Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Fronting Insurer Underwriting Accountability and Program Capacity

Abstract. Fronting insurance distributes the economically important functions of a primary insurance transaction among a licensed issuing insurer, program manager, reinsurer, and often captive, collateral provider, and claims administrator. This research monograph investigates how the issuing insurer's underwriting accountability persists when risk selection, daily binding activity, claims work, and much of the expected loss are assigned to other participants. It distinguishes legal policyholder obligations, contractual risk transfer, economically recoverable reinsurance, collateral security, immediate claims liquidity, and operational ability to intervene. Its source base includes the International Association of Insurance Supervisors' Insurance Core Principles, National Association of Insurance Commissioners materials on reinsurance and insurer receivership, the European Union's Solvency II requirements, selected state statutory provisions, Lloyd's delegated-authority requirements, and research on agency costs and reinsurance information asymmetry. A transparent synthetic quota-share program illustrates the divergence between net modeled loss and gross payment exposure when recoveries, collateral, or administration fail. Original analysis synthesizes existing legal, actuarial, financial, and governance concepts without creating a new legal doctrine, actuarial model, or empirically estimated risk frequency. The principal finding is that safe program capacity depends on the simultaneous enforceability and operational effectiveness of policy authorization, underwriting limits, treaty coverage, security, liquidity, and claims continuity.

Keywords: Fronting insurer; fronting insurance; underwriting accountability; insurance program capacity; managing general agent; delegated underwriting; reinsurance; quota share; collateral; counterparty credit risk; claims liquidity; policyholder protection; insurance governance; runoff; supervisory regulation.

Introduction

Insurance fronting is a contractually organized separation of policy issuance from much of the underlying program's economic risk-bearing and day-to-day administration. A carrier licensed to issue insurance policies may permit a specialized managing general agent (MGA) to bind particular risks and may reinsure a very large proportion of the resulting losses to another licensed or otherwise eligible risk-bearing entity. The program can deliver market access, specialized underwriting expertise, capital diversification, captive participation, or an operating platform that the participants could not efficiently establish independently. These economic benefits make fronting an important institutional arrangement in contemporary insurance markets.

The arrangement has an equally important structural consequence. The policyholder has contracted with the issuing insurer, yet the principal information, execution, and reimbursement channels can reside outside that insurer. The National Association of Insurance Commissioners (NAIC), in its reinsurance chapter of the receivership handbook, identifies substantial cession and delegated underwriting and claims authority as key fronting characteristics and emphasizes the fronting insurer's continuing financial obligation to policyholders. That official discussion makes direct liability the starting point of the analysis.

The research asks three questions. What due diligence and underwriting authority structure must an insurer maintain when an MGA operates a fronted program? How should reinsurance recoverability, counterparty credit, and security determine permissible capacity? Which claims, reporting, and runoff controls enable an issuing insurer to perform its legal obligations when another participant is delayed, insolvent, or no longer available? The answers require analysis of primary policy language, treaty terms, delegated authority, prudential capital, operational controls, and the timing of money movement.

The word capacity denotes several different constraints that should be examined together. Legal capacity concerns permitted classes and territories. Underwriting capacity concerns accepted exposures and authority limits. Financial capacity concerns retained and contingent losses in relation to usable resources. Reinsurance capacity concerns enforceable recovery and its credit or security support. Liquidity capacity concerns cash available when claims fall due. Operational capacity concerns the ability to bind, record, adjust, supervise, and service policies. A carrier can meet one of these constraints and fail another.

This dossier uses an international comparative research horizon while identifying the force of individual sources. IAIS principles are international supervisory standards; the NAIC's model laws provide draft frameworks requiring state enactment; North Carolina law is a specific state example; Solvency II applies within its legal scope; Lloyd's requirements govern its own market. The research makes functional comparisons among these regimes without treating their legal rules as interchangeable. The NAIC's overview of reinsurance purposes explains the role of cession in expanding insurance capacity, while the analysis here identifies the conditions that make that expansion reliable for the legal carrier and policyholder.

Pre-Bind Diligence and Underwriting Accountability

A substantive approval of the business

Underwriting accountability begins before the first policy is bound. The fronting carrier should know the covered population, product terms, rate methodology, exposure distribution, underwriting selection rules, aggregate loss potential, territory and licensing constraints, claims development pattern, financial assumptions, and treaty treatment. It should independently decide whether the business falls within its appetite and whether it has sufficient underwriting, actuarial, compliance, finance, and claims resources to supervise it.

Lloyd's delegated underwriting guidance identifies the importance of appropriate authority, pricing parameters, referral triggers, active underwriting resources, and continuing accountability. Those are Lloyd's institutional expectations rather than a global legal code. Their relevance is analytical: an insurer can only supervise a delegate effectively if its permissions are specific, testable, and connected to the underlying risk profile.

A program approval should document the material loss drivers and explanations for them. For catastrophe-exposed property, this may involve concentration, construction, hazard, valuation, and event correlation. For liability business, trigger type, attachment, limits, defense treatment, latency, and judicial or inflation sensitivity become important. For specialty or emerging products, reliance on proxy datasets should be identified, along with uncertainty in coverage interpretation and future claim emergence.

Historical loss ratios can mislead when claim maturity, deductibles, inflation, rate adjustments, underwriting-year mix, or policy wording differ. A five-year sample is informative only if its exposure denominator and reserve development are intelligible. It is especially important to distinguish claims reported, claims paid, incurred losses, and independently developed ultimate losses. New programs may have little directly relevant experience; in such cases, a sensitivity envelope and limited initial authority are more defensible than an unsupported point estimate.

MGA suitability and control infrastructure

Diligence on the program manager should cover corporate identity, ownership, licensing, financial viability, staff experience, underwriting systems, information security, funds segregation, claims capability, complaints procedures, outsourcing dependencies, audit history, and prior performance. Lloyd's third-party due diligence expectations call for independent challenge in approval, evidence of expertise, financial controls, operational resilience, and continued suitability. A fronting carrier can adopt comparable questions as a research-informed governance practice while checking the actual legal requirements applicable to its own jurisdiction.

The insurer should test the MGA's rating and binding platform against approved rules. Test cases need to include missing fields, invalid territories, excessive limits, prohibited class combinations, override permissions, referral expiration, duplicate policies, and aggregation after manual amendments. Human reviewers should be able to reconstruct why a given risk was accepted. An algorithm's vendor ownership or proprietary status does not make its performance unknowable; contractual audit and data-access arrangements are part of diligence.

Contractual scope should specify the allowed insured classes, locations, limits, deductibles, rate approaches, form editions, aggregate premium cap, maximum concentration, referral triggers, claims powers, authority start and end dates, permitted subdelegation, record standards, financial controls, and termination rights. The Lloyd's contracts of delegated authority provide a concrete illustration of how premium caps, aggregate limits, territories, reporting, funds settlement, and post-termination duties can be specified in an enforceable document.

Timing and the right to bind

The dates of the policy form approval, MGA authority, insurer authorization, reinsurance treaty, and collateral effectiveness should be reconciled before the program begins binding. A short interval between the commencement of policy writing and treaty inception can create unanticipated gross exposure. A treaty may also exclude business improperly bound beyond an MGA limit even if the direct policy remains enforceable. Automated controls should use the current authorized guideline rather than an outdated manual, and exception authority should be explicitly assigned.

Underwriting profitability requires more than a fronting-fee calculation. The issuer may be paid a fee while retaining a contractual loss share, maintaining capital and liquidity against contingent exposures, processing claims, bearing reinsurance credit risk, and supervising the delegate. An appropriate economic evaluation examines the fronting fee, acquisition and servicing costs, retained premium and loss, potential uncollectible recoveries, administration cost in runoff, and capital consumption. Pure volume-based remuneration can produce a production bias; profit sharing can reduce some incentive conflicts but may encourage optimistic initial reserves unless the accounting includes later development.

An empirical article by Garven, Hilliard, and Grace reports adverse selection in reinsurance markets using a panel of U.S. property-liability insurers for 1993–2012. Its abstract reports associations between relationship tenure, reinsurance activity, profitability, and insurer credit quality. It does not estimate the causal effects of fronting-specific audit practices or MGA compensation. The relevant conclusion for this dossier is narrower: information asymmetry and the time dimension of trading relationships matter when designing independent diligence and ongoing oversight.

Changes during the program

Initial approval is only the first decision. The carrier should define reapproval when the business mix, geographic distribution, rates, policy wording, treaty, manager ownership, systems, reinsurer strength, collateral, or volume changes materially. Thresholds should be linked to possible financial effects. A quarterly report can arrive too late if catastrophe aggregation or policy limit breaches develop within days. A combination of automated near-real-time controls and slower actuarial development analysis better matches the different rates at which risks evolve.

The insurer should also establish criteria for stopping new business without abandoning existing insureds. Expired security, missing premium reconciliations, unresolved audit findings, claims service deterioration, and repeated unauthorized binding are plausible triggers. The action taken should be documented and should preserve the insurer's continuing ability to service the outstanding book. Underwriting accountability is established by what the carrier can verify and change, not only by what its initial binder agreement declares.

Program Capacity and Portfolio Economics

Capacity as a set of simultaneous constraints

Program capacity is commonly represented as a gross written premium cap, insured limit, or maximum retained loss. These quantities measure different phenomena. An authorized carrier may be able to issue a policy with a USD 10 million insured limit and transfer 90 percent of its defined covered losses under a quota-share treaty. That leaves a nominal retained share of USD 1 million for a full-limit indemnity claim under simplified assumptions, but the direct policy may require the issuer to fund the entire USD 10 million before recovery. If the treaty does not cover some expenses, if the reinsurer disputes a claim, or if security cannot be called quickly, the carrier's actual financial exposure is larger than the nominal retention.

A credible capacity approval should therefore document the boundaries of legal permission, portfolio exposure, retained underwriting loss, reinsurer credit exposure, secured recoverables, and claims liquidity. Operational capacity is equally important: if the program manager's systems cannot provide accurate policies and claims records under stress, contractual financial resources may be difficult to mobilize. The IAIS supervisory treatment of reinsurance connects strategy, appetite, counterparties, effectiveness of risk transfer, and management of recoverables within a control framework.

The binding constraint can change over time. At launch, the maximum premium amount may determine what the MGA may write. After several months, a catastrophe accumulation cap or inadequate loss ratio may become decisive. During a concentrated large-loss event, claims liquidity or reinsurer recovery timing can dominate. In runoff, the available claims administrator, completeness of records, and survival of reinsurance protection may determine whether the issuing insurer can perform. Capacity should be reconsidered as these conditions evolve rather than treated as permanently established at program inception.

Distinguishing exposure metrics

Gross written premium measures sales written during a period; it does not by itself measure the size or timing of possible claims. Earned premium follows the period over which coverage is provided. Gross incurred loss includes relevant claim costs and reserve estimates. Ultimate loss estimates depend on the claims development pattern, coverage, and selected assumptions. Policy limits establish contractual maximums for specified coverages, though aggregate and other limits can modify exposures. The insurer needs consistent definitions before computing ratios or comparing policies and treaties.

Gross and net loss analyses should be presented with the same loss basis. If a carrier's gross results include allocated loss-adjustment expenses but the treaty excludes them, applying a quota-share fraction to the entire gross amount produces a misleading net estimate. The same problem occurs with extra-contractual liabilities, claims-made reporting periods, deductibles, aggregates, and exclusions. The treaty should be mapped to policy coverage categories and claims processes, with an exception register that identifies potentially material gaps.

A commercial program can contain geographically dispersed policies but remain highly exposed to one common peril, economic trend, liability interpretation, or reinsurance counterparty. Regional diversification is not sufficient if one storm system, industry practice, or judicial development affects many insureds. The insurer should study accumulation by location, event definition, insured group, sector, peril, and policy term. Catastrophe models and reserve development analyses should be challenged for the presence of independent data and the plausibility of their assumptions.

Capital, growth, and concentration

The NAIC explanation of reinsurance capacity describes the use of reinsurance to expand primary insurance risk-writing capacity and stabilize underwriting results. A program can take advantage of this capacity while producing a concentrated receivable against one reinsurer. The carrier's apparent diversification of insured risks may thus coexist with concentration of counterparty risk. Assessments should distinguish gross risk reduction from the quality and timing of the expected reimbursement.

Growth creates an additional governance problem. MGAs can originate thousands of policies through scalable platforms, while human oversight and claims settlement resources may expand more slowly. Changes in insured mix and rates may not be visible to a carrier until after substantial policies are issued. The fronting insurer should therefore impose authority limits that can be enforced by the system at the time of binding and should make the rate of growth itself a monitored risk variable.

Fronting fees and retained premiums also require an economic comparison with costs. A small fee percentage applied to large written premium can appear profitable before retained claims, expenses, credit impairment, regulatory capital, liquidity facilities, and runoff commitments are considered. Fee adequacy should be tested under expected and adverse loss experience, including the cost of replacing an administrator and the cost of delayed reinsurance collection. No particular fronting fee, retained percentage, or premium cap is identified here as universally sufficient; those values are contract- and portfolio-specific.

Information and contract incentives

Jean-Baptiste and Santomero's study of private reinsurance contract design analyzes contractual responses to information asymmetry and the importance of relationships over time. Its implications for fronting are interpretive: retaining some economic interest, sharing information, demanding transparent loss data, and preserving audit access can help align the parties' incentives. The work does not establish an optimal quota share for every line of business or show that a fronting carrier can dispense with underwriting control once a reinsurer bears most expected losses.

Underwriting guidelines are therefore an economic as well as a legal instrument. They specify the circumstances under which premiums and risks enter a shared portfolio, determine which exceptions require human review, and limit the capacity of one participant to expose others to unapproved loss distributions. Effective governance must also consider the timing of profit commissions and reserve reporting. Incentives that reward apparent early profitability may be inconsistent with development of long-tail losses. Independent actuarial review and transparent development-period accounting can reduce, though not eliminate, this distortion.

Reinsurance, Security, and Recovery

Recoverability and its conditions

The quota-share percentage in a reinsurance treaty is an allocation term, not a certificate of financial protection. Actual recovery requires that the direct loss falls within treaty subject business; relevant contract conditions have been met; the cedent has an enforceable claim; the counterparty can pay; and any security is accessible as intended. The timing of recovery matters because direct claims may be paid before the insurer receives reimbursement. Each condition should be tested separately at the program-approval stage.

The NAIC's reinsurance receivership handbook distinguishes quota-share, excess-of-loss, facultative, and treaty reinsurance and examines letters of credit, trust funds, cut-throughs, and the treatment of recoveries in insurer insolvency. Those distinctions show why fronting structures require exact wording review rather than a generic label of ceded risk. A treaty might respond to a claim while a collateral arrangement offers no immediate funds; a bank letter of credit might be drawable even before the reinsurer has made a voluntary cash payment, subject to the instrument's actual terms and governing law.

Regulatory credit and collateral regimes

U.S. reinsurance collateral rules have changed over time and depend on the status of the counterparty and applicable state adoption. The NAIC's covered-agreement implementation account explains revisions to the Credit for Reinsurance Model Law (#785) and Model Regulation (#786), including arrangements for qualifying reciprocal-jurisdiction reinsurers. It follows that the phrase unlicensed foreign reinsurer cannot by itself determine the collateral percentage applicable to a particular treaty. A legally recognized reduction in collateral does not eliminate the cedent's need to assess financial strength, payment behavior, and concentration.

The NAIC model law catalog provides the model-law identity, but legally operative reinsurance credit rules must be established from the cedent's own domicile and relevant implementing rules. An internal approval document should identify the reinsurer's category, jurisdiction, authorization or recognition status, collateral requirement, type of permitted security, beneficiary, legal rights, revaluation terms, expiry or replenishment provisions, and the consequences of ratings deterioration or nonperformance. Model-law text and domestic enactment need to be distinguished, especially when the program operates across more than one state.

The European Union's Solvency II rulebook uses a different institutional architecture. Article 209 on qualitative recognition ties risk mitigation recognition to legally effective and enforceable arrangements, continuing monitoring, defined rights in counterparty default, and avoidance of double counting. Article 210 on effective risk transfer addresses the effectiveness and clarity of the transfer itself. Article 211 on reinsurance contracts adds applicable criteria for reinsurance and special purpose vehicles. Their detailed legal application requires the current operative EU rules and the facts of the arrangement, including the jurisdiction of the relevant undertaking.

The Solvency II SCR recognition framework also requires appropriate reflection of credit and other risks arising from mitigation techniques. The resulting conceptual point is straightforward: a cedent should not take economic or prudential credit for a contractual risk transfer without considering its enforceability and the new risks produced by that transfer. Regulatory recognition, expected recovery, collateral value, and liquid cash are related but distinct assessments.

Forms of protection

Cash collateral held under an appropriate, enforceable arrangement can provide ready funds, but its availability depends on legal segregation, control, beneficiary rights, and withdrawal conditions. Reinsurance trusts may contain eligible securities whose market value and liquidity change. Bank letters of credit depend on issuing bank credit quality, irrevocability, draw conditions, expiry, renewal, and the ability to present a conforming demand in time. Parental guarantees and contractual promises depend on the guarantor's ability and legal duty to perform. None of these instruments should be described as equivalent merely because a spreadsheet assigns them the same face amount.

The NAIC discussion of reinsurance trusts and letters of credit identifies distinct security mechanisms and specific contractual features affecting access. The decisive program question is whether the carrier can convert a legally available security interest into payment capacity when claims are due. Counting the same balance both as unrestricted carrier cash and as independent collateral support would overstate protection. Similarly, discounting a secured receivable for counterparty default and then applying the same full default loss to that balance would double count the stress.

An insurer should maintain a dated reconciliation between estimated ultimate ceded liability, reported recoverable balances, actual funds withheld or collateral security, uncollected premiums, disputed items, and available draw rights. The balance is dynamic. A program that has been appropriately collateralized at inception may become undersecured after sharp loss development or rapid portfolio growth. Security sufficiency should be measured against the contractual liabilities it is intended to protect, with a clear treatment of margins and asset valuation.

Liquidity and timing

Suppose a large covered claim becomes payable to an insured on day 20 while the reinsurer is expected to reimburse the carrier on day 90 after completion of documentation and settlement cycles. The cedent must bridge the period using accessible resources. The accounting presentation of a reinsurance recoverable may be entirely consistent with a material short-term liquidity problem. The IAIS's reinsurance supervision standards discuss the need to assess the economic effect of risk transfer, including risk concentrations and liquidity implications.

Claims-payment liquidity should be modeled by horizon. Relevant components include gross cash claims, available premium receipts, expected reinsurer reimbursement by timing and certainty, immediately callable security, cash balances, committed facilities, operating cash expenses, and settlement costs. Stress assumptions should address synchronized claim payments, reinsurer delay, bank credit impairment, collateral-market movements, and the operational time required to draw security. A company that is solvent in an ultimate-net-loss model can still be unable to settle claims promptly if cash timing is adverse.

The effect is especially pronounced in event-driven or long-tail programs. A catastrophe can generate a surge of immediate payments and concurrent impairment of multiple reinsurers or collateral investments. Long-tail liability programs may instead create slow, recurring cash needs over many years, including periods after the MGA has ceased operations. The economic capital and liquidity frameworks must therefore analyze different time profiles.

Counterparty surveillance

Credit analysis requires identification of the actual legal reinsurer, its ownership, regulatory status, capital resources, ratings where available, loss reserves, liquidity, business concentration, and exposure to the same catastrophe or economic stress affecting the fronting program. A panel of reinsurers may reduce concentration, but only if recoveries are sufficiently independent and the contractual shares are enforceable. Affiliates of one parent, reinsurers using common retrocession, and entities holding similar collateral assets can share material dependencies.

Counterparty management should also include recoverable aging, dispute causes, frequency and size of collateral calls, settlement delays, and the completeness of treaty premium accounts. A reinsurer with a strong external rating may still refuse a claim that was bound beyond authority or outside treaty coverage. Conversely, a technically valid claim on a deteriorating reinsurer may have high ultimate uncertainty. The carrier needs both legal coverage analysis and financial collectibility monitoring.

Runoff and insolvency

The direct insurance policy can outlive the underwriting authority agreement and the commercial fronting relationship. Even where new business stops, existing claims and unearned obligations may continue. The NAIC's receiver-oriented treatment of reinsurance explains why the identity of the cedent, reinsurer, receiver, collateral beneficiary, and any cut-through payee matters in insolvency. The effect of each clause should be established under the law governing the relevant agreement and estate, not inferred from a generic description of fronting.

The carrier should test whether reinsurance continues to cover the original insured obligations after the MGA appointment expires, whether reporting and record rights survive, and whether losses can be billed accurately in a transfer to a new claims administrator. Commutation or novation may change the legal structure and economics; they require specific analysis. A program is not fully funded for its insured commitments if the only practical access to treaty records ends with the disappearance of the administrator.

Governance, Reporting, and Delegated Operations

Delegation as an auditable decision process

The legal right to bind a policy must be translated into a functioning control system. Every material underwriting decision should be traceable to an approved policy form, valid authority period, risk classification, pricing basis, jurisdiction, individual limit, aggregate capacity, and any required referral. The fronting insurer should be able to reproduce a selected policy from original submission through approval, issuance, reinsurance cession, and subsequent claims experience. The traceability requirement protects not only compliance but also empirical understanding of whether loss deterioration reflects selection, pricing, claims development, data error, or contractual coverage.

Lloyd's oversight and monitoring requirements call for governance frameworks, continuing management information, escalation, and board reporting. For a carrier, receiving a monthly bordereau is a beginning, not proof of effective oversight. Records should be reconciled against original issued policies, premium bank flows, aggregate limits, treaty declarations, and claim systems. Exceptions must be assessed and corrected by people with defined authority.

A useful operational record contains a unique policy identifier, program and MGA identifiers, approved product and wording, jurisdiction, dates, insured risk data, premium calculation, manual overrides, binder authority version, insurer referral outcome, exposure location and limit, and treaty allocation. Claims data should connect back to these fields. If an MGA's technology cannot produce the necessary history, the carrier's ability to evaluate underwriting quality is weakened. Missing data are not automatically evidence of poor underwriting selection, but they limit the strength of any claim that underwriting is being controlled.

Independence and escalation

Business production, risk management, actuarial review, compliance, claims, finance, and internal audit have different responsibilities. A commercial team may recommend a program; independent specialists should challenge pricing, model assumptions, reserves, collateral coverage, legal permissions, claims controls, and contingent liquidity. Lloyd's third-party approval guidance requires an element of independent challenge in approval and calls for continuing assessment of suitability. That requirement supplies a concrete comparison for evaluating how conflicts of interest can be reduced.

Committee minutes should identify questions, evidence, decisions, responsible parties, deadlines, and exception approvals. A decision to increase authority needs an explicit analysis of changing accumulations, limit exhaustion, treaty capacity, claims trend, and financial resources. Generic statements that the account is progressing well are insufficient if the record cannot show why an increased risk limit was justified.

Escalation must have consequences. A breach of approved rates or limits should trigger an immediate authority review; persistent late bordereaux should trigger testing of data and premium records; rising catastrophe accumulations should trigger limit controls; a reinsurer downgrade or collateral shortfall should trigger credit and liquidity review. The carrier should be able to suspend new binding, reduce authority, demand additional information or security, and direct corrective action while preserving legal obligations already incurred.

Monitoring frequency and denominator integrity

Risks evolve at different speeds. A new binder may exceed a policy limit within seconds; geographic accumulation can grow materially within days; emerging claims frequency may be visible monthly; reserve adequacy can require several development periods. The control schedule should match these processes, using real-time or near-real-time blocks for authority breaches and slower independent reviews for long-tail actuarial performance.

The Lloyd's delegated reporting standards establish common risk, premium, and claims reporting expectations and recognize additional data needs depending on jurisdiction and risk. Reliable reporting requires clear denominators: gross written or earned premium, risk count, policy limits, exposure units, reported or incurred loss, development period, currency, and exchange-rate date. A program's gross loss ratio is incomparable across periods if one report uses paid losses and another uses incurred ultimate estimates.

Bordereau reconciliation should include duplicate detection, policy status, cancellations, endorsements, premium receipts, cession percentages, and outstanding claims. Claims should be linked to policy and coverage fields so that the insurer can test the treaty's subject-business requirements. Sampling must include adverse or unusual transactions rather than selecting only readily verifiable normal policies. Audit results should lead to measurable remediation, with an explicit follow-up test rather than mere management assurances.

Subdelegation and technology dependencies

An MGA may depend on subproducers, software platforms, call centers, third-party administrators, external actuarial consultants, and payment processors. Some perform clerical tasks; others make binding or claims decisions with financial consequences. Each function requires identification of the decision owner, data access, outsourcing permissions, service continuity obligations, and regulatory context.

Lloyd's contract layering and subdelegation guidance illustrates how layered authority is controlled in its own market and emphasizes continuing managing-agent responsibility. The insurer should ensure that a delegate cannot silently extend underwriting powers to another entity outside the approved arrangement. Contractual approval should address both the existence and scope of any subdelegation, while audit rights should reach the actual records and operations necessary to evaluate performance.

A vendor failure can impair multiple program functions simultaneously: underwriting approvals, policy documents, premium accounting, claims files, and reinsurance notices. Operational resilience is therefore a component of insurance capacity rather than a separate information technology concern. The carrier's continuity arrangements should specify recoverable data formats, service restoration targets, backups, alternative personnel, and a pathway to transfer functions when the primary administrator is unavailable.

Claims, Policyholders, and Runoff

The claims function tests direct accountability

The practical test of fronting accountability occurs when a claimant invokes the policy. The carrier's direct liability is meaningful only if it has the authority, information, resources, and operational means to ensure timely and lawful payment of covered claims. A delegated claims administrator may supply expertise and scale, but the issuing insurer must understand and supervise the service. It should identify claims authority levels, referral thresholds, loss adjustment expenses, reserve approvals, dispute procedures, settlement authority, defense management, and protection of claim funds.

Lloyd's delegated claims administrator framework provides a market-specific example of claims outsourcing and its oversight expectations. Detailed delegated claims contract requirements cover authority levels, investigation, claims estimates, complaints, access to records, and continuing duties. The underlying research question is whether the carrier can reconstruct and direct a material claim, especially when claims servicing and treaty recovery involve different organizations.

A claims control file should identify the insured contract, alleged event, timing and coverage trigger, reserve evolution, communications, adjustment authority, legal expenditure, settlement decision, payment date, reinsurer notice, and collectible amount. It should record when a claim is outside expected treaty cover, even if the insurer must pay under the primary contract. A program that measures claims service only by net retained cost may neglect the claimant's experience or the legal consequences of gross obligation.

Disputes, late recoveries, and claims inflation

Claims disputes can arise from policy interpretation, disputed facts, valuation, reserving, coverage notice, or treaty wording. A reinsurer might question whether the MGA was authorized to write a risk, whether a claim meets treaty definitions, whether defense expenses are covered, or whether notification was timely. A prudential analysis should not equate the existence of an insurance claim with an immediately collectible reinsurance receivable.

Late recoveries impose financing costs and can constrain the insurer's ability to pay other policyholders. An adverse claims event may involve a cluster of policies under the same management platform, with the same reinsurer, collateral bank, or legal interpretation. This interconnectedness makes it necessary to test cash flow at the insurer level rather than calculate only expected net ultimate loss.

Claims service quality should be measured through time to acknowledgement, time to settlement, complaints, reopened files, paid-to-reported lag, adequacy of large-loss reserving, claim file completeness, and recovery aging, with denominators defined consistently. For long-tail liability, early claim payment statistics provide only partial evidence of ultimate development. Operational metrics should be interpreted alongside actuarial and coverage information.

Continuity after the end of new business

Ending the MGA contract does not extinguish policies already issued. Occurrence-triggered losses, claims-made reporting periods, long-tail reserves, reopened claims, unearned premium liabilities, and contractual notice obligations may persist. The fronting carrier should maintain claims and record access, reinsurance billing rights, collateral protection, complaints handling, and payment capability after the producing arrangement terminates.

Lloyd's runoff oversight and monitoring guidance addresses the need for a continuing service plan and transfer of administration where the outgoing third party cannot perform. This is an instructive functional standard: a capable fronting carrier should test its ability to transfer files and authority before an emergency arises. Data portability must include accurate policy versions, claim documents, reserve histories, premium and reinsurance accounts, and customer contact records in usable formats.

A hypothetical run-off test can simulate sudden insolvency of the MGA. The carrier should be asked to identify all in-force contracts, obtain all claim files, contact the claims administrator, restrict further unauthorized binding, preserve premium funds, report treaty exposures, and appoint qualified replacement personnel. If those actions depend on the failing MGA's sole cooperation, the continuity plan is not executable. This test can be performed without claiming to predict the probability of a real insolvency.

Policyholder protection and the identity of the insurer

Customer communications should clearly identify the legal issuing insurer and the authorized claims contact. Program branding may emphasize the MGA or distributor, but it should not create confusion about the contractual insurer. Claims handling standards, complaints, and disclosure obligations vary by jurisdiction and policy type; the regulatory requirements must be checked before a particular practice is declared compliant.

The NAIC receivership handbook's treatment of reinsurance illustrates why the protection of insureds depends on the legal relationship between direct policy, reinsurance, insolvency estate, and collateral. A cut-through or special arrangement can change relevant rights, but no universal outcome should be inferred without the governing text. Administrative continuity, financial security, and clear policy documentation jointly support the insured's ability to use the protection purchased.

Comparative Supervisory Analysis

International principles

The IAIS Insurance Core Principles provide the international supervisory framework for evaluating governance and reinsurance. The relevant principles require insurers to manage material risks and to ensure that reinsurance arrangements are understood, controlled, and effectively incorporated into financial management. They are intended for supervisory implementation. Their application as binding law depends on national authorities and their respective legal instruments.

Across jurisdictions, the functional sequence is similar. A carrier accepts an insurance obligation, delegates some activity, transfers some risk, receives and evaluates information, and must intervene when the arrangement deteriorates. The sequence is analytically comparable without asserting that every regulator uses the same legal language or enforcement powers. This distinction is essential in research with an international scope.

United States

State insurance law determines the legal powers and obligations of U.S. insurers within the relevant jurisdiction. The NAIC's model-law directory identifies the Managing General Agents Act and credit for reinsurance models as regulatory reference instruments. The operative rules must be located in state statutes and regulations, with their effective dates, definitions, exemptions, and supervisory interpretation. The NAIC's own models should not be described as federally binding requirements.

North Carolina provides a targeted statutory illustration. Its management-contract filing provision addresses insurer management arrangements, including disclosure concerning compensation and potential transfers of substantial control. The related grounds for contract disapproval describe a state supervisory response to certain arrangements. These sections are not a complete national framework for MGAs and do not determine whether every particular fronting contract falls within their scope.

On reinsurance security, the NAIC's covered-agreement and reciprocal-jurisdiction account documents changes in model treatment of collateral and qualifying reinsurers. A real underwriting approval should identify the ceding insurer's domicile, the actual state enactment, the particular reinsurer's status, and any relevant covered-agreement conditions. An account of these distinctions is more accurate than a global statement that every nonadmitted or foreign reinsurer is subject to the same collateral requirement.

European Union

Article 49 of the Solvency II Directive contains an express retained responsibility rule for outsourcing. The insurer remains responsible for the obligations applicable to the outsourced activity. Article 274 of the Delegated Regulation provides outsourcing policy and contract requirements, including service-provider diligence, contractual access to information, regulatory access, monitoring, and contingency controls for critical or important functions. Whether an individual program activity is classified as critical or important is a fact-sensitive regulatory question.

EU recognition of reinsurance as a risk mitigation technique requires the applicable contractual and legal conditions to be satisfied. Article 209 on qualitative criteria and the related effective-transfer and contract provisions illustrate the relationship between risk transfer, counterparty exposures, and capital calculations. These requirements should not be transposed automatically to non-EU insurers or used to determine the exact capital requirement of a specific carrier without its financial and regulatory information.

Lloyd's delegated-authority market

Lloyd's represents a distinctive market structure with syndicates, managing agents, coverholders, and separate delegated claims administration. Its contractual authority requirements and ongoing oversight expectations create a particularly detailed operational comparison for fronting governance. Managing agents must establish and supervise permitted powers, premium and liability limits, reporting, escalation, and termination arrangements.

The Lloyd's structure should not be labeled identical to national fronting where a licensed carrier cedes risk to a captive or reinsurer. However, its technical design demonstrates how a delegation of binding authority can be constrained by a verifiable record of permissions and how responsibility for performance remains with the institution empowered to delegate. This functional comparison supports a general governance interpretation while preserving the distinct legal identities of the markets.

Comparative conclusion

The regulatory comparison identifies agreement at the level of economic and operational functions, while the precise legal mechanisms vary. The direct issuer's obligations, authorized discretion of the delegate, enforceability of reinsurance, capital recognition, claims-payment ability, and continuing oversight can be assessed using a common set of research questions. The legal answer to each question must be anchored in the applicable contract and jurisdiction.

The international evidence thus supports a disciplined sequence of inquiry: identify the primary policy obligor; reconstruct delegation; test treaty and collateral enforceability; verify financial and operational capacity; and assess continuity when a program participant fails. This sequence is an analytical synthesis of existing requirements, not a proposed new supervisory code.

Reproducible Stress Illustrations

Synthetic program assumptions

A deterministic example makes the difference between legal policy liability and financial reinsurance protection visible. Consider a synthetic annual property and casualty program with USD 100 million gross written premium. Expected ultimate covered losses are assumed to equal USD 65 million. The insurer has signed a 90 percent quota-share treaty covering every modeled indemnity loss, with no exclusions or treaty aggregate exhaustion for this first illustration. The retained percentage is 10 percent. To isolate the effect of cession and collection, acquisition costs, fronting fees, taxes, regulatory capital, and nonindemnity expenses are initially excluded.

Under these assumptions, expected ceded covered losses are 0.90 × USD 65 million = USD 58.5 million. Expected retained covered losses are 0.10 × USD 65 million = USD 6.5 million. The primary insurer nevertheless has the policyholder-facing covered loss obligation represented by the USD 65 million gross scenario. The ceded amount is an expected contractual reimbursement, not a transfer of the direct policy's identity. This arithmetic is an explicit scenario calculation, not an inference from a proprietary insurer dataset.

Gross loss and counterparty failure

Consider an adverse event producing USD 120 million covered gross ultimate loss. If every relevant claim is covered by the treaty, the contractual reinsurer share is 0.90 × 120 = USD 108 million and the retained share is USD 12 million. Suppose the reinsurer fails to pay 30 percent of the ceded claim, with no available security covering this lost amount. The uncollected ceded loss is 0.30 × 108 = USD 32.4 million. The resulting modeled financial burden on the direct insurer is USD 44.4 million: USD 12 million retained loss plus USD 32.4 million failed recovery.

The modeled burden is 3.7 times the initial USD 12 million retained loss under the same gross-loss stress. The multiple is produced by a stipulated 30 percent recovery loss and is not a prediction of a 30 percent default probability. It demonstrates the scale of contingent counterparty exposure relative to nominal retention in a highly ceded arrangement.

The sensitivity is transparent:

Gross covered loss (USD m) Ceded share (USD m) Retained loss (USD m) Ceded shortfall assumption Ceded shortfall (USD m) Total carrier loss (USD m)
65 58.5 6.5 0% 0 6.5
120 108 12 0% 0 12
120 108 12 10% 10.8 22.8
120 108 12 30% 32.4 44.4
120 108 12 50% 54 66

The same equations can be reproduced independently:

  • Ceded loss = gross eligible loss × cession fraction.
  • Retained loss = gross eligible loss × (1 − cession fraction).
  • Failed recovery = ceded loss × unrecovered fraction.
  • Carrier economic loss = retained loss + failed recovery, if these categories are mutually exclusive.

Any actual model must also account for policy deductibles, treaty deductibles, aggregate limits, recoveries from salvage or subrogation, covered expenses, reserves, and collateral rules. The equations are deliberately simple to reveal one mechanism without claiming to model a real carrier's capital requirement.

Collateral sensitivity without double counting

Retain the USD 120 million gross-loss scenario and the USD 108 million ceded treaty claim. Assume 70 percent of ceded liability, USD 75.6 million, is backed by immediately drawable and legally effective cash collateral. The remaining USD 32.4 million is unsecured. If all unsecured recoveries fail and collateral performs exactly as assumed, the carrier's modeled loss is USD 12 million retained plus USD 32.4 million unsecured loss, or USD 44.4 million. If collateral delivers only 90 percent of its recorded value, the additional security shortfall is USD 7.56 million and modeled loss becomes USD 51.96 million.

This example treats the unsecured loss and collateral haircut as separate amounts. It does not impose a second default loss on the already impaired collateral balance. If, instead, the unsecured USD 32.4 million yields 95 percent recovery and the secured amount performs in full, the unsecured loss is USD 1.62 million and modeled total carrier loss is USD 13.62 million. These values show how contractual security and actual collectibility affect economic outcomes. They do not establish that a specified collateral haircut is frequent or regulatorily required.

Financial analysis should also ask who owns the collateral, which entity may draw it, what documentation a bank or trustee requires, when the instrument expires, what laws govern the account, and whether collateral value correlates with the loss event. It is possible for fully recognized collateral to have a cash-access problem; the legal recognition and immediate liquidity tests must be performed separately.

A claims-payment liquidity gap

Suppose that USD 50 million of covered gross claims must be paid by day 30, while reinsurance settlement occurs only after day 60. Available unrestricted insurer cash and legally accessible committed facilities total USD 35 million through day 30. The day-30 cash gap is USD 15 million. The ultimate recoverable balance could still be USD 108 million in the overall scenario, but those future funds do not eliminate the interim need for liquid resources.

A general horizon-dependent expression is: cash gap at horizon h = max(0, cumulative legally payable claims through h − cumulative actually accessible cash inflows and facilities through h). The timing convention must prevent counting a collateral draw both as unrestricted cash and as an additional facility. A stress with different settlement lags can be evaluated by changing the cash date without changing the ultimate covered loss.

This distinction connects directly to the IAIS treatment of reinsurance risk transfer, under which the insurer's economic effects and related risks require examination. Net loss expectations, regulatory solvency calculations, and claims-payment cash projections answer different questions and can produce different binding constraints on a program.

Basis risk between policy and treaty

Now assume the primary policies require payment of USD 120 million in covered indemnity and USD 8 million of additional loss-adjustment expense, while the 90 percent quota share applies only to indemnity. Expected treaty reimbursement is USD 108 million. The carrier bears USD 12 million retained indemnity and USD 8 million expenses outside the assumed treaty cover, totaling USD 20 million before any credit deterioration or collection delay.

The example establishes why a reinsurance percentage must be applied to the legally relevant category of underlying loss. If one were to calculate 90 percent of USD 128 million, the result would overstate the assumed reinsurance recovery by USD 7.2 million. This is a contractual mismatch example, not a claim that fronting treaties uniformly exclude claims costs.

A reproducible spreadsheet implementation needs the following input fields: primary covered indemnity, ancillary covered payments, treaty-eligible share of each payment type, quota share, treaty aggregate remaining, collateral balance and haircut, unsecured recovery fraction, claim due date, reinsurer collection date, and accessible carrier cash by date. Model outputs can be checked against the arithmetic above. The source record must preserve the assumptions so a reader can distinguish demonstrable calculations from hypothetical magnitudes.

Method, Limitations, and Research Agenda

Comparative documentary method

The study employs purposive, primary-source documentary analysis rather than a statistical sample of insurers. Institutional sources were selected to address the mechanism of fronting: legal responsibility, underwriting delegation, risk transfer, collateral and recoverability, information reporting, and claims continuity. The NAIC's explanatory publications and model-law catalog establish U.S. supervisory context, but not the enactment of a specific provision in all states. North Carolina law supplies a narrow domestic statutory illustration. Solvency II and Lloyd's documents supply detailed EU and market-specific comparators. The IAIS provides international supervisory principles.

Documentary sources have different evidentiary weight. The NAIC receiver's reinsurance chapter is an official explanatory document and supports the fronting definition and insolvency distinctions within its scope. The EIOPA outsourcing requirements reproduce specified regulatory provisions. Lloyd's published requirements document market obligations and expectations. Academic articles support theoretical explanations or reported empirical results only within the boundaries of the inspected material. Each type of evidence is assigned the claim it can substantiate.

The research was conducted against public materials available by November 21, 2026. It has no direct access to confidential MGA underwriting agreements, actual fronting treaties, carrier balance sheets, internal control reports, regulatory examination files, or insurer transaction-level bordereaux. The synthetic examples use declared assumptions. No causal inference from insurer microdata, estimation of reinsurer default frequency, or numerical capital adequacy opinion for an actual carrier is asserted. Jurisdiction-specific duties require review of currently operative domestic law and contracts at the point of application.

Relationship to established research

The agency-cost work of Jensen and Meckling provides an established economic language for diverging incentives between a principal and its agent. The Garven, Hilliard, and Grace reinsurance study reports relationships between insurer–reinsurer tenure and selected outcomes in a U.S. historical insurer panel; its visible abstract does not support claims about fronting-specific treatment effects. Jean-Baptiste and Santomero's contract analysis examines reinsurance design under information asymmetry. The present manuscript integrates the conceptual implications of these works with current supervisory and contract evidence. This is an original interpretive synthesis, without attribution of findings those works did not report.

The institutional evidence permits several competing explanations of a poor fronting outcome. One involves weak risk selection or inadequate pricing. Another involves policy and treaty coverage mismatch. A third involves reinsurer credit failure or delayed collection. A fourth involves deficient data and loss reserves. A fifth involves administrative failure after program termination. A carrier's deteriorating combined ratio cannot alone distinguish these causes. Empirical work should decompose underwriting loss, expense, counterparty impairment, and liquidity delay before inferring the effectiveness of a particular governance practice.

A future empirical research design

A credible empirical extension could construct a carrier–program–reinsurer–quarter panel, with program identifiers preserved across changes in issuing company or delegate. Potential explanatory variables include initial underwriting authority breadth, referral and override frequency, collateral type, unrecovered balances, treaty wording changes, reinsurer concentration, bordereau delays, audit timing, and MGA remuneration design. Outcomes could include developed gross and net loss ratios, claim payment times, reinsurance collection lag, adverse reserve development, disputes, complaints, and verified continuity incidents.

The identification design must address selection. Riskier programs may be audited more often; financially weaker carriers may purchase more reinsurance; catastrophe-exposed products may demand more collateral. Simple correlations between intensive oversight and adverse performance could therefore reverse cause and effect. Event studies around documented regulatory or contractual changes would require credible parallel-trend assumptions, stable program composition, and attention to differing claim maturity. Data sharing, confidentiality, privacy, and survivorship bias would need explicit treatment.

Another project could compare national legislative texts concerning outsourced insurance functions, admitted policy issuance, reinsurance credit, licensed intermediaries, and insurer insolvency. It should record the effective date and precise provision for each legal claim, distinguish regulatory guidance from binding law, and avoid inferring jurisdiction-specific legality from an international standard. Such research would extend rather than retroactively validate the intentionally bounded comparison here.

Operational tests of the interpretation

Three audits offer a practical method for evaluating the accountabilities described in this study. A binding audit selects policies and reconstructs risk data, pricing, approved guidelines, referral decisions, aggregated authority, and issuance. A reinsurance cash audit traces major claims from direct policy coverage through settlement, treaty notice, recoverable accounting, collateral draw rights, and actual cash collection. A discontinuity audit simulates loss of the MGA and verifies the carrier's ability to identify every in-force policy, reconstruct files, stop unauthorized issuance, and arrange continuous claim service.

These are tests of existing contractual, underwriting, and prudential obligations. The suggested approach does not amount to a new theoretical model; it operationalizes known controls into verifiable records. The interpretation would be weakened in a particular program if the proposed problems proved immaterial after examination of exact treaty terms, financial security, timing of cash, and effective operational rights. Its explanatory force lies in its ability to isolate the conditions under which large reinsured capacity remains consistent with the primary insurer's obligations.

Conclusion

Fronting insurance coordinates licensed policy issuance, specialized risk selection, reinsurance finance, and distributed administrative expertise. The central institutional problem is that the insured's direct contractual claim remains attached to the issuing insurer while much of the risk financing and operational discretion belongs to other participants. A carrier's underwriting accountability is therefore demonstrated through substantive approval, precisely delimited binding rights, reconstructible transaction evidence, effective monitoring, recoverable and legally secured reinsurance, claims liquidity, and the ability to maintain service when a program partner exits or fails.

The documentary evidence supports five connected findings. The first is the persistence of policyholder obligations despite substantial cession, highlighted in the NAIC's explanation of fronting. The second is the need for enforceable authority limits and continuing oversight, illustrated by Lloyd's delegated underwriting guidance. The third is the dependence of recognized reinsurance mitigation on legally effective risk transfer and assessment of associated risks, as in Solvency II's recognition criteria. The fourth is the distinct importance of immediate cash availability alongside ultimate contractual reimbursement. The fifth is the continuation of claims and servicing obligations into runoff even when new policy issuance ceases.

The reproducible example shows why a carrier retaining 10 percent of a loss can nevertheless face a much larger economic burden when the reinsurer does not perform, and why a solvent final-net-loss projection may coexist with an immediate cash shortfall. The numerical magnitudes are hypothetical; the underlying legal and financial distinctions are grounded in the cited sources. No universal cession ratio, optimal collateral percentage, or new doctrine follows.

The resulting conclusion is that defensible fronting capacity is the scale of insurance obligations that the issuing carrier can lawfully authorize, technically understand, monitor, finance, and service under realistic contractual and counterparty stresses. This is an analytical synthesis of established responsibilities and risk mechanics. Applying it to any individual insurer requires the actual contracts, balance sheet, claims data, applicable law, and an independent assessment of operational performance.

References

  • National Association of Insurance Commissioners (2022). Receiver's Handbook for Insurance Company Insolvencies, Chapter 7, 'Reinsurance,' especially pages 398–399, 426–427. Official institutional publication. Source

  • National Association of Insurance Commissioners (accessed 2026). 'Insurance Topics: Reinsurance.' Official explanatory resource. Source

  • National Association of Insurance Commissioners (accessed 2026). 'Insurance Topics: Covered Agreement.' Reinsurance collateral model revisions and reciprocal-jurisdiction conditions. Source

  • National Association of Insurance Commissioners (accessed 2026). 'Model Laws.' Official model-law catalogue, including the Managing General Agents Act. Source

  • International Association of Insurance Supervisors (2024; online tool consulted 2026). Insurance Core Principles and ComFrame, especially ICP 8 and ICP 13, Reinsurance and Other Forms of Risk Transfer. Source

  • Lloyd's (2026). 'Delegated Underwriting.' Market Knowledge, last reviewed April 27, 2026. Source

  • Lloyd's (accessed 2026). 'Oversight & Monitoring of Contracts of Delegation of Underwriting Authority.' Market guidance and requirements. Source

  • Lloyd's (accessed 2026). 'Contracts of Delegated Authority.' Requirements of the Council and Intermediaries Byelaw provisions. Source

  • Lloyd's (accessed 2026). 'Managing Agent Due Diligence of Third Parties and Contracts of Delegation.' Market Knowledge. Source

  • Lloyd's (2026). 'Reporting Standards.' Delegated Authority Market Knowledge, reviewed August 10, 2026. Source

  • Lloyd's (accessed 2026). 'Coverholders.' Official definition of underwriting authority at Lloyd's. Source

  • Lloyd's (accessed 2026). 'Delegated Claims Administrators.' Claims delegation overview. Source

  • Lloyd's (2026). 'Contract Layering & Sub-Delegation.' Market Knowledge. Source

  • European Parliament and Council (2009, as amended). Directive 2009/138/EC (Solvency II), Article 49, 'Outsourcing.' EIOPA Single Rulebook. Source

  • European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 274, 'Outsourcing.' EIOPA Single Rulebook. Source

  • European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 209, 'Qualitative Criteria.' EIOPA Single Rulebook. Source

  • European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 210, 'Effective Transfer of Risk.' EIOPA Single Rulebook. Source

  • European Commission (2015, as amended). Commission Delegated Regulation (EU) 2015/35, Article 211, 'Risk-Mitigation Techniques Using Reinsurance Contracts or Special Purpose Vehicles.' EIOPA Single Rulebook. Source

  • European Parliament and Council (2009, as amended). Directive 2009/138/EC, Article 101, 'Calculation of the Solvency Capital Requirement.' EIOPA Single Rulebook. Source

  • North Carolina General Assembly (accessed 2026). North Carolina General Statutes § 58-34-10. Management contract and custodial agreement filing. Source

  • North Carolina General Assembly (accessed 2026). North Carolina General Statutes § 58-34-15. Grounds for disapproval. Source

  • 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. Source

  • Garven, James R., James I. Hilliard, and Martin F. Grace (2014). 'Adverse Selection in Reinsurance Markets.' The Geneva Risk and Insurance Review 39(2): 222–253. Author abstract at SSRN. Source

  • Jean-Baptiste, Eslyn L., and Anthony M. Santomero (2000). 'The Design of Private Reinsurance Contracts.' Journal of Financial Intermediation 9(3): 274–297. DOI 10.1006/jfin.2000.0291. Source