Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Insurance Underwriting Cycles and Price Dynamics

Abstract. Insurance underwriting cycles describe recurrent changes in the price, availability, profitability, and contractual conditions of risk transfer. Their empirical manifestations span property, casualty, reinsurance, and internationally traded commercial coverage, but the relevant variables do not move together or display a universal period. This monograph examines the economic and institutional mechanisms connecting expected claims, capital constraints, costly external finance, investment conditions, pricing lags, competitive conduct, reinsurance, and supervisory requirements. It reconstructs the claims and limitations of major scholarly interpretations, from rational-expectations and partial-adjustment accounts to capacity-constraint and financial-quality models. It then reconciles those interpretations with the observable distinction between a premium rate, a premium volume, a combined ratio, a reinsurance rate on line, and the quality and quantity of insurance protection. Primary institutional evidence for 2024–2026 illustrates sharp differences among geography, business line, observation period, and data source: declining commercial renewal rates can coexist with positive aggregate premium growth, strong insurer capital, and persistent casualty severity pressures. The analysis establishes an empirical identification agenda and examines how inflation, reserve revision, correlated losses, capital mobility, contractual adjustment, and supervision interact. Its original contribution is a source-grounded synthesis and explicit set of measurement and causal distinctions, rather than a newly named theory or an original econometric finding. The result is a historically informed, international framework for understanding what underwriting-cycle evidence actually proves and what remains uncertain.

Keywords: Underwriting cycle; insurance pricing; hard market; soft market; property-casualty insurance; reinsurance; insurance capacity; insurer capital; risk-adjusted rate; claims inflation; loss reserves; combined ratio; external finance; insurance competition; solvency; empirical identification; insurance economics.

Introduction

The price of insurance is a contingent promise priced before its full cost is known. The insurer receives a premium today and undertakes to pay covered future losses, subject to contract definitions, limits, deductibles, claims settlement, and its continuing ability to perform. This temporal separation makes an insurance quotation simultaneously a forecast, a financing decision, a contractual allocation of risk, and an offer of future claims-paying capacity. Price movements therefore need to be examined together with changes in the quantity and quality of protection offered.

An underwriting cycle refers to temporally patterned changes in insurance-market terms and underwriting results, commonly described as transitions between hard and soft markets. A hard market generally combines tighter underwriting, higher risk-adjusted prices, smaller offered lines or limits, and stricter conditions; a soft market more often combines competition for premium, lower rates, and broadened terms. These descriptions are empirical tendencies rather than universal bundles. A policyholder may encounter a hard casualty market and a soft property market at the same renewal. One insurer may reduce a quote while another withdraws. A nominal premium increase may represent a fall in the price per insured unit if replacement costs rise faster.

The classic literature establishes both the scientific importance and the contested explanation of these phenomena. Cummins and Outreville (1987) linked observed international underwriting patterns to rational expectations and institutional intervention. Doherty and Kang (1988) explained apparent oscillation through partial adjustment to equilibrium prices, while Gron (1994) examined capacity constraints across insurance lines. Winter's competitive-market dynamics (1994) placed dependent insured risks and costly external equity inside the mechanism. These are different, testable accounts of pricing adjustment. Their evidentiary disagreement is part of the subject itself.

The scope of this dossier encompasses underwriting dynamics in private insurance and reinsurance, with principal empirical attention to non-life and property-casualty lines. The historical scholarly studies largely analyze particular United States or industrialized-country samples; current supervisory and broker reports describe separate populations and periods. Life insurance is relevant when capital and asset-liability mechanisms overlap, but its long-duration liability valuation requires separate treatment. The legal status of coverage and the policyholder's welfare enter where changing prices, limits, and exclusions alter the economic allocation of loss.

The method is structured documentary synthesis. Each proposition is checked against original research or institutional data with explicit attention to reporting period, denominator, region, product line, observational unit, and source access. Algebraic relationships below are accounting identities or transparent illustrative exercises. Historical articles are interpreted within the limits of the verified original article or publisher record. Published market statistics are reported as the source defines them, with no claim to possession of the underlying account-level samples. The conclusion distinguishes robust descriptive regularities from disputed causal attribution and establishes a replicable research agenda.

Historical Underwriting Cycles

A historical problem of timing and interpretation

Descriptions of hard and soft insurance markets predate modern econometrics, but the scholarly problem is more precise: what series oscillates, why does it oscillate, and which institution controls the timing? Published premium, underwriting margin, capacity, loss development, and reserve figures embody different observation rules. In particular, an annual statutory underwriting profit is affected by premiums earned from earlier policies and by revisions to claims from several preceding accident years. Its fluctuations are not automatically contemporaneous movements in new-business quotations.

The international study by Cummins and Outreville (1987) demonstrated that rational-expectations reasoning can coexist with observed patterns in accounting underwriting profits once regulatory, institutional, and reporting delays are admitted. Their international underwriting-cycle analysis did not establish a single invariant period for every country, line, or era. It identified a possibility: a rational response to new information can acquire a cyclical appearance after it passes through heterogeneous institutional adjustment mechanisms. This is particularly important where price filings, annual renewals, claims reporting, and reserve reviews operate on different calendars.

Doherty and Kang (1988) used a capital-market pricing framework to distinguish a contemporaneous equilibrium price from the path toward it. Their interest-rate and price-cycle model treats informational, regulatory, and contractual lags as causes of partial adjustment. The paper's reference to a roughly six-year property-liability cycle describes the historical evidence and research discussion available to its authors, not a universal law of insurance markets. A deterministic sinusoid built into a forecast because an earlier regression found a six-year period would mistake a sample-specific regularity for a stable structural parameter.

The contrast between apparent cycling and temporary capital scarcity sharpened in Gron's 1994 work. In the RAND capacity-constraint study, four property-casualty lines provide evidence broadly consistent with an inverse relation between available capacity and underwriting margins. In a separate 1949–1990 capacity examination, Gron reported that declines in capacity were associated with subsequent increases in price-payment margins, and that later capacity growth was positively associated with earlier margin growth. The studies illuminate adjustment after financial depletion. The reported evidence does not mean every price increase is caused by scarce capital: simultaneously rising claim costs, coverage changes, and changes in risk mix can also raise observed premiums.

Winter (1994) developed an account in which correlated insured losses make net worth relevant to the aggregate supply of credible insurance promises, while raising new external equity can cost more than using retained earnings. The resulting dynamic competitive-market explanation shows how hard-market pricing can arise without assuming that firms collectively choose an arbitrary date for restricting supply. The important historical shift is analytical: a competitive market may display temporary price-cost wedges when risk-bearing capital is costly and risk cannot be diversified away at negligible expense.

From a supposed universal cycle to heterogeneous regimes

Later work complicates any simple triumph of one explanation. Lamm-Tennant and Weiss (1997) examined international rational-expectations and institutional effects and associated cycle features with differences in market and institutional structures. Choi, Hardigree, and Thistle (2002), in their comparison of alternative pricing models, found the actuarial and capacity-constraint interpretations consistent with their 1935–1997 stock-insurer data among six specifications tested. Higgins and Thistle (2000) used smooth-transition regression evidence to distinguish constrained from unconstrained regimes in historical United States data. The last study's finding of cycling in constrained regimes should be read as a sample-bound regime result rather than a timeless timetable.

Three implications follow from the literature as a whole. First, observed oscillation requires an explicit variable and measurement process. Second, multiple mechanisms can operate together, and evidence for one does not automatically exclude the others. Third, a change in capital adequacy, institutional rules, investment returns, claims-cost expectations, or reinsurance supply may alter the amplitude, persistence, or existence of a cycle. It is therefore more defensible to speak of insurance-cycle mechanisms and line-specific episodes than to assign every market to a fixed universal sequence.

Historical institutional heterogeneity

Property and catastrophe insurance are disproportionately sensitive to concentrated loss events and replacement-cost uncertainty. Long-tail liability insurance is disproportionately sensitive to delayed reporting, litigation, social inflation, changes in legal interpretation, and revisions to reserves. Workers' compensation may respond to employment, wages, occupational injury frequency, statutory benefit rules, and rate regulation. Credit, agricultural, cyber, and marine insurance each contain further risk, information, and concentration structures. Even within a single corporation, the capital fungibility between these portfolios can be incomplete.

International comparisons require similar care. Insurers in different jurisdictions face different licensing arrangements, allowable pricing practices, reinsurance access, investment regulation, solvency rules, natural-hazard exposure, liability law, and claims inflation. The rational-expectations literature offers one reason institutional calendars matter; the capital-constraint literature offers another reason the same financial shock may yield different local price paths. An observed national cycle cannot simply be assigned to a global market without accounting for those intervening arrangements.

The historical record consequently motivates an empirical principle of interpretation, rather than a new authorial doctrine: first reconstruct the time and risk unit of each series; then compare mechanisms that generate its movement. If the premium written today belongs to one underwriting year, the loss recognized today to several past accident years, and the reserve release to yet another set of policies, their simple correlation is already an economically compound observation.

Pricing and Claims Shocks

What precisely is the price?

The insurance premium paid by a policyholder can be written for analytical purposes as a function of exposure, expected loss cost, claims adjustment expense, operating and acquisition expense, reinsurance cost, capital remuneration, taxes, and the net effect of competition and contract design. This decomposition is descriptive and does not require a unique additive pricing model. The actuarial indication, marketed quotation, premium actually bound, and reported accounting premium are distinct observations.

A simple accounting identity illustrates the first distinction: aggregate written premium is the sum of contract-level written premiums. An increase in this sum can follow from new business, greater insured values, expanded coverage, changes in geographic or risk composition, or an increase in rate per standardized unit. A standardized renewal-rate change, by contrast, attempts to compare the price of a comparable risk at consecutive renewals. A market index of renewal quotes and a national premium-growth statistic cannot be substituted for each other.

The OECD's 2026 analysis of non-life premiums reports average 2025 nominal non-life premium growth of 8.7% and real growth of 4.1% among its reporting populations. It attributes much of the growth to rate adjustments following claims-cost increases, while documenting variation in policy numbers, prices, and claim experience by jurisdiction. This is directly compatible with commercial renewal-rate declines in other samples during 2026: the observation year, insured population, weighting, and measurement target differ. There is no empirical contradiction to explain away.

The lag from claim shock to quoted premium

A loss shock can be immediate, but its recognized cost unfolds over time. An earthquake can produce a prompt cluster of reported insured losses, followed by repair-cost revisions and litigation; a liability event can enter insurers' reported experience only after years. Underwriters may raise expected future severity as soon as information arrives. Rate filing or renewal constraints may defer the revised price. Insurers may also recalibrate deductibles, limits, waiting periods, exclusions, sublimits, percentage-based deductibles, and aggregate caps.

Where claims are long-tailed, incurred losses and loss-reserve development require separate accounting. An adverse reserve development today increases the estimated cost of policies written in prior periods, potentially weakening current reported profit and capital. It need not imply the same percentage deterioration in the prospective loss ratio for a newly underwritten risk. A favorable reserve release similarly improves calendar-year underwriting performance without providing proof that today's new underwriting is priced at an adequate prospective margin. The Lloyd's 2025 results report both a 1.7-percentage-point benefit from prior-year reserve releases to the combined ratio and differing underlying and headline underwriting measures. They illustrate why price adequacy must be assessed using the correct loss period and contract population.

Inflation enters through multiple channels. General consumer-price inflation may be a poor proxy for building reconstruction costs, vehicle parts, specialized labor, health services, legal expenses, or court awards. Wage and medical inflation can develop differently from a national CPI series; replacement values may be updated on a schedule. Increased asset values enlarge both insured sums and the economic consequences of catastrophe loss. Higher nominal sums insured can increase written premiums even when rate per unit remains flat. Claims-cost inflation can therefore raise the actuarially required premium while price competition pushes the observed renewal index down.

Profit measures and an accounting example

The combined ratio is conventionally the ratio of incurred losses, loss-adjustment expenses, and underwriting expenses to earned premium, subject to accounting-basis differences in expense denominators and treatment of dividends. A 100% value indicates underwriting break-even before investment results under a consistent convention. The NAIC first-half 2026 report records a United States property-casualty combined ratio of 92.5% for the six months ended June 30, 2026, versus 96.4% in the corresponding prior-year half, alongside a $38.6 billion underwriting gain. Those are industry aggregate half-year accounting results, not rate changes for a representative commercial contract or a claim that every line was profitable.

For a deliberately hypothetical portfolio with earned premium of $100 million, incurred losses and adjustment expenses of $72 million, and underwriting expenses of $25 million, the combined ratio is (72 + 25)/100 = 97%. If expected losses for the next comparable underwriting cohort rise 12% while the future premium for identical exposure increases only 4%, the simple projected loss amount becomes $80.64 million and prospective earned premium $104 million; if expenses for illustration remain $25 million, the forward ratio is ($80.64 million + $25 million)/$104 million = approximately 101.58%. The explicit denominator matters, and the arithmetic shows how an apparently healthy historical ratio can coexist with prospective inadequacy. These figures are a constructed teaching example, not observed loss development.

The reverse can also occur. A past-year catastrophe depresses the current calendar-year combined ratio while current quotations already reflect new risk assessment and improved future terms. This is why reserve adequacy, catastrophe load, current rate level, and accounting earnings must each be examined before labeling the market hard or soft. The relationship between prices and results is dynamic, with cohort and valuation lags rather than one-to-one simultaneity.

The economic cost of protection

A policyholder purchases a distribution of retained and transferred losses, not an abstract rate. Let a covered claim amount be L, a deductible D, a contractual limit M, and ignore exclusions for illustration. The insurer's simple payment is min(max(L-D, 0), M), and the policyholder retains the remainder. A lower premium can accompany a higher deductible or lower M; that combination may increase the policyholder's expected total cost or tail exposure. Conversely, the same premium may buy more protection when deductibles fall and exclusions narrow.

Consider two hypothetical policies against the same $20 million loss: Contract A has a $2 million deductible and an $18 million maximum insurer payment; Contract B has a $5 million deductible and a $10 million payment limit. The covered insurer payment is $18 million under A and $10 million under B. The 44.4% reduction in the illustrative claim payment is much larger than a small nominal premium difference would convey. A proper price-of-protection measure should therefore normalize attachment, limit, insured values, peril definition, coverage territory, exclusions, and claims service before comparing renewal rates.

The central interpretive result is precise: insurance price dynamics consist of movements in the tariff or risk-adjusted rate, the quantity and quality of promises, and the insurer's expected and realized cost of fulfilling them. Each has its own data source, adjustment lag, and economic meaning.

Capital and External Financing

Why insurance capital affects the supply of promises

An insurer's capital supports the credibility of a promise whose payoff is uncertain in size and date. Capital absorbs adverse deviations of losses, reserve revisions, asset deterioration, and operational events; its required amount depends on the distribution and dependence of exposures, asset-liability risks, and the applicable supervisory regime. The economic quantity of market capacity is consequently not just the book value of equity. It also depends on capital quality, concentration, willingness to deploy it, reinsurance access, rating constraints, policyholder confidence, and management limits. Two insurers reporting equal surplus may have different economically usable capacities after catastrophe accumulation, long-tail liabilities, or investment losses are considered.

The capital-constraint literature identifies one mechanism by which prices can rise following a shock that reduces insurers' financial slack. In Gron's evidence on capacity and price-payment margins (1994), capacity decreases are associated with later margin growth. The theoretical interpretation depends upon imperfectly elastic expansion of supply: if all depleted capital could be replaced immediately at a constant cost without information, taxation, regulatory or agency frictions, a market-wide shock would not produce a persistent scarcity rent. This conditional argument is more informative than the blanket statement that insurers raise prices whenever an adverse event occurs.

Winter (1994) considers insured risks that are dependent and external equity that is more costly to obtain than internally generated funds. His model of competitive insurance dynamics derives transitional pricing effects from real financial frictions. The distinction between a transitory marginal cost of risk-bearing capital and an economy-wide market price of financial risk is crucial. The former may fluctuate sharply after correlated underwriting losses; the latter can change with interest rates and investor risk appetite even without a catastrophic insurance loss.

Demand for insurer financial quality

The quality of an insurance promise is partly determined by the probability and consequences of insurer failure. Cummins and Danzon (1997) model an insurer whose customers value financial quality alongside price. Their price, financial quality, and capital-flow analysis predicts an inverse relationship between default risk and insurance price under specified conditions and shows that the direction of price response to capital depletion can vary with demand conditions. Their firm-level tests provide evidence of a negative association between price and default risk, and report declining prices following particular mid-1980s loss shocks. This is an essential qualification to a simple shortage-of-capital story: a less creditworthy promise may command a lower price even as the insurer's cost of capital rises.

Suppose the expected discounted claim and expense cost of a standardized policy is $900, and its market premium is $1,050. If an insurer's estimated nonpayment risk increases, a cautious policyholder can reduce willingness to pay even if the actuarial loss distribution is unchanged. A competitor with better credit quality might charge $1,100 and nevertheless offer higher economic value. An observed premium spread of $50 therefore cannot be classified as a pure underwriting-cycle spread without identifying the default-risk difference and the contract's indemnity terms.

The price-quality relation has implications for empirical work. Company fixed effects remove constant issuer traits, but changes in financial strength over time require direct measurement; ratings can react after capital impairment. In addition, a portfolio selected by price-sensitive buyers may change risk composition after a rating downgrade. The observed premium of a surviving company is not a randomly assigned experiment in capital scarcity.

Capital is replenished through several channels

Capital can be rebuilt through retained underwriting earnings, investment income, new equity, subordinated instruments where eligible, strategic capital transfers, risk reduction, or reinsurance. Each mechanism has timing and cost. Retained earnings ordinarily accumulate over reporting periods. New external capital can arrive faster during a profitable hard market but is conditioned on investors' information and the terms of prospective returns. Reinsurance can make existing direct-insurance capital more productive by reducing net tail exposure, but the reinsurer requires its own capital and may revise attachment points and terms.

The 2025 global review by the International Association of Insurance Supervisors records differentiated non-life solvency and underwriting trajectories across geographic regions in its Global Insurance Market Report. Its aggregated combined-ratio series and supervisory monitoring categories are informative about broad conditions but are not interchangeable with a representative risk-adjusted quote. Regional capital standards and reporting populations differ, and the report expressly cautions against simplistic cross-jurisdiction comparisons of some solvency ratios.

The European supervisory picture also illustrates why the capital-position story must be dated and qualified. EIOPA's October 2026 Insurance Risk Dashboard reports Q2 2026 median solvency ratios of 217.6% for insurance groups and 221.4% for non-life insurers in its reported population, while the median non-life combined ratio deteriorated slightly to 95.6%. A high sector median does not demonstrate that every exposed catastrophe zone, liability layer, or individual company has abundant deployable capital. Medians also do not reveal the tail of the insurer distribution.

Investment income, interest rates, and reserves

Interest rates alter insurance pricing through multiple channels. The discount value of future claims can decline when relevant interest rates rise, while the market value of fixed-income assets may fall immediately; asset-liability duration mismatch determines the net impact. New fixed-income purchases can provide higher running yield, while valuation changes can reduce reported or economic capital. For long-tail policies, the time structure and uncertainty of claim payment are critical. For catastrophe-heavy short-tail lines, the scale and correlation of loss shocks and reinsurance costs may dominate the investment effect.

The dynamic equilibrium concept in Doherty and Kang's interest-rate adjustment study (1988) makes explicit why observed premium can lag changes in financial pricing inputs. The capital perspective complements it: even if the present value of expected claims declines, the available funding to support aggregate risk may also decline after asset losses. A forecast that treats a single benchmark bond yield as a sufficient driver of all underwriting-cycle phases overlooks the insurer's investment portfolio, claim payment schedule, and loss covariance.

A transparent capital-use illustration helps. Assume an insurer has $100 million of eligible risk-bearing capital and applies an internal constraint requiring $1 of allocated capital for each $4 of a standardized net exposure metric. The illustrative capacity is $400 million of that metric. After a $15 million capital depletion, and with the same internal multiplier, it becomes $340 million. If $10 million is raised externally, it becomes $380 million. These are arithmetic consequences of an assumed constraint, not regulatory capital ratios, actual solvency requirements, or estimates of real market capacity. The example isolates the conditional supply effect from the empirically separate question of how prices react to reduced supply.

Capital depletion need not produce synchronized hard markets

Capital can be geographically and legally segmented. A diversified insurer may substitute capacity between lines, but subsidiaries, local licensing, risk accumulation limits, reinsurer credit restrictions, and rating-agency constraints can restrict substitution. Conversely, global capital, insurance-linked securities, and new underwriting platforms may supply a line-specific market before retained earnings fully replenish the incumbent carriers' balance sheets. In the contemporary market, capital mobility is a variable to be measured rather than a binary assumption.

Claims shocks can also be anticipatory rather than realized. An updated catastrophe model, a new view of cyber aggregation, or the recognition of litigation risk can increase required risk capital before actual losses arrive. The insurance market can harden because expected tail risk has risen while accounting capital is unchanged. Conversely, an actual loss can be followed by softer prices if capital remains abundant, demand for coverage weakens, or the loss causes reassessment of the quantity of risk transferred. These implications require a combined price, credit-quality, and exposure-accounting interpretation.

Capacity and Reinsurance Cycles

Risk transfer beyond the direct insurance balance sheet

Reinsurance allows direct insurers to redistribute insured risks and modify their net retained loss distribution. The economic effect depends on the treaty form, attachment, limit, exclusions, hours clauses, reinstatement provisions, collateral, reinsurer credit risk, and price. Quota-share reinsurance affects the ceded proportion of premium and loss, while excess-of-loss arrangements may protect specified portions of tail outcomes. The resulting reduction in net exposure may support greater direct underwriting volume or the maintenance of coverage under a local capital constraint.

The IAIS's Insurance Core Principles and ComFrame, especially ICP 13 and ICP 17, place reinsurance risk transfer and capital adequacy within supervisory systems. Reinsurance reduces certain insurance exposures while creating or retaining counterparty, basis, operational, liquidity, and concentration risk. Consequently, a market's headline reinsurance limit is not automatically equivalent to the direct insurer's freely available risk capacity. Contract recovery conditions and enforceability matter.

Reinsurance prices can rise after catastrophe losses, adverse model changes, or capital withdrawals, causing primary insurers to retain more risk or transfer it at higher cost. Direct insurance prices may then adjust, but the transmission is mediated by renewal dates, preexisting treaties, cedent retention strategies, treaty structures, and competition. A reinsurer's decision to lift attachment points can move more medium-sized losses back onto direct insurers even if the treaty's nominal premium falls.

The price of a reinsurance layer

A common property-catastrophe market indicator is rate on line, defined in simple terms as reinsurance premium divided by the contract's limit. For a hypothetical layer with a $12 million premium and $100 million limit, its rate on line is 12%. If an otherwise identical layer renews for $10 million, its rate on line becomes 10%, a relative reduction of 16.67%. This computation describes price per nominal limit. To call it a risk-adjusted rate decline of equal magnitude, one must also hold attachment, portfolio risk, peril footprint, hours clauses, reinstatements, and terms sufficiently constant.

Consider two superficially similar treaties. One covers $100 million excess of $100 million; another covers $100 million excess of $200 million. The second generally attaches at a less frequent loss level under an unchanged exposure distribution. Even equal nominal rate on line does not represent equal compensation per unit of risk. In addition, the source material may report an index of normalized renewal terms rather than raw premium divided by stated contractual limit. A market-level index should be cited and described using the broker's own methodology.

The Guy Carpenter January 2026 renewal resource center describes expanded reinsurance capacity and accelerated pricing softening at the January 1 renewals. This contemporaneous broker interpretation is consistent with the broad capacity mechanism, although the published material alone does not identify an exogenous causal effect of capital on prices. Reinsurance-broker indices involve negotiated transactions and changing portfolio mixes, and individual renewal opportunities may be highly selected by catastrophe exposure and loss record.

Primary insurance and reinsurance may turn at different times

Commercial insurance and reinsurance renew on different calendars and serve different risk pools. Large direct insureds may renegotiate annually, sometimes with layers placed in different markets. Reinsurance negotiations are often concentrated on principal renewal dates and can incorporate multiple cedent portfolios and geographies. A softer global property-catastrophe reinsurance renewal can coexist with persistent hard terms for wildfire-exposed property, United States casualty excess layers, or specific cyber aggregations. The timing of primary price transmission is therefore empirically open.

Marsh's third-quarter 2026 market index records a 5% decrease in its global composite commercial renewal-rate measure, including property -10%, casualty +1%, financial and professional -3%, and cyber -4%. It reports a ninth consecutive quarter of global composite declines, while noting persistent constraint in some United States casualty layers. The index is based on Marsh's business, not a probability sample of all insurance purchases or a measure of all nationally written premiums. It establishes heterogeneity within the apparent soft market.

The 2025 Lloyd's results add another useful distinction. In the market's annual results announcement, gross written premium rose 4.2% to £57.9 billion, with reported 10.3% volume growth, a -2.4% foreign-exchange effect, and a -3.7% price contribution. The figures are sourced from Lloyd's reporting and should retain their stated component conventions. Rising gross written premium here did not mean that price was increasing. In its first-half 2026 results, Lloyd's reported 6.9% growth in written premiums to £34.7 billion despite a -6.7% market-wide price change. This combination is an especially direct warning against inferring the direction of a rate cycle from premium volume alone.

Catastrophe risk, liability risk, and the geographical allocation of capital

The temporal pattern of catastrophe losses makes reinsurance an important transmission channel between disaster experience and prices. A large disaster can impair capital, prompt model revision, and increase demand for ceded protection. But an isolated catastrophe does not necessarily produce a worldwide hard market, because available capital, prior reserves, existing coverage, catastrophe distribution, risk-adjusted pricing, and investment supply vary. A single global indicator can mask both sharply hardened high-risk locations and falling prices elsewhere.

Liability insurance requires a different analytical timeline. Adverse court judgments, social inflation, long-tail reserve strengthening, and litigation finance can change expected loss costs gradually and with substantial estimation uncertainty. The NAIC's 2026 first-half industry assessment notes continuing pressure in commercial liability even as property-related experience improves. An analyst should not force the casualty trend onto the same phase as a property-catastrophe reinsurance index merely because both belong to the broad non-life sector.

Agricultural markets are affected by seasonal weather correlation, crop prices, public programs, and basis risk; marine and aviation lines can respond to concentrated hull, war, logistics, and liability shocks. Emerging risks and catastrophe models can change the measured distribution of insured losses before sufficient observed claims data exist to estimate a classical time series. Cross-market comparability therefore requires specifying the unit of economic protection and each line's risk-and-contract technology.

The limits of global capital aggregation

Insurer and reinsurer capital can replenish through retained earnings, institutional investors, and alternative capital, but international risk-transfer networks introduce interdependence. Counterparty concentration, correlated portfolios, common catastrophe models, and collateral calls may reduce the actual diversification obtained. Well-capitalized global markets can still withdraw from a tightly defined exposure if the expected return on that exposure is poor. Insurance-market capacity is an allocation decision under uncertainty, not a mechanically fixed multiple of industry surplus.

Supervisory data reinforce this distinction. The IAIS 2025 global monitoring evidence describes profitability and solvency by region, with important differences in business mix and risk profile. The OECD 2026 cross-jurisdictional findings document continuing premium growth and generally favorable underwriting performance in 2025 alongside changing insurer reinsurance retention. None of these summaries constitutes transaction-level evidence that an increase in solvency capital caused a particular renewal discount. That causal proposition requires additional identification.

Empirical Cycle Tests

The identification problem

A convincing empirical account must decide whether the target is a cycle in premium rates, underwriting margins, loss ratios, capacity, contractual restrictions, or the financing cost of insured risk. Each variable differs in sample coverage, timing, and accounting treatment. The familiar evidence of repeated peaks and troughs is a starting description. Causal inference additionally requires a credible counterfactual: how would price or capacity have changed without the hypothesized shock, after accounting for expected claims, demand, investment returns, competitive entry, and contract design?

Historical tests show the difficulty. Gron's four-line analysis of capacity and margins compared predictions of capacity and arbitrage interpretations, while her industry time-series evidence examined lead-lag relations. Choi, Hardigree, and Thistle's six-model comparison identified specifications consistent with historical insurer data. Higgins and Thistle's regime-switching evidence found different dynamics under constrained and unconstrained capacity conditions. These studies provide substantively important evidence but do not eliminate confounding from changing regulation, portfolio composition, economic shocks, or measurement practice across long samples.

A price-cost margin can rise because premium rates increase, because expected costs fall, because reserve releases lower reported losses, or because the mix of risks written changes. It can also rise if high-cost risks cease to be insured, causing the observed insured population to become safer. This latter selection effect is especially relevant when market hardening includes nonrenewal and withdrawal. An analysis using only policies that remain on the books cannot estimate the full market impact on accessibility.

Defining a reproducible panel

An appropriately structured study would identify at least the insurer or syndicate, underwriting unit, product line, jurisdiction, risk class, renewal date, and exposure unit. For a comparable contract, it would record quoted and bound premium, insured value, policy limit, deductible, attachment point, covered perils, exclusions, claims-handling conditions, and policy term. It would separately record accident-year ultimate-loss projections, paid and incurred loss development, and calendar-year result. On the financing side it would track economic and regulatory capital, reinsurance purchases, net retention, asset duration, realized and unrealized investment gains, credit quality, and capital transactions.

The distinction between quote and bind is essential. Quoted risk is not the same population as accepted business; an insurer can tighten underwriting by refusing particular accounts without changing the rate on those it accepts. Published broker renewal indices reflect observed placements with their own client population and methodology. Marsh explicitly states that its 2026 renewal-rate figures reflect the segment mix of its client portfolio. A national statistical survey of direct gross premiums, such as the OECD's non-life market tables, measures a different population and target variable.

The most useful panel would retain unsuccessful requests and withdrawn exposures, because availability is an outcome rather than missing data. It would also preserve rejected quotes, claims-history information available at quotation, and the exact timing of new risk-model releases or supervisory actions. Without these data, observed pricing may partly reflect endogenous selection. Insurer-by-quarter aggregate data, while valuable, cannot fully reconstruct the value or accessibility of identical contracts.

Persistence, periodicity, and spectral claims

Statistical detection of a cyclical component depends on sample frequency and length, detrending method, seasonality, aggregation, structural breaks, and error structure. An autoregressive coefficient pattern can imply a damped oscillatory response without proving a permanent self-sustaining clock. The characteristic roots of an autoregressive process can be complex, while an unrelated regime break may generate a similar periodogram peak over a finite sample. Aggregating geographically asynchronous series can either blur genuinely local cycles or create spurious national regularity.

Consider an illustrative autoregression in a deseasonalized underwriting-margin series, m_t = a + phi1 m_(t-1) + phi2 m_(t-2) + u_t. An oscillatory impulse response arises for certain coefficient configurations, but the estimated period depends on both coefficients, the measurement frequency, and parameter stability. The paper by Cummins and Outreville (1987) discusses historical forecasting and cycle findings, including earlier autoregressive interpretations. An analyst should report uncertainty around any estimated periodicity, conduct out-of-sample validation, and compare stationary-cycle, structural-break, and regime-switching alternatives rather than label any visible wave a stable cycle.

A further issue is real versus nominal trend. Inflation can give nominal premiums a rising trend while the rate per unit of real exposure follows another path. Written premium may also experience seasonal spikes at annual renewals. Frequency-domain results should therefore be supported by exposure-adjusted price and cost evidence. When a series changes accounting regime or geographic composition, stable parameters cannot simply be presumed.

Causal tests of the capital channel

The capital mechanism predicts conditional effects. A negative capital shock, with demand and expected losses otherwise comparable, should constrain supply or raise the marginal price of credible risk-bearing capacity when replenishment is costly. Stronger effects may occur among insurers with thin excess capital, high catastrophe concentration, less access to parent capital, or restricted reinsurance. But the same catastrophe that reduces insurer capital also increases expected future claims and may change policyholder demand. A naïve regression of postcatastrophe rates on catastrophe losses mixes several mechanisms.

A stronger design would separate the insurer's ex ante capital exposure to an event from the insured risk's exposure to that event. For example, a catastrophe can impair a diversified insurer through a different geographic portfolio from that of a comparison account. With careful controls, that variation may help distinguish balance-sheet spillovers from risk-specific repricing. Yet the exclusion restriction remains demanding: catastrophe exposure can change reinsurer prices, ratings, or managerial appetite and may correlate with risk choices. Instrumental-variable design must document these channels, while event studies must test pretrends and allow for anticipation.

Another approach is to examine an exogenous change in permissible capital treatment or collateral requirements across otherwise comparable insurers and lines. Such a design can estimate the effect of a particular regulatory rule under its local institutional conditions; it need not identify a universal underwriting-cycle elasticity. Similarly, difference-in-differences around a catastrophe-model revision might measure repricing if the treated and comparison risks plausibly followed parallel trends and if insurers did not change the insured sample before the announcement. Negative controls and specification sensitivity are part of the evidence, not optional decoration.

Testing adjustment lags and financial-quality effects

The rational partial-adjustment hypothesis makes predictions about the speed of price convergence to a changing target. An empirical study needs a defensible estimate of target risk-adjusted price. This involves expected discounted losses and expenses, contract terms, capital charges, and exposure composition. The estimated target should not be inferred exclusively from the same observed premium series whose adjustment is under examination, because that construction can manufacture apparent convergence.

Financial-quality explanations have a different prediction: a deterioration in credit quality may reduce the price customers will pay for an otherwise similar contractual promise. Cummins and Danzon's firm-level price and default-risk findings provide an historical motivation for joint estimation of price, demand elasticity, insurer quality, and capital flows. The identification problem is simultaneous determination: more competitive quotes alter premium income, insurer risk mix affects both capital adequacy and customer demand, and a rating downgrade may follow previously anticipated losses.

Firm-level designs can include time-varying quality measures, but these may incorporate the same information as pricing. Where available, pre-shock ratings, public capital-raising events, insurer-level market share, policyholder retention, and bound transaction prices can help untangle mechanisms. The contribution of the method lies in coherent joint interpretation, not in a single mechanical sign on a capital ratio coefficient.

Data-source audit: a contemporary worked comparison

Four published observations illustrate why measurement must precede synthesis. First, the OECD reports 8.7% nominal non-life premium growth in 2025 across its reporting jurisdictions, averaging country observations under its reporting method. Second, the NAIC reports a 92.5% first-half 2026 combined ratio for United States property-casualty insurers. Third, the Marsh Q3 2026 commercial renewal index reports global -5% composite rate movement for its client segment mix. Fourth, Lloyd's reported a first-half 2026 price decline of 6.7% and positive growth in gross written premium. These are simultaneously plausible descriptions because they cover different time windows, accounting units, coverage, and insurance populations.

No legitimate algebra combines these four percentages into a universal price elasticity, profit margin, or stage of a world underwriting cycle. They are observations within a documentary synthesis. A causal or global statistical estimate would require the underlying records, standardized definitions, matched exposures, and explicit model uncertainty. The available publications support a bounded conclusion: price reductions in substantial commercial segments and reinsurance markets in 2026 coexist with strong reported profitability in some wider non-life populations and continuing hard conditions for selected risks.

Scientific limitations and reproducibility

The present dossier does not estimate original coefficients, possess individual renewal quotations or insurer supervisory microdata, or conduct a systematic full-text meta-analysis of the entire literature. Some classic journal results have been reviewed through publisher abstracts and archived scholarly records, while Cummins and Outreville's original article was inspected in archival full text. Regulatory, market, and supervisory facts are checked in the directly cited official reports. Sources differ in methodological transparency, population selection, data access, vintage, and financial-reporting convention. These limitations constrain claims about universal effects and exact decomposition, even as they allow substantive comparison of documented mechanisms and observations.

A reproducible empirical extension would publish a source inventory, data dictionary, common exposure definitions, jurisdiction and line classifications, premium and loss development vintages, selection rules, reproducible code, alternative specifications, and falsification tests. Its central aim would be to estimate the conditional contribution of each mechanism rather than to fit a universal cycle to an insufficiently specified average.

Competition and Supervisory Effects

Competition shapes price and access together

Competition in insurance markets acts through prices, contract terms, insurer financial strength, acquisition service, underwriting appetite, and willingness to supply capacity. When risk is uncertain and claims may be paid years later, competitors are differentiated even where policy labels are identical. Their capital condition, risk models, distribution systems, reinsurance, geographic portfolios, and credibility differ. A policyholder's choice of insurer therefore allocates default and contract-enforcement risk as well as premium expense.

Competitive entry can accelerate softening when new firms, delegated underwriters, or alternative capital perceive attractive returns. During a hard market, available incumbents can earn higher margins if risks cannot be transferred elsewhere at the same cost. But new entry involves approval, systems, expertise, distribution relationships, and funding; it may be concentrated in attractive sectors. A nominal abundance of industry capital is consistent with localized supply shortages or selective exclusions in high-risk areas. Market power and capital scarcity can also coexist; the same observed markup may reflect costs of financial backing, market concentration, information frictions, or a change in the loss distribution.

The comparison of alternative models by Choi, Hardigree, and Thistle (2002) is a reminder that several economic explanations can imply closely related observed pricing equations. Competition policy therefore requires identification of the underlying mechanism before attributing cyclical markups to collusion, irrational price cutting, or scarcity rents. A public allegation of market power should not rest on a historical chart of underwriting ratios alone.

Consumers bear the cost of nonprice tightening

A hardening market can change welfare even where the average rate index moves only slightly. Higher deductibles increase retention at moderate losses; narrower limits increase uncovered tail risk; withdrawals reduce the option of purchasing any cover. Increased documentation, inspections, risk engineering and preventive requirements may improve expected loss outcomes but also increase transaction costs and impede access for small buyers. Differences in premium affordability can affect investment, housing, entrepreneurship, and recovery from catastrophe.

A total-cost comparison should distinguish expected premium, expected retained loss, financing and liquidity costs from extreme retained loss, contract uncertainty, and counterparty quality. An insurer may offer a lower premium by reducing coverage; the apparent price saving cannot be treated as a welfare gain without evaluating the changed distribution of consumer outcomes. A policyholder may rationally accept a higher premium for a better-capitalized insurer or broader policy. The response to price movement is conditioned by risk aversion, liquidity, information, and available alternatives.

Insurance scarcity also has external effects. If property owners remain uninsured, disaster recovery may require additional household debt, public assistance, or loss-absorbing mechanisms. If insurers condition renewal on verified mitigation, those measures may reduce expected loss, though the distribution of mitigation costs may fall unevenly across wealth and property classes. A social welfare evaluation requires measuring who retains risk, who pays for prevention, and which exposures disappear from the insured pool.

Solvency regulation and risk-sensitive pricing

Solvency supervision affects the capital available to back contractual promises and the incentives to accumulate or transfer exposure. The IAIS capital adequacy and reinsurance principles articulate international supervisory expectations, including the management of reinsurance and risk-bearing capital. These principles do not dictate one world premium rate or identical national regulatory requirements. Their actual implementation depends on the domestic supervisory framework and its legal authority.

Risk-sensitive capital charges may make underwriting capacity more expensive for concentrated, correlated, long-tail, or poorly modeled risks, which can contribute to higher risk-adjusted rates or tighter terms. They can also strengthen claims-paying security and make future capacity less vulnerable to financial distress. Lower minimum capital charges might appear to increase capacity in the short term while exposing policyholders to greater default risk after a correlated event. The relevant welfare tradeoff includes insurer solvency, pricing adequacy, coverage availability, and the costs of capital.

The IAIS 2025 monitoring report and EIOPA's October 2026 risk dashboard supply complementary supervisory views. The former aggregates participating jurisdictions and broad business trends; the latter reports specific Solvency II-based European data. Their populations and ratios should be kept distinct when evaluating international underwriting capacity. A regulatory solvency ratio is a prudential measure under defined valuation and capital rules; it is not itself a direct observation of a policyholder's next renewal price.

Rate regulation, information, and temporal adjustment

Insurance rate regulation can alter the timing of rate changes relative to claims-cost information. Prior approval, use-and-file, file-and-use, and other approaches differ in jurisdiction and line; the economic effect depends on legal details, administrative processing, competition, and the pace of inflation. When insurers cannot adjust adequately to projected loss costs, capacity can contract through reduced appetite or nonrenewal. When filing constraints are weak or absent, other frictions, such as annual contract duration and slow claims development, still induce adjustment lags.

The partial-adjustment analysis of Doherty and Kang (1988) and the international study by Cummins and Outreville identify institutional timing as a potentially important mechanism. They do not license the assertion that regulatory delay always causes a cycle or that removing price oversight always raises welfare. Oversight may address unfairly discriminatory rates, inadequate pricing, consumer information, or solvency concerns. An evaluation needs a precise legal setting, credible comparison group, and observed effect on price, coverage, claim costs, and entry.

Information quality is another policy-relevant determinant. Supervisors can require capital reporting, reserving governance, and insurer disclosure, while market participants may improve risk classification, claims experience, and catastrophe modeling. Better information may stabilize adjustments by reducing forecast error; it may also disclose previously underrecognized risk and initially produce large price increases or coverage withdrawals. The direction of the immediate price effect cannot be inferred from the general desirability of better information.

Market monitoring should use multiple indicators

A practical supervisory and scientific dashboard should preserve several independent indicators: standardized renewal-rate movements, written and earned premium volume, policy count and nonrenewal rate, limits and deductibles, insurer or reinsurer lines signed, accident-year loss projections, reserve development, combined ratios, net reinsurance retention, insurer capital and solvency, credit quality, and investment results. The indicators should be stratified by business line, peril, geography, insurer type, and market segment, and dated by the relevant loss and policy cohort.

One indicator can offset another. A low combined ratio may reflect favorable catastrophe experience or reserve releases, while risk-adjusted rates have already fallen. A high regulatory solvency median may mask difficult insurance availability in a hazard-prone region. An increase in direct premium may reflect rising property values or medical cost inflation rather than an increase in economic protection. The joint interpretation is more informative than any single reported cycle chart.

For example, the Lloyd's 2025 reporting combines strong underwriting results, positive premium volume, and a negative reported price contribution. The Marsh third-quarter 2026 commentary describes both widespread softening and continued United States casualty constraints. The NAIC's first-half 2026 results show an improved aggregate combined ratio alongside worsening direct loss experience in some liability lines. A multi-indicator interpretation accommodates all three, while the assertion that the world is in one uniformly soft phase would discard material evidence.

Policy implications and boundaries of the interpretation

The insurance underwriting cycle is a governance challenge because prices and contract terms respond to uncertain future loss and constrained financial capital. Supervisory objectives include adequate claims-paying resources, credible risk disclosure, orderly market functioning, and protection against practices that undermine policyholder security. Policy responses should target demonstrable failures: capital weakness, lack of risk data, contractual opacity, barriers to entry, correlated accumulation, reserve inadequacy, or risk that is difficult to pool privately. Different failures call for different interventions.

For governments contemplating subsidies or residual market mechanisms, the relevant question is whether private premiums exceed an affordable threshold because of temporary capacity scarcity, structurally high expected losses, information problems, concentrated market power, or long-term uninsurability at socially acceptable prices. Each diagnosis implies a different distribution of costs and incentives. Subsidizing a rate without addressing the physical concentration of catastrophe risk can shift the loss burden to taxpayers; providing reliable risk information and supporting mitigation can change the underlying risk itself. Any conclusion about a specific policy must take account of local law and public finance, topics beyond this international comparative dossier.

For insurers and reinsurers, the analysis supports disciplined separation of prospective and historical profitability, forward-looking risk accumulation, reinsurance effectiveness, financial strength, and contract value. For buyers, it supports comparing actual retained loss and insurer performance rather than focusing solely on nominal rate change. For researchers, it demands a combination of historical theory, transparent accounting, and credible identification of shocks, not a narrative that treats every renewal-rate turn as proof of a single cause.

Conclusion

Insurance underwriting cycles arise within an economic system that prices contingent future liabilities using incomplete information and finite, costly risk-bearing capital. Historical scholarship establishes several credible mechanisms: adjustment delays and rational expectations, dependence among insured risks, temporary capacity constraints, costly external financing, and policyholder valuation of insurer financial quality. The literature also documents empirical heterogeneity and competing interpretations. The evidence supports investigation of conditional mechanisms, rather than the assumption of an immutable cycle length or a uniform world-market phase.

The analysis yields five source-grounded conclusions. First, the object being measured determines the meaning of a cycle. Renewal rate, written premium, combined ratio, capital, reinsurance rate on line, and coverage availability are distinct. Second, claims shocks and reserve revision are transmitted through timing and valuation; the underwriting year and the accounting year should not be conflated. Third, capital depletion can raise the marginal price of insurance capacity when replenishment is costly, while deterioration of insurer financial quality can also depress customer willingness to pay for an exposed promise. The net observed price effect depends on both channels and on contract and demand conditions. Fourth, reinsurance, alternative capital, and geographic substitution transmit and sometimes buffer supply shocks, but capital remains segmented by risk and contractual terms. Fifth, the measurement and causal standards for underwriting-cycle research must confront selection, endogenous capacity, changes in coverage, and differences in institutional regimes.

The 2025–2026 institutional evidence illustrates these results without implying a newly discovered causal coefficient. OECD non-life premium growth, NAIC accounting profitability, Lloyd's reported volume and pricing decomposition, EIOPA solvency measures, IAIS global monitoring, and Marsh commercial renewal-rate declines all address different populations and economic objects. Their apparent differences become analytically useful once time, geography, portfolio composition, and denominators are preserved. Contemporary softening in substantial commercial sectors therefore can coexist with profitable aggregates and persistent hard conditions for selected casualty or catastrophe exposures.

The original analytical contribution of this monograph lies in reconciling those distinct measurements, identifying the causal ambiguity of their movements, and specifying the evidence required to discriminate among explanations. It creates no new named theory, proprietary market dataset, or original regression estimate. Its conclusions are bounded by the directly checked research and institutional sources. A mature science of underwriting cycles advances through matched exposure-level prices and terms, dated risk and capital information, cross-jurisdictional institutional comparison, and empirical designs that can distinguish shock transmission from simultaneous changes in expected losses, coverage, and insurer quality.

References

  • Choi, Seungmook, Don Hardigree, and Paul D. Thistle. 2002. “The Property/Liability Insurance Cycle: A Comparison of Alternative Models.” Southern Economic Journal 68 (3): 530–548. Publisher article and DOI.

  • Cummins, J. David, and Patricia M. Danzon. 1997. “Price, Financial Quality, and Capital Flows in Insurance Markets.” Journal of Financial Intermediation 6 (1): 3–38. Publisher article record.

  • Cummins, J. David, and J. François Outreville. 1987. “An International Analysis of Underwriting Cycles in Property-Liability Insurance.” Journal of Risk and Insurance 54 (2): 246–262. Original archived article.

  • Doherty, Neil A., and Han Bin Kang. 1988. “Interest Rates and Insurance Price Cycles.” Journal of Banking & Finance 12 (2): 199–214. Publisher article record. DOI: 10.1016/0378-4266(88)90035-0.

  • European Insurance and Occupational Pensions Authority (EIOPA). 2026. Insurance Risk Dashboard: October 2026. EIOPA-BoS-26-495; Q2 2026 quarterly and 2025 year-end data. Official online dashboard.

  • Gron, Anne. 1994a. “Capacity Constraints and Cycles in Property-Casualty Insurance Markets.” RAND Journal of Economics 25 (1): 110–127. Scholarly bibliographic record.

  • Gron, Anne. 1994b. “Evidence of Capacity Constraints in Insurance Markets.” Journal of Law and Economics 37 (2): 349–377. Publisher DOI record.

  • Guy Carpenter. 2026. January 1, 2026 Renewals: Reinsurance Buyers Experience Market Softening as Reinsurers Grow Capital Following Strong Returns. Renewal resource center. Official renewal analysis.

  • Higgins, Matthew L., and Paul D. Thistle. 2000. “Capacity Constraints and the Dynamics of Underwriting Profits.” Economic Inquiry 38 (3): 442–457. Publisher DOI record.

  • International Association of Insurance Supervisors (IAIS). 2024. Insurance Core Principles and Common Framework for the Supervision of Internationally Active Insurance Groups. December 2024 edition; ICP 13, ICP 17. IAIS authoritative principles.

  • International Association of Insurance Supervisors (IAIS). 2025. Global Insurance Market Report (GIMAR) 2025. December. IAIS global monitoring report.

  • Lamm-Tennant, Joan, and Mary A. Weiss. 1997. “International Insurance Cycles: Rational Expectations/Institutional Intervention.” Journal of Risk and Insurance 64 (3). Scholarly article record.

  • Lloyd's. 2026a. “Lloyd's Market Delivers Strong Full Year Performance; Very Strong Balance Sheet; Increased Capital.” March 19; results for 2025. Official annual-results announcement.

  • Lloyd's. 2026b. “Lloyd’s Market Delivers Solid First Half Performance, Despite Softening Rate Environment and Rising Geopolitical Tensions.” September 3; results for first half 2026. Official interim-results announcement.

  • Marsh. 2026. Global Insurance Market Index: Third Quarter 2026. Global commercial client renewal-rate index, with methodology and segmentation disclosures. Official market index.

  • National Association of Insurance Commissioners (NAIC). 2026. U.S. Property & Casualty and Title Insurance Industries—2026 First Half Results. Industry Analysis Report; six months ended June 30, 2026. NAIC statistical report.

  • Organisation for Economic Co-operation and Development (OECD). 2026. Global Insurance Market Trends 2026. Paris: OECD Publishing. Published November 2, 2026. DOI: 10.1787/c8be6585-en. Chapter 2, “Premium Growth and Underwriting Performance in the Non-life Insurance Sector.” Official report chapter.

  • Winter, Ralph A. 1994. “The Dynamics of Competitive Insurance Markets.” Journal of Financial Intermediation 3 (4): 379–415. Publisher article record. DOI: 10.1006/jfin.1994.1011.