RESEARCH DOSSIER
Captive Insurance Domicile Choice and Regulatory Arbitrage
Abstract. Captive insurance domicile choice is an institutional decision concerning the legal location, regulatory classification, capitalization, management, and financial operation of an insurer controlled by its insured enterprise or participating owners. It connects corporate risk financing with international competition among regulatory jurisdictions. This monograph investigates the distinction between productive regulatory specialization and regulatory arbitrage that shifts costs or vulnerabilities beyond the entity receiving the advantage. It combines comparative institutional analysis of Vermont, Bermuda, the Cayman Islands, and the European supervisory setting with United States reserve-financing and tax evidence, international transfer-pricing guidance, and explicitly bounded empirical literature. Official year-end licensing statistics for Vermont and Cayman are examined as administrative observations rather than measures of globally comparable insurance capacity. Worked financial illustrations distinguish captive-level obligations from consolidated enterprise losses, minimum legal capital from economic capital, and nominal liquidity from assets available during joint distress. The analysis finds that the economic effects of a domicile depend on the entire risk-bearing chain, including policy issuance, reinsurance, collateral, asset investment, governance, tax residence, and insolvency exposure. A less costly regulatory regime can be compatible with robust risk financing when its requirements correspond to the underlying business. Conversely, a reduction in recognized capital or tax expense can leave the consolidated loss distribution unchanged and impair recovery prospects. The conclusions establish an evidence-based comparative approach without asserting a universal ranking, a causal effect of domicile selection, or a new proprietary theory.
Keywords: Captive insurance; domicile choice; regulatory arbitrage; insurance supervision; enterprise risk financing; reinsurance; proportionality; insurance solvency; transfer pricing; fronting; group capital; regulatory competition.
Introduction
A captive insurer establishes a regulated mechanism through which an enterprise or group of participating owners finances exposures associated with its operations. The choice of domicile determines the legal and supervisory environment in which that mechanism is constituted. It influences the eligibility of the proposed insurance business, the treatment of capital and assets, the cost of retaining risk, and the conditions under which the insurer may enter relationships with insurers and reinsurers in other jurisdictions. Its scientific importance lies in the interaction between organizational boundaries and geographical boundaries: business activity, insurance obligation, control, capital, and claims payment can have distinct legal locations.
The central question is how differences among supervisory regimes affect the economic and institutional substance of captive risk financing. Three issues organize the investigation. First, which observable characteristics of jurisdictions and captive forms affect the enterprise's feasible choices? Second, what risk is transferred when an operating company insures with an affiliated insurer, and what risk remains within the group? Third, under what circumstances does selection of a more accommodating regime represent proportional supervision or efficient institutional specialization, and under what circumstances does it reduce the resources available to meet essentially unchanged obligations?
These questions require comparison at the level of specific transactions. A nominal capital requirement is meaningful only for the applicable license category, liability portfolio, admissible assets, additional solvency obligations, and supervisory discretion. A low tax rate is economically significant only in conjunction with the treatment of the transaction in every relevant jurisdiction. A captive count measures units defined by the reporting authority, not the financial intensity, diversity, or quality of their business. The problem is therefore especially suitable for institutional insurance science, where legal forms and economic outcomes are linked through identifiable contracts and balance sheets.
International scope is achieved through deliberately contrasting jurisdictions, not through an assertion that every jurisdiction is represented. Vermont illustrates a United States state-based captive framework with multiple organizational forms. Bermuda and Cayman offer distinct class-based frameworks for international insurance. The European Union supervisory perspective supplies a dated example of convergence within a common prudential architecture. International tax guidance connects the resulting legal structures to economically substantive intragroup transactions. The US life reserve-financing case supplies a counterpoint to operating-company captives. The inquiry concerns insurance institutions and public policy rather than individualized domicile, contract, or tax advice.
Concepts, Institutions and Historical Development
Captive organization and the location of risk
The economic basis of a captive is the deliberate organization of risk financing around the controlling or participating insureds. The insurer may be owned by one enterprise, several unrelated firms, an association, or a specialized sponsor. Its license and insurance contracts give the insurer distinct rights and duties. These legal obligations can produce important effects for individual insured entities even when the associated cash flows remain within a larger consolidated group.
Vermont's regulator describes a range of structures in its formation and licensing overview (Vermont DFR, n.d.). A pure captive principally insures the parent and its affiliates within the permitted class of business. Group and association approaches allow participating enterprises to pool defined exposures. A sponsored structure provides a platform through which separate participants finance liabilities through cells. A branch and an affiliated reinsurance company address other organizational relationships. The institutional vocabulary must remain attached to its jurisdiction: a similarly named type in another regime need not have identical permitted business, liability segregation, or capital conditions.
A captive domicile is the jurisdiction responsible for the insurer's constitution or primary insurance authorization in the arrangement under study. It is analytically distinct from the situs of the insured property or activity, the place of policy issuance, the location of directors or management, the tax residence of the parties, and the jurisdiction of a third-party reinsurer. These attributes may coincide in a purely domestic arrangement; they may be distributed across multiple territories in an international program. Each attribute has separate economic consequences.
A direct captive writes coverage to an insured under an enforceable insurance arrangement. A reinsurance captive instead assumes obligations under a treaty or facultative reinsurance contract with a cedent. In the second case the original insured's rights ordinarily arise from the primary policy, and the cedent's recovery arises from the reinsurance contract. The contractual chain creates exposure to both loss severity and counterparty performance. Whether a particular program gives the policyholder direct recovery from another party depends on the governing policy, contract, and law. A general diagram cannot settle an actual coverage dispute.
The organizational spectrum also includes cells and pools. A cell reduces the need to build an entire stand-alone infrastructure for each participant, subject to the relevant legal segregation and contractual arrangements. Sharing administration can lower fixed overhead while introducing dependencies on the platform, sponsor, manager, and common service providers. An association or group captive may obtain diversification across firms, but the existence of multiple owners is no assurance of low correlation. Industry concentration, shared geographic hazards, and common liability trends can create severe aggregate exposures.
Institutional evolution and competing motivations
Captive development is associated with enterprises seeking greater influence over retained risk, funding, coverage design, claims administration, and insurance procurement. Formation becomes attractive when the conventional insurance program fails to match the enterprise's preferred combination of retention, coverage availability, and price. A captive can provide a durable location for loss data and reserve management, consolidate insurance decisions across subsidiaries, and support consistent protection for unusual exposures. Such benefits are hypotheses about organizational mechanisms; their magnitude must be established in the enterprise examined.
The design of captive classes reveals a historical transition toward specialized regulation. The Bermuda multi-license system (BMA, n.d.) separates captive general insurance classes from commercial and other categories. Cayman similarly distinguishes license subclasses through the characteristics of the business, as explained in the regulator's Class B licensing FAQs (CIMA, n.d.). In these arrangements regulation is differentiated partly by the relation between policyholders and owners. A commercial insurer can undertake substantial obligations to unrelated customers with limited ability to inspect or control its management. A captive's insured owners may possess more information and control, but employees, claimants, fronting insurers, and creditors can retain economically important exposures.
The same institutional flexibility also facilitates transfers between related insurers. The NAIC captive insurance overview describes historical use of captive transactions within US life insurance groups to finance specified statutory reserves. The regulatory problems of reserve financing differ substantially from the retention of industrial property or liability losses by a noninsurance parent. The relevant question in the life case is how obligations arising from external policies are funded after intragroup reinsurance. In the industrial case the principal question is how the enterprise organizes payment for operational losses. Similar corporate forms do not make the two populations interchangeable.
Regulatory competition and regulatory arbitrage
Regulatory competition denotes the interaction among jurisdictions that develop and maintain institutional environments attractive to mobile insurance activities. A jurisdiction can invest in specialized officials, predictable approval procedures, professional service networks, clear legislation, and appropriate risk-based requirements. An increase in the attractiveness of that jurisdiction may reflect a reduction in administrative friction without changing the resources needed for loss payment.
Regulatory arbitrage, in the analytical use adopted here, refers to structuring or locating economically related activities to obtain favorable treatment under differing rules, classifications, or supervisory perimeters. The term describes a mechanism and does not itself determine legal validity or welfare effects. A difference that corrects excessive duplication can increase productive efficiency; a difference that weakens the effective backing of liabilities can redistribute downside risks. Tax arbitrage concerns tax treatment, whereas prudential arbitrage concerns requirements designed to support insurance performance. The two can coexist, but their factual and legal tests differ.
The distinction is observable through counterfactual comparison. Would the same loss-generating exposures, contractual claims, risk distribution, and asset resources be present under the alternative arrangement? Which regulatory cost changes, and why? Who benefits from the change? Who bears losses in adverse states? These questions separate the advantages attributable to improved organization from those attributable primarily to changes in recognition or enforceability.
Primary Market Sources and Comparative Research Design
Evidence boundaries and comparability
The comparative study uses licensing regimes as institutional cases, administrative observations as descriptive data, selected peer-reviewed findings as bounded evidence, and financial arithmetic as explicitly hypothetical demonstrations. The jurisdictions were selected purposively to expose different classification and supervisory mechanisms. They are not a random sample; the analysis estimates neither a worldwide average effect nor a statistical coefficient of regulation on domicile formation.
Source authority follows the nature of the claim. Licensing pages describe the relevant regulator's published requirements at the retrieved snapshot. Statutory excerpts establish specifically inspected legal language, subject to version and retrieval limitations. An opinion issued by a European authority is interpreted as supervisory guidance for its stated framework. An administrative table establishes the reported numbers for the designated dates and units. An academic abstract supports only the results explicitly reported in that abstract. A professional accounting FAQ supports the limited accounting example it describes. These evidence types should not be merged into a single undifferentiated claim of regulatory compliance.
The international reference point is the IAIS ICPs and ComFrame standard-setting framework. Its institutional role is distinct from the binding requirements adopted by any individual jurisdiction. A study of implementation requires identifying domestic legal authority, scope, effective dates, supervisory practice, and the status of the regulated entity. The monograph does not assign a national compliance rating from the existence of the international framework.
The principal common-unit observations used below are end-of-year license counts. Their comparative interpretation is deliberately narrow. For jurisdiction j and date t, define N(j,t) as the number of licenses included in that regulator's specified administrative series. The change N(j,t) minus N(j,t-1) is a net stock change. It can combine new licenses, cancellations, transitions of status, category changes, and transfers. New approvals report only one kind of flow. The reporting unit can include a cell-bearing insurer without counting every cell as another insurer. These features matter even before the issue of international currency or insurance accounting arises.
Administrative observations for Vermont and Cayman
The Vermont DFR 2025 Aggregate Report (2026, pp. 2, 4) reports 707 licensed captives at year-end 2025, comprising 677 active and 30 dormant. New licenses during 2025 numbered 51. The end-2024 stock was 683. The same report gives 316 cells and separate accounts in a separate series. The analysis treats the reported active and dormant categories as mutually exclusive and does not add cells to the company-license total.
The CIMA insurance licensee statistics table reports 670 Class B insurers at year-end 2024 and 693 at year-end 2025. Its 2025 total includes 138 segregated portfolio companies. The reported B(i), B(ii) and B(iii) subclass counts at the later date are 471, 14, and 208 respectively. Their sum is 693, a basic internal cross-check. The earlier subclass counts of 463, 13 and 194 sum to 670.
| Administrative series | End 2024 | End 2025 | Net change |
|---|---|---|---|
| Vermont licensed captives | 683 | 707 | 24 |
| Cayman Class B licensees | 670 | 693 | 23 |
| Cayman Class B(i) | 463 | 471 | 8 |
| Cayman Class B(ii) | 13 | 14 | 1 |
| Cayman Class B(iii) | 194 | 208 | 14 |
Source: Vermont DFR, 2025 Aggregate Report; CIMA, Insurance Statistics. All figures are license-count observations within the authority's own reporting population, not monetary market shares. Cayman subclass rows are included in the Class B total.
The corresponding changes relative to the beginning stocks are approximately 3.51% for Vermont and 3.43% for Cayman Class B. These descriptive percentages are neither a difference-in-differences estimate nor evidence that the two markets attracted similar kinds of parent enterprise. Vermont's active share in its licensed population is approximately 95.76%; Cayman B(i)'s share of Class B in 2025 is approximately 67.97%. Those percentages have different denominators and different meanings.
The difference between Vermont's 51 new licenses and net increase of 24 is 27. It is an arithmetic residual of the provided stock and inflow observations. Assigning all of the residual to insolvencies, departures, or cancellations would require independent flow records. The same reasoning applies to increases in Cayman's subclasses. Growth in B(iii) can arise through new entrants or reclassification. The table alone does not identify the contributing mechanisms.
A global domicile league table would require harmonized units, observation dates, rules for inactive entities and branches, treatment of cells, and a definition of what constitutes a captive across classification systems. Premium series would introduce further issues: gross versus net, written versus earned, coverage period, currency conversion, affiliated versus external contracts, and elimination on consolidation. The present evidence supports a matched-period comparison of administrative movement; it does not support ranking supervisory quality or global premium capacity.
Scientific literature and limits of inference
Russell, Chen, and Chang (2026), in their captive insurance cash-flow study, examine the 2020 constituents of the S&P Global 100. Their published abstract reports insufficient support for a general improvement in annual corporate cash flow from captives. This finding challenges a simple inference from retention of premium cash to improved enterprise cash generation. It does not estimate a causal effect of changing domicile, and the present dossier does not reconstruct the paper's sample construction or regression specification from its abstract.
The second academic case has a different population. Koijen and Yogo (2016), in Shadow Insurance, report substantial growth of identified US life and annuity liabilities ceded to affiliated shadow reinsurers between 2002 and 2012. Their published abstract reports both potential efficiency benefits from relaxed capital requirements and potential solvency consequences. It supplies evidence that the regulatory perimeter can affect the economics of reinsurance. It does not determine contemporary capital adequacy for operating-company captives. Abstract access limits this dossier to the stated finding and mechanism, without independent validation of the study's estimation results.
A stronger empirical study of domicile choice would use an entity-year or enterprise-program-year panel with corporate ownership, new licenses, exits, historical rule changes, exposures, premium and reserve definitions, fronting arrangements, ultimate external risk transfer, and observed financial resources. It would include enterprises that considered but did not form a captive if such data could be obtained. Observing only formed entities creates selection bias; observing only survivors adds survivorship bias. Measurement of exposures and feasible alternatives is indispensable for causal interpretation.
Captive Domicile Incentives
Regulatory categories and minimum capital
Domicile choice begins with legal feasibility. A proposal must fit a license class that permits its intended insured relationships, coverage, direct or reinsurance business, and governance arrangements. The minimum legal capital figure is a lower boundary within a defined regime, rather than an estimate of the resources needed for every insurer in that category.
The State of Vermont's captive solvency capital guidance identifies a USD 250,000 minimum for a pure captive and describes the commissioner's discretion to prescribe additional surplus according to the business. The CICA Vermont domicile profile corroborates the stated amount. The comparison uses the published executive guidance rather than an independently certified consolidated statute; it neither equates the threshold with economic capital nor resolves additional entity-specific requirements.
Bermuda's general insurance license classes specify a $120,000 floor for Class 1, which covers the stated owner and affiliate risks. The minimum solvency margin determined through premium and reserve formulas can be higher. Class 2 has a $250,000 floor and permits the stated multi-owner business and certain other business within the relevant unrelated-risk threshold. The result is a classification-dependent minimum, not an unconditional flat capital cost.
CIMA's captive licensing capital schedule distinguishes minimum capital requirements (MCR) from prescribed capital requirements (PCR). For general business in Class B(i), the published schedule identifies both at US$100,000. Its general-business Class B(ii) and B(iii) schedules apply increasing premium-related requirements across tranches, in addition to the stated minimum floor. Assessment of the submitted business plan can generate additional capital requirements.
| Jurisdiction and identified class | General-business minimum or floor | Qualification |
|---|---|---|
| Vermont pure captive | USD 250,000 | Government domicile guidance; additional surplus discretionary |
| Bermuda Class 1 | 120,000 in regulator schedule | Formula-based margin can exceed the floor |
| Cayman Class B(i) | USD 100,000 | Published PCR equals MCR; business-plan capital may be additional |
These observations refer to different legally defined categories. The table demonstrates heterogeneity in thresholds, without treating the categories as equivalent or recommending a domicile based on the amounts. It excludes reserve recognition, asset admissibility, operational cost, tax, guarantees, and third-party insurance requirements.
Consider an explicitly hypothetical Cayman general-business license class to illustrate the published premium formula rather than a real capital assessment. If net earned premium equals US$12 million, the Class B(ii) schedule implies 10% of the first US$5 million plus 5% of the next US$7 million, or US$850,000. The corresponding Class B(iii) arithmetic is 15% of US$5 million plus 7.5% of US$7 million, or US$1,275,000. Both exceed their respective minimum amounts of US$150,000 and US$200,000. The calculation assumes the quoted tier schedule, general business, no additional supervisory requirements, and the same illustrative premium in each alternative; actual class eligibility is independent of the arithmetic. It shows why comparison of headline minimums fails for larger portfolios.
Operating costs, service networks, and institutional credibility
A complete cost comparison includes annual licensing and regulatory cost, manager and audit expenses, legal services, actuarial analysis, claims administration, bank and collateral expenses, local infrastructure, and the costs of changing domicile or exiting the program. Specialist service networks can reduce entry and recurring expenses through shared capabilities. Regulator experience can reduce uncertainty about permissible business and documentary requirements. These benefits have economic substance when they release resources that were otherwise consumed in unnecessary duplication or improve the administration of insured risk.
An institutional ecosystem also creates dependency. A manager serving many captives can supply efficient technical infrastructure, but disruption of that manager or poor oversight may affect multiple entities. A cluster of banks, auditors, actuaries, and fronting intermediaries can provide expertise and competitive capacity while creating concentration in service provision. The relevant unit of operational resilience extends beyond the legal entity.
Domicile-specific substance is illustrated by CIMA's licensing application requirements, which identify financial projections, business plans, director and owner information, and appointment of local licensed insurance management. The administrative burden has more than one purpose. It enables review of feasibility, accountability, and the ability to service obligations. Some functions can be performed by an external manager; the regulated insurer still needs an effective governance relationship with the outsourced provider. A comparison of fees alone cannot capture those institutional capabilities.
The practical value of predictable regulation derives partly from the option to amend the program when exposures change. Expansion into a new line, acceptance of external insureds, alteration of reinsurance, or restructuring of assets can change the license category and applicable conditions. The expected cost of these contingent changes belongs to the long-term choice of domicile. A low initial cost may be outweighed by expensive transition or run-off arrangements. Conversely, an expensive initial approval can be efficient when the resulting legal architecture supports a durable program.
Economic cost rather than headline tax
Tax treatment can influence insurance financing at several points: the recognition of insurance premiums and losses in the operating company, income of the captive, allocations of intragroup profits, taxes attached to premium or risk location, and taxation of the controlling owner. Cross-border reinsurance may add further constraints. These components depend on the actual parties, contract, jurisdictions, and periods. A nominal statutory tax rate for the domicile by itself does not establish the consolidated tax effect.
The OECD's 2020 financial transactions guidance addresses captive transactions through accurate delineation and arm's-length pricing. The existence of a legally registered insurer and formal contract is one piece of evidence; assessment of actual functions, capacity, and risk assumption is also necessary. Comparable premiums require comparable coverage, risk profile, and contractual terms. Where the relevant comparison is unavailable, an actuarial analysis may be appropriate within the stated transfer-pricing framework. Actual tax consequences depend on applicable domestic law and its interpretation.
This creates a separation between the choice of a regulated insurance domicile and the choice of a tax treatment. A business can have genuine insurance operations in a specialized jurisdiction even when a specific tax benefit is unavailable. It can also be subject to a tax-reporting requirement without that requirement deciding every issue about the substantive insurance contract. Institutional research should identify the question answered by each authority rather than treating registration, tax deductibility, and prudential soundness as one indivisible status.
Parent Enterprise Risk Transfer
Entity-level obligations and consolidated losses
The central financial distinction is between legal transfer and transfer beyond the economic reporting perimeter. At the operating-subsidiary level, the captive's policy can create a claim against a separate insurer. At the group level, a payment between two wholly consolidated affiliates is a transfer inside the group. The ICAEW intragroup insurance accounting guidance explains this difference for IFRS reporting: qualifying issued contracts are considered at the issuer's separate reporting level, while intragroup contracts are eliminated on consolidation. External insurance has a different perimeter effect.
Assume a manufacturing group with one operating company and a wholly owned consolidated captive. In a simplified one-year arrangement, the operating entity pays 5 million monetary units in captive premium. The captive pays 4 million in covered losses to that entity. The underlying operational loss is 4 million. In separate accounting records the premium and claim have entity-level consequences. Across the consolidated group those internal premium and indemnity cash flows cancel. The group has still suffered the 4 million operational loss, before external expenses and any independent recoveries.
This simplification ignores investment income, timing, taxes, separate financial-reporting measurement, expenses, and minority interests. It identifies only the internal flow cancellation. A captive can improve allocation and timing of funds, provide legal certainty to individual subsidiaries, and influence actual prevention or administration. These benefits can be valuable without the internal policy alone transferring the consolidated loss to independent capital.
Suppose the captive purchases external reinsurance covering losses above 3 million, up to a further 5 million. If an eligible covered loss is 7 million, a fully performing reinsurer pays 4 million under the assumed terms. The group's retained covered loss is then 3 million before premium, expenses, exclusion, credit, and settlement timing effects. If the reinsurer pays only half of the 4 million recoverable, the retained economic loss becomes 5 million before those additions. The reinsurance contract modifies the group's loss distribution only to the extent that an independent party accepts and performs the obligation.
A change of domicile can affect access to such independent risk-bearing capacity. It can change recognition of reinsurance, collateral eligibility, documentation, and counterparty arrangements. Those changes may have substantive value when they improve enforceable protection or reduce inefficient restrictions, and they may be hazardous when they remove economically significant resources while leaving the group exposed to the same catastrophe.
Diversification, dependence, and economic capital
Pooling contributes to insurance efficiency when risks are imperfectly dependent. Consider n equal exposures with loss variance v and common pairwise correlation rho. For total loss S, variance is n v + n(n-1) rho v. Dividing by the square of n, variance of average loss is v times [1/n + (n-1)rho/n]. This elementary identity follows from the sum of individual variances and twice each pairwise covariance. It assumes finite second moments, equal variance, and a valid equicorrelation structure.
For ten hypothetical exposures each having standard deviation one million, independent losses produce aggregate standard deviation sqrt(10) million, approximately 3.162 million. At correlation 0.25, aggregate variance is 32.5 squared million-unit terms, with standard deviation approximately 5.701 million. At perfect positive dependence aggregate standard deviation is ten million. Independence versus partial dependence makes a large difference even when every exposure has the same expected loss.
The illustration does not assume normally distributed insurance losses or interpret standard deviation as regulatory solvency capital. Catastrophe severity, long-tail claim development, and common legal or inflation shocks can generate much heavier tails than a second-moment summary captures. Scenario analysis, appropriate distributional models, and dependence sensitivity are needed to assess actual economic capital. A domicile classification based on owner relationship or premium amount cannot substitute for those exposure characteristics.
OECD (2020), in its discussion of risk assumption and diversification, recognizes that internal and external portfolios can provide different diversification opportunities and ties economic substance to the relevant facts. The existence of separate corporate insureds is therefore an incomplete measure of diversification. Geographic dispersion can be ineffective against globally correlated exposures, and different business activities can still depend on the same critical supplier or technology.
Fronting, collateral, and credit risk
Fronting separates the insurer that issues the primary policy from the captive that assumes a defined part of the underwriting risk. OECD (2020), in its fronting transaction discussion, explains the separate insurance and reinsurance contracts and the retention of compensation by the fronting carrier. The arrangement can accommodate restrictions on direct policy issuance and use the fronting carrier's operational capacity. It also creates a chain of distinct obligations and costs.
A fronting carrier ordinarily evaluates the captive's capacity to pay recoverables and can require contractual security. The effective risk retained by the front can depend on collateral quality, the priority of claims, permissible setoff, recoverable timing, and reinsurance wording. A letter of credit or trust arrangement has an economic cost, but its face amount is not automatically equivalent to freely available funds. The institution providing the security, restrictions on drawing, currency, and replenishment conditions matter.
A stylized balance-sheet scenario illustrates the difference. Assume a captive has 20 million of assets, 15 million of insurance liabilities, and 5 million of book equity. Assets consist of 8 million in independent liquid instruments and 12 million in loans to the parent group. Immediately before a 6 million claim payment becomes due, the parent suffers severe financial distress and 50% of the intercompany loan value is lost. Asset value falls to 14 million; liabilities remain 15 million before payment; equity becomes negative one million. The liquid investments could fund the immediate 6 million payment, but leave only 2 million immediately accessible and an impaired residual loan. Equity remains negative after payment, because assets and liabilities both decline by the amount settled.
The example assumes no external guarantees, recoveries, changes in reserve estimates, taxes or resolution costs; it is not a reported insurer event. It illustrates that solvency and immediate liquidity are distinct, and that a captive's asset impairment can occur in the same state in which the enterprise needs insurance funding. A lower stated minimum capital can leave this conditional dependency unchanged.
Regulatory Choice
Proportionality, governance, and supervisory convergence
The regulatory case for captive-specific rules rests partly on the different ownership and information relationships in captive insurance. A parent may have more insight into its insurer's retained exposures than a dispersed set of unrelated policyholders. Regulatory specialization may recognize the lower cost of particular conduct safeguards when commercial retail distribution is absent. That reasoning changes if the insurer assumes substantial unrelated liabilities, owes claims to parties outside the controlling enterprise, or relies on resources unavailable in stress.
EIOPA's 2024 Opinion on the Supervision of Captives addresses proportional supervision within its stated European framework. Its analysis of intra-group liquidity arrangements supplies a particularly useful example of substance-sensitive supervision. Asset classification affects the applicable risk treatment; supervisory assessment also considers liquidity, concentration, counterparty exposure, governance, and the possibility of contagion from group deterioration. A classification label cannot by itself settle the economic characteristics of the asset.
Group membership does not automatically make the group's funds available to the insurance entity. A treasury center may use captive assets for the group's general financing; a promise of repayment can encounter liquidity constraints and competing claims. The supervisor therefore needs sufficient information about the terms and availability of funds, the identity of the counterparty, and the exposure in stressed circumstances. The analytical significance of the European opinion is its treatment of risk arising from economic substance. It is dated guidance under its identified framework, not a uniform law applicable to Vermont, Bermuda, or Cayman.
Reserve financing and the recognition of capital
A life insurance reserve-financing transaction moves a specified insurance exposure or financial obligation through reinsurance. The operating insurer may recognize a different reserve or capital position if the transaction meets its applicable rules. The reinsurer may be subject to another regulatory measurement. The policyholder's original claim and the consolidated exposure may remain materially dependent on the same insurance group. The scientific question becomes whether the resulting arrangement provides additional enforceable resources, diversifies risk, or predominantly changes the regulatory measurement.
Koijen and Yogo's 2016 life-insurance study reports in its abstract that identified US life and annuity liabilities ceded to shadow reinsurers increased from USD 11 billion in 2002 to USD 364 billion in 2012. This historical evidence concerns the research authors' definition of shadow reinsurers. It cannot be used as a current global captive volume estimate. It demonstrates the scale at which differences in supervisory treatment can matter for financial intermediation, within the population and period reported.
The NAIC's reserve-financing regulatory account documents the historical response through Actuarial Guideline XLVIII and Model Regulation #787, together with the changing incentives associated with principle-based reserving. Model standards and accreditation mechanisms require attention to actual implementation. This dossier does not take the existence of a model regulation as proof of every state's identical legal requirements. The case is valuable because regulatory treatment evolved in response to a particular transaction type rather than to captive ownership in general.
A simplified thought experiment isolates the accounting-perimeter effect. An insurer that owes claims with an economic adverse-state resource need of 100 units faces a measured capital requirement of 25 units under an initial set of rules. Suppose an intragroup reinsurance arrangement reduces the combined measured requirement to 15 units, while the economic adverse-state exposure and all realizable external resources remain unchanged. The ten-unit reduction is a release of recognized regulatory capital, not an automatic reduction in the underlying loss. If the amount is distributed from the group, the actual buffer available to absorb losses can fall.
This numerical example is an assumption, not a claim about a named reserve formula, and it deliberately leaves taxes and transaction costs aside. A different arrangement could provide independent collateral or improve the allocation of real capital and thereby reduce economic cost without an equivalent deterioration in claim recovery. The counterfactual comparison must include these changes. A policy response that treats every intragroup transaction identically can therefore miss both legitimate efficiencies and genuine vulnerabilities.
Tax classification and the evidentiary test
A distinct regulatory interface arises in tax administration. The 2025 United States Treasury and IRS micro-captive final regulations designated specified categories of arrangements for transaction disclosure. This is a dated regulatory event within the US federal tax framework. The fact of classification under reporting rules does not establish the tax characterization of every captive, and the historical discussion does not assess subsequent judicial challenges, amendments, or any taxpayer's current obligations.
A scientifically meaningful distinction is between a transaction that finances an uncertain insurable loss and a transaction in which the principal observed benefit is unrelated to adequate insurance performance. Evidence relevant to this distinction includes the coverage wording, underwriting records, exposure data, actuarial basis of pricing, actual claims, policy limits, resources available for indemnity, utilization of specialist functions, and contractual obligations to unrelated parties. Absence of claims during a short period is not sufficient evidence against insurance because low-frequency loss exposure may be genuine. Premium volume likewise does not establish that a risk has been properly priced or transferred.
Cross-border tax analysis must specify which entity performs the underwriting control functions and possesses the capacity to assume the risk. The OECD transfer-pricing analysis of captives explicitly distinguishes the allocation of risk from the pricing of the resulting transaction. A purely contractual allocation, unsupported by actual functions and financial capacity, raises a different question from market pricing of a genuinely assumed risk. Appropriate methods depend on case-specific evidence. The scientific value of this distinction is to prevent an apparently competitive premium from obscuring the prior question of what economic transaction occurred.
Comparative total-cost and stress sensitivity
Consider two hypothetical alternatives with identical insured exposures and coverage terms. Monetary values are in millions per year, and the comparison excludes taxes and investment income. Alternative A has expected retained loss 4.0, external reinsurance premium 1.2, operating and supervisory expenses 0.5, and capital carrying cost 0.6. Alternative B has expected retained loss 4.0, external premium 1.2, operating expense 0.35 and capital carrying cost 0.35. The resulting expected cost sums are 6.30 and 5.90 respectively, so B is 0.40 cheaper under those assumptions.
These numbers are hypothetical and assume the retained loss distribution already accounts for the specified reinsurance recovery. The premium is added because it is the cost of the external protection. The capital cost is an opportunity cost, not the capital principal itself. The result does not rely on interpreting statutory minimums as actual capital requirements.
Now consider an adverse-state comparison in which B's arrangement generates a further expected uninsured credit or recovery cost of 0.55 relative to A, with all other assumptions held constant. The adjusted expected cost of B is 6.45. A becomes cheaper by 0.15. The break-even additional cost is 0.40. If the relevant difference instead consists of adverse tail exposure without a comparable difference in mean cost, an expected-cost comparison remains incomplete: liquidity requirements, risk tolerance, and insolvency consequences may dominate the mean savings.
This transparent sensitivity analysis does not propose a new domicile-selection model. It shows why research must account for the cost components that change with legal classification and institutional form. A lower capital carrying cost can represent more efficient financing of a constant risk, reduced available loss-absorbing resources, or a mixture of both. Its scientific interpretation depends on the real exposure and the structure of guarantees, reinsurance, and capital.
Integrated Scholarly Synthesis and Research Frontiers
The location of supervision and the location of loss
The comparative evidence supports an account of captive domicile choice as a choice over the institutional organization of exposure. Bermuda and Cayman demonstrate the significance of license classification. Vermont demonstrates the diversity of captive ownership and operating forms. EIOPA demonstrates the significance of related-party asset and liquidity arrangements. Historical reserve financing illustrates the consequences of transferring insurance obligations between different regulatory perimeters. Taken together, these cases identify the sources of both productivity and fragility in regulatory choice.
The decisive analytical unit is the chain from the insured loss to the ultimate party bearing its consequences. The chain includes the insured, primary carrier, captive, external reinsurer, guarantee or collateral provider, and parties whose claims compete with insurance liabilities. Territory matters because each link has enforceability conditions, insolvency rules, supervisory access, or capital constraints. Group organization matters because ownership creates correlated sources of financial capacity. The relation between territory and ownership determines whether the apparent diversification of entities represents an economic diversification of resources.
This interpretation places the captive's own capital and liquidity in their proper context. An entity with ample book equity may face a shortage of assets available to pay claims. Another may have access to reliable external risk transfer, yet suffer from an inadequate immediate cash buffer. A larger nominal capital requirement can coexist with poor exposure control, and a lower minimum can coexist with adequate additional capital and robust governance. The empirical object is the resulting distribution of claim recovery, not any single number in the license schedule.
Counterfactual judgment and public policy
Regulatory competition can improve the match between supervisory intensity and risk characteristics. The scope for efficiency includes elimination of duplicate reporting, development of specialized licensing pathways, reduced administrative uncertainty, and improved access to professional insurance infrastructure. These effects reduce real resource consumption or improve performance. Their measurement requires observation of actual expenses, administrative outcomes, risk transfer, and funding.
Regulatory arbitrage can also redistribute benefits and costs. If recognition of a transaction releases funds that would otherwise back obligations, the beneficiaries can include owners receiving additional distributions. Potentially affected parties can include policyholders, cedents, workers, external creditors, or a resolution system. The distributional consequences are particularly significant where such parties are not represented in the domicile decision. Research should measure which obligations remain enforceable and which assets are accessible in adverse states before assigning a welfare conclusion.
An integrated supervisory response involves identification of the risks created by contracts and assets and coordination among authorities where information or enforcement is divided. Jurisdictional diversity can coexist with proportionate protection when the competent authorities can observe the relevant dependencies and assess the real risk-bearing capacity. Uniform minimum capital amounts alone would neither resolve the diversity of exposures nor guarantee adequate claim recovery. Conversely, unrestricted competition based on the least demanding nominal condition would fail to account for the external costs of insufficiently funded liabilities.
These are analytical implications derived from the documented institutional mechanisms, the abstract-limited empirical literature, and the financial examples. They establish no novel named doctrine, empirical law, or claim of precedence. A complete welfare analysis of a specific regulatory reform would require estimated effects on coverage prices, underlying risk reduction, capital allocation, claim recovery, and costs imposed on noncontrolling parties.
Future empirical identification
The next research frontier is the construction of comparable enterprise-program histories. A credible dataset would identify the parent enterprise, domicile, license class, formation and withdrawal events, changes in business permissions, major risk exposures, external reinsurance, assets held within and outside the group, and economic outcomes. It should distinguish license issuance from actual underwriting, separate entity from cell and group, and track redomiciliation as a transition of the same economic arrangement wherever identifiable.
A reform study could compare changes in domicile flows or program structures around a documented legal change. A difference-in-differences design would require plausible parallel trends and consideration of anticipation, simultaneous tax or insurance-market shocks, and changes in which entities appear in the data. Domicile reforms are potentially endogenous to anticipated market growth. A design that ignored this could attribute to regulation growth that motivated the reform. Event-study plots would help diagnose preexisting dynamics but would not prove identification on their own.
A discrete-choice study could represent enterprises as choosing among feasible domiciles and the option of no captive. Relevant attributes would include permitted business, risk-adjusted capital cost, fixed expenses, supervisory expertise, and expected access to reinsurance and service networks. Feasible alternatives would differ across enterprises. Corporate size, industry, country of ownership, and loss experience would influence both choice and outcomes. Identification of preferences from observed choices alone would be vulnerable to omitted alternatives and unobserved company characteristics.
For prudential research, the important outcome is conditional recoverability under joint stress. Exposure scenarios should connect operational loss, parent distress, investment impairment, counterparty default, collateral restrictions, and payment delays. Standard errors around average loss are insufficient for strongly dependent catastrophic exposures. Information on actual asset location, contractual restrictions, and enforceable guarantees would support more realistic sensitivity tests. Results should identify which inputs are observed, assumed, modeled, or inferred.
The limitations of the present source base are substantive boundaries of the conclusions. The legal regimes are illustrated through selected official licensing and supervisory texts; no exhaustive law-by-law comparison or certification of current tax compliance is claimed. The Vermont capital comparison uses directly inspected state executive domicile guidance corroborated by an industry profile; it is not a consolidated statutory interpretation. The Vermont aggregate report's extracted numerical tables were accessible, while visual rendering of the table page was unavailable. The two academic articles support only their inspected abstracts. No firm-level confidential data, causal regression, full solvency model, or externally adjudicated welfare estimate is supplied. Further work would expand the sample and validate these sources at the operative legal versions required for any current legal application.
Conclusion
Captive insurance domicile choice is a decision about the organization of insurance obligations, risk-bearing capital, governance, and supervisory authority across legal jurisdictions. Its economic significance is determined by the substance of the program: the losses covered, the parties exposed to those losses, the resources available to pay claims, and the legal and financial relationships that permit those resources to be used. Regulatory variation influences these conditions through classification, capital, investment rules, governance, licensing, and the treatment of cross-border transactions.
The comparative cases demonstrate that significant differences exist in the reported license structures and capital schedules of captive regimes. The examined statistics document administrative growth within two selected populations at matched reporting dates, but their differing coverage prevents a unified measure of market share or regulatory quality. The numerical illustrations demonstrate how dependence, internal asset impairment, reinsurance performance, and capital treatment affect underlying enterprise risk. Their assumptions distinguish financial reasoning from empirical evidence.
The central substantive conclusion is that a reduction in the costs of financing a given risk must be assessed jointly with preservation or improvement of the effective ability to pay claims. Efficient regulatory specialization and cost-saving organizational innovation are compatible with strong supervision. Regulatory arbitrage introduces vulnerability when the measured burden declines while financially meaningful exposures remain, especially when the resulting costs fall on parties outside the controlling enterprise.
The evidence also establishes a methodological conclusion. Domicile choice cannot be adequately studied through nominal capital differences, tax rates, or license counts alone. It requires comparative legal analysis, accounting and risk-perimeter reconciliation, investigation of reinsurance and asset dependencies, and appropriate counterfactual evidence. The international science of insurance benefits from this combination because it explains the individual institution within the wider relationships that determine the performance of insurance obligations.
References
Vermont Department of Financial Regulation. (n.d.). Captive Formation & Licensing. Captive Insurance Division, State of Vermont. Full reference
State of Vermont, Agency of Commerce and Community Development. (n.d.). Captive basics. Vermont Captive Insurance. Retrieved November 19, 2026. Full reference
Captive Insurance Companies Association. (n.d.). Vermont. Domicile profiles. Retrieved November 19, 2026. Full reference
Vermont Department of Financial Regulation. (2026). 2025 Aggregate Report: All Captive Insurance Companies. Captive Insurance Division, pp. 2, 4. Full reference
Bermuda Monetary Authority. (n.d.). Insurance Licensing Requirements. The Multi-Licence System, general insurance classes. Full reference
Cayman Islands Monetary Authority. (n.d.). Insurance Licensing Requirements. Forming a Class B and Class C Insurance Company. Full reference
Cayman Islands Monetary Authority. (n.d.). Insurance FAQs. Related and Unrelated Business; Class B Insurers. Full reference
Cayman Islands Monetary Authority. (n.d.). Insurance Statistics. Insurance Licensees table, observation dates 31 December 2024 and 31 December 2025. Full reference
European Insurance and Occupational Pensions Authority. (2024, July 2). Opinion on the Supervision of Captives. EIOPA-BoS-24/176, Frankfurt, §§ 2–3. Full reference
OECD. (2020). Transfer Pricing Guidance on Financial Transactions: Inclusive Framework on BEPS: Actions 4, 8–10. OECD Publishing, Paris. doi:10.1787/794bcddd-en. Section E, pp. 35–41. Full reference
National Association of Insurance Commissioners. (n.d.). Captive Insurance Companies. Insurance Topics. United States; source includes August 2025 implementation snapshot. Full reference
U.S. Department of the Treasury & Internal Revenue Service. (2025). Micro-Captive Listed Transactions and Micro-Captive Transactions of Interest. T.D. 10029; Internal Revenue Bulletin 2025-09; final regulations published January 14, 2025, 90 FR 3534. Full reference
Koijen, R. S. J., & Yogo, M. (2016). Shadow Insurance. Econometrica, 84(3), 1265–1287. doi:10.3982/ECTA12401. Princeton University research publication record. Full reference
Russell, D., Chen, J.-L., & Chang, M.-S. (2026). Exploring the Connection Between Captive Insurance and Cash Flow: Evidence from S&P Global 100 Companies. Asia-Pacific Journal of Risk and Insurance, 20(1), 55–86. doi:10.1515/apjri-2024-0061. Full reference
Institute of Chartered Accountants in England and Wales. (n.d.). IFRS 17 Insurance Contracts for Non-insurers. FAQ 8: intragroup insurance contracts. Full reference
International Association of Insurance Supervisors. (n.d.). ICPs and ComFrame. Standard Setting, Basel. Full reference