Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Offshore Captive Insurance Market Structure

Abstract. Offshore captive insurance is an institutional market for organizing, financing, and selectively transferring the risks of enterprises and groups through licensed insurers located outside the principal territorial markets of their sponsors. This monograph investigates its market structure using comparative institutional analysis, regulatory statistics, insurance cash-flow analysis, and critically bounded empirical literature. Bermuda, the Cayman Islands, and Guernsey establish three distinct observational regimes; Vermont supplies an onshore benchmark. The comparison shows that captive market composition depends on ownership, related-business classification, cellular organization, direct insurance and reinsurance channels, and the distribution of service functions. The study reconstructs the captive-specific composition in Bermuda's 2024 annual report, distinguishes Cayman international insurer counts from a narrow captive population, and demonstrates the effect of cellular units on Guernsey market measurement. It explains how legal-entity counts can diverge from the distribution of underwriting volume, and why fronting and management dependencies require separate concentration measures. The principal analytical contribution is a comparative interpretation of observed market perimeters and the relationship between sponsored risk retention and external market access. The available evidence permits descriptive conclusions concerning market structure; causal effects of domicile, tax treatment or captive ownership, global market shares and provider-level competition remain unidentified. The findings establish a reproducible foundation for further study of international insurance markets without introducing a new named theory.

Keywords: Captive insurance; offshore insurance; insurance market structure; insurance domiciles; risk retention; fronting; reinsurance; cellular insurers; market concentration; insurance supervision; international insurance institutions.

Introduction

The international captive insurance market is a market in institutional capacity. Its participants purchase or supply risk-bearing balance sheets, corporate governance, insurance contracts, supervisory authorization, professional management, reinsurance, and access to the territorial markets where insured obligations arise. An enterprise establishing a captive becomes an owner of an insurer while remaining a buyer of various services from the wider insurance sector. This coexistence of internal production and external exchange is central to the structure of the market.

The conventional market for commercial insurance is usually described through buyers, sellers, premiums, product categories, and geographical territories. Those measures are relevant to captives, but their interpretation changes when the policyholder and insurer belong to a common group. A premium can be revenue to the insurer and a cost to its affiliated insured; it can represent an economically important budget allocation even when eliminated within consolidated accounts. A large premium volume inside a captive domicile may therefore coexist with relatively little contestable business sold to independent customers. Market structure must be understood at the level appropriate to the question.

Three questions organize the research. First, which institutional forms and reporting categories constitute the observable market? Second, how do the populations reported by domicile authorities differ in substantive composition, including legal carriers, licensed classes, business cells, and service suppliers? Third, which mechanisms explain the interaction of captive ownership with commercial insurance and reinsurance, and which conclusions can actually be established with the available evidence?

The principal observations use a 2024 common benchmark, with 2025 developments used where comparable reporting is available. The study selects centers for documentary contrast rather than statistical representativeness. Bermuda supplies a company-type and financial-size decomposition; Cayman supplies an international license census and related-business classifications; Guernsey supplies separate counts of cellular and non-cellular operations; Vermont establishes the relevance of captive infrastructure within a major domestic insurance jurisdiction. Each observation retains its source population and date. The study does not aggregate these non-equivalent categories into an estimate of world market share.

The research belongs to Global Insurance Markets and Institutions. Detailed corporate captive design, the optimization of domicile choice, captive capitalization rules, loss experience, insolvency and runoff belong to related research objects. They enter this study insofar as they shape the organization of market participants, access to independent capacity, and the meaning of published statistics. The research concerns financial and legal offshore arrangements rather than insurance of physical offshore energy installations.

Concepts and Historical Development

Captive ownership and the market boundary

A captive insurer is an insurance undertaking established or controlled by one or more sponsors principally to finance risks associated with their enterprise, membership, or business network. The identity of the qualifying insured population and the permissibility of unrelated business differ by jurisdiction. The definition is thus functional for comparative research, with local legal definitions recorded separately. An offshore captive is classified here by the relationship between its domicile and the principal operations or residence of its sponsors. The word offshore describes a cross-jurisdictional institutional arrangement; a geographic location alone does not establish offshore status for every carrier licensed there.

In its captive supervision paper, the IAIS (2015) considers specialized supervisory treatment and different ownership structures. Its discussion is an appropriate starting point for a cross-jurisdictional inquiry because the scope of insurance regulation depends in part on whose risks an undertaking accepts. The legal status of the insurer and the economic relationship between insurer and insured are complementary dimensions. A separately constituted insurer can participate in internal group risk financing while assuming real contractual liabilities.

Four arrangements illustrate this institutional range. A pure captive primarily serves the parent and affiliates. A group captive brings multiple sponsors into a common underwriting arrangement and introduces incentives concerning member selection and pooled losses. An association captive connects ownership or eligibility to a collective organization. A sponsored or cellular arrangement permits separate participants to use a common infrastructure while allocating assets, liabilities, and contractual obligations under the governing law. These categories may intersect. A cell can serve a single enterprise, and a multi-owner insurer can underwrite related risks across several participating enterprises.

The captive insurance overview of the Insurance Information Institute (n.d.) describes the evolution of captive forms, including rent-a-captive and protected cell arrangements. Its historical account locates the adoption of the protected cell concept in Guernsey in 1997. This innovation matters to market measurement: a platform potentially serves many insurance programs without requiring one new conventional company per program. Historical claims about priority should be understood as institutional chronology in the cited industry account rather than an exhaustive adjudication of all antecedent legal forms.

The emergence of specialized domiciles

Specialization develops when a jurisdiction makes the operating costs and supervisory treatment of captive insurance predictable. A captive requires an organizational framework capable of supporting underwriting, claims settlement, financial reporting, investment, and orderly termination. An established domicile can supply these activities through a professional network. Because the sponsor's insured exposures need not arise where the captive is licensed, a relatively small local economy can support a substantial international underwriting population.

The twentieth-century expansion of captives was tied to the development of corporate risk management and periodic changes in commercial insurance availability. The historical industry account above identifies the liability insurance difficulties of the 1980s as an important period of expansion. Those conditions illustrate a general mechanism: when the price or contractual availability of external capacity changes, the value of organized internal retention changes as well. The mechanism does not imply that every captive formation during a hard market was induced by premium increases. Tax rules, accumulated expertise, corporate restructuring, line-specific exposures and strategic preferences can act simultaneously.

A subsequent development was the emergence of domestic captive regimes. This changes the competitive geography. The decision to finance a risk through a captive and the decision to license that captive offshore are distinct decisions. An enterprise may enter the captive sector while choosing its home country or another domestic jurisdiction. A decline in one offshore domicile might reflect entry elsewhere, consolidation, migration, cancellation, or changes in category definitions rather than a contraction of total captive activity.

Captive reinsurance and reserve financing

The word captive also occurs in arrangements established by insurance groups to reinsure liabilities. Some of these transactions concern the financing of reserves associated with life insurance. The NAIC's captive insurance topic (n.d.) treats reserve financing and efforts to harmonize aspects of US captive reinsurance regulation, including Actuarial Guideline XLVIII. These transactions have a different insured population and different prudential transmission channels from an industrial sponsor's property or liability captive. The present comparative census keeps mixed classifications visible rather than automatically treating every limited-purpose insurance undertaking as an industrial captive.

Research Design and Statistical Perimeters

Units of observation

The unit of observation determines the meaning of market structure. At least six observable units need distinction: a licensed carrier, an active underwriting entity, an independently controlled parent group, a cellular portfolio, an insurance contract or exposure unit, and a professional service provider. The first is usually obtainable from supervisory records; the others require additional data. A license may remain in force while new underwriting has ceased. A parent may control several licensed carriers. A single legal carrier can support many programs through cells. An external manager may serve a portfolio of legally unrelated captives.

The study uses public regulatory categories as reported. For ownership analysis, the economic unit would ideally be the ultimate controlling group with a consistent control definition. For risk-bearing capacity, the relevant unit may instead be a separate pool of assets and liabilities, depending on enforceability and contract terms. For service competition, the unit is the economic provider group rather than every local subsidiary. Combining observations across these levels generates false market shares.

This distinction is especially important for jurisdictional comparison. A denominator containing pure captives, commercial reinsurers, and special-purpose structures may be meaningful for an international insurance sector but inappropriate for estimating the prevalence of enterprise-owned insurers. Therefore the analysis distinguishes narrow and broad census figures. A narrow census includes only the expressly designated captive categories available in the source. A broad census includes the regulator's wider international insurer population. Both can be useful if the label specifies their scope.

Flow measures and financial quantities

The study reports monetary amounts in the currency and units supplied by the authority. Premiums describe activity over a period; assets and capital represent balance-sheet stocks at a reporting date. Their relation can indicate differences in observed scale but cannot establish solvency adequacy without liability composition and applicable risk-based capital requirements.

Gross premiums generally include amounts written before outward reinsurance effects; net premiums reflect the reporting entity's retained premium after relevant cessions under the source definition. Gross premiums are not identical to recognized insurance service revenue, collected cash, aggregate insured value, or risk-adjusted exposure. A ratio between gross and net premiums is sensitive to commissions, contract terms, reporting convention and timing. The analysis uses arithmetic ratios only as descriptions of the reported measures.

The comparison also preserves reporting versions. When a later statistical table restates the opening year, comparison uses that stated version for both endpoints. The study does not force reconciliation of separately released counts obtained at different dates. Dynamic online statistics need dated preservation for long-term replication. The source ledger identifies the accessible table and the specific observations used rather than representing a live page as an immutable file.

Empirical strategy

The design is descriptive and comparative. It uses tabular arithmetic and a structured comparison of institutional mechanisms, supplemented by published empirical studies of captive ownership. Selection of jurisdictions follows reporting diversity and established relevance. The resulting observations are neither a random sample nor an exhaustive population of offshore centers.

A causal inquiry into the impact of a captive law would require exposure to an identifiable policy change, a suitable comparison population, control for concurrent changes and a defensible treatment definition. A causal inquiry into captive formation would require data describing enterprises before and after entry, including comparable firms that did not enter, and a strategy for endogenous selection. Neither exercise is claimed here. The descriptive analysis instead establishes which observations would be needed for such work and how misleading results arise when units are mixed.

The Organizational Architecture of the Captive Market

Sponsorship and ownership

In a single-parent arrangement the sponsor internalizes a defined part of its insurance financing. Underwriting decisions can reflect knowledge of the parent's own exposures, and the captive can make loss experience financially salient within the group. The owner also supplies capital that would otherwise be available for alternative uses. The organization's position as both insured and shareholder changes the relationship between insurance price and ultimate economic cost.

Group captives have a different incentive structure. Participants can obtain pooling economies and jointly maintain a specialized insurer, but the design of participation and loss-sharing becomes material to each member. A low-risk member may object to supporting a high-risk member if premiums and assessments do not reflect expected costs or the value of joint protection. A withdrawal decision can alter the risk composition of the remaining insurer. Association captives introduce similar questions of eligibility and membership.

The sponsored-platform model reduces the fixed setup costs of entering captive risk financing by sharing infrastructure. The value of that shared infrastructure depends upon the enforceable allocation of liabilities, the resources of the core, and contractual relationships with participants. A cell is neither automatically equivalent to an independently licensed corporation nor merely an accounting entry. The possibility of separate risk pools within shared legal infrastructure explains why the population of programs can exceed the number of carrier companies.

Related and unrelated business

The degree of related-party underwriting organizes the boundary between captive and commercial markets. When premiums originate only from the sponsoring enterprise, ordinary customer acquisition is constrained by ownership. The captive competes principally with other ways of financing the parent's risk. As independent underwriting expands, the insurer participates more directly in distribution and price competition for external business.

This continuum has supervisory significance. Independent policyholders may have weaker information about the sponsor and less control over the insurer than an affiliated policyholder. The regulator's distinction between classes addresses the business mix and the consequences of failure for policyholders. Exact classifications must remain jurisdiction-specific. An insurer described commercially as a captive may occupy a category that admits significant third-party business, while a narrow legal definition elsewhere could exclude that business.

Market organization beyond insurance licenses

The supply side consists of more than the risk-bearing carrier population. Managers, actuarial firms, accountants, legal advisers, brokers, fronting insurers, claims administrators, reinsurers and custodians supply specialized functions. Participants may purchase these services separately or as a bundled offering. A manager can be independent of the captive's sponsor, associated with a broker or part of an international provider group.

A domicile may consequently have two very different competitive structures: many sponsor-controlled carriers and a small set of service organizations. Independence of captive ownership does not imply dispersion of operational dependencies. An analysis of provider access and switching costs needs contracts, ultimate provider ownership, the number and type of managed clients, and the allocation of responsibility for critical activities. Counting the number of authorized service entities alone cannot answer whether any one firm has market power.

Comparative Domicile Evidence

Bermuda: composition and size of the captive population

Bermuda's insurance licensing requirements (BMA, n.d.) distinguish specialized and commercial categories in general insurance, long-term insurance and other business. For an empirical decomposition, the 2024 company-type statistics (BMA, 2025, p. 67) are preferable to a blanket sum of license classes. The table explicitly identifies three captive groups and separately reports special-purpose insurers. The classification is retained in the table below.

BMA company type, 2024 report Licenses Gross premiums, USD million Net premiums, USD million Assets, USD million Capital and surplus, USD million
Captives writing third-party business 186 17,589 12,437 66,549 31,102
Captives writing connected or related business 266 10,304 8,696 56,448 32,328
Single-parent captives 179 2,956 1,996 25,439 19,027
Captive rows summed by author 631 30,849 23,129 148,436 82,457

Source: BMA (2025), Annual Report 2024, p. 67; last row is calculated from the three source rows. This is the published report's company-type table, not a claim about the latest revised register. The report's limited-purpose subtotal also includes 209 special-purpose insurers and therefore is not the captive-specific denominator used here.

The distribution is materially different when measured by license count and reported premium. Captives reporting third-party business constitute 186/631, or 29.5% of the three-row license total, yet account for 17,589/30,849, or 57.0% of reported gross premiums. The connected/related category accounts for approximately 42.2% of licenses and 33.4% of gross premiums. Single-parent captives represent 28.4% of licenses and 9.6% of premiums. Rounded shares do not necessarily sum to exactly 100%.

These results establish a composition difference, not carrier concentration. The third-party category has greater reported mean gross premium per license: approximately USD 94.6 million against USD 38.7 million for connected/related captives and USD 16.5 million for the single-parent category. A few unusually large carriers could account for much of the difference; the table does not reveal the distribution within each category. Underwriting lines, turnover, retrocession, and differences in obligations can also explain why mean size varies.

The reported ratio of net to gross premium across the three rows is approximately 75.0%. It provides a second perspective on the aggregate reporting perimeter. A net premium share cannot be interpreted as a universal proportion of insured losses ultimately financed by Bermuda captive shareholders: premiums are prices associated with coverage and reinsurance arrangements, not states of the underlying loss distribution.

The adjacent beneficial-owner table in the annual report covers limited-purpose insurers including special-purpose insurers. Its geographical distribution is therefore evidence about the broader population rather than a captive-only ownership census. It demonstrates that owner residence can be recorded separately from insurer domicile. Inferring captive-specific regional ownership from that pooled table would require a category-by-region cross-tabulation.

Cayman Islands: the international insurer perimeter

The Cayman market illustrates the need to distinguish a regulatory class from ownership-based captive identity. CIMA's insurance licensing classifications (n.d.) describe the Class B international insurer categories with reference to the related-business share of net written premiums. B(i) covers the at-least-95% related-business category; B(ii) involves a majority of related business below the first category; B(iii) involves half or less. The economic character of a B(iii) undertaking may therefore differ materially from that of a pure own-risk captive.

The regulator's historical insurance statistics (CIMA, n.d.) report the following year-end observations. Class C and Class D are separate international insurance classifications. The table includes only entities in the international licensing perimeter; Class A domestic business is excluded.

International license category December 2024 December 2025
B(i) 463 471
B(ii) 13 14
B(iii) 194 208
All Class B 670 693
Class C 18 16
Class D 9 11
International total B + C + D 697 720

Source: CIMA, Insurance statistics, historical columns; international total calculated by adding Classes B–D and cross-checked against the licensing activity table. This is the accessible later snapshot rather than a contemporaneously archived Q4 2024 release.

In 2024, B(i) represented 69.1% of Class B and 66.4% of the full international license population. The difference arises solely from the chosen denominator. B(iii) represented 29.0% of Class B and 27.8% of the international population. Neither percentage is a precisely identified industrial captive market share because related premiums and ultimate ownership are distinct attributes.

Cellular structure creates another layer of measurement. In December 2024, the regulator identifies 139 Class B segregated portfolio companies within the 670 Class B total and separately lists 576 segregated portfolios and 61 portfolio insurance companies. These variables describe different organizational units. Adding them to the international license count would require a documented reconciliation of the statutory status and potential overlaps. The numbers establish that the carrier census alone does not reveal how many distinct insurance programs operate through the market.

Guernsey: cellular and non-cellular organization

The insurance market statistics of the GFSC (n.d.) distinguish non-cellular and cell insurer categories. In the 2025 movement table, which notes restatement of some 2024 figures, the non-cellular captive population moved from 197 to 196 while captive cells moved from 122 to 126.

Reported captive units December 2024 December 2025
Non-cellular captives 197 196
Captive cells 122 126
Arithmetic sum of these categories 319 322

The combined arithmetic sum is an indicator of reported captive-related operating units, not a count of independent legal insurers. The cell share of the combined observation rises from 38.2% to 39.1%. With a non-cellular census alone, the change is negative; with the two-category operating-unit indicator it is positive. This is a concrete reporting-perimeter effect, independent of any judgment concerning domicile attractiveness.

Guernsey's international insurance aggregates include business beyond captive insurance. A reported total premium figure for all international insurers cannot be allocated to the captive classes without an appropriate breakdown. This exclusion is consequential because a comparison using Bermuda's captive-specific premium figure and Guernsey's all-international figure would silently change the business perimeter. The comparison therefore uses Guernsey primarily to establish organizational composition.

Vermont: the onshore counterpoint

Vermont's 2024 formation announcement (Vermont Agency of Commerce and Community Development, 2025) reports 683 licensed captives at year end, including 654 active and 29 dormant. The reported 41 new licenses comprise 30 pure captives, nine sponsored captives, one special-purpose finance insurer and one risk retention group. These are institutional categories in a US state, not a replica of Bermuda's or Cayman's categories.

The active share is 654/683, approximately 95.8%. A comparison that substitutes 683 for active carriers overstates the active population by 29 under this reporting definition. In a jurisdiction that reports only actively writing insurers, a comparison by the term licensed alone would thus need adjustment. Vermont establishes that the supply of specialized captive infrastructure can attain substantial scale inside a major home-market legal system.

Summary of comparative observation

Domicile 2024 observation selected for this study Essential scope qualification
Bermuda 631 captive-category licenses Three company-type rows in the 2024 report
Cayman Islands 697 international licenses B, C and D; not a narrow captive-only total
Guernsey 197 non-cellular captives and 122 captive cells Cells and carriers distinguished; restated base
Vermont 683 licensed captives, 654 active Includes dormant and specialized forms; onshore benchmark

The table is neither a ranking of global captive domiciles nor an additive census. It exposes the classification problem that precedes a defensible ranking. In the absence of common firm identifiers, ownership consolidation, active-status definitions and standardized cell treatment, a published total should retain its native observation label.

Licensing Flows and the Dynamics of Market Structure

A market population is a stock of surviving legal entities. Its annual movement can be represented by opening population plus additions minus cancellations and other exits, adjusted for reclassifications and corrections. Formation statistics alone measure one side of this process. A jurisdiction could experience substantial entry at the same time as a stable or falling year-end population.

CIMA's licensing activity series provides a directly reconcilable example. The 2024 opening international population was 683. There were 42 licenses issued and 28 cancelled; the resulting closing population was 697. Net increase was 14, approximately 2.0% of the opening population. In 2025 the same series gives 697 + 42 − 19 = 720. Net increase was 23, or approximately 3.3%. Gross additions were unchanged in these two periods, while fewer recorded cancellations explain the difference in net change. This is a descriptive identity, not evidence of the motives behind either formation or cancellation.

Technical reclassification can also alter apparent composition. A change in the mix of related and unrelated business can modify the applicable class without establishing a new captive. A move from a conventional captive into a sponsored platform may reduce the number of legal carriers while preserving the underlying coverage activity. A group merger can eliminate duplicate entities while retaining almost the same insured exposures. An insurer entering runoff may remain licensed with a declining premium flow. Researchers should distinguish these events before attributing population changes to supply or demand.

Licensing dynamics have economic consequences even without a large net change. New entrants generate demand for legal and actuarial design, capital planning and implementation; exits generate runoff and transfer services. A market with considerable gross turnover and stable stock may support a large professional-services sector. Conversely, a population dominated by old, self-managed insurers may generate lower external management demand per license. Neither relationship is quantified here because consistent fee and activity data are absent.

Intermediation, Fronting, and Reinsurance Networks

The contractual chain

An offshore captive often participates in a multinational insurance program through an intermediary insurer licensed to issue the required local policy. The fronting insurer takes contractual responsibility toward the insured according to the primary policy and cedes some or all defined exposure to the captive. The captive may purchase further reinsurance. Contractual risk allocation is determined by the complete chain, including retained layers, limits, exclusions, deductibles, collateral and recovery conditions.

The OECD's financial transactions guidance (2020, section E.2.4) describes fronting as a form of captive reinsurance and examines the distinct functions in the chain. The arrangement is economically useful because access to a local policy, management of claims and acceptance of risk can be supplied by different institutions. The fact that risk is largely ceded does not necessarily extinguish the direct insurer's policy obligations. Conversely, the captive's assumption of insurance risk depends on the applicable reinsurance contract and its capacity to perform.

Consider a hypothetical one-period chain. The insured pays a gross premium of USD 12 million to an external fronting insurer; the fronting insurer cedes a USD 10 million premium to the captive. The captive in turn pays USD 3 million to an independent reinsurer. Ignore taxes, commissions, brokerage, interest and other adjustments for this arithmetic example. The premiums reported gross across the three legal insurers would be USD 12 + 10 + 3 = 25 million if each reports the specified incoming premium. The sponsor's consolidated economic expenditure on the chain, excluding its own captive's costs and losses, is USD 12 − 10 + 3 = 5 million. The same coverage chain produces very different totals depending on whether the observer adds insurer-reported premiums or measures cash leaving the sponsor group.

Suppose further that a covered claim of USD 9 million is paid by the fronter; the captive is obliged to reimburse USD 8 million and collects USD 3 million from the independent reinsurer. Assuming complete, timely performance, the ultimate net claim allocation is USD 1 million to the fronting insurer, USD 5 million to the captive and USD 3 million to the independent reinsurer. The obligations sum to the original loss; the chain has not generated three independent losses. If the captive defaults on its reimbursement obligation, the claim allocation changes according to contracts, collateral recovery, and applicable law. The direct insurer may still owe the insured the original policy payment.

This illustration establishes why an analysis of market structure must follow both premium and loss flows. The nominal domicile of a carrier identifies one node in the chain. It does not establish the final geographical allocation of losses or capital.

Managers, intermediaries, and operational concentration

The IAIS (2015) discusses supervision of managers and the possibility that the failure of a fronting insurer harms a captive. This is important for market structure because the network of operational and financial dependencies can be concentrated even when the direct ownership network is dispersed. A common fronting insurer may provide local regulatory access for many captives, while a single management group may provide accounting, compliance and underwriting administration across several domiciles.

Public regulator census data illustrate the distinction. The Cayman December 2024 count shows 22 licensed insurance managers, while the GFSC 2024 base shows 24 authorized managers. These figures provide information about the number of institutions within the respective licensing populations, but neither implies a distribution of managed captives by provider. Some licensees may have a modest or specialized portfolio; others may belong to the same global group. A provider-market study would collect ultimate manager ownership, the number of client programs, premium or fee weights where appropriate, and substitutability of critical functions.

Dependence arises across geographical boundaries. A group manager may use unified technology, a common claims platform or shared compliance processes. A disruption can affect clients in several jurisdictions at once. The same interdependence can be beneficial through consistent reporting and economies of scale. The empirical question is whether the network has adequate alternatives and resilience, which cannot be settled by counting insurers or managers.

Concentration, Competition, and Dependence

Distinct objects of competition

The captive market operates through several connected but separately definable competitive domains. Domiciles offer institutional environments; managers offer operating services; fronting insurers offer policy access and contractual capacity; reinsurers offer contingent loss protection; and captive owners allocate underwriting to affiliated balance sheets. Competition in one domain need not establish competition in another.

A concentrated domicile distribution might be driven by specialized expertise, historical entry patterns, recognition by other authorities or administrative efficiency. A concentrated management sector might arise from scale economies, acquisitions, distribution relationships or scarce expertise. A concentrated reinsurance network might arise from risk appetite and contractual capital requirements. These mechanisms have different implications. The use of one aggregate market share obscures which institution possesses the capacity to change prices or restrict access.

Herfindahl measures and aggregation bias

Let s_i be a participant's proportion of an explicitly specified market variable, with shares summing to one. The Herfindahl-Hirschman index on the unit scale is H = sum of squared shares. The customary 0–10,000 representation multiplies H by 10,000. It is essential to state whether i denotes a legal carrier, ultimate corporate group, domicile, or service provider, and whether the weight is gross premium, retained premium, fee revenue, or another variable.

For the Bermuda table, grouping the three captive categories as if they were observations produces a category concentration statistic of approximately 0.3451 using counts and 0.4458 using gross premiums. These are calculated from (186, 266, 179)/631 and (17,589, 10,304, 2,956)/30,849, respectively. They are measures of category composition, not regulatory competition HHIs. The larger premium statistic reflects the prominence of one underwriting category in the reported premium total.

The difference between group and firm aggregation can be demonstrated algebraically. If a group contains two participants with shares a and b, replacing their separate squares by (a + b)^2 raises the index by 2ab. Thus the result based on consolidated ownership may exceed the legal-entity measure even when every underlying premium is unchanged. Conversely, splitting one platform into several independently counted cells can depress a unit-weighted concentration statistic without showing greater independent control.

Suppose, hypothetically, that four manager groups supply services for 40%, 30%, 20% and 10% of a defined population. Their H is 0.30, or 3,000 on the conventional scale. If the largest is recorded as two legal subsidiaries with shares 20% and 20%, the legal-entity statistic becomes 0.22, or 2,200. The underlying controlling interests are unchanged. This example does not assert that any of the studied domiciles has this distribution; it isolates the effect of the relevant observation unit.

Firm-level premium distributions within the Bermuda categories are not available in the cited table. Consequently, neither an observed firm-level HHI nor a cross-domicile HHI is estimated. The category totals could support a conditional bound under additional assumptions about the distribution of premiums within their groups, but such a bound would be of limited practical use without active issuer counts and ultimate group identifiers. Avoiding false precision is part of the scientific result.

Switching costs and contestability

The number of licensed providers does not directly identify the conditions faced by a captive attempting to replace one. A sponsor may need to transfer accumulated underwriting data, collateral agreements, claims systems, contracts and governance arrangements. The loss of continuity or the time needed for regulatory approval can impose costs even where alternatives formally exist.

Market contestability is therefore an empirical issue involving the quality of alternatives, contractual lock-in and the costs of migration or substitution. Domicile changes may require a new regulatory authorization or restructuring under relevant laws; they are not a simple renewal of a standard retail policy. Some services may be more readily substituted than the carrier location itself. A useful competition inquiry should examine which elements of the service bundle are separable and whether the client can change suppliers while preserving continuity of insurance obligations.

The Economics of Market Boundaries

Internal risk financing and external resource costs

For an enterprise, captive ownership changes its allocation of insurance functions. Expected loss does not disappear when a premium moves between affiliates. External premiums, investment income, loss prevention, administrative expenditures, capital opportunity costs, taxes and default effects determine the economic consequences of the arrangement. The relevant comparison holds coverage and the distribution of losses as constant as possible while analyzing the change in financing and service organization.

In an illustrative calculation, assume a sponsor can purchase an outside insurance program for USD 11 million per year or retain the relevant loss layer through a captive. Suppose expected retained claims are USD 6 million, outside reinsurance costs USD 2 million, outside administration costs USD 0.8 million and incremental annual opportunity cost of captive capital is USD 1.2 million. The simple expected annual captive economic cost is USD 10 million. Under these assumptions it is USD 1 million below the outside premium. If expected retained claims rise by USD 1.5 million, captive cost rises to USD 11.5 million. This arithmetic shows sensitivity to loss assumptions. It omits taxes, investment returns, transition costs, differences in coverage, collateral, loss timing and the distribution of unexpected losses; it is not an investment recommendation or observed profitability estimate.

The captive demand decision is endogenous. Larger enterprises may have sufficient exposure volume, reliable loss data and specialized staff to make internal financing attractive. In the S&P 500 captive study, Chang and Chen (2018) report associations between captive ownership and firm characteristics in panel data covering 2000–2016, including company size. Their results support treating sponsor characteristics as potential selection variables. They do not establish that establishing a captive itself changes those characteristics.

Diversification and capital demand

If a captive aggregates two identically distributed loss portfolios with standard deviation 10 monetary units each, their combined variance is 100 + 100 + 2 × 100ρ, where ρ is correlation. At correlation zero the standard deviation of aggregate losses is about 14.14; at correlation one it is 20. Relative to the sum of standalone standard deviations, the numerical diversification effect disappears as correlation approaches one. This hypothetical calculation demonstrates that arranging separate legal policies does not guarantee meaningful risk diversification.

Deriving required regulatory capital would require the applicable formula, the full distribution or stress assumptions, assets, reserves, reinsurance credit and other risks. Moreover, the parent may suffer the insured operating losses at the same time as its ability to support the captive deteriorates. The two-unit calculation deliberately excludes these interactions. Its purpose is to establish why insured risk correlation and actual economic capacity, rather than the number of related entities, matter to risk-financing structure.

Evidence on shareholder returns and cash flow

The benefits of captive ownership should be tested against evidence rather than presumed from retained premiums. Adams and Hillier's (2000) UK captive formation study uses firms with Guernsey captives and examines market returns around formation. The available publisher abstract and conclusion excerpts do not substantiate positive abnormal returns or changes in the parent companies' risk measures. The study offers a historical example of how the financial consequences of captive formation can be examined empirically.

Russell, Chen, and Chang (2026), in their international cash-flow study, report that evidence from S&P Global 100 constituents does not establish an annual cash-flow improvement from captive ownership. The abstract distinguishes retention of funds inside the group at a particular time from improvement in annual corporate cash generation. That distinction is consistent with the contractual-chain analysis of this monograph.

The studies use different populations, periods and outcome variables. Their results do not establish that all captives fail to produce economic value. A captive may supply coverage continuity, governance, or loss information that is difficult to measure in a short market-return window or in annual cash flows. Equally, such possibilities do not constitute measured benefits in these studies. Market growth is compatible with a heterogeneous distribution of sponsor outcomes, including unsuccessful programs and those retained for reasons other than immediate financial return.

Domicile competition and the interpretation of tax

Domicile choice affects the cost and availability of institutional services, and tax treatment can be among the relevant considerations. The presence of a captive in an offshore center does not itself establish the sponsor's motive, the amount of tax saved, or the lawfulness of its arrangements. Identifying any of those outcomes requires transaction-level information and applicable law.

The OECD (2020) distinguishes the commercial rationale for captive insurance from the accurate delineation of actual transactions. It relates the latter to financial capacity, risk assumption and control of economically significant functions. For the present market-structure inquiry, the implication is that a jurisdiction's licensed population cannot be converted into a measure of tax avoidance or tax efficiency. A study of tax incidence would need taxable income, effective rules, related transactions and a credible counterfactual. These variables are not contained in the supervisory census.

Competition among domiciles can encourage improvement of specialized institutional services. It may also create incentives to relax oversight or exploit discontinuities between legal systems. The relative importance of these responses is an empirical matter, conditioned on the quality of licensing and supervision, the types of admitted insurance, the enforceability of claims and the interaction with other regulators. This monograph identifies these channels without making a causal claim concerning any jurisdiction's policy.

Supervisory and Institutional Implications

Supervision and the perimeter of external exposure

The organization of a captive generates different supervisory interests according to the insured population and counterparties. An affiliated policyholder may be able to influence the captive through ownership and governance. An independent policyholder may lack comparable ability to observe or influence underwriting and capital decisions. A fronting insurer can become exposed to the captive's failure through reinsurance recoverables. A group captive exposes participating members to the operation of a collective risk-bearing organization. These distinctions give institutional content to proportional supervision.

The international supervisory guidance of the IAIS (2015) provides a basis for considering specialized captive arrangements, including managerial and cellular organization. Its application paper is guidance for supervisors; operative licensing and insolvency requirements remain jurisdiction-specific. Principles of proportionality should be applied to the actual nature and scale of exposures and the obligations of the insurer. A legal label is insufficient evidence that a business carries little risk.

Interdependence and reporting architecture

A useful supervisory dataset would associate each licensed insurer with the ultimate controlling group, license status, principal classes of business, related-party share, cell structure and critical service providers. It would also distinguish premiums by direct and assumed business, outward risk transfer, reserves and capital. In cross-border arrangements the locations of insured exposures and external counterparties should be recorded separately from license domicile. A harmonized public aggregate could preserve confidentiality while providing more informative denominators.

The benefits of such a design are specific. Ownership-based identifiers would permit researchers to compare corporate-group concentration with legal-entity concentration. Activity status would clarify the distinction between license stock and new underwriting. A stable cell identifier would permit analysis of program counts without counting a platform twice. A dated classification history would distinguish formations from license migrations. Comparable reporting of cessions would reduce premium chain double counting.

None of these improvements requires treating unlike national forms as legally identical. Harmonization can retain a translation table between each local category and a clearly delimited research category. A translation with uncertain cases should flag them, rather than allocate them mechanically to a purported common population. This is particularly important for insurer groups employing specialized reinsurance structures.

Resilience and the distribution of risk

A mature domicile's institutional resilience depends not merely on retaining many licenses but on the continuity of underwriting, management, financing, and claims payment. Common dependencies can be identified at the service, fronting, reinsurance and corporate owner levels. In a severe event, a captive may require additional resources precisely when its sponsor faces operating losses. Across different sponsors, a common reinsurer or fronting carrier may be subject to correlated claims. Cross-border supervision must take account of those relationships where they are material.

The relevant indicator is ultimately a distribution of contractual obligations and available resources under stress. Premium volume alone supplies limited information. Captives writing predictable first layers may depend differently on market capacity from those writing low-frequency severe losses. A jurisdiction with many small carriers can be exposed through a concentrated service supplier. These observations motivate collection of additional evidence; they do not establish a measured incidence of systemic instability.

Research Limitations and Reproducibility

The research reconstructs specific source populations rather than claiming a full worldwide census. Bermuda's observations are transcribed from the published 2024 annual report, including the original company-type table and its exact reported monetary units. Cayman and Guernsey statistics are dynamic pages with explicit historical columns; both may be revised. Their dated retrieval references and individual data claims are preserved in the accompanying evidence ledger. Vermont's official announcement gives a single benchmark date and reported active status.

Public financial aggregates cannot reveal the entire distribution of company sizes, correlations, ownership, program-level risk and service provider shares. A worldwide study would need a harmonized list of domiciles and rules for counting active captives, cells, group control and mixed-business insurers. Without these, the selection of four reporting jurisdictions is informative for structural comparison but not sufficient for global market estimation.

The published scholarly papers were used only to the extent of accessible original publisher or institutional abstracts and excerpts. The present research does not claim an independent re-estimation of their regressions or event studies. Its original quantitative work consists of transparent arithmetic on public observations and explicitly hypothetical examples. The institutional analysis integrates separate levels of comparison; it does not claim priority for a new theoretical system.

For replication, the three Bermuda category rows may be copied into a table with count, gross and net premiums, assets, and capital; then summed independently and divided by the relevant total. The Cayman licensing balance can be recomputed directly from its opening, issuance and cancellation series. The Guernsey cell share can be reproduced by dividing the expressly identified captive cell count by the sum of that count and non-cellular captive count. The hypothetical concentration and cash-flow illustrations can be derived entirely from their stated assumptions. Each operation has an observable input and a defined output. Reviewers should preserve the table version or capture an immutable dated source copy before comparing later revisions.

The substantive boundary of the study is therefore explicit. There is evidence of different organizational and reporting structures, heterogeneous scale among captive categories, and coexistence of internal risk financing with externally supplied capacity. There is insufficient evidence here for global captive market shares, a ranking of competition, firm-specific solvency conclusions or causal effects of domiciliation. Further research must obtain the data appropriate to those questions.

Conclusion

Offshore captive insurance constitutes a specialized institutional segment of the international insurance system. The market combines enterprise-controlled insurance capacity with externally supplied professional services, licensed policy access, and independent risk transfer. It consequently contains several connected competitive domains, each requiring its own unit of observation and denominator.

Bermuda's published 2024 statistics establish that the distribution of underwriting volume differs from the distribution of captive-category licenses. Cayman demonstrates that an international insurance population includes distinct related-business classes and other insurance categories. Guernsey establishes that cellular arrangements materially affect the measure of captive activity, while Vermont demonstrates the significance of onshore alternatives and the distinction between active and dormant licenses. These findings derive from separate, identified supervisory populations; they cannot be converted into a worldwide league table.

The monograph's central analytical conclusion is that captive market structure should be described simultaneously through ownership, legal organization, business mix, underwriting flows and external institutional connections. These are dimensions of the observed market, not the elements of a new named theory. The decomposition explains why published license counts, gross premiums and provider populations can yield different pictures of the same institutional setting.

A captive provides an organizational means of allocating risk-financing functions. Any resulting economic benefit depends on realized losses, the terms of outside insurance and reinsurance, capital costs, risk management and the capacity to honor claims. The evidence concerning shareholder returns and annual cash flow supports caution in attributing uniform financial benefits to captive ownership. Tax and regulatory motives require transaction-specific evidence.

The scientific basis for comparative captive market research is disciplined measurement of the relevant contractual and institutional perimeter. A robust international dataset must preserve the distinction between carrier and cell, between domicile and ownership, between premium and ultimate loss, and between licensed provider and independent provider group. The findings of this study establish the value of those distinctions and the concrete data required for the next stage of empirical research.

References