transactions and limits. The arithmetic demonstrates mechanisms and does not estimate their prevalence. This distinction is particularly important for decisions about cyber exposures, catastrophe accumulation, and regulatory compliance, where data availability and contractual wording shape the validity of a conclusion.
Bordereaux record design
Risk identities and exposure observations
The record design begins with the unit of observation. A policy number identifies an instrument of coverage; one policy may insure several locations, vehicles, insured persons, or contractual interests. A risk record can represent a location, a coverage section, an insured item, or a modeled exposure. A policy term may contain multiple transactions after inception. A claim file may contain several coverage components and several transactions for payments, reserve adjustments, or recoveries. These identities have different cardinalities, and each reporting grain requires its own stable key.
A risk bordereau should permit reconstruction of exposure at the required level of detail. For a property portfolio this can require location, occupancy, construction characteristics, applicable perils, effective dates, currency, relevant insured values, limits, deductibles, participation and territory. Liability risks need fields appropriate to the business activity, jurisdiction, coverage trigger, retroactive date when relevant, coverage section, occurrence or aggregate limit, and period of insurance. A cyber book may require industry, revenue band, relevant dependency identifiers, security attributes and contract terms. The schema is purpose-dependent: a physical address might be indispensable for flood accumulation but inappropriate for an aggregated commercial reporting task with no lawful need for personal identifiers.
The market baseline supplied by Lloyd's (n.d.-c) Reporting Standards identifies risk, premium, claims, tax and regulatory information within its reporting program. The page describes the application of core requirements to binding authority and coverholder appointment agreements incepting from July 2017. Its detailed user guide and applicable country and class rules determine actual mandatory fields. The record recommendations developed here are analytical design choices; they are not presented as a reproduction of that guide or as a globally mandatory field list.
Lloyd's (2019), in the Version 5.2 user guide, distinguishes risk information from paid premium transactions and permits reuse of information already supplied (p. 8). Its field definitions distinguish a certificate reference and risk inception (p. 51) from an endorsement or cancellation effective date (p. 61). The dated guide illustrates the semantic differences between contract identity, coverage time and transaction time; subsequent applicable requirements must be determined separately.
An insurer should distinguish the legal contracting party from the insured risk identifier, the producer and delegated underwriter from one another, and the carrier or syndicate participation from the whole-market placement. Identifiers need scoped uniqueness. Policy 000123 at Coverholder A and policy 000123 at Coverholder B can be distinct legal contracts. Assigning a global key from the policy number alone collapses them. Creating a composite key from provider, authority agreement, policy, section and an appropriate term identifier can preserve their distinction. Keys must remain stable when descriptive attributes change, including names, addresses, or amended risk classifications.
Premium and claims records
A premium record represents a defined financial measure or transition. Written premium records the consideration attributable to written coverage according to the adopted reporting convention; earned premium allocates the amount to the expired exposure period under an appropriate methodology; collected premium concerns received cash. Commission, taxes, fees, cancellations, return premiums, installment schedules and remittances can alter the apparent relation between these measures. A coherent dataset specifies the premium basis rather than labeling all three simply as premium.
A claims bordereau can communicate loss date, report date, claim identity, policy and coverage reference, status, payments, case reserves, recoveries, expense amounts and currency. An insurer needs the convention by which losses are reported: incremental movements versus cumulative amounts; gross versus net of recoveries or reinsurance; inclusion or exclusion of allocated expenses; and valuation date. A change in case reserve may be negative in an incremental report even when the outstanding liability remains positive.
The relation between reported claims and financial risk is nontrivial. The sum of claim-level paid and outstanding values provides a measure of reported incurred losses on a compatible valuation basis. It excludes any separately estimated liabilities for unreported claims and future development when those components are absent. The complete actuarial estimate of ultimate cost therefore requires further analysis. A claim-level reporting feed that is satisfactory for individual claim administration can still lack the earned exposure denominators and maturity history required for pricing.
Under a delegated arrangement the policy register, financial schedule, and claims register should support cross-reference without forcing all observations into one flat table. A row-per-risk table joined directly to a row-per-payment table creates multiplication whenever a policy contains multiple insured locations. At the natural grain of each table, exposure, premium, and claims can be aggregated deliberately before joining. Where premium allocation across insured items is required, the allocation rule and residual amounts must be specified.
Data dictionary and reporting responsibilities
The following conceptual dictionary specifies purposes rather than a proprietary market template.
| Record group | Illustrative key or grain | Essential interpretive attributes |
|---|---|---|
| Authority | Agreement, period, version | Effective dates, delegated scope, approved limits, participants |
| Risk | Policy term and insured item or coverage | Exposure units, peril, location or activity, contract terms, status |
| Premium | Unique financial movement or defined balance | Signed amount, currency, transaction date, premium basis, commission and tax treatment |
| Claim | Claim and coverage component | Loss and report dates, status, link to policy |
| Claim movement | Unique movement and valuation date | Payment, reserve or recovery, signed value, currency, expense basis |
| Participation | Risk or contract and insurer share | Effective share, layer, gross or signed basis |
| Submission | Provider, reporting period, revision | Receipt time, schema version, control totals, reconciliation status |
Each mandatory item should have a named generating system, accountable owner and validation obligation. The person producing the report may not control the original exposure classification; the reporting team may possess only a mapping of a value created in the underwriting system. Correcting that value in the insurer's reporting warehouse alone leaves the operational source incorrect. Data stewardship therefore requires a distinction between the owner of the economic fact, the custodian of the reporting transmission, and the owner of the carrier's derived analysis.
Data mapping and standards
Semantic interoperability
Data integration involves translating the meaning of observations as well as their syntax. Two parties can use identical columns named inception date while one reports the inception of a binding authority and the other the effective date of an individual risk. A standardized date format does not resolve this mismatch. Similarly, a territory field may refer to regulatory risk location, policyholder domicile, or coverage geography. The wording expectations for delegation published by Lloyd's (n.d.-d) distinguish these territorial concepts and address information sharing with following markets.
Mapping should document origin field, destination field, source definition, code conversion, unit, currency, date basis, treatment of null values, derivation and version. Preserve the raw submission and the normalized representation. A transformation that alters risk class from a local code to an insurer code should be reproducible with the mapping table that applied at ingestion. Ambiguous or unmapped codes belong in explicitly visible exception categories. A default category that silently absorbs unknown values can alter underwriting limits, catastrophe mapping and performance assessments.
Wang and Strong (1996), in Beyond Accuracy, investigated dimensions of data quality using survey and sorting methods. Their accessible abstract distinguishes intrinsic, contextual, representational and accessibility dimensions. Applied to bordereaux, a valid numeric field can be inadequate for a decision if it uses the wrong currency or refers to the wrong effective period. A valid report available only after a portfolio restriction should have been triggered is deficient for that decision. These are authorial applications of their general results, not claims that the original investigation studied delegated insurance.
A shared schema reduces the number of bespoke mappings only when participants implement definitions consistently. Where ten submitters each supply a different local occupancy classification, a carrier can maintain ten mappings to a common reference dictionary. However, a reference dictionary with broad categories may erase distinctions needed for catastrophe or underwriting models. The design should preserve both original and normalized codes and record information loss. Mapping quality is then assessable against the actual downstream uses rather than merely against the proportion of rows assigned a code.
Versions, currencies, precision and missingness
A reporting agreement needs schema-effective dates, migration procedures, accepted file or message formats and compatibility rules. Changes should be tested against representative historic observations. A new field may be mandatory for newly bound risks yet unavailable for the in-force book. Reporting a field as unavailable with a reason distinguishes structural missingness from an incomplete observation. Zero must be reserved for an actual zero value. An unknown deductible, unknown loss amount, or unrecorded exposure cannot be safely encoded as zero.
Currency conversion introduces another semantic issue. Original amount and original currency should remain available, accompanied by a reporting currency, exchange rate, conversion date, rate source and rounding convention where converted values are reported. Written premium and cash reconciliations can legitimately use different exchange rates. An unexplained difference between converted ledgers may be exchange remeasurement rather than unreported premium. Separate local-currency control totals help detect true missing transactions before exchange movements obscure them.
Time should be represented at several levels: the effective time of the insurance fact, the date of recording by the delegate, the insurer's receipt time, and the time of validation or acceptance. For an endorsement effective April 1 but received May 15, the system should support both the restated April exposure and the knowledge that the insurer actually possessed at April month-end. This distinction is indispensable in evaluating a historical underwriting decision without giving the decision-maker information obtained later.
The Actuarial Standards Board (2016), in ASOP No. 23, treats suitability, reasonableness, sufficiency and the implications of material limitations as relevant to actuarial data work. Its review requirements are distinct from a formal audit. Accordingly, an insurer's actuary may conduct a defensible analysis using delegated data with documented constraints while the operational reconciliation team investigates underlying records. Approval of an actuarial analysis does not establish independent certification of every source observation.
Transport and lineage
A batch file can be exchanged through a controlled portal or secure transfer mechanism; an API can convey individual events or grouped submissions. Either transport requires explicit identity, access control, receipt acknowledgments, accepted schema revision, confidentiality protection, and replay handling. API transmission reduces some forms of delivery delay, but an event created late in the underwriting platform remains late even if transmitted instantly. File-based reporting can be reliable if the production process, controls and deadline fit the risk.
A submission manifest should identify the agreement, reporting period, creation time, row count, control totals and revision. A cryptographic digest can detect unintended modification of exact file bytes after generation. A digest establishes correspondence between bytes, not the accuracy of the economic facts encoded in them. Lineage links original records to validated observations, derived measures, corrections and downstream decisions. The preservation of lineage is particularly important when delegated records become inputs to regulatory reporting, actuarial estimates or automated restrictions.
Reporting timeliness and accuracy
Completeness and the independent denominator
Completeness cannot be established using a submitted file as its own population. One thousand reported risks with valid fields provide no evidence that only one thousand risks were bound. Completeness requires comparison with an independently identified expected population. Appropriate evidence can include issuance registers, binding-event logs, numbered policy sequences, premium collection ledgers, claims notification registers and independent transaction extracts. Each source has limitations: a gap in a numbering sequence may represent a voided quotation, and a premium ledger may omit a valid policy with payment not yet due.
A bidirectional reconciliation tests reported items against the source population and source items against the report. The first direction reveals unmatched or duplicated submissions; the second reveals omissions. It is crucial to distinguish missing rows from missing fields because the controls have different denominators. A field completeness rate can reach 100 percent for a submission that omits half of the active risks.
Let N be the number of independently identified expected active risks at a specified effective cut-off; M is the number of distinct expected risks matched to an accepted report. The count coverage ratio is M/N when N is positive. Exposure-weighted coverage can be calculated only when a trustworthy weight is known for the expected population, including unmatched records. A denominator derived from only reported insured values cannot establish exposure coverage. Where the expected population itself is uncertain, the metric must disclose the limitation rather than invent a definitive percentage.
The insurer's reconciliation should classify differences: missing record, duplicate record, inconsistent amount, incompatible reporting grain, timing difference, legitimate exclusion, unsupported adjustment, and unresolved identity. Financial net differences alone can conceal compensating errors. Detection should include record-level comparison, unmatched counts, absolute financial discrepancies and signed totals.
A worked premium reconciliation
Consider a hypothetical agreement denominated in US dollars. Every row describes the same financial basis, and cancellation amounts are negative. The independent source ledger records four distinct signed written-premium movements.
| Movement | Independent ledger | Submitted ledger |
|---|---|---|
| New policy A | 12,000 | 12,000 |
| New policy B | 8,000 | 0 |
| Endorsement C | 4,000 | 12,000 |
| Cancellation D | -2,000 | -2,000 |
| Total | 22,000 | 22,000 |
The submitted zero for policy B is a reconciliation display convention for the absent movement, not an actual reported zero transaction. Endorsement C is overstated by $8,000. A totals-only comparison yields zero variance. The absolute discrepancy is $16,000, obtained by adding the magnitudes of the two $8,000 errors. Three of four expected movements are present, yielding 75 percent count completeness. If the absolute deviation is divided by the sum of absolute independent movements, the result is 16,000 / 26,000, or approximately 61.54 percent. This example demonstrates why cancellation signs and denominator definition matter.
Now assume the accurate written premium of $22,000 relates to a hypothetical cash settlement in which all amounts have been collected, contractual commission is 20 percent of signed premium, and there are no taxes, fees or other deductions. Commission is $4,400 and the net remittance is $17,600. These amounts say nothing about earned premium or the time when collections become due. If only $18,000 had been collected, the remittance basis and timing would require the actual contract and collection ledger; applying the same calculation mechanically to written premium could generate a fictitious receivable or a misleading delinquency.
The control objective is an explainable reconciliation across written obligations, invoiced amounts, collected amounts, intermediary-held funds, remittances, and carrier posting. A simple cash bridge begins with opening receivables, adds billed amounts and adjustments, subtracts collections and write-offs where authorized, and reaches closing receivables. Separately, funds held for the carrier reconcile from opening balances through cash received and remitted, with contractual deductions and other authorized disbursements separately identified. The two balances answer different questions and may diverge legitimately.
Delay, aging and unobserved risk
Delivery timeliness should measure elapsed time from the relevant contractual or operational event to receipt, acceptance, and usability. For risk accumulation, the meaningful period begins when the binding becomes effective or when actionable exposure information first becomes available. For a payment reconciliation it may begin at settlement cut-off. A single monthly on-time percentage can obscure extreme delays in individual exposures.
Useful indicators include the median and upper percentile of event-to-receipt delay, age of the oldest unmatched material observation, exposure count and amount pending validation, time to resolution, and the fraction of submissions received by the agreed deadline. Percentiles must be accompanied by sample size and observation window. Unreceived events are right-censored at the evaluation date and must not be silently excluded from a timeliness statistic. An aging report based on independently known pending events is often preferable to a misleading distribution limited to completed submissions.
A cadence should reflect the speed with which new business can consume capacity, change exposure concentration or alter the carrier's obligations. A low-volume homogeneous portfolio might support periodic batch reporting supplemented by urgent notifications. A rapidly growing catastrophe-exposed portfolio may require shorter reporting intervals and exposure-specific escalation. Frequency is a control parameter, not a universal indicator of quality. Faster production without adequate corrections, independent checks and actionable escalation can transfer errors more quickly to the insurer.
Corrections and exceptions
Lifecycle and accountability
An exception is an observed failure of a defined reconciliation, validation, timeliness, contract or information requirement. Its severity depends on potential consequences rather than merely on the field involved. An unmatched material claim, prohibited territory, duplicate policy obligation, unknown limit, or unexplained cash shortfall can require immediate attention. Formatting issues may be corrected without suspending decision use if they leave all material meanings intact.
The delegated underwriter and carrier should assign exception ownership, detection timestamp, affected records, materiality, evidence, provisional controls, response deadline, resolution decision and independent verification of the correction. An exception remains open if a corrected file was delivered but not reconciled to the new expected population. Status distinctions such as received, technically accepted, reconciled, restricted and superseded prevent a successful upload from becoming a false assurance signal.
Lloyd's (n.d.-b) oversight and monitoring guidance places data integrity among the information domains relevant to delegated oversight. The inference for bordereaux operations is that exceptions become significant when the organization cannot establish the actual portfolio state or identify the limits of its knowledge. Reporting the amount of outstanding corrective work together with its economic significance gives senior management a basis for proportionate action.
Replacement, adjustments and reproducibility
There are two common correction forms. A full replacement submission restates a previously reported period. An incremental correction conveys explicitly identified reversing and replacement movements, or a defined adjustment to an earlier record. The receiving system must distinguish these modes before processing. Adding a replacement snapshot to the previous snapshot duplicates obligations. Treating a delta as a replacement can delete valid historical business.
A practical correction protocol identifies the original submission, version and transaction; states whether the previous observation is superseded, reversed or adjusted; and records the new amount, effective date, reason and authorization. Repeated delivery should have an idempotent outcome when the identity and version are identical. Different content with a repeated identity should trigger investigation rather than automatic acceptance. A rejected item should retain its rejection reason and source reference.
The original and corrected knowledge states have different research value. The revised exposure history informs the best available reconstruction of the portfolio. The original accepted history reveals what information was available when prior decisions were made. Overwriting the prior state destroys evidence needed to identify why an insurer exceeded limits, underestimated a liability, or acted too late. Access to historical versions must nevertheless be governed by applicable retention and confidentiality requirements.
Exception reports should distinguish isolated errors from systematic mechanisms. A single wrong exchange rate may affect thousands of transactions; ten independently missing risks may represent ten separate failures. Counts of records are useful for remediation workload, but counts of root causes are often more informative for control redesign. Recurrence after an apparently successful correction warrants examination of the source process and deployment of a permanent validation rule rather than a repeated manual repair.
Assurance and continuity
The control environment includes operational validation by the teams responsible for producing and accepting the reports, independent risk or compliance challenge as appropriate, and audit assessment of design and operation. An audit sample selected exclusively from accepted rows cannot measure omission risk. Independent sampling from the source population and coverage of anomalous and high-value observations are necessary complements.
Continuity planning should address provider inability to generate or transmit data, loss of access credentials, corrupt source extracts, platform outages, delegated business termination and migration. The carrier's ability to service existing policies and claims survives the loss of the reporting relationship. Exit arrangements consequently need data access, documented semantics, historical identifiers, authoritative open-claim balances, outstanding premium and reconciliation records. Tests should establish that the insurer can actually use an export rather than merely establish that an export file exists.
Portfolio aggregation and controls
Concentration, shares and layers
Bordereaux enable aggregation only after the insurer establishes a valid observation grain and contract basis. Let the stated exposure of risk i be E(i), and let s(i) denote its effective carrier participation as a fraction of the represented whole for a simple proportional placement. The corresponding signed exposure measure is E(i) multiplied by s(i). For a geographical segment G, a simple signed total is the sum across its constituent risk observations. This calculation is valid only for the stated exposure measure under the proportional assumptions. It is not expected loss, probable maximum loss or capital requirement.
Property values, policy limits, occurrence layers, annual aggregates and deductibles are different quantities. A $20 million property value under a $5 million policy limit produces distinct measures of physical insured value and nominal policy liability. For an excess layer with attachment a and limit l, an individual covered loss x produces a simple layer payment of min(max(x - a, 0), l), before participation and other contractual adjustments. The insurer must respect occurrence grouping, reinstatements and aggregate exhaustion before treating sums of policy limits as event-loss exposure.
Subscription arrangements demand careful separation between whole-market records and the carrier's share. A risk received through both the lead and a broker can refer to the same placement. It may be correctly counted once at whole-market level and allocated once to each participating insurer. Different policy layers or coinsurance shares are legitimate distinct obligations; a deduplication rule based only on customer name or address may erase them. Lloyd's (n.d.-d) addresses participant reporting responsibilities in its contract wording guidance. Transparent allocation therefore depends on both unique placement identity and contract-specific participation information.
Worked concentration example
Suppose a hypothetical carrier applies an internal signed-limit aggregate threshold of $20 million to one flood zone. The threshold is a simplified underwriting metric and does not represent a modeled flood loss. Four active placements have the following limits and carrier shares.
| Placement | Whole-market limit | Carrier share | Signed limit | Reporting status |
|---|---|---|---|---|
| A | $12m | 50% | $6m | Accepted |
| B | $10m | 60% | $6m | Accepted |
| C | $8m | 50% | $4m | Accepted |
| D | $12m | 50% | $6m | Delayed |
| Total | $42m | — | $22m | — |
The carrier's accepted report shows $16 million, or 80 percent of threshold. The complete exposure population shows $22 million, or 110 percent of threshold. The unreported signed amount is $6 million. The apparently available headroom of $4 million becomes a threshold excess of $2 million once the delayed placement is incorporated. Count coverage is 3/4 = 75 percent. Signed-limit coverage is 16/22, approximately 72.73 percent. Both measures are conditional on the independent register identifying placement D correctly.
The example isolates the effect of reporting latency, with no assumed modeling error or catastrophe occurrence. If location data were missing for a reported placement, it might still appear in the overall carrier total while being excluded inadvertently from the flood-zone aggregate. A portfolio control should therefore track unallocated exposure and assess conservative scenario allocations where missing locations are material. That is a sensitivity analysis, not a claim that every unallocated exposure is located in the zone.
If the insurer adopts a conservative provisional restriction on new business pending reconciliation, it should define the trigger and expected resolution. Broad restrictions can impede insurance supply, including for eligible risks that would otherwise be written. The economic tradeoff is between temporary opportunity cost and the potential cost of unknown accumulation. No uniform tolerance follows from the synthetic example. The carrier should connect tolerances to contract type, accumulation speed, plausible loss severity and available verification capability.
Claims information and underwriting performance
Claims data connect the exposures written under a delegation with observed losses. A simple reported incurred ratio divides compatible reported incurred loss by earned premium for a defined cohort and valuation date. If earned premium is $100,000 and paid claims are $30,000 with $20,000 in case reserves, the illustrative reported incurred ratio is 50 percent. If the same cohort's earned premium is actually $125,000 and $25,000 is missing from the reported denominator, the compatible ratio is 40 percent. If instead a $15,000 case reserve is omitted while earned premium remains $100,000, the apparent ratio is 35 percent rather than 50 percent. The direction of distortion depends on which population or measure is deficient.
An underwriting-year cohort should be compared with claims and earned premium aligned to that cohort; calendar-year accounts, accident-year losses, and underwriting-year results answer different questions. Newly written business can have little reported loss because policy exposure and claim reporting develop over time. Comparative performance requires appropriate maturity, exposure mix, and any applicable adjustment for reporting and settlement development. An insurer should retain historical valuations to construct development observations and assess changes in reporting delay.
Bordereaux should support the monitoring of more than short-term profitability. Cancellation rates, claim notification delays, complaint records, claim reopening, referral breaches and changes in insured risk mix can indicate operational or customer-outcome problems. Such indicators need clear denominators, verified definitions and independent sources where possible. A rise in complaints can reflect improved recording rather than deteriorating outcomes; a falling complaint count can reflect reduced accessibility. Classification changes must be separated from underlying behavioral changes.
Comparative institutional and legal analysis
International insurance supervision
The IAIS (2024) Insurance Core Principles and ComFrame provide an international supervisory framework. ICP 8 addresses risk management and internal controls, including standard 8.8 for oversight and accountability of outsourced material activities. The insurer's management remains concerned with the risk profile, contractual arrangements, provider capability, continuity and periodic reporting. The IAIS framework supports convergence of supervisory objectives; enforceable obligations arise through relevant jurisdictional instruments and their application to insurers. ComFrame provisions apply to internationally active insurance groups within that framework and should not automatically be assigned to every small intermediary.
In information terms, accountable oversight implies the ability to evaluate outsourced performance using trustworthy observations and to respond to material deviations. An insurer may engage another actor for data production, normalization or remediation, but its ability to understand and act on the resulting information remains pertinent to its oversight responsibilities. This analytical implication does not establish a universal monthly reporting period or prescribed data format.
Lloyd's and historical UK evidence
Lloyd's market guidance provides a comparatively specific account of the delegated underwriting relationship because it addresses the operation of market participants and the reporting arrangements for syndicates. The market's core data requirements (Lloyd's n.d.-c) and underwriting guidance (Lloyd's n.d.-a) are evidence of established arrangements for that market; their wording needs to be interpreted with the actual delegated agreement and applicable regulatory law. They do not by themselves supply a statutory obligation for an insurer outside the market.
The FCA's (2015a) delegated authority thematic review examined arrangements at 12 insurers and activities at 19 firms holding delegated underwriting or claims authority. The official summary identified weaknesses in the treatment of delegation as outsourcing, oversight and attention to customer outcomes. This historical sample demonstrates that assigning responsibilities contractually can coexist with shortcomings in actual monitoring. It cannot establish a current UK prevalence estimate or quantify the specific incidence of bordereaux errors. The summary and selected sections of the original report were consulted; the review is not treated as statistically representative.
The original thematic review (FCA, 2015b, paras. 1.13–1.14) identifies uneven management information and excessive reliance on third-party audits. These observations concern governance and the evidence used to monitor outcomes. An audit cannot substitute for decision-relevant information about a portfolio that lies outside its scope.
Where an intermediary participates in product design and customer communication as well as underwriting, data reporting may be needed for several purposes. The design must separate contractual risk reporting from conduct monitoring while maintaining meaningful cross-reference. An aggregate premium report cannot establish whether claims outcomes are appropriate for individual groups. Reviewing profitability and reviewing the customer experience require different evidence even when both concern the same delegated portfolio.
European Union
Commission Delegated Regulation (EU) 2015/35, Article 274, is presented in EIOPA's (2024) outsourcing rulebook text. It links written outsourcing policy with reporting and monitoring, and sets requirements for selection and contractual terms for critical or important functions. The inspected text includes information access, disclosure of material developments, protection of confidentiality, sub-outsourcing conditions and contingency considerations. It refers to Article 49 of Directive 2009/138/EC. The interpretation here concerns the rulebook text inspected as updated April 12, 2024, not a comprehensive consolidation audit of all later EU and national implementing measures.
In a delegated underwriting arrangement that falls within these outsourcing provisions, a contractual right to receive information becomes operationally meaningful when the carrier can access usable records, recognize changes and integrate the activities into its risk control system. The precise classification of a particular activity requires analysis of the arrangement and applicable rules. Data sharing also requires attention to lawful access, permitted purposes and confidentiality. A standardized bordereau should not automatically be replicated to every stakeholder simply because the data are technically portable.
United States: the Virginia illustration
The Commonwealth of Virginia (n.d.), in Code of Virginia section 38.2-1360 prescribes contractual provisions for insurers acting through managing general agents within the statutory scope. The inspected statute requires the agent to render accounts of transactions and remit funds due at least monthly, requires separate records of business, and provides access to relevant records. Where claims settlement is delegated, it also requires timely reporting and identifies circumstances requiring transmission of claim files. The provision includes requirements concerning electronic claims files and profit-sharing timing in specified circumstances.
The monthly interval concerns the statutory accounting and remittance requirement. It does not establish that a monthly risk-exposure update is sufficient for any type of underwriting authority. An insurer might need faster exposure information when its capacity restrictions require it. Conversely, the existence of a reporting contract under a US market arrangement does not necessarily mean that the intermediary meets every legal element of a managing general agent definition. State classification and applicability must be examined separately.
Virginia is a concrete state-law example, not a proxy for a single uniform federal MGA law. The comparison illuminates the relationship between legal access to business records and operational reconciliation: both the content of reporting and the ability to compare it with an authoritative register matter for control.
Australia
The Australian Prudential Regulation Authority's (2026) CPS 230 Operational Risk Management determination examined here commenced on July 1, 2026. The standard applies to defined APRA-regulated entities and treats underwriting and claims management among services whose providers are ordinarily classified as material for insurers unless another classification is justified. It establishes expectations for provider management, reporting to governance bodies, service arrangements, operational controls and continuity. The standard has broad operational-risk purposes beyond delegated underwriting.
The Australian approach makes the potential loss of access to delegated data part of the wider question of operational resilience. A reporting arrangement might be accurate in routine conditions and fragile during a system outage or provider exit. The governance assessment therefore includes operational recovery, continued access to necessary business records and alternative ways to serve policyholders. These are implications of the standard for the subject of this monograph rather than a claim that it mandates a particular bordereau format.
| Source regime | Legal or institutional character | Reporting implication for this study |
|---|---|---|
| IAIS ICP 8 | International supervisory standard | Retained accountability and assessment of outsourced material activities |
| Lloyd's | Market requirements and guidance | Specified reporting domains and participant oversight |
| FCA thematic review | Historical supervisory observation | Actual oversight and customer outcomes require evidence |
| EU Article 274 | Delegated regulatory provision, subject to applicability | Written monitoring arrangements and effective information access |
| Virginia section 38.2-1360 | State statute for scoped MGA contracts | Monthly accounting and remittance; records and claims access |
| APRA CPS 230 | Australian prudential standard | Material provider oversight, resilience and risk reporting |
The documents differ in legal force, institutional scope, subject and observation date. Their convergence is substantive: delegated activity requires an information relationship through which the accountable insurer can observe and control relevant risks. Their differences remain consequential. A carrier cannot substitute a market reporting template for applicable law, or cite a general supervisory objective as if it prescribed a particular reporting interval.
Comparative synthesis, limitations, and reproducibility
The control significance of information
The preceding analysis answers the first research question by relating the report content to the actual decision being governed. A risk record requires an interpretable exposure identity, temporal scope and contractual characteristics sufficient for the intended accumulation and authority control. Premium records describe signed financial transactions on identified bases. Claims records describe reported liabilities and movements at defined valuation dates. These information streams share identities but preserve their own observation grains. Control over delegated underwriting depends on both completeness and semantic compatibility.
The second question requires separation between report validation and population verification. Schema rules confirm that reported observations satisfy formal conditions. Reconciliation to an independent population establishes whether the report represents the intended business and whether individual financial values correspond. The premium example establishes a deterministic result: equal net totals coexist with substantial transaction-level discrepancies. The ratio of matches to expected records measures a different property from the discrepancy in monetary values. Using both measures reduces the risk that compensating errors remain invisible.
The third question concerns temporal governance. Corrections, provenance and escalation allow revised information to repair current control without destroying the historical decision context. Exposure can be validly bound before receipt of the information needed to assess its concentration. The flood-zone example demonstrates how even a single delayed risk can transform apparent headroom into actual threshold excess. The insurer's policy should therefore relate reporting requirements to the potential change in obligations during the unobserved interval.
The economic significance of these results is the allocation of verification effort. Data-quality investment should be directed at uncertainty that can change underwriting, claims, financial or supervisory decisions. Some data errors are inexpensive to repair and immaterial to policyholder outcomes; others affect the validity of an entire loss denominator or the carrier's capacity assessment. An organization can reduce unnecessary transmission by measuring which information is needed for each decision and designing access accordingly. Retaining useful data without indiscriminate disclosure is compatible with both operational oversight and confidentiality.
Competing interpretations and tradeoffs
One interpretation identifies inadequate data governance primarily as a technical standardization problem. Standardization removes friction, makes comparison possible, and improves automated validation. Its weakness as a complete explanation is that missing events and delayed sources can produce technically valid reports. Another interpretation emphasizes agency incentives and independent assurance. It accounts for strategic reporting or distorted performance measures, yet it can overstate conflict when deficiencies arise from incompatible systems or limited resources. A third interpretation emphasizes operational continuity and the resilience of the reporting chain. It identifies common dependencies capable of disrupting many delegated arrangements at once, but service availability alone does not establish semantic accuracy.
These interpretations address different causal mechanisms. Improvement requires diagnosis of whether the problem concerns unrecorded economic events, inconsistent representation, incentives, transmission failure, verification or organizational response. A more frequent feed may resolve transmission delay but leave source omission intact. A financial incentive may encourage complete reporting but cannot restore interoperability by itself. A fully automated check can enforce an incorrect data dictionary consistently across all participants. The appropriate control is the one that addresses the diagnosed failure mechanism and its consequences.
A further limitation concerns small delegations. The fixed cost of a detailed record model can be substantial relative to portfolio scale. Standardized exports and shared infrastructure can reduce this cost, while proportional reporting arrangements can accommodate low-materiality risks. Proportionality does not eliminate the need for decision-relevant information. A small book may contain a severe concentrated exposure; premium volume alone is an inadequate indicator of the materiality of unknown risk.
A reproducible empirical research design
Further empirical work would require matched event histories from carriers and delegates with explicit contractual and technical reporting definitions. The unit of analysis could be a binding event, policy term, exposure observation, financial movement, claim valuation or submission. The selection depends on the question. A study of exposure reporting delay should sample the independently recorded binding population and preserve pending observations; a study of financial accuracy should sample both accepted and source transactions with a defined currency and accounting basis.
A proposed observational design stratifies agreements by line of insurance, reporting method, size, authority scope and geography. Within each stratum, select agreements under explicit criteria, then draw random samples from independently established event populations with supplemental targeted examination of high-materiality exceptions. Estimate omission proportions with denominators and uncertainty intervals; analyze latency with appropriate treatment of right-censored events; measure value discrepancies on comparable accounting bases. Document any duplicated identities, exclusions, changes to classification, or loss of records. Where sample design is clustered by agreement, statistical inference must account for within-agreement dependence.
A prospective evaluation of reporting changes could compare the same agreements before and after an implementation and, where feasible, matched agreements with stable reporting methods. Candidate outcomes include event-to-usable-data delay, omission rate, value discrepancy, exception age and detection lead time for material risk thresholds. Changes in product mix, provider staffing, seasonality, catastrophe activity and simultaneous system migration can confound attribution. Selection into the new reporting method may depend on existing data maturity, which makes a naive comparison biased. Pre-specifying outcome measures, observation periods and matching or adjustment strategy would strengthen interpretability. These are research design recommendations rather than a demonstrated causal result.
The examples in this monograph are exactly reproducible from their tables. The premium control totals use signed sums; the absolute discrepancy uses the magnitudes of mismatches and a denominator of absolute independent movements. The concentration calculation multiplies each placement limit by the carrier fraction, sums accepted and independently identified placements separately, and compares both totals with a fixed hypothetical threshold. The claims illustration uses fixed reported loss and earned-premium amounts to isolate denominator and numerator errors. No simulation randomness or estimated parameters are involved.
Evidence limits and updates
The market guidance pages are contemporary accessible institutional documents, but full template-by-template compliance interrogation requires the current user guides, schema releases and actual delegated contracts. The cited academic literature is used within inspected abstract and bibliographic scope; the original studies' full datasets and empirical estimates were not reanalyzed. The FCA evidence describes selected historical firms. The EU comparison is bounded by the inspected rulebook edition. Proprietary insurer and intermediary records were unavailable. Consequently, no publication-level inference concerning current market-wide error rates, average reporting latency, causal effects of automation, or universal reporting thresholds is warranted.
Changes to market reporting standards, applicable outsourcing law, professional actuarial standards or data transport arrangements are explicit review triggers. The documentary identity, date or access basis, and proposition supported by each cited source are maintained in the accompanying evidence register. Future empirical studies should retain the same distinction between researched claims, synthetic demonstrations and new observed results.
Conclusion
Delegated underwriting reporting is an institutional control over the insurance obligations created through distributed underwriting decisions. Bordereaux organize the insurer's knowledge of risks, premiums, claims and portfolio development, but their usefulness depends upon the economic meaning, population coverage, temporal accuracy and provenance of the underlying observations.
The central finding is that an insurer must be able to connect its received reports to independent evidence of the obligations actually created. A structured report with compatible definitions, stable identifiers and accurate financial values supports control only within its observed population. An independently anchored reconciliation exposes omissions and duplicate or incorrect records that aggregate totals cannot reveal. The premium example establishes that a balanced total can coexist with material error; the concentration example establishes that reporting latency can conceal a threshold breach. Neither example supplies an empirical estimate of market frequency.
The comparative investigation identifies retained insurer accountability, usable information rights, active monitoring and proportionate controls as recurring institutional concerns. Different legal and market regimes operationalize those concerns through requirements of different scope and force. A responsible synthesis preserves these differences while recognizing the common importance of the ability to observe, verify and respond.
The governance of bordereaux therefore extends from reporting contract design through source records, semantic mappings, independent reconciliation, correction and portfolio analysis to decisions about delegated capacity. A successful transmission is one step in this sequence. The final test is whether the insurer can reconstruct its obligations, recognize limitations in its information, and take a timely decision grounded in the actual portfolio.
References
Actuarial Standards Board. 2016. Actuarial Standard of Practice No. 23: Data Quality. December 2016 revision. Official standard.
Australian Prudential Regulation Authority (APRA). 2026. Prudential Standard CPS 230 Operational Risk Management. Determination dated April 23, 2026; commenced July 1, 2026. Official prudential standard.
European Insurance and Occupational Pensions Authority (EIOPA). 2024. Solvency II Single Rulebook, Commission Delegated Regulation (EU) 2015/35, Article 274: Outsourcing. Rulebook page last updated April 12, 2024. Article 274.
Financial Conduct Authority (FCA). 2015a. TR15/7: Delegated Authority: Outsourcing in the General Insurance Market. June 1, 2015; official summary page updated August 29, 2025. Official review summary.
Financial Conduct Authority (FCA). 2015b. TR15/7: Delegated Authority: Outsourcing in the General Insurance Market. June 2015. Full report, paras. 1.13–1.14. Original report.
International Association of Insurance Supervisors (IAIS). 2024. Insurance Core Principles and Common Framework for the Supervision of Internationally Active Insurance Groups. ICP 8, especially standard 8.8. Online tool, December 2024 framework. Official online text.
Jensen, Michael C., and William H. Meckling. 1976. “Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure.” Journal of Financial Economics 3 (4): 305–360. DOI: 10.1016/0304-405X(76)90026-X. Bibliographic record and abstract.
Lloyd's. 2019. Coverholder Reporting Standards: User Guide. Version 5.2, August 20, 2019, pp. 8, 51, 61. Official user guide.
Lloyd's. n.d.-a. Delegated Underwriting. Delegated authority market knowledge. Market guidance.
Lloyd's. n.d.-b. Oversight & Monitoring of Contracts of Delegation of Underwriting Authority. Delegated authority market knowledge. Oversight guidance.
Lloyd's. n.d.-c. Reporting Standards. Delegated authority market knowledge. Market reporting requirements.
Lloyd's. n.d.-d. Wording Expectations for Contracts of Delegation. Delegated authority market knowledge. Contract wording guidance.
Virginia, Commonwealth of. n.d. Code of Virginia, Title 38.2, Chapter 13, § 38.2-1360: Required Contract Provisions. Current online statutory text retrieved November 19, 2026. Official statute.
Wang, Richard Y., and Diane M. Strong. 1996. “Beyond Accuracy: What Data Quality Means to Data Consumers.” Journal of Management Information Systems 12 (4): 5–33. DOI: 10.1080/07421222.1996.11518099. Bibliographic record and abstract.