Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Economies of Scale and Scope in Insurance

Abstract. Economies of scale and economies of scope are central explanations for the organization of insurance production, the persistence of specialist and diversified carriers, and the changing boundaries of insurance groups. Their evaluation requires a more demanding account than the proposition that larger insurers have lower expense ratios. An insurer jointly produces underwriting, claims settlement, risk information, distribution services, investment intermediation, and financial promises contingent on uncertain events. Increasing output can spread fixed administrative and technological expenditures, improve access to risk and capital markets, and permit specialized labor. It can also increase coordination requirements, distribution conflicts, model concentration, governance burdens, and correlated exposure. Combining lines of insurance can share inputs and diversify losses while impairing revenue performance or raising capital and complexity costs. This monograph develops a source-critical synthesis of multiproduct cost theory, insurance risk-bearing economics, efficiency-frontier methods, and comparative empirical research. It defines ray scale economies, cost and profit scope, subadditivity, capital diversification, and measured technical efficiency as distinct objects; examines what empirical studies can identify; and assesses organizational, digital, supervisory, and policy consequences. Evidence from historical U.S. insurer samples, Spanish consolidation studies, and multicountry efficiency comparisons supports conditional rather than universal economies. The principal conclusion is that the economic frontier of an insurance organization is determined jointly by product composition, scale, risk covariance, contract quality, governance, technology, capital access, and regulatory setting. Corporate size, concentration, and diversification are observable characteristics; sustainable economic efficiency remains a separately demonstrated property.

Keywords: Economies of scale; economies of scope; insurance economics; multiproduct cost function; ray scale economies; cost subadditivity; insurer efficiency; stochastic frontier analysis; data envelopment analysis; risk pooling; underwriting diversification; reinsurance; capital allocation; distribution economics; conglomeration; strategic focus; stock and mutual insurers; insurer mergers; regulatory capital; insurance technology; expense ratios; organizational complexity; international insurance markets.

Introduction

Insurance is a multiproduct financial and contractual activity. The insurer commits resources in the present to assess and accept risks, administer contractual obligations, invest premiums, and deliver claims payments or services in uncertain future states. Production therefore includes both operating services and the capacity to honor contingent liabilities. A company with twice as many policies has not necessarily doubled economically comparable output. The policies may involve different sums insured, durations, risks, deductibles, expenses, claim frequencies, and service requirements. The cost of financial promises may differ even when reported administrative expense appears similar.

An integrated assessment of scale and scope must therefore ask four questions. What outputs are produced? Which inputs and fixed capabilities can be shared? Which risks and capital obligations arise from expanding or combining those outputs? How is performance evaluated relative to credible alternatives, holding output quality and operating environment sufficiently comparable? These questions connect industrial organization, actuarial modeling, financial intermediation, corporate finance, and insurance supervision.

The distinction between scale and scope entered modern multiproduct cost analysis through the work of Panzar and Willig (1981), whose economies-of-scope analysis concerns the cost consequences of producing different outputs together. Scale instead concerns proportional expansion of a specified output bundle. These ideas become operationally complicated in insurance because an underwriting line is both a service product and a contingent portfolio of losses. The economic cost of enlarging it reflects administration, acquisition and claims handling together with the risk-bearing and funding services that sustain the contract.

The insurer-efficiency literature provides instructive counterweights to easy generalization. Cummins and Weiss (1993), in their stochastic frontier study, found scale-related possibilities for cost improvement in small and intermediate U.S. property-liability insurers over the 1980–1988 sample. Berger, Cummins, Weiss, and Zi (2000) examined conglomeration versus strategic focus through profit scope rather than administration alone. Cummins, Weiss, Xie, and Zi (2010) later found in their U.S. scope-efficiency comparison that cost economies could be offset by revenue diseconomies. These findings describe particular populations, periods, output definitions, and methods. Their joint significance is the requirement to distinguish gross cost savings, net value creation, and the appropriate organizational boundary.

International developments add another layer. The Organisation for Economic Co-operation and Development's Global Insurance Market Trends 2026 reports on insurance activity across 67 jurisdictions using 2025 data. The International Association of Insurance Supervisors (IAIS), in its 2025 Global Insurance Market Report, monitors systemic vulnerabilities and large insurance groups rather than estimating a universal cost curve. The two institutional sources illuminate market context and group risk; they do not identify whether a merger or a larger carrier caused productive efficiency. This monograph keeps those evidentiary roles separate.

Research object, design, and source controls

The research object is the relationship between insurer organizational size, product combination, shared production inputs, risk and capital interactions, and the resulting cost, revenue, profit, and solvency outcomes. The primary disciplinary setting is Insurance Economics. Relations to Insurance Finance and Asset-Liability Management, Enterprise Risk Management in Insurance, Global Insurance Markets and Institutions, and Insurance Technology are analytical cross-memberships rather than additional canonical identities.

This study uses critical reconstruction of established multiproduct theory, close interpretation of published empirical results, comparison of efficiency methods, and source-based examination of supervisory standards. The research cutoff is November 14, 2026. Bibliographic identity was checked against journal and institutional records; evidence entries distinguish accessible abstracts, publisher records, and institutional full texts. No raw insurer microdata have been obtained or independently reestimated. Numerical examples are explicitly illustrative and do not purport to reproduce any cited data. The manuscript establishes analytical deductions from specified assumptions and does not announce a new named theory, principle, or proprietary empirical finding.

Scale and Scope Definitions

Economic output in insurance production

Output measurement determines the meaning of every scale result. Written premium measures a financial flow affected by insurance prices, insured risks, inflation, regulatory rules, business mix, and commercial strategy. Policy count measures contractual units without controlling for the services and contingent protection delivered. Claims paid may rise because of poorer risk experience rather than improved output. Reserves describe obligations and accounting estimates; assets capture investments and accumulated financial resources. Each is useful for some purposes and inadequate as a universal quantity of insurer production.

A theoretically appropriate output vector can distinguish underwriting or risk-bearing services, loss settlement, intermediation or financial services, and separate lines or policy groups. Let y=(y1,…,ym)y=(y_1,\ldots,y_m) be a vector of positively defined, quality-adjusted insurance outputs, ww an input-price vector, zz external conditions, and qq relevant quality or solvency constraints. A conditional cost frontier may be written as

C∗(y,w,z,q)=min⁡x∈F(y,z,q)w⊤x.C^*(y,w,z,q)=\min_{x \in \mathcal F(y,z,q)}w^\top x.

The feasible set F\mathcal F includes the labor, systems, claims-administration, governance, and financial-resource requirements imposed by the output vector and its contractual obligations. This is a standard cost-minimization representation. It becomes an insurance model only after the economic definitions and cost treatment of contingent liabilities are specified.

One often-observed accounting outcome, the expense ratio, divides certain acquisition and operating expenses by earned or written premiums, according to the applicable definition. It can decrease merely because premium rates rise while the number and complexity of contracts remain fixed. Conversely, an insurer writing low-premium policies can have a high expense ratio even with efficient administration per policy. The loss ratio and combined ratio add underwriting information but likewise do not identify full economic cost efficiency, especially for life products and long-duration liabilities.

Ray economies of scale and local versus global inference

For a fixed composition of outputs, proportional expansion is represented by tyt y, where t>0t>0. Holding input prices, institutional conditions, and output standards fixed, define the local ray cost elasticity as

εC(y)=dln⁡C∗(ty,w,z,q)dln⁡t∣t=1.\varepsilon_C(y)=\left.\frac{d\ln C^*(ty,w,z,q)}{d\ln t}\right|_{t=1}.

If εC<1\varepsilon_C<1, cost rises less than proportionately with the chosen output bundle: there are local ray economies of scale. Equality indicates constant ray returns for the comparison, while a value greater than one indicates ray diseconomies. This derivative concerns a neighborhood of the observed bundle. It cannot establish that indefinite growth remains economical.

In a differentiable multiproduct representation, the ray elasticity equals the sum of the output-specific cost elasticities, εC=∑j∂ln⁡C/∂ln⁡yj\varepsilon_C=\sum_j \partial\ln C/\partial\ln y_j, with other arguments fixed. A useful ray scale index is S=1/εCS=1/\varepsilon_C, provided elasticity is positive. The value of SS depends on the output mix and the point of evaluation. Accordingly, a finding of economies in small carriers does not prove the existence of economies at the scale of a large diversified group, and a cross-sectional average cannot locate the minimum-efficient scale of every product configuration.

A simple illustration clarifies the fixed-cost mechanism. Assume identical policies, a constant variable cost vv, and a fixed cost F>0F>0. Then C(n)=F+vnC(n)=F+vn and C(n)/n=F/n+vC(n)/n=F/n+v. Average cost falls as nn rises, because a larger policy base spreads the fixed capability. Introduce administrative congestion an2a n^2, with a>0a>0, and C(n)/n=F/n+v+anC(n)/n=F/n+v+an. The minimum occurs at n=F/an=\sqrt{F/a}. This is a mathematical illustration of a potentially U-shaped average cost curve; it is neither an estimated insurance production function nor an assertion that the same convex term describes actual carriers.

Cost economies of scope

For two outputs at specified quality and input prices, the conventional cost-scope comparison is

SC(y1,y2)=C∗(y1,0)+C∗(0,y2)−C∗(y1,y2)C∗(y1,y2).SC(y_1,y_2)= \frac{C^*(y_1,0)+C^*(0,y_2)-C^*(y_1,y_2)} {C^*(y_1,y_2)}.

Positive values indicate a lower joint cost than the sum of separate specialist production costs for the two output bundles, under the maintained assumptions and common measurement convention. This is the basic logic of the Panzar–Willig treatment. Shared underwriting analytics, customer records, claims systems, treasury, governance, and distribution may produce such economies.

The formula has a demanding informational requirement. Costs at (y1,0)(y_1,0), (0,y2)(0,y_2), and (y1,y2)(y_1,y_2) must be defined for feasible comparable organizations. If a dataset contains no credible specialist firms at the necessary size, the endpoints can require extrapolation beyond observed technology. A fitted translog function with logarithms of outputs cannot simply insert a zero output without suitable treatment. Results can become artifacts of functional form, sample support, or imposed adjustment rather than identified economies.

Cost scope also differs from cost complementarity. In a twice differentiable cost function, a negative cross-partial derivative ∂2C/∂y1∂y2<0\partial^2 C/\partial y_1\partial y_2<0 means increasing one output reduces the marginal cost of the other locally. It is suggestive of shared production but neither alone necessary nor generally sufficient for positive aggregate cost scope over the entire relevant output range. The derivative is local; the scope statistic compares distinct bundles.

Subadditivity and organizational boundaries

A cost function is subadditive over a relevant output region when producing the total output in one organization costs less than allocating the same total among feasible separate producers. For any admissible partition y=∑k=1Kyky=\sum_{k=1}^K y^k, the condition is C∗(y)≤∑kC∗(yk)C^*(y)\leq\sum_k C^*(y^k). This is a stronger organizational comparison than observing lower average cost along one ray or positive scope for one two-product split. It speaks to the relative cost of integrated production versus an admissible set of alternative producer configurations.

Insurance firms need not integrate every shared activity to capture its benefit. A specialist can obtain scale in catastrophe modeling, claims administration, cloud infrastructure, or reinsurance through external purchasing or consortium arrangements. Therefore an observed technological economy is not automatically an argument for common ownership. Transaction costs, contractual hazards, proprietary data, counterparty concentration, and operational resilience determine whether purchasing, cooperative production, outsourcing, or corporate integration best captures the economy.

Revenue and profit scope

Cost scope evaluates inputs at specified output quantities. A diversified insurer might share platforms efficiently yet lose specialist distribution relationships, brand clarity, underwriting discipline, or the ability to price products appropriately. Revenue scope compares achievable revenues under joint and separate production, subject to pricing and market conditions. Profit scope compares the attainable profits of the integrated organization with the profit opportunities of specialists. These comparisons also depend on market power, demand interdependence, risk acceptance, and the relevant counterfactual.

Berger et al. (2000) explicitly used profit scope economies to test whether conglomeration or strategic focus provided the stronger explanation across insurer types. A firm may display lower costs from sharing systems and still have lower profit because of impaired revenues or higher economic capital charges. Positive cost scope and negative profit scope are logically consistent observations, and careful empirical inquiry must allow both.

Shared Risk and Administrative Inputs

Production indivisibilities and administrative platforms

The first identifiable source of scale is the indivisibility of certain necessary capabilities. A regulated insurer must maintain governance, compliance, accounting, claims administration, data security, financial reporting, and methods of assessing contractual liabilities. Even a small book of business requires minimum information systems, trained staff, documentation, and supervisory interfaces. Larger books can spread these expenditures over more transactions, provided the extra business is sufficiently homogeneous.

The fixed-cost argument is strongest when a platform can process a large increase in comparable work with modest incremental expenditure. It weakens as products multiply, regulation diverges, legacy systems accumulate, and personalized risk assessment becomes necessary. Cost accounting often assigns shared overhead to product lines by premium, headcount, or policy count. These allocations can be useful internally but do not measure the avoidable incremental cost of a line. To decide whether a line benefits from scope, one needs the cost that would actually disappear if the line ceased or was produced separately.

Claims handling makes the issue concrete. Common intake, document management, fraud screening, payment processing, and complaint handling can serve many insurance products. The clinical adjudication of health claims, bodily injury litigation, catastrophe loss adjustment, and long-term life benefit administration, however, have distinct expertise and time horizons. A shared software layer may yield economies while specialist adjusters remain dedicated to particular claims. Aggregating the two under a single overhead category obscures the production boundary.

Information accumulation and actuarial credibility

A larger exposure base can provide more granular frequency, severity, lapse, longevity, and behavioral data, increasing the potential precision of estimates for relevant homogeneous risk groups. The value depends on data quality, exposure comparability, stable definitions, and the treatment of tail dependence. Pooling heterogeneous observations without modeling differences may reduce credibility. Economies in statistical learning are therefore conditional on informative aggregation rather than simply on the number of records.

To illustrate ordinary sampling precision, suppose independent and identically distributed claim observations XiX_i have variance σ2\sigma^2. The sample mean has variance σ2/n\sigma^2/n, which declines as the sample grows. If observations share a positive common correlation ρ\rho and equal marginal variance, the mean variance becomes σ2[1+(n−1)ρ]/n\sigma^2[1+(n-1)\rho]/n. The common-risk component persists when nn increases. This identity explains why exposures subject to common weather, inflation, cyber, legal, or economic shocks require a different analysis from independent idiosyncratic claims.

The analogy between more policyholders and stronger insurance solvency becomes valid only after the dependence structure, tails, contract design, and capital basis have been examined. Data pooling can be obtained through industry databases, reinsurers, statistical partnerships, and purchased models. A large insurer may possess proprietary information advantages, while a small specialist can possess superior segment knowledge. Neither organizational size nor ownership form determines informational accuracy by itself.

Labor specialization, learning, and quality

Scale can support specialized actuarial teams, claims litigators, investment managers, catastrophe analysts, and technology-security staff. Repetition creates opportunities for procedural learning, quality assurance, and automation. Specialization also adds handoffs and requires coordination among pricing, reserving, underwriting, claims, asset-liability management, and compliance. The insurer gains when marginal specialization benefits exceed the transaction and coordination costs of the more complicated organization.

Quality is an essential dimension of that calculation. A claims operation that reduces staffing expenses by delaying legitimate claims may appear more efficient in a narrow cost ledger yet diminish policyholder value and potentially increase litigation and regulatory expenses. A low acquisition-cost channel may produce poorly understood products, cancellations, or complaints. For economic comparisons, prompt and accurate claims settlement, contractual fulfillment, complaint resolution, data security, and solvency should be treated as outputs or constraints whenever material.

Cummins and Weiss (1993) explicitly treated loss settlement as an output in their property-liability efficiency analysis. Their result illustrates why measured underwriting-production efficiency depends on the services considered, not solely on written premiums. It does not establish that any given claims-management scale or settlement speed is socially optimal today.

Organizational form and ownership

Mutual and stock insurers differ in governance, ownership claims, capital access, and incentives. Their cost-minimizing technologies may differ because they serve different market segments, employ different risk appetites, or value member benefits differently. Cummins, Weiss, and Zi (1999), using cross-frontier organizational analysis, found evidence that stock and mutual U.S. property-liability carriers operated on distinct production and cost frontiers. A single pooled frontier can therefore classify genuine organizational differences as inefficiency.

Organizational form also interacts with financing. A rapidly growing stock insurer may raise equity but face shareholder return requirements; a mutual may rely more heavily on retained surplus or alternative capital structures. The comparative cost is related to available risk-bearing capacity, governance and customer composition. A lower observed expense ratio does not by itself settle which form offers more effective insurance at the same quality and financial security.

Economies beyond the legal entity

A corporate insurance group may centralize risk modeling, investment operations, procurement and technology while retaining separate licensed entities. Several legally independent insurers may belong to a common group, use the same underwriting platform, or buy services from an external administrator. Conversely, a single legal company may operate different brands and largely autonomous systems. Empirical work must identify the decision-making and cost-bearing unit appropriate for each economic mechanism.

This is especially important for international groups. Consolidated reporting may net intragroup transactions, reclassify shared service charges, and combine business that is regulated in multiple jurisdictions. A group can realize procurement efficiencies without unrestricted capital mobility; capital may be required and ring-fenced in individual jurisdictions or legal entities. A group-level cost advantage and a legal-entity-level solvency obligation can coexist. Comparative analysis needs to retain both levels of observation.

Diversification versus Cost Synergy

Risk pooling within a line and across lines

Pooling insurable risks is foundational to underwriting, but actuarial diversification has a different denominator and mechanism from administrative scale. Let L1L_1 and L2L_2 be losses with variances σ12\sigma_1^2 and σ22\sigma_2^2, and correlation ρ\rho. Then

Var⁡(L1+L2)=σ12+σ22+2ρσ1σ2.\operatorname{Var}(L_1+L_2)=\sigma_1^2+\sigma_2^2+2\rho\sigma_1\sigma_2.

The combined standard deviation is below the sum of the standalone standard deviations when correlation is less than one and both standard deviations are positive. Under a stylized capital rule proportional to standard deviation, combining imperfectly correlated lines can reduce aggregate required risk capital relative to the sum of standalone calculations. This is a property of the assumed dependence and capital metric, not an empirical assertion about any particular regulator's formula.

For illustration, if two portfolios each have standard deviation 100 and correlation 0.25, their combined standard deviation is 25000\sqrt{25000}, approximately 158.1, compared with 200 as the sum of standalone standard deviations. The 41.9 difference is a reduction in that particular dispersion measure; it is not an automatic monetary saving, and would change under different distributions, confidence levels, risk margins, and capital definitions. The example isolates the covariance mechanism rather than measuring a company.

For catastrophe, liability, pandemic, cyber, inflation, interest-rate, and longevity exposures, correlations can change under stress and tail dependence may be more important than ordinary linear correlation. An insurer may also accumulate apparently separate lines whose losses emerge together in a catastrophe or a market crisis. A diversification conclusion must therefore be scenario-dependent and tested against joint extreme outcomes.

Economic capital, cost of capital, and risk prices

Reducing aggregate volatility can create shareholder or policyholder value if internal capital is costly, external capital is imperfect, distress entails real costs, or better portfolio balance improves the availability of underwriting. Froot and Stein (1998) incorporated funding frictions into an integrated capital-allocation analysis for financial institutions. Froot (2007) extended the analysis to insurers and reinsurers, examining how product and capital-market imperfections and asymmetric loss distributions affect risk allocation and capital decisions.

These studies supply the economic rationale for treating risk-bearing capacity as potentially scarce and costly. They also impose discipline on simplistic corporate diversification arguments. Under frictionless capital-market assumptions, diversified investors can sometimes achieve portfolio diversification without an insurer combining business lines. A corporate diversification premium requires a specific mechanism that investors or contractual counterparties cannot replicate at equal cost. Distress costs, underwriting information, financial frictions, regulatory recognition, and contractual commitment are plausible mechanisms and must be evaluated, rather than presumed.

Capital allocated to a line is also method-dependent. Allocating standalone risk capital overstates the group's total if a diversification benefit is recognized, while distributing all benefit mechanically by premium can misprice the marginal contribution of correlated risks. Incremental risk, stress scenarios, risk appetite, and group fungibility need to inform internal decisions. The optimization objective includes expected underwriting return, claims and operational cost, capital charges, and constraints on future liquidity and solvency.

Reinsurance as an organizational alternative

A specialist can transfer part of its risk to diversified reinsurers. Reinsurance shares tail exposures and can stabilize capacity without requiring all primary risks to be underwritten by a single conglomerate. The costs include reinsurance premium, counterparty credit exposure, contract basis risk, reinstatement terms, possible limits on recoverability, and negotiation and monitoring expenditure. Economies of risk pooling therefore exist at multiple layers: insured, direct insurer, reinsurer, capital market, and public risk-sharing facility.

Cummins, Doherty, and Lo (2002) examined catastrophe financing capacity in the U.S. property-liability sector, highlighting how industry resources and the distribution of losses affect the capacity to respond to extreme events. Their capacity framework does not imply that concentrated ownership maximizes social insurance capacity; it illustrates that the allocation and availability of resources in severe states matter as much as aggregate totals.

The IAIS's 2025 global monitoring report examines reinsurance alongside macrofinancial and insurer-group risk. Reinsurance-market concentration, group linkages, and correlated financial exposures are relevant to a complete account of diversification. A larger combined book can reduce idiosyncratic uncertainty while increasing common exposure to the same reinsurance counterparties, service providers, or financial markets.

Limits of regulatory diversification credit

Regulatory diversification benefit is a property of a prescribed capital framework as applied to specified legal entities and recognized risk relations. It is distinct from an accounting expense saving and from any capital amount that can actually be moved across jurisdictions. The IAIS Insurance Core Principles and ComFrame establish supervisory expectations for governance, capital, risk management and group oversight; actual quantitative recognition depends on the applicable national or regional system.

For a diversified insurance group, prudential assessment must address legal-entity solvency, group-wide risk, intragroup exposures, risk concentration, governance, and capital transfer restrictions. A group may have a reduced modeled aggregate capital requirement while preserving material local capital requirements. It may also confront higher capital charges from complex noninsurance exposures, opaque structures, or stress dependence. Consequently the relevant net benefit is the attainable risk-bearing value after funding, execution, and supervisory constraints.

A complete net-benefit comparison

The organization decision can be represented as a transparent decomposition rather than as a new theory:

ΔV=ΔVshared operations+ΔVrisk capacity+ΔVcustomer services−ΔCcoordination−ΔCintegration−ΔCgovernance−ΔCfinancial frictions.\Delta V = \Delta V_{\mathrm{shared\ operations}} +\Delta V_{\mathrm{risk\ capacity}} +\Delta V_{\mathrm{customer\ services}} -\Delta C_{\mathrm{coordination}} -\Delta C_{\mathrm{integration}} -\Delta C_{\mathrm{governance}} -\Delta C_{\mathrm{financial\ frictions}}.

The terms are categories in a decision accounting exercise. They should be evaluated against a defined specialist, contractual, or integrated alternative with consistent assumptions about demand, liability quality, and risk. Several terms may overlap in actual data; double counting must be avoided. The equation's purpose is to force recognition that shared underwriting platforms, lower risk capital, and higher consolidated profit are distinct potential outcomes.

Distribution and Technology Scale

Customer acquisition and marketing

Insurance distribution has fixed and variable elements. Building a recognizable brand, developing a sales platform, training agents, negotiating partnerships, maintaining regulatory disclosures, and creating customer-support systems can involve upfront costs spread over many policies. A larger portfolio may support wider marketing experiments, route optimization, and investments in consumer understanding. Yet acquisition costs can rise as a company expands beyond its strongest segments and moves into more expensive markets.

A wide portfolio can support cross-selling and lower repeated identity-verification and onboarding costs. Scope benefits are credible when customers actually value complementary products, underwriting conditions remain suitable, and service quality is maintained. An apparent cross-selling gain may reflect bundling discounts, differential pricing, transfer of acquisition charges between lines, or market power instead of production efficiency. Analysis should distinguish reduced real resource costs from redistributed prices.

Independent agents, brokers, bancassurance partners, aggregators, employer channels, and direct digital sales have different cost structures. A direct channel may have higher initial software expense but lower marginal processing cost; an intermediated channel may reduce information and search costs for complex contracts. Larger scale does not identify the superior channel absent customer and product comparability. A low distribution-cost insurer serving relatively simple standardized risks is not a direct frontier benchmark for a specialist advising on layered commercial programs.

Software, automation, and learning curves

Digital underwriting, document classification, claims routing, customer authentication, fraud analytics, actuarial computation, cybersecurity monitoring, and data infrastructure can make fixed investment economically important. When an automated process has a large development cost but small marginal processing cost, the volume over which its fixed expenditure is amortized determines average cost. This is a plausible scale mechanism with a measurable engineering budget and throughput, not a claim that every algorithm reduces total insurance cost.

The gains depend on accurate inputs, model calibration, auditability, human escalation, cybersecurity, resilience, and the cost of system replacement. A shared model may improve consistency, while the same shared model can transmit errors across many products or affiliates. Cloud services can permit small carriers to purchase processing capability on a usage basis, reducing the need for corporate integration. External scale economies and internal firm-scale economies should be studied separately.

The European Insurance and Occupational Pensions Authority's October 2026 risk dashboard identifies digitalization and cyber risks among observed sector vulnerabilities. This supervisory evidence establishes that technology introduces an additional risk domain; it does not estimate a causal productivity effect of any particular AI system. Technology-related business cases should report both operating benefits and the cost of governance, validation, resilience, incident response, and vendor concentration.

Legacy systems and integration diseconomies

Mergers and line expansion can generate duplicated policy-administration systems, incompatible customer and claims records, divergent accounting taxonomies, and overlapping service contracts. Integration may require years of migration, reconciliation, testing, and employee retraining. The timing of expenditures and benefits matters: a merger that eventually shares platforms can reduce near-term profits and elevate operational risk while systems coexist.

Project evaluations should distinguish recurring run-rate cost savings from one-time implementation expenditures and transition risks. An ordinary net-present-value calculation can compare the discounted expected cost streams of integration versus continued separation. In either case, risk-adjusted forecasts should account for delivery probability and for the possibility that regulatory or contractual restrictions prevent the desired operational consolidation. Stated merger synergies are managerial forecasts until independently evidenced.

Customer outcomes and scale

Economic welfare concerns extend beyond insurer cost. The relevant consumer outcome includes risk protection, price, coverage terms, continuity, claims service, privacy, and redress. Size can finance better catastrophe-response capacity or broader service hours, but can also produce less transparent products, standardized processes poorly suited to vulnerable consumers, or weaker competitive alternatives. A cost improvement is a social benefit to the extent resources are actually saved and the relevant quality constraints are respected; the allocation of that saving to policyholders depends on competition and conduct rules.

The OECD's 2026 insurance-market report documents considerable diversity in the composition of life and non-life insurance markets. This variation cautions against treating a single operating-cost metric as universally comparable. A cross-country analysis of digital scale must control for compulsory insurance, channel structure, product mix, income, claims inflation, local regulatory requirements, and the role of social protection.

Conglomerate Complexity

Scope, specialization, and strategic focus

The conglomeration hypothesis holds that integrated production can create value through shared production, information, distribution, risk transfer, and financial resources. The strategic focus hypothesis holds that specialization can improve expertise, governance, incentives, underwriting selection, brand position, and capital discipline. Neither is a logical universal. The observable organizational landscape can contain diversified and specialist insurers because their advantages differ by product, jurisdiction, market condition, and inherited operating technology.

Berger, Cummins, Weiss, and Zi (2000) investigated joint production and strategic focus in U.S. insurance and reported that the relative advantages depended on the type of provider examined. The inference is not that diversification has no economic value. The appropriate inference is that profit scope must be measured for the actual production arrangement, because cost, revenue and risk-bearing interactions can favor different organizational structures.

Cummins, Weiss, Xie, and Zi (2010), studying 1993–2006 U.S. data in their DEA scope-economies investigation, found that property-liability insurers' cost scope advantages were more than offset by revenue scope disadvantages, while life-health insurers showed both cost and revenue scope disadvantages in their examined combination. They accordingly favored strategic focus in that studied setting. The result is specific to a life-health versus property-liability comparison and should not be extended mechanically to all combinations within property insurance, all current markets, or all technologies.

Agency problems and incentives

Large organizations may experience managerial discretion, managerial empire-building, internal capital-market cross-subsidies, and soft budget constraints. A line with high true incremental cost can persist if overhead allocations conceal its economics or if its weak performance is obscured by stronger affiliates. Conversely, a corporate center can impose underwriting discipline, create transparent enterprise risk limits, and remove duplicated management. Both mechanisms depend on the quality of governance and internal information.

Firm-level technical efficiency, profit efficiency, and governance quality should therefore be evaluated jointly. A parent company's consolidated profit does not identify whether a specialist subsidiary is subsidized or whether the group creates shared customer and financial value. A group may need line-specific profit and risk metrics, explicit transfer pricing, and documented allocation of fixed and capital costs to avoid false scope claims.

Asset-liability mismatches and correlated shocks

Diversifying product offerings changes the timing and risk properties of liabilities. Life and annuity business may feature longevity, lapse, guaranteed-rate and duration exposures; property-liability lines may feature catastrophe, liability-development and claims-inflation exposures. The associated assets and hedging requirements differ. A joint balance sheet can create new opportunities for liquidity management and duration matching but can also concentrate exposure to interest rates, credit spreads, private assets, or difficult-to-liquidate investments.

The IAIS's Global Insurance Market Report 2025 examines financial-market vulnerabilities, group-wide exposures, and reinsurance. The report's monitoring coverage includes large international insurance groups and sector-wide supervisory information. It provides context for studying conglomerate risk, while its aggregate data do not show that every larger group is more or less efficient than a specialist.

The EIOPA 2026 insurance-risk assessment similarly draws on Solvency II insurer and group reporting to characterize risk levels. Such dashboards describe the condition of supervised markets; identifying causal economies from scale would require a different design with insurer-specific output, input, and counterfactual evidence.

Operational and supervisory complexity

As insurers enter new jurisdictions and businesses, they must manage additional contract rules, tax arrangements, local capital requirements, data localization rules, distribution authorizations, conduct standards, group reporting, and supervisory coordination. Some of these requirements can be centrally administered, while others necessarily remain local. Cost elasticity depends on the incremental compliance load as well as fixed central capability.

The IAIS Insurance Core Principles make group governance, ERM, supervisory reporting, capital and group-wide oversight part of the prudential framework. The presence of supervisory obligations does not demonstrate diseconomies: effective oversight also protects policyholders and financial stability. The relevant question is how to implement the necessary controls efficiently while accounting for the risk that complexity makes monitoring and intervention more difficult.

Concentration and competitive effects

A cost-efficient large insurer can create consumer benefits through lower prices, broader coverage, and better service. Market concentration can also change pricing power, entry barriers, bargaining positions, or the access of intermediaries to alternative providers. Observing concentration after consolidation therefore cannot reveal whether economies caused the merger or whether market-power motives, regulation, capital-market conditions, or strategic positioning were decisive.

Scale efficiencies and market concentration occupy distinct research objects. A transaction can increase concentration without improving productive efficiency; it can also improve efficiency without raising the degree of national market concentration in a meaningful way. Competition authorities require market definition, contestability, and consumer-outcome assessment beyond the insurer production-cost model. This dossier focuses on organization and production, while the separate insurance-market-structure research addresses competition and concentration directly.

Empirical Efficiency Comparisons

The United States: property-liability cost frontiers

Cummins and Weiss (1993) estimated stochastic translog cost frontiers using 1980–1988 U.S. property-liability insurer data split by size. Their published frontier analysis reported that large insurers clustered near roughly 90% of their size-specific cost frontier, while medium and small samples had average estimates near 80% and 88%, respectively. They also found economies of scale among small and intermediate insurers. These values compare observations to particular estimated frontiers and cannot be read as the proportion of total industry expense that could be saved through universal consolidation.

A cross-size interpretation faces an immediate caveat. Size-stratified frontiers can differ, and a carrier's distance from its own reference frontier does not mean it could operate on another group's frontier without changing its outputs and technology. A high efficiency score among large firms can coexist with high absolute cost if the measured frontier itself includes complex activities; a smaller insurer can operate effectively in a narrow segment even if scale economies exist elsewhere.

The study's historical period precedes contemporary digital administration, cyber governance, current catastrophe models, and modern capital standards. Its enduring contribution is methodological and conditional: it shows that operational efficiency can vary within size groups and that scope for scale-driven improvement is not uniform across the size distribution.

The United States: life consolidation and ownership structure

Cummins, Tennyson, and Weiss (1999) studied consolidation and efficiency in U.S. life insurance, using 1988–1995 observations and frontier-based productivity measures. They reported that acquired firms experienced greater efficiency gains than firms outside merger and acquisition activity, and that firms with non-decreasing returns and financial vulnerability were more likely to be acquired. These associations are evidence about the studied consolidation process, not proof that all future acquisitions generate causal benefits.

Selection is central. Acquirers may target carriers with particular characteristics; merger integration may accompany technological change or restructuring unrelated to ownership; and surviving firms differ from exits. A proper counterfactual would compare the acquired firm against the outcome that it would have achieved absent acquisition, under comparable macroeconomic and industry conditions. Frontier movements, selection, and measured post-merger efficiency must be distinguished.

Cummins, Weiss, and Zi (1999) found in their stock-and-mutual cross-frontier analysis that the two organizational forms could operate on different technologies. This warns against attributing every intergroup cost difference to size. Corporate form, customer segment, pricing behavior, and objective functions can shape the economically relevant frontier.

Spain: deregulation, restructuring, and production choices

Cummins, Rubio-Misas, and Zi (2004), studying Spanish insurers over 1989–1997, found that stock and mutual organizations operated on distinct production, cost, and revenue frontiers in their organizational-structure comparison. They also considered comparative organizational advantages rather than treating the market as a single uniform producer technology. The implication for scale research is that institutional setting and ownership form belong among its conditioning variables.

Cummins and Rubio-Misas (2006) examined Spanish deregulation and consolidation around the European insurance-market liberalization period. Their research described a substantial decline in firm counts, growth in average size, and changes in efficiency and unit prices across the observed industry. The historical conjunction of deregulation, merger activity, and efficiency outcomes is substantively informative; it does not permit all price changes to be attributed to scale savings.

Cross-country transposition must account for the Spanish market's regulatory transition and product composition. Findings from a liberalization episode need not apply with equal strength to mature markets with already integrated operations, to compulsory insurance systems, or to jurisdictions with different solvency, distribution, and ownership arrangements.

International evidence: heterogeneity in efficiency

Eling and Luhnen (2010) analyzed 6,462 insurers across 36 countries in their international efficiency comparison, comparing multiple approaches and firm characteristics during 2002–2006. They reported substantial country differences and changes in efficiency over time. The sample is unusually broad for insurer-level comparative research, yet a broad international dataset does not itself solve the difficult problem of matching product and service quality across jurisdictions.

Environmental differences can shape measured performance through wage levels, catastrophe exposures, compulsory coverage, taxation, underwriting cycles, solvency requirements, and the role of life savings products. Exchange rates and inflation can alter financial comparisons. Firm size may correlate with better information systems or access to finance, but also with serving countries in which the business mix naturally generates larger balance sheets. Proper inference should state whether estimates compare firms within the same country and line, across pooled groups, or relative to separate country frontiers.

The OECD's 2026 market-trends report gives current macroeconomic context but does not reproduce the historical insurer-level production panel. In 2025, its reported average penetration was 6.6% of GDP for OECD countries and 4.2% among non-OECD reporting jurisdictions. These are premium-to-GDP measures and cannot be transformed into evidence of carrier size-related cost efficiency. They illustrate the market composition differences that any comparative production study must consider.

Contrasting the U.S. scope studies

The central scope evidence should be read in its historical and methodological order. Berger et al. (2000) investigated profit scope across insurance providers and reported heterogeneous support for conglomeration and strategic focus. Cummins et al. (2010) measured cost, revenue and profit efficiency with a later U.S. panel and concluded that focus dominated for the major life-health and property-liability combinations they studied. The two articles address related questions using different data and techniques; their findings should not be collapsed into one numerical verdict.

A coherent interpretation is that shared costs, demand effects, business mix and institutional constraints can move independently. A specialist may outperform a conglomerate on revenue even when the latter can spread fixed expenses across business lines. A diversified producer may benefit financially from a low-correlation portfolio without enjoying operating cost scope. Comparative empirical work must preserve those distinct endpoints.

Evidence matrix

Source and studied setting Observed design or method Source-supported finding Key interpretive boundary
Cummins and Weiss (1993), U.S. property-liability, 1980–1988 Stochastic translog size-stratified cost frontiers Small and intermediate samples exhibited scale economies; efficiency differed by size Historical sample and separate estimated frontiers
Cummins, Tennyson, and Weiss (1999), U.S. life, 1988–1995 DEA and productivity change in M&A context Acquired insurers had greater observed efficiency gains Merger selection and counterfactual identification
Cummins, Weiss, and Zi (1999), U.S. property-liability Cross-frontier organizational comparison Stocks and mutuals had distinct estimated frontiers Organizational form and technology differences
Berger et al. (2000), U.S. insurers Profit scope comparison Conglomeration and focus advantages varied by provider type Heterogeneous outputs and profit measurement
Cummins, Rubio-Misas, and Zi (2004), Spain, 1989–1997 DEA organization-specific frontiers Different technologies across organizational forms Deregulation-era institutional setting
Cummins and Rubio-Misas (2006), Spain, 1989–1998 Frontier and productivity analysis around deregulation Consolidation accompanied restructuring and efficiency changes Multiple simultaneous policy and market shocks
Eling and Luhnen (2010), 36 countries, 2002–2006 Cross-country frontier efficiency comparisons Large international differences and temporal efficiency growth Input/output comparability and environment
Cummins et al. (2010), U.S., 1993–2006 DEA and scope-oriented efficiency comparisons Cost benefits can be offset by revenue losses; focus favored in studied combinations Historical business-line combinations
OECD (2026), 67 jurisdictions, 2025 market data Official national-authority market statistics Geographic variation in premiums, business mix and performance Sector aggregates are not insurer cost frontiers
IAIS (2025), global supervisory monitoring Individual group and sector-wide supervisory reporting Observed group-level vulnerabilities and systemic exposures Monitoring does not identify efficiency causation

The matrix intentionally compares research designs and admissible conclusions, not a fabricated pooled numerical effect size. Its study periods differ, inputs and outputs are not fully harmonized, and methodologies do not estimate an identical causal parameter. An arithmetic average of their findings would have no clear economic interpretation.

Methodological Findings and Research Implications

Cost, technical, allocative, and profit efficiency

Technical efficiency concerns the amount of input used to produce a given output bundle relative to a feasible production frontier. Cost efficiency includes input prices and the choice of input mix, so a technically efficient producer may remain allocatively inefficient. Revenue efficiency concerns output mix and attainable revenue under specified prices and demand. Profit efficiency combines these components with revenue and cost opportunities. An insurer can be technically efficient without operating at an optimal scale or producing an economically advantageous business mix.

The familiar data envelopment analysis methodology originated in the Charnes–Cooper–Rhodes study of multiple-input, multiple-output decision units. DEA envelops observed producers under specified returns-to-scale and disposability assumptions. It provides relative efficiency measures conditional on the sample and production definitions; it does not automatically identify a population technology or prove that frontier firms themselves are economically optimal.

Stochastic frontier analysis models departures from a parametric frontier using assumptions about random noise and inefficiency. The Aigner–Lovell–Schmidt formulation is a classic reference for the composed-error approach. It can accommodate statistical noise, but inference is sensitive to functional form, distributional assumptions, omitted environmental heterogeneity, and the treatment of measurement error. Stochastic and nonparametric models address different uncertainty structures, and agreement between them is informative only when input, output and sample definitions are sufficiently aligned.

Simar and Wilson (2007), in their two-stage efficiency-inference analysis, showed why conventional regressions of estimated nonparametric efficiency scores on environmental variables may provide invalid inference and developed bootstrap-based procedures under explicit data-generating assumptions. Consequently a statistically significant coefficient on insurer size in a second-stage regression is not automatically a reliable test of scale effects. Research designs must state sampling assumptions and how uncertainty in estimated frontiers is propagated.

Endogeneity and the merger counterfactual

Insurer size is an endogenous organizational result. Efficient firms may grow because lower costs enable competitive pricing; managers may choose to merge because they anticipate future changes in demand or capital requirements; better-funded companies may acquire struggling firms; and regulators may intervene in distressed mergers. A regression of efficiency on size therefore mixes potential causal effects of scale with selection into size.

A credible causal strategy requires a defined intervention and counterfactual. Prospective options include merger-event designs with comparable nonmerged firms, carefully justified difference-in-differences, variation in regulatory or market access conditions, and within-firm analyses of genuinely exogenous shocks to scale. Each demands assumptions about trends, spillovers, timing and treatment selection. Merger announcement dates do not mark instant integration; financial outcomes may be anticipated, realized gradually, or distorted by restructuring charges.

A productive research program should estimate multiple endpoints: total economic cost, administrative unit cost, underwriting profitability, capital cost, customer prices, coverage quality, complaints, claims timing, and solvency robustness. Lower expense alone cannot be interpreted as improved insurance efficiency when the output bundle or contractual performance changed.

Data construction and the unit of analysis

An empirically usable insurer panel should identify consistent legal entity and group codes, mergers and portfolio transfers, product-line definitions, business written and earned, earned exposure, claims and loss-adjustment services, acquisition expenses, general administration, input prices, risk capital, reinsurance, investment activity, solvency measures, and observable quality. Monetary values should be deflated using suitable price measures, and structural changes in accounting standards need explicit bridge methods.

Researchers should specify whether business volume means policy count, exposure units, premium volume, insured amount, reserves, or a constructed value-added measure. An output index should incorporate meaningful service complexity and should not increase when the insurer simply raises prices. Risk-bearing cost should distinguish expected losses priced into contracts from the incremental resources and capital services required to provide insurance.

Firm groups are not always independent statistical units. Multiple subsidiaries can share costs, customers, and control systems. A consolidated group may supply common inputs to affiliates with different geographic or product mandates. Statistical designs should model hierarchical dependence and avoid comparing group observations against subsidiaries as if those units were distinct production alternatives.

Estimating economies of scope responsibly

Scope estimation often fails at the comparison endpoints. Single-line insurers may not operate at the volume or quality represented by a diversified group's separated lines. An analyst should report support for each output combination, minimum sample density, and the sensitivity to near-zero or zero outputs. Any counterfactual obtained through extrapolation must be labeled as model-based rather than directly observed.

The choice among a translog, a flexible composite cost function, a directional frontier, and DEA changes the assumptions. Empirical estimates should be tested for monotonicity and economically sensible input-price homogeneity where relevant, together with sensitivity to functional form and the treatment of output zeros. Cost frontiers must reflect the cost of quality and risk-bearing services consistently across specialist and joint producers.

When alternative ownership forms use different technologies, pooled frontier results should be compared with group-specific or meta-frontier analyses. The Cummins–Weiss–Zi findings make this problem especially salient for mutual and stock insurers. Inference should distinguish efficiency differences from structural differences in feasible production or business objectives.

Policy, regulatory and management implications

For insurer management, the relevant decision is whether an additional line, acquisition, digital investment, or shared-service consolidation creates risk-adjusted net economic value relative to the strongest feasible alternative. Business cases should include integration time, transition risk, retained service quality, loss covariance, capital fungibility, operational resilience, reinsurance costs, and any revenue effects. Scale claims should be validated against observable process-level costs as well as enterprise-level efficiency.

For supervisors, the appropriate task is to assess governance, risk accumulation, control effectiveness, intragroup transactions, and policyholder protection in proportion to the insurer's actual risk and complexity. The IAIS global supervisory framework provides a reference for risk management and group-wide supervision. A regulator should not infer lower risk merely from a larger number of policies, nor impose a presumption of inefficiency upon all diversified groups. The meaningful assessment is of actual dependence, available capital, operational controls, and contractual obligations.

For competition and public policy, a claimed economy must be converted into a credible resource-saving estimate and considered alongside changes in consumer outcomes and market structure. A merger that saves duplicated infrastructure may also restrict choices or raise barriers for specialist firms. Outsourcing, interoperability, cooperative data systems, and reinsurance may capture some scale benefits while preserving independently competing underwriters. The economically efficient configuration can therefore be plural rather than a single dominant organizational form.

Source quality and limitations

The empirical articles cited here have identifiable journal or institutional records, but several full papers were accessible only through abstracts and publisher metadata in the present research environment. Their detailed econometric coefficients and robustness tables have not been reestimated. The report does not construct a harmonized quantitative meta-analysis, a 2026 global insurer-size regression, or an audit of the financial statements of any named insurance company. The cited OECD and IAIS datasets operate at different units of analysis and are not silently converted into an insurer cost-efficiency panel.

These limitations define the reach of each statement. Theoretical expressions establish conditional relationships; historical frontier papers establish reported sample findings; contemporary supervisory sources describe monitoring frameworks and market conditions. None alone proves the universal optimal size of an insurance company. The substantive synthesis rests on the consistency of their definitions and differences, with source-to-claim links retained for independent inspection.

Conclusion

Economies of scale arise when proportionately expanding an economically comparable insurance output bundle increases cost by a smaller proportion. Economies of scope arise when producing distinct outputs jointly reduces the resources needed relative to feasible separate production. The two properties are conditional on output definition, quality, technology, input prices, market conditions, and the specific point or production comparison. Risk diversification is an additional mechanism, governed by loss dependence, financial frictions, capital rules, and the ability to use capital where obligations arise.

The insurance production process offers genuine opportunities to share information systems, risk modeling, governance functions, claims infrastructure, distribution, specialized skills, and financial capabilities. Those opportunities can be realized through corporate integration, group services, contractual outsourcing, reinsurance, or cooperative arrangements. An economy in one input or one business function does not decide which ownership boundary creates the greatest total economic value.

The historical evidence demonstrates that size, efficiency and scope can diverge. Small and medium insurers can exhibit potential scale efficiencies, while larger firms may face different frontiers and market conditions. Acquired firms can show efficiency gains without settling the causal role of the merger. Stock and mutual insurers may occupy distinct production technologies. Diversified groups can save operating cost while losing revenue efficiency; the joint effect on profit must be assessed independently. International comparisons reveal differences that require country, line, period, output and institutional controls.

The most defensible evaluation of insurance scale is therefore a multiproduct, quality-adjusted, risk-aware, and institutionally situated comparison. Every empirical claim needs to name its unit of production, its feasible alternative, its cost and output measures, and its identification assumptions. Every managerial claim of synergy requires a net present-value case after integration and risk costs. Every supervisory assessment of diversification requires scrutiny of tail dependence, group structures, and legal-entity capital. These conclusions organize established research into a coherent analytical account and identify testable questions for subsequent empirical work without instituting a new authorial theory.

References