Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Health Insurance Rate Development and Medical Trend

Abstract. Health insurance rate development is the prospective estimation of the financial resources required to cover a specified population, medical benefit contract, and period of risk. Medical trend is the forecast change in the cost of providing insured care after accounting for the quantity of care, negotiated unit prices, service mix, clinical intensity, pharmaceutical innovation, and the insured population. This monograph investigates the relationship between those two problems by integrating exposure measurement, incurred-claim completion, morbidity and utilization models, service-specific price-volume decomposition, population risk change, benefit design, provider contracting, risk adjustment, expense allocation, capital margins, regulatory constraints, and uncertainty. A fully specified synthetic numerical example carries an experience-period medical cost through a two-year projection, explicit population and benefit factors, risk-transfer cash flows, and gross-premium construction. It then tests unit-price assumptions and separates actuarial indication from legally permitted rates and observed market premiums. The comparative analysis examines the United States' single-risk-pool and filing methodology, Australia's private health insurance premium approval process, and the broader health-financing institutions described by the World Health Organization and OECD. It distinguishes health-care inflation, insurer-paid claims trend, morbidity change, and premium trend as separate empirically identifiable quantities. The central conclusion is that a defensible rate indication is an internally reconciled forecast of insured cash obligations under a clearly specified contract and population; changes in premiums can be economically justified only after the interaction between medical claims, redistribution, benefits, expenses, and financing is made explicit. Numerical examples are constructed analytical illustrations, not insurer observations, filed rates, or jurisdiction-specific advice. The study develops an original synthesis of existing actuarial and health economics literature without proposing a new theory, doctrine, or actuarial standard.

Keywords: Health insurance; actuarial ratemaking; rate development; medical trend; morbidity; medical utilization; unit price; case mix; risk adjustment; risk equalization; incurred claims; completion factors; premium per member per month; experience rating; credibility; health care inflation; insurance regulation; health financing; provider contracting; cost sharing; pricing uncertainty; sensitivity analysis; medical loss ratio.

Introduction

Health insurance prices financial protection against medical expenses whose occurrence, intensity, provider, timing, and monetary valuation are uncertain when the premium is set. Rates govern access to a collectively financed medical benefit and also the insurer's capacity to settle claims. Health insurance ratemaking therefore requires a translation among three objects: a medical event or service, an insurer's contractual obligation for that service, and the future member or premium exposure over which the obligation is financed. The translation changes across public systems, commercial group coverage, individual regulated markets, and supplemental insurance.

The health regulatory filing standard, ASOP 8 (Actuarial Standards Board [ASB], 2014) provides a particularly instructive account of prospective health filings: experience, population, product design, medical and provider conditions, expenses, regulation, and forecast assumptions are each material. Its jurisdiction and professional scope must remain explicit. In the United States, the federal 2027 Unified Rate Review Instructions (Centers for Medicare & Medicaid Services [CMS], 2026) implement a specific filing framework for applicable individual and small-group products. Neither document supplies a universal mathematical rate formula for all global health coverage.

Four different changes are often loosely called "medical trend." First, a given service can become more expensive at an unchanged treatment intensity. Second, individuals can receive a different number of services. Third, the relative distribution of treatments can move toward intensive settings, new procedures, or costly medicines. Fourth, the people insured can become older, sicker, or otherwise different. The first three contribute to service and medical cost trend under a fixed population; the fourth is primarily population risk change and must be identified separately when the purpose is comparison at constant population mix. Product benefits and member cost sharing introduce an additional transfer between total care expenditures and insurer payments.

The primary analytical questions are: What is the correct exposure and experience basis? How should morbidity and utilization be distinguished from service prices and intensity? How should risk adjustment, cost sharing, claims maturation, and expenses enter a prospective estimate without duplication? What constitutes an adequate rate, and what does its accuracy mean when health regulations control permissible factors and the distribution of premiums? The answers require equations whose terms have operational definitions, historical data whose development is understood, and external evidence matched to claims of appropriate scope.

The work uses actuarial accounting identities, comparative institutional analysis, targeted review of primary actuarial and regulatory documents, empirical research on care demand and insurance markets, and independently constructed arithmetic examples. Professional standards are treated as professional obligations within their stated application. Peer-reviewed studies are evaluated as evidence about their own settings, not as automatic forecasts for a different insured population.

Health Rating Exposure and Population

The unit of exposure

For a short-duration medical plan, the member-month is a natural denominator. Let E denote the total eligible member-months in the observation period, C the ultimate covered incurred claims attributable to that period, and c = C/E the paid-benefit claim cost per member per month (PMPM). For 12 months of continuous membership one person contributes 12 member-months; a person covered for four months contributes four. For a group plan quoting employee or contract premiums, family composition must also be modeled because a contract-month and a member-month are different units. Rates per employee, household, policy, and covered life are connected through the expected covered-life distribution, not through a universal multiplier.

A correct exposure record identifies when eligibility starts and ends, whether newborns are included, the effect of retroactive enrollment and termination, calendar versus policy year, duplicate member identifiers, and benefit responsibility during changes of insurer. Average head count multiplied by 12 approximates exposure only if entry, exit, and measurement conventions are consistent. A changing population can distort claims PMPM when average headcount is substituted for earned member-months.

If service categories are indexed by s, define C_s as ultimate insurer-paid service costs and c_s = C_s/E. The elementary reconciliation is c = sum_s c_s. For every amount, one must declare whether it is allowed provider charges, an insurer's negotiated allowed amount, paid net of deductible or copayment, cash disbursement, incurred claim cost, or a capitation obligation. Adding these concepts as though they had identical measurement bases creates double counting. The ASOP 23 data quality guidance (ASB, 2016) addresses relevance, sufficiency, review and limitations of source data, while distinguishing appropriate review from exhaustive auditing.

Risk population and time alignment

Let x denote a defined cell of demographics, morbidity, geography, benefit eligibility, or other lawful classification, and let w_x,t be its share of member-month exposure in period t. The average expected insured medical cost is c_t = sum_x w_x,t m_x,t, where m_x,t is expected medical cost for cell x under the same payment and benefit definitions. This equation separates the amount spent for comparable risks from the changing distribution of risks.

A decomposition of change can be performed as c_1 - c_0 = sum_x w_x,0(m_x,1 - m_x,0) + sum_x (w_x,1 - w_x,0)m_x,1. The first term holds the prior population mix constant and isolates within-cell cost change. The second uses current cell costs to value the membership shift. The ordering is a convention; reversing the period used for weights produces a different but equally reconcilable index-number decomposition. For exact multiplicative analysis use a declared Laspeyres, Paasche, Fisher, or other index methodology rather than combining approximate percentages without explaining residual interactions.

Insurance morbidity is not synonymous with biological disease prevalence. A morbidity index can measure diagnoses, clinical severity, future cost, hierarchical condition categories, or resource consumption. Changes in coding intensity can affect a diagnosis-based risk score without changing physiological illness. Likewise, a health insurer's insured mix reflects enrollment incentives and eligible population rules as well as morbidity. The ASOP 12 risk classification standard (ASB, 2011) helps clarify the professional nature of classification; statutory restrictions on health rating may prohibit using variables that can still matter for pooled cost forecasting.

Selection and the projection population

Prospective rates must forecast who will remain insured after the rate and plan changes. A rate increase can change enrollment, and selection may change the average insured cost. This is an endogenous pricing problem. The historical average cost of an old cohort is not necessarily the expected cost of purchasers at the new price. Einav, Finkelstein, and Cullen's price-variation analysis (2010) illustrates how demand and selection can jointly affect insurance-market welfare. Their empirical application is an insurance market study, not a direct estimate of any contemporary health carrier's premium elasticity.

The model therefore defines the priced population at the future coverage date, including forecast enrollment distribution and the institutional rules that constrain it. Premium adequacy and distributional effects depend on whether a risk adjustment mechanism reimburses predictable differences, whether enrollment remains open, and whether subsidies stabilize participation.

Morbidity and Utilization

Frequency, quantity, intensity, and covered claims

Medical claim cost can be represented through conditional expectations. For an enrollee i and service group s, let N_is denote the number of covered episodes, Q_isk the billable units within episode k, P_isk the negotiated allowed price per unit, and B_isk the insurer's contractual payment function after the benefit rules. Then individual benefit cost is L_i = sum_s sum_k B_isk(Q_isk P_isk, other episode attributes). The cohort's insurer liability is the expectation of the sum of L_i over covered members.

Under restrictive homogeneity and conditional independence assumptions, expected allowed service spending PMPM can be approximated by episode frequency times units per episode times average negotiated unit price. When episode frequency, severity, diagnosis, price, and treatment choices are associated, the correct identity is an expectation of their joint product. The product of separately estimated marginal means can be biased because E[QP] differs from E[Q]E[P] when covariance is nonzero. A transparent model identifies whether the price index holds treatment complexity constant.

The phrase "claim frequency" is ambiguous in medical coverage: one hospitalization can generate facility, physician, laboratory, imaging and medication claims; a single claim may contain numerous service lines. An observed rise in processed claim count may reflect a billing-system change. Stable analysis instead defines clinically interpretable episodes or consistent units of service and documents mapping changes. Member-months, admissions per thousand, prescription fills per thousand, equivalent service units, visits per member, or standardized episodes answer different questions.

Conditional morbidity models

Useful statistical approaches include generalized linear models for count and severity outcomes; two-part models with a probability of any utilization followed by conditional expenditure; zero-inflated models when indicated by the data-generating process; hierarchical models for provider and locality; and collective risk or compound distributions for aggregate cost. The selection among models follows the question, exposure definition and empirical fit. Disease-specific models may forecast counts of encounters, while aggregate PMPM models can conceal substitution between inpatient and outpatient care.

A Poisson model sets Var(N|x)=E(N|x) and can understate observed overdispersion. A negative binomial count model allows additional variation. For positive expenditures, gamma-log models, lognormal models, and flexible mixture or quantile approaches have different tail and retransformation properties. Zero medical expenditures and catastrophic high-cost claims frequently coexist, so modeling average cost with a single homoscedastic normal equation is a poor default. The ASOP 56 modeling standard (ASB, 2019) establishes professional considerations for selecting, using, validating, and communicating models; it does not prescribe a single universal health claims distribution.

Demand under cost sharing

The relevant service volume also depends on the patient's price at the point of care. The randomized RAND Health Insurance Experiment findings (Manning et al., 1987) document a meaningful response of care demand to cost sharing within the experimental context, including a reported approximate medical-demand price elasticity of -0.2 for its studied setting. Applying that number directly to a new specialty drug, emergency service, insurance population, or modern network would assume external validity that the study cannot establish.

The Oregon Medicaid Experiment clinical outcomes study (Baicker et al., 2013) likewise connects insurance expansion to access, utilization, and financial protection for a specific eligibility population and study window. It shows why enrollment and benefit changes can create utilization shifts that should be separately estimated from medical service unit-price inflation. Coverage changes may also produce delayed care, adherence improvements, and long-run health changes whose direction and pricing implications depend on follow-up duration.

Care substitution and treatment technology

Utilization trends may be offsetting: fewer inpatient days alongside more outpatient procedures, a greater use of generic drugs alongside greater specialty-drug treatment, and lower avoidable admissions alongside higher preventive-service use. Aggregating claims before examining service composition can turn clinically different mechanisms into the same percentage. A credible forecast specifies causal or accounting mechanisms for these movements and tests whether the forecast can actually identify them from observed data.

Clinical innovation may increase unit cost while reducing later utilization, or expand treatment to previously untreated patients. Its effect on premiums is therefore a net present or coverage-period financing question. Forecasts should state the horizon and whether savings occur inside the same rating period or are realized by other payers or future policyholders.

Medical Unit Cost and Trend

Defining medical cost trend

For service s, an interpretable projected cost change can be written as

c_s,t+1 / c_s,t = (1 + u_s)(1 + p_s)(1 + a_s)(1 + z_s),

where u_s is the utilization frequency change, p_s a negotiated like-for-like unit-price change, a_s a treatment-intensity or service-complexity change, and z_s any separately defined contractual, coding, or residual adjustment. Every factor requires a stable measurement specification, and the formula assumes the effects can be represented multiplicatively for that service group. It is a useful forecasting identity rather than an empirical assertion of causal independence. The combined change is the product minus one, not the sum of component percentages.

For example, a service group with 3% more visits and a 4% higher equivalent price has an exact 7.12% combined change if intensity and population remain unchanged: 1.03 x 1.04 - 1 = 0.0712. A simplistic 7% estimate suppresses the interaction. For a multiyear forecast with annually compounded factors, both the number of years and the start and end dates must be specified.

Prices and health spending are different quantities

The OECD's Health at a Glance 2025 price comparison (OECD, 2025) separates relative medical price levels from the volume of care across countries and employs health-specific purchasing power parities for meaningful price comparison. These cross-sectional international price-level differences are not annual insurer claims inflation rates. A country can have high prices and low price growth, or the reverse; differences in benefit scope and negotiated payments add further distinctions.

Health-care consumer price indexes, pharmaceutical list prices, net prices after rebates, provider contract escalators, national health expenditure projections, and individual insurance premium changes also measure different baskets. For example, CMS National Health Expenditure projections describe national financing aggregates with multiple payers and sources of spending, so they cannot directly replace a particular plan's medical trend. The plan requires insurer-paid allowed costs for its own coverage and provider relationships.

Contracted prices and provider mix

A hospital contract can change from per diem to diagnosis-related group reimbursement, from fee-for-service to capitation, or from a percentage of billed charges to fixed rates. Such changes alter payment timing and the definition of "unit." A valid trend study should bridge old and new payment methods through common episodes or a consistent projected claims-cost calculation. It should not apply a fee schedule escalator to costs already settled through a population capitation contract.

Changes in provider mix can occur even when all provider-specific prices are unchanged. If care shifts toward higher-priced hospitals or specialized centers, an insurer's average price per admission rises. Conversely, a narrower network may reduce average negotiated payments but change access, utilization patterns, and selection. Provider and clinical case mix deserve separate factors when material. Even with encounter-level data, the decomposition can be order-dependent because there is no unique numerical way to allocate all interaction terms to price, volume, or intensity without an index convention.

Pharmaceuticals and high-cost treatment

Prescription trend requires analysis of manufacturer prices, negotiated pharmacy payments, rebates, utilization, generic substitution, formulary controls, emerging therapies, patents and biosimilars. Gross claims and net-of-rebate expenditure differ in timing, allocation, and uncertainty. A new drug can increase member treatment options and total insured expense through additional eligible patients while replacing a more expensive treatment for others. Forecasting one aggregate drug trend from a broad inflation rate can miss these simultaneous changes.

For extremely high-cost care, exposure pooling, reinsurance terms, annual maximums, and pooling thresholds matter to the insurer's retained cost. Trend applied to gross claims before reinsurance must be reconciled to ceding amounts and recoveries. A ceded claim is still a medical event; the reinsurance settlement changes who finances it. A projection should never count the full medical claim and the reinsurance recoverable as two independent benefit costs.

Calendar and cohort comparisons

A claim incurred in the final month of an experience year may be paid several months later. A service price renegotiated midyear affects only the portion of future utilization after its effective date. Similarly, a January benefit revision may change utilization before medical trend assumptions take effect. Exact date alignment requires projecting the center of the historical exposure distribution to the center of the rating period, or modeling the monthly calendar directly. Rate filings often use an experience period, bridge period, and projection period; their durations should be visible in the actuarial memorandum. The CMS 2027 rate-review instructions describe this structured separation in their U.S. filing context (CMS, 2026).

Benefit Cost Sharing

Allowed spending and insurer liability

Let A be an episode's negotiated allowed amount and define B(A) as insurer responsibility under the contract. The function B depends on deductibles accumulated earlier in the coverage year, coinsurance, copayments, out-of-pocket limits, benefit exclusions, service category, network tier, prior authorization rules, and coordination with another payer. Aggregate insurer claims equal sum of B across members and episodes; in general, E[B(A)] cannot be recovered from E[A] by multiplying by a single fixed percentage.

An actuarial value approximation can be useful at the portfolio level if the utilization distribution and contract structure are stable. It can fail when changes to deductibles, copayments, or maximums shift patient behavior, concentrate liability among high users, or cause a different utilization distribution. A plan paying 80% of a historical allowed-cost total is not necessarily expected to pay 80% of a future total after its deductible, enrollment, and service mixture change.

Behavioral responses and benefit effects

The benefit change factor should isolate the change in insurer-paid expense under the new plan, holding other identified drivers constant. A full benefit adjustment includes both the mechanical transfer of allowed costs from insurer to enrollee and the response of utilization to the newly faced price. Manning et al. (1987) provide causal evidence for a cost-sharing utilization effect in the RAND setting. The direction of a utilization response for a specific modern insured service still depends on price salience, income, health need, clinical substitutes, access, and maximum out-of-pocket protection.

A deductible increase may reduce insurer-paid cost mechanically even with unchanged care, but higher patient cost can also lead to deferred care, changes in treatment adherence, and eventual later claims. The rating-year forecast should identify which effects are included and which extend beyond the policy period. Equity consequences arise because identical cost-sharing changes affect low- and high-income households differently.

Payment models and quality safeguards

Under a capitation agreement, the insurer's primary payment may be a predetermined per-member amount rather than fee-for-service episode reimbursements, with risk-sharing, quality adjustments, stop-loss provisions, and exclusions. A clinical utilization forecast is then relevant to the provider's cost and the sustainability of negotiated payment even when claims booked by the payer remain capitated. Global budgets similarly create a financing exposure whose observed cash payment is less directly responsive to service counts.

Under bundled payment, the provider may internalize several services that formerly appeared as separate claims. Comparing claim counts across the payment-method transition exaggerates an apparent utilization decline. Analysis must preserve a clinically consistent episode definition and measure both the payer obligation and quality/access outcomes.

Effective access and coverage

A formally insured population may face network shortages, denied services, or high cost sharing that prevents effective use. The WHO's health financing and pooling framework situates financing alongside access and financial protection (WHO, accessed 2026). A lower observed claims trend can thus reflect successful prevention and bargaining, but also foregone necessary care. Pricing interpretation requires service quality and access context rather than automatically treating lower paid claims as evidence of improved system performance.

Expenses and Risk Margins

A prospective financing identity

For a defined premium unit, let M denote expected net insurer-paid medical costs after properly signed risk-adjustment transfers and reinsurance recoveries, F fixed administrative expenses per premium unit, v a proportional expense ratio applied to gross premium, t proportional taxes and assessments, and q the indicated provision for contingency, capital financing, or surplus as a fraction of gross premium. Then the simplified gross indicated premium is

P = (M + F) / (1 - v - t - q),

provided the denominator is positive, all terms use the same exposure unit and payment period, and statutory rules permit that method of allocation. It is a transparent prospective accounting convention for the illustrations here. Other products require different timing and premium recognition conventions; profit, capital and taxation cannot be assumed to have identical regulatory meanings across jurisdictions.

Fixed and proportional expenses must be kept separate. A per-member administration budget can change with scale and service mix, while premium taxes or distribution commissions may be percentages of written or earned premium. A single “load” inserted without defining its base can be materially incorrect. Reinsurance premium, recoveries, insurer risk adjustment transfers, and regulatory assessments must be assigned consistent signs and periods.

Risk transfer and experience fluctuation

A mean expected-claims estimate does not by itself establish the probability that reserves and capital will cover adverse outcomes. A portfolio forecast can therefore evaluate dispersion, catastrophic treatment exposure, concentration in a specialty provider, high-cost claims and correlated shocks. The risk margin or financing allowance reflects the contractual capital plan, risk appetite, capital cost, and applicable regulatory requirements. It should not be interpreted as an automatic authorization to add any requested margin to a regulated consumer rate.

A distribution-based approach can use simulation, stress scenarios, and tail-risk diagnostics to evaluate claims volatility, including dependence between care inflation and morbidity shocks. For example, a new therapy can change both service eligibility and per-patient expense in the same period; modeling the two effects as independent may materially misstate tail spending. The model should report the sensitivity to correlation assumptions and the effects of reinsurance attachment points and claim aggregation.

Pricing versus reserving and accounting

Premium rate development forecasts a future coverage period, whereas incurred-claim reserves estimate payment liabilities for coverage already provided. Claims completion informs rate development but remains analytically distinct from valuation. The ASOP 5 incurred health claims standard (ASB, current publisher record) addresses estimation of already incurred health and disability claim liabilities; an insurer should not equate the reserve for that past obligation with next year's premium indication.

IFRS 17 Insurance Contracts (International Accounting Standards Board [IASB], 2017, as amended) establishes a financial reporting measurement framework, including fulfilment cash flows and risk adjustment for nonfinancial risk. Its reporting risk adjustment is an accounting measurement concept, not a universal tariff loading or the statutory U.S. medical loss ratio. In the same way, regulatory solvency capital is a constraint on financial resilience rather than a per-person claim cost.

Cash-flow timing and revised standards

Long-duration health products, claims settlements, investment income, and financing lags may require discounted cash flows and asset-liability consistency. The revised ASOP 7 life or health cash-flow standard (ASB, revised December 2025; effective June 1, 2026) addresses the relevant health actuarial cash-flow analysis where its scope applies. The contemporaneously revised ASOP 20 property/casualty cash-flow standard (ASB, effective June 1, 2026) applies to property/casualty actuarial services. This subject-matter separation matters: the word “discounting” in a prior research outline does not establish ASOP 20 as the governing health standard.

A one-year, uncomplicated illustration may omit discounting explicitly for pedagogical purposes. The omission is an assumption, not a conclusion that insurance cash-flow timing is financially irrelevant.

Prospective Rate Adequacy

From projected medical cost to indicated price

The indicated rate should reconcile: a documented baseline of ultimate experience-period cost; exposure and membership changes; projection-period unit-price, utilization, and service-mix trends; risk changes; benefit and network modifications; taxes and operating expenses; risk transfer; and the approved financial margin methodology. The process is directional: historical paid cash must first be associated with the appropriate incurred service and exposure, while future assumptions are applied to a defined prospective population and contract.

A rate may be adequate for the insurer's expected financial obligations, excessive under a legal review standard, unfairly discriminatory under permissible-classification rules, unaffordable to some households, or inconsistent with market equilibrium. These predicates answer different questions. The ASOP 8 health filing guidance explicitly discusses rate adequacy, non-excessiveness, unfair discrimination and projected loss ratio as distinct possible regulatory benchmarks (ASB, 2014).

Experience and manual rates

For established products, a subject-experience forecast gives local responsiveness but may be volatile. A manual or reference rate developed from a larger relevant dataset is more stable but can fail to reflect a small group's distinctive morbidity, provider prices or utilization. Credibility weights should reflect the estimation objective, comparability, and forecast performance, not merely a convenient member-count threshold. The ASOP 25 credibility procedures (ASB, current issuer record) provides the applicable professional framework when credibility is used.

A conventional linear blend is R = Z R_subject + (1-Z) R_manual, 0 <= Z <= 1. It is a valid weighted estimate when both inputs are constructed on compatible coverage, time, trend, exposure and benefit bases. The weight does not retroactively make incompatible data comparable. In other applications, Bayesian shrinkage, hierarchical regression or risk-adjusted pooling is appropriate. The selected estimator should be tested on prospective performance and its bias under population shifts.

Expected adequacy and tail adequacy

A premium equal to expected benefits plus operating expenses can be insufficient for a portfolio requiring capital investment and protection against tail risk. Conversely, a margin chosen without reference to capital and legal constraints can make a filing excessive or less accessible. The actuarial conclusion should state the expected underwriting result, stress outcomes, capital assumptions, and definitions of acceptable deviation. Financial soundness can also depend on the timing of claims and investment income.

Rate “adequacy” may also be assessed over multiple rating periods. A temporarily inadequate rate funded through legally permitted risk corridors, reinsurance, subsidies or capital can be sustainable under a documented financing arrangement. If those instruments expire, historical rates are no longer an appropriate starting point without adjustment. The public payer or taxpayer may bear part of the same economic risk even when the member's invoiced premium is unchanged.

Claims Experience, Completion, and Forecast Data

Paid, incurred and ultimate claims

Claims for care incurred in one period are reported, adjudicated, disputed, adjusted, and paid over time. A recent paid-claims report systematically understates incurred-period costs where unreported or unpaid claims remain. For an experience month m at valuation lag l, a completion factor k_m,l can be estimated from comparable historical development, and ultimate claims approximated as U_m = paid_m,l / k_m,l for a positive completion fraction. Choosing completion factors requires treatment of changing adjudication speed, high-cost claim reporting, seasonal treatment, provider billing practices, and operational disruptions.

For a portfolio with $9.0 million in observed paid claims and a reliable 90% completion fraction under a declared development definition, the arithmetic ultimate estimate is $10.0 million. Applying a further 10% “IBNR factor” after dividing by 0.90 double counts the missing $1.0 million. If experience is already incurred and complete, adding an incurred-but-not-reported provision again is also wrong. Estimated ultimate costs should have paid-to-incurred reconciliation and reporting-lag diagnostics.

Provider invoices and diagnosis data

Paid claims have strong accounting relevance but may have inadequate clinical granularity. Encounter records, diagnoses, authorization systems, pharmacy utilization, capitation shadow claims, stop-loss recoveries and external population data can improve interpretation. Each dataset needs a lineage record: who generated it, the observation date, coverage definition, missing fields, code version, currency, price basis, privacy status, and linkage algorithm. The audit should compare raw and transformed aggregates, outliers, enrollment denominator, duplicates, negative adjustments, and the balance between recorded payments and claim liability.

Morbidity scores can suffer from differential coding across providers. Changes in diagnosis capture after a reimbursement incentive may raise predicted risk without an equal change in actual clinical burden. The measure may remain useful for revenue transfers under the governing program, yet the analyst should not call it a pure clinical-incidence measure. The empirical separation of treatment need, diagnosis intensity, and payment incentive is part of identification.

Inflation and experience-period comparability

Prices, technology, demographics, and utilization can move during the experience period. A simple annual historical cost divided by 12 member-months can combine different monthly enrollment mixes and provider contracts. If seasonality is material, the exposure-adjusted monthly series should be inspected and projected into the planned premium period. Forecasting on a price-normalized base is often preferable where contractual unit price increases are documented. Expense and benefit changes need their own bridge assumptions.

The minimum practical evidence package for a rate indication therefore includes: reproducible earned exposure, member segmentation, ultimate claim cost, service group splits, contemporaneous provider and pharmaceutical changes, benefit assumptions, risk adjustment amounts, expense schedules, and assumption change logs. The ASOP 41 communications standard (ASB, 2010 and later issuer guidance) guides the clarity of professional actuarial disclosures.

Population Risk Adjustment and Credibility

Predictive risk and insurer transfers

Health risk adjustment redistributes financing to account for expected spending differences associated with enrollment composition. If an insurer enrolls higher-risk individuals, a risk-adjusted transfer may reduce its net insurance obligation; another insurer or public funding system finances the transfer. A predictive cost score can be used in the calculation without being an individually permitted premium factor. The distinction between risk score for payments and lawful risk factor for charging the member is fundamental.

The U.S. single-risk-pool regulation (45 CFR 156.80, reviewed 2026) defines pool-level index-rate requirements and permissible plan-level adjustments for applicable U.S. markets. It does not authorize a plan to set each consumer's rate equal to that person's expected medical claims. The 2027 HHS benefit and payment final rule (CMS, 2026) is a separate plan-year regulation whose risk-adjustment specifications must be evaluated for the applicable coverage period.

Transfers, reinsurance and the claims base

For a net payer, a positive transfer charge raises financing cost; for a recipient, a transfer receipt reduces it. A clear convention is M_net = C_gross - RA_receipts + RA_charges - Reinsurance_recoveries + Reinsurance_premium, after removing amounts already embedded elsewhere. If receivables are uncertain, the forecast must consider settlement risk. Actual transfers can depend on market-wide relative risk and premiums, creating dependence on behavior of other insurers; an insurer-only forecast should recognize that uncertainty.

Risk corridors and public risk-sharing can partly redistribute deviations of observed costs from targets. A program should be characterized by the percentage shared, threshold, period, budget neutrality and counterparty; using a generic “risk adjustment” line for every stabilization mechanism obscures the economic operation. In socially financed health systems, equalization among funds can serve a similar financing purpose under different legal rules. van de Ven and Ellis's health-plan risk-adjustment framework (2000) discusses the broader relation between risk adjustment, selection and competition.

Forecasting remaining selection

An insurer may respond to transfer rules with more attractive services for certain risk groups or adjusted provider networks, creating residual selection even where individual medical underwriting is restricted. Predicted costs and actual transfer payments therefore belong to a joint market analysis. If premium increases lead lower-risk members to leave, the forecast member-mix change can alter the needed premium again. A multi-period scenario may be required to examine this feedback. It is methodologically unsound to mechanically multiply a pooled risk-adjustment transfer estimate by every individual's morbidity factor.

Comparative Regulatory and Institutional Systems

United States: market and plan rate construction

For applicable U.S. individual and small-group ACA markets, the single risk pool under 45 CFR 156.80 requires eligible claims experience to be pooled within each relevant state market, with the regulation's permitted market-wide and plan-level adjustments. The 2027 CMS Unified Rate Review Template framework then organizes experience, projections, plan product information, actuarial memoranda, and consumer justifications. Rates are developed inside this legal architecture; the generic individual risk-score pricing identity is not a substitute for the filed methodology.

The U.S. minimum medical loss ratio provisions at 45 CFR 158.210 set 80% minimum MLR and rebate consequences in the applicable individual and small-group markets, and 85% in the large-group market, subject to the rule's qualifications. The statutory MLR uses defined numerator, denominator, adjustments and rebate calculations; it is not equal to any arbitrary accounting quotient of claims and invoiced gross premium. A synthetic illustration's simple medical-cost percentage therefore must not be described as statutory MLR compliance.

A rate increase is an actuarial price change, while a consumer's paid premium also reflects permitted rating adjustments and applicable premium support. A change in federal subsidies can materially change what members actually pay without causing an equivalent change in the filed gross premium. Conversely, gross medical cost trend can be offset by risk transfers or benefit modifications.

Australia: centrally reviewed private insurance premiums

Australian private health insurance uses a different approval architecture. The Department of Health's 2027 private health premium-round notice (Australian Government, 2026) establishes a specific annual application process, while the Australian Prudential Regulation Authority's description of its premium-round role explains its advisory and financial-sustainability contribution. This is a national institutional process, not the U.S. single-risk-pool index-rate formula.

Premium proposals may be assessed in connection with insurer finances, member affordability, health costs and value. A published average national premium increase is a weighted outcome of approved insurer changes; it is not necessarily equal to future national medical price inflation or a specific member's renewal increase. Comparisons across the United States and Australia should state what is being compared: insurer-filed gross rates, paid member premiums, premium increases, claims trends or macroeconomic health spending.

Public and social health financing

Across global public or social health insurance arrangements, revenues can arise from payroll contributions, general taxation, mandatory or voluntary premiums, transfers, subsidies, and earmarked levies. The WHO's health financing topic framework treats revenue raising, pooling and purchasing as distinct institutional functions. The price paid by an individual enrollee can be deliberately dissociated from that person's expected medical cost. In a tax-funded scheme, the annual “rate” analogue may be a capitation budget, provider payment tariff, contribution rate, or public finance allocation rather than a retail insurance premium.

Therefore the actuarial concepts of expected covered cost, utilization, provider payment, morbidity, and financial volatility are internationally transferable, while the authority to set rates, the eligible financing base, redistribution, claims-payer arrangements, and consumer price vary with the system. A globally valid research result must disclose which parts of the analysis are mathematical identities and which depend on institutional law.

International solvency and reporting standards

The International Association of Insurance Supervisors' Insurance Core Principles framework (IAIS, current online framework) gives a supervisory perspective on valuation, governance, financial resilience, and market conduct. The IFRS Foundation's IFRS 17 insurance contracts standard addresses financial reporting for entities applying the standard. Neither document replaces the detailed rate-filing law of a given jurisdiction. Distinct accounting, prudential, and premium-setting purposes should appear explicitly in the actuarial workpaper.

Integrated Reproducible Numerical Case Study

Scope and units

This original case is an annual renewable medical policy with a stable benefit basis in 2025 and a projected 2027 rating year. The numerical example is synthetic. Every starting cost and trend assumption is an invented demonstration value chosen to illustrate the method; none is attributed to an actual insurer, government filing, national health system or observed medical price forecast. PMPM is U.S. dollars of insurer-paid covered claims per eligible member-month. We assume fully matured 2025 incurred claims, no separate claims completion factor, no change in exchange rates, and no discounting over the illustration's short coverage horizon.

The member-weighted baseline in 2025 is $480 PMPM, partitioned into hospitalization $165, physician and other professional care $95, prescription medicines $110, and other covered services $110. The annual assumption for each group comprises utilization growth u, unit-price growth p and case/service-mix or intensity growth a. The cost index is compounded over two years from a specified mid-2025 base to the comparable mid-2027 rating year. The hypothetical forecast assumes stable benefit coding and the same initial population mix until a separate risk-mix adjustment.

Covered service group 2025 PMPM Annual utilization Annual unit-price change Annual intensity/mix Two-year projected PMPM
Hospitalization $165.00 2% 4% 1% $189.41
Professional care $95.00 3% 3% 0% $106.92
Prescription drugs $110.00 5% 6% 1% $139.00
Other medical services $110.00 1% 3% 0% $119.04
Total $480.00 service-specific service-specific service-specific $554.38

For example, hospitalization cost is $165 x (1.02 x 1.04 x 1.01)^2 = $189.41 after display rounding. Calculations retain unrounded values until the final result. The full sum is $554.377487 PMPM, which is 15.4953% higher than the $480 baseline over two years. This percentage is the constructed compounded covered medical claims trend on a fixed population and benefit basis, not the future gross premium increase.

Population and benefit transitions

Next assume the 2027 member risk composition has an expected 4% higher claims cost than the fixed-2025 composition, after accounting for enrollment and legal pooling; multiply $554.377487 by 1.04 to obtain $576.552586 PMPM. This 4% is a separately declared population factor and is not additional medical provider inflation. The prospective benefit design, projected utilization response included, is modeled as a net insurer liability factor of 0.98. Applying it yields $565.021535 PMPM in projected medical benefits.

Assume a separately expected net receipt under an admissible market risk-transfer program of $12 PMPM and a net cost of reinsurance financing of $6 PMPM. The net retained medical and risk financing amount becomes $565.021535 - $12 + $6 = $559.021535 PMPM. The transfer is applied once; no other line silently incorporates the same receipt or recoveries. In a real regulated market, whether these transfers belong in the allowed index-rate calculation must be checked against current filing rules.

Expenses, margin and gross rate

Assume $34 PMPM of fixed administration, variable expenses equal to 8% of gross premiums, taxes or assessments equal to 2%, and an illustrative target contingency/capital margin equal to 3% of gross premium. We assume all of these amounts apply on the same earning basis. The indicated gross premium is

P = ($559.021535 + $34.000000) / (1 - 0.08 - 0.02 - 0.03) = $593.021535 / 0.87 = $681.633948 PMPM.

Annualizing gives $8,179.607376 per covered member-year under these assumptions. Rounded to cents, the indication is $681.63 PMPM. This is a constructed financial indication, not a filed rate, a permissible individual premium, or a statement about an actual U.S. or Australian plan.

The numerical reconciliation is simple: at an indicated $681.633948 premium, 8% variable expense is $54.530716, 2% taxes are $13.632679, and 3% capital/contingency margin is $20.449018. The remaining 87%, $593.021535, exactly funds the stated $559.021535 retained medical-risk cost and $34 fixed expense, subject only to displayed rounding. The ratio $559.021535/$681.633948 is an ordinary constructed accounting ratio; it is not the legal medical loss ratio under 45 CFR Part 158.

Unit-price sensitivity

To isolate unit-price uncertainty, adjust every annual service-specific unit-price trend by plus or minus two percentage points while keeping utilization, service intensity, population risk, benefit design, fixed expenses, risk transfers and loadings unchanged. Recompound over the same two-year horizon. The low unit-price scenario produces approximately $533.276500 PMPM in 2027 benefits before the 4% risk and 2% benefit adjustments, and a $656.91 indicated gross premium. The high scenario produces approximately $575.887934 and a $706.83 premium. These deviations show that a seemingly small annual unit-price error compounds and propagates through financing loadings.

Scenario Projected 2027 medical trend claims PMPM before risk and benefit changes Indicated gross premium PMPM
Unit-price growth minus 2 percentage points per year $533.28 $656.91
Base assumptions $554.38 $681.63
Unit-price growth plus 2 percentage points per year $575.89 $706.83

Because the scenario holds utilization fixed, it is not an estimate of how care demand would respond to a higher price. Because the margin is proportional to premium, a dollar of additional projected medical cost increases indicated gross premium by more than one dollar after expenses and loadings. These findings follow directly from the declared identity and do not require extrapolation to actual market conditions.

Interpretation and reproducibility

This example demonstrates three distinct bridges: 2025 fixed-mix medical costs to 2027 fixed-mix claims, fixed-mix claims to the new population and benefit basis, and net retained covered costs to gross indicated premium. Collapsing those bridges into a single “medical trend” percentage would incorrectly attribute risk mix, benefit generosity, risk transfers and business expenses to care inflation. A production workpaper should reproduce the group-level multipliers and retain the unrounded arithmetic, including each source, base period and parameter owner. The identity also exposes exactly where a regulator, external reviewer or reader can disagree with an assumption without disputing the whole calculation.

Sensitivity, Identification, and Model Governance

Statistical uncertainty and scenario design

The numerical case varied a single family of assumptions. An actual forecast also confronts estimation error in historical claim completion, morbidity segmentation, pharmaceutical costs, demographic selection, provider contracts, coding, risk adjustment receipts, expenses and membership volume. Model uncertainty is distinct from process variance. A conventional confidence interval on historical average claims does not automatically cover changes in contract prices, new services, legislation, or a market exit.

A robust scenario set varies at least three mechanisms independently: negotiated price inflation, care utilization and intensity, and forecast member risk mix. It also examines coordinated changes such as economic distress increasing enrollment churn at the same time as administrative cost PMPM rises. Scenario tests should report probability only when its basis can be supported by data; otherwise they are defined conditional stress cases.

Causality and attribution

Observed claims PMPM is the joint result of care need, payment, benefits, and population. A historical price increase may coincide with a new high-risk enrollee cohort; regressing PMPM on time alone cannot identify the marginal unit-price contribution. Similarly, a new utilization-management program may be implemented during a benefit revision, confounding the program's effect. Identification can sometimes improve through matched cohorts, difference-in-differences with validated parallel trends, randomized interventions, instrumented contract changes, or standardization across demographic cells, but each design has assumptions requiring case-specific testing.

The RAND and Oregon experiments provide examples of better identified policy effects because allocation mechanisms help address confounding in their settings. Their results remain tied to historical products, participants and outcomes. A rate trend assumption is a forecast of a future insured portfolio, not an unrestricted causal effect of an insurance policy.

Credibility and cross-validation

Models should be assessed on out-of-time samples that match the intended future use; random record-level splits can leak calendar prices and provider contract information into validation data. Claims from the same insured person or provider are correlated. A hierarchical split or forward-chaining validation better tests the forecast problem than a random claim-line split. Evaluation can include calibrated PMPM, error by age and risk group, tail claim underprediction, interval coverage, and robustness to coding changes. A perfectly calibrated overall mean can conceal material underestimation for a clinically important subgroup.

Large subgroup differences may affect risk transfers, equity, and market behavior. Predictive accuracy, legality of classification, fairness, and financial adequacy must be considered separately. A model using a protected or restricted variable to explain historical differences may be useful for methodological audit even where law forbids its use in consumer rating.

Traceability and disclosure

An auditable rating workpaper links every table to a data cutoff, extract, transformation, trend assumption, experience base and actuarial decision. It specifies missing-data treatment, normalization, the calculation of completion, and the relationship between premium units and member-months. The ASOP 41 disclosure guidance and ASOP 56 modeling framework provide professional anchors for documenting assumptions and modeling limitations in applicable U.S. actuarial practice. A final indication is reproducible only when adjustments can be followed from base evidence to priced unit.

Distribution, Affordability, and Public Policy

Premium incidence and financing

Medical costs generate a resource obligation, while the design of an insurance system determines how it is financed. In community-rated coverage, a portion of higher expected health costs can be pooled across members rather than allocated to the exact person receiving care. Employer, government and household contributions may each bear a portion. Risk adjustment can reduce the incentive of individual insurers to avoid high-cost members without removing the underlying medical obligation. The WHO's financial-protection and pooling approach is relevant to the institutional objective, but it does not dictate one premium factor or national subsidy schedule.

A premium can increase while the average cost of medical services is stable if enrollment risk rises, a subsidy expires, administrative costs change, coverage expands, or required capital increases. A premium can remain constant while provider prices rise if public transfers, benefit restrictions, member cost sharing or insurer margins absorb the change. These are accounting and institutional possibilities. It follows that public explanations should name the component responsible for a change rather than using “medical trend” as a residual justification.

Access, selection and dynamic adequacy

A high premium can reduce take-up among healthy or price-sensitive members, intensifying selection and the next period's required premium. Subsidies, automatic enrollment, guaranteed-issue regulation and risk adjustment can mitigate different aspects of the mechanism. However, each intervention has different budgetary and incentive consequences. van de Ven and Ellis (2000) examine risk adjustment in competitive plan markets; the result is an institutional design problem, not simply a more accurate medical cost model.

In cost-sharing designs, lower claims spending may shift expenses to households, reduce necessary utilization, or improve care efficiency. The direction must be investigated rather than inferred from the insurer's ledger alone. Consumer protection requires attention to out-of-pocket exposure, provider access and benefit usability. Good actuarial rates can coexist with poor financial protection if the insured contract or financing distribution is badly designed.

Transparency to members and regulators

Public explanations are strongest when they report a short reconciliation: starting incurred claims per member, adjustment for maturation, trend over the specified months, expected change in population morbidity, benefit and provider changes, forecast risk transfers, expense and capital provisions, and resulting indicated rate. Regulatory disclosures should identify whether a percentage is per member, per contract, weighted portfolio-wide, or an individual premium increase. The U.S. 2027 CMS consumer justification instructions illustrate how actuarial filings and public-facing explanations can serve different audiences while relying on the same underlying assumptions.

The central practical implication is that an insurer can defend its prospective indication analytically while leaving the political and distributive choice about funding transparently visible. A subsidy, contribution rule, benefit redesign or regulatory cap changes the financing arrangement and should be recorded as such.

Research Method, Source Boundaries, and Limitations

Research design

This monograph is a structured analytic synthesis. Primary-source analysis covers U.S. professional actuarial standards, federal insurance rating and filing texts, Australian premium-round institutional documents, international reporting and supervisory frameworks, and WHO and OECD health-financing sources. Empirical research on care demand and insurance selection is used for contextualized inference. The equations and arithmetic cases are independently constructed from stated assumptions. No confidential insurer data, patient records, interviews, randomized study replication, or comprehensive meta-analysis were performed.

The research distinguishes three evidentiary levels: authoritative statements of current professional or legal requirements within stated jurisdiction, empirical claims drawn from published work and limited to their study populations, and original analytical deductions whose validity follows from assumptions and arithmetic. The appendix-like example is in the third category. The References provide full bibliographic identities for the external evidence, and the companion EVIDENCE.json maps the verified URLs to claims and limitations.

Limitations of available evidence

Historical observed claims need source access, full development triangles, benefit detail, contract terms and risk-adjustment settlement data to support a real insurer's rate indication. Aggregate published projections cannot supply those inputs. The numerical demonstration assumes stable annual growth parameters and a single appropriately defined benefit factor. It does not separately estimate dynamic demand response, claim incidence variance, capital at risk, morbidity coding drift, or the uncertainty of public transfer settlements.

The professional standards reviewed are not interchangeable across countries or actuarial practice areas. ASOP 8, ASOP 23 and ASOP 56 apply within their stated U.S. professional scope; ASOP 7 and ASOP 20 have importantly distinct subject scopes effective June 2026. U.S. 2027 rate-review documents apply to designated products and plan years and do not establish requirements for all employer, public or international health insurance. Australian annual premium processes are jurisdiction-specific, and future regulatory changes can alter applicability. IFRS 17 governs financial reporting, while IAIS principles address supervision; neither is a universal consumer ratemaking equation.

Scientific sources are read within their accessible primary or publisher-record scope. When a journal article's complete article is not examined, the bibliographic record and accessible abstract support only appropriately limited statements. The study makes no claim of originality for the underlying cost identities, risk adjustment concepts, insurer pricing theory, professional practice standards or empirical findings. Its independent contribution lies in the explicit reconciliation of these established components, critical identification of common conceptual conflations, international scope comparison, and fully stated synthetic quantitative calculation.

Conclusion

Health insurance rate development is the prospective financing of defined medical benefits for a defined insured population. The scientific calculation begins with earned exposure and ultimate incurred claims, separates utilization and price from treatment intensity and changing morbidity, adjusts for the actual contract and providers, incorporates risk transfers and operational expenses, and reports a transparent indication under a declared capital and regulatory framework. Medical trend is one important input within that chain. Its measured amount depends on the service basket, price definition, enrollment mix, treatment changes and projection horizon.

The decomposition demonstrates why health spending inflation, fixed-population insurer medical claims trend, changes in population risk, and consumer premium changes must be identified separately. Their numerical differences are meaningful, and their interactions compound. The synthetic case shows a $480 PMPM experience base becoming $554.38 of fixed-population trended medical cost, then $565.02 of prospective benefits after population and benefit adjustments, $559.02 of retained medical and risk-financing cost, and an illustrated $681.63 gross premium after the stated loads. Each transformation is auditable, and altering unit-price assumptions exposes the premium's sensitivity without disguising uncertainty as observed data.

Medical cost forecasting has an institutional dimension: who pools the costs, who may set rates, which risk variables may legally be used, how financial protection is subsidized, and what public standards govern actuarial transparency. The United States' single-risk-pool rate-review structure, Australia's premium application process and public health-financing arrangements use overlapping economic ingredients but different pricing authorities and allocation rules. Legal permission, predictive accuracy, adequacy and affordability are separate questions in the final judgment.

A durable scholarly and regulatory account of health premiums therefore joins medical evidence, actuarial methods, financial obligations and distributive design through explicit assumptions, source-linked claims and reproducible calculations. That is the central analytical result of this Research Dossier. It is a synthesis and application of established science, and proposes no new named model, theory or doctrine.

References

  • Actuarial Standards Board. (2014). Actuarial Standard of Practice No. 8: Regulatory Filings for Health Benefits, Accident and Health Insurance, and Entities Providing Health Benefits. Last revised March 2014; effective September 1, 2014. Official ASOP 8 text.

  • Actuarial Standards Board. (2016). Actuarial Standard of Practice No. 23: Data Quality. Official ASOP 23 text.

  • Actuarial Standards Board. (2019). Actuarial Standard of Practice No. 56: Modeling. Official ASOP 56 text.

  • Actuarial Standards Board. (2025). Actuarial Standard of Practice No. 7: Life or Health Cash Flow Analysis. Revised December 2025, effective June 1, 2026. Official ASOP 7 text.

  • Actuarial Standards Board. (2025). Actuarial Standard of Practice No. 20: Analysis of Property/Casualty Cash Flows, Including Discounting. Revised December 2025, effective June 1, 2026. Official ASOP 20 text.

  • Actuarial Standards Board. (2017). Actuarial Standard of Practice No. 5: Incurred Health and Disability Claims. Official ASOP 5 text.

  • Actuarial Standards Board. (2011). Actuarial Standard of Practice No. 12: Risk Classification. Official ASOP 12 text.

  • Actuarial Standards Board. (2013). Actuarial Standard of Practice No. 25: Credibility Procedures. Official ASOP 25 text.

  • Actuarial Standards Board. (2010). Actuarial Standard of Practice No. 41: Actuarial Communications. Official ASOP 41 text.

  • Australian Government, Department of Health, Disability and Ageing. (2026). PHI 78/26: 2027 Private Health Insurance Premium Round Applications. Official 2027 premium-round notice.

  • Australian Prudential Regulation Authority. (2026). APRA's Role in the Private Health Insurance Premium Round. Official APRA guidance.

  • Baicker, K., Taubman, S. L., Allen, H. L., Bernstein, M., Gruber, J. H., Newhouse, J. P., Schneider, E. C., Wright, B. J., Zaslavsky, A. M., & Finkelstein, A. N. (2013). The Oregon experiment—Effects of Medicaid on clinical outcomes. New England Journal of Medicine, 368, 1713–1722. https://doi.org/10.1056/NEJMsa1212321.

  • Centers for Medicare & Medicaid Services. (2026). Unified Rate Review Instructions: Rate Filing Justification, Parts I, II, and III, Effective for Plan Years Starting on or after January 1, 2027. U.S. Department of Health and Human Services. 2027 Unified Rate Review Instructions.

  • Centers for Medicare & Medicaid Services. (2026). HHS Notice of Benefit and Payment Parameters for 2027 Final Rule. U.S. Department of Health and Human Services. Official 2027 final-rule summary.

  • Centers for Medicare & Medicaid Services. (n.d.). National Health Expenditure Projections. U.S. Department of Health and Human Services. Official projections portal.

  • Einav, L., Finkelstein, A., & Cullen, M. R. (2010). Estimating welfare in insurance markets using variation in prices. The Quarterly Journal of Economics, 125(3), 877–921. Public research record and text.

  • International Accounting Standards Board. (2017, as amended). IFRS 17 Insurance Contracts. IFRS Foundation. Official IFRS 17 standard page.

  • International Association of Insurance Supervisors. (n.d.). Insurance Core Principles and Common Framework: ICP Online Tool. Official IAIS ICP tool.

  • Manning, W. G., Newhouse, J. P., Duan, N., Keeler, E. B., Leibowitz, A., & Marquis, M. S. (1987). Health insurance and the demand for medical care: Evidence from a randomized experiment. American Economic Review, 77(3), 251–277. PubMed bibliographic record and abstract.

  • Organisation for Economic Co-operation and Development. (2025). Health at a Glance 2025: OECD Indicators, chapter “Prices in the Health Sector.” OECD Publishing, Paris. OECD 2025 price comparison.

  • United States. (2026). 45 CFR § 156.80: Single Risk Pool. Code of Federal Regulations, U.S. Department of Health and Human Services. Single-risk-pool regulatory text.

  • United States. (2026). 45 CFR § 158.210: Minimum Medical Loss Ratio. Code of Federal Regulations, U.S. Department of Health and Human Services. Minimum-MLR regulatory text.

  • van de Ven, W. P. M. M., & Ellis, R. P. (2000). Risk adjustment in competitive health plan markets. In A. J. Culyer & J. P. Newhouse (Eds.), Handbook of Health Economics (Vol. 1, pp. 755–845). Elsevier. University research bibliographic record.

  • World Health Organization. (n.d.). Health Financing: Overview of Revenue Raising, Pooling, and Purchasing. WHO health financing portal.