RESEARCH DOSSIER
Health Insurance Risk Equalization Transfers and Incentives
Abstract. Health insurance risk equalization reallocates resources across insurers or health funds to offset predictable differences in enrolled populations' expected health expenditures and, in doing so, changes the incentives surrounding enrollment, coverage, coding, prevention, and clinical quality. This monograph offers a comparative international examination of insurer-level transfers and risk-adjusted contributions, the mathematics of pooled financial balance, and the remaining channels of selection after formal compensation. The analysis separates the accounting identity of a zero-sum transfer pool from normative fairness, actuarial prediction from economic incentives, ex ante compensation from ex post risk sharing, and regulatory design from causal evidence. Official schemes in the United States, the Netherlands, Germany, Switzerland, Ireland, and Australia are interpreted in their respective legal and population boundaries. The documentary review is combined with an independently derived three-insurer illustrative transfer calculation, explicit comparisons of residual expected profit, marginal care incentives and variance exposure, and a structured identification plan for program evaluation. Across systems, transfer neutrality governs the disposition of financial flows, whereas the capacity to reduce risk selection depends on risk predictors, contract and premium rules, quality incentives, coding controls and the distribution of residual uncertainty. The conclusions establish conditions for interpreting observed redistribution without treating transfer magnitude, statistical prediction or cross-country program labels as sufficient measures of public performance.
Keywords: health insurance; risk equalization; risk adjustment; insurer transfers; budget neutrality; community rating; adverse selection; morbidity adjustment; predictable profits; residual risk; quality incentives; reinsurance; coding intensity; comparative health financing.
Introduction
The central problem of regulated competitive health insurance is the separation between the premium that an insurer can charge and the expected medical spending generated by those it insures. When acceptance obligations, community-rated premiums, standardized benefits, or mobility rights constrain risk pricing, the financial profitability of an enrollee may depend on the enrollee's risk class rather than on the quality of the care purchased on that person's behalf. A transfer system modifies that profitability calculation by assigning compensation to high-risk enrollments and financing it through assessments, pooled contributions or lower allocations elsewhere.
Risk equalization is best studied as a joint institution of actuarial forecasting and market organization. The authors of the goal of risk equalization study (van de Ven et al., 2023) describe the persistent debate over whether eliminating predictable over- and undercompensation is the sole objective or whether efficiency is an independent design criterion. This debate is practical: a predictive factor may improve accuracy yet reward avoidable utilization; a transfer that encourages appropriate access may reduce one form of selection while increasing another form of administrative gaming. A health financing institution must therefore be assessed through the contractual and political choices surrounding the transfer.
The canonical research object here is financial adjustment among health insurers or comparable insurance funds based on differing expected risk and market characteristics. Premium subsidies for households, general intergovernmental fiscal equalization, and provider risk-adjusted capitation are connected phenomena but remain separate objects unless they change the insurer-level financial incentives being examined. This distinction permits a common comparative vocabulary while retaining the legal specificity of national schemes.
The document follows an explicit evidence hierarchy. Primary rule text and competent agency documentation establish program mechanics and effective dates. Agency accounts and published annual data establish administrative aggregates. Peer-reviewed analyses support claims about mechanisms and observed outcomes within their study designs. The independently calculated examples are designated illustrative and do not reproduce any country's proprietary or statutory complete formula. The research cutoff is November 14, 2026.
Definition, Scope, and Historical Foundations
Risk equalization denotes a rule-based redistribution of resources among competing or otherwise distinct insurance-bearing entities in response to the expected cost characteristics of their enrolled risk pools. Risk adjustment commonly denotes the predictive classification and calculation that helps determine that redistribution. In some systems the two expressions are used interchangeably; here they are distinguished whenever the distinction clarifies how scores become financial transfers.
The elementary actuarial statement is that expected cost varies across people. Age, observed morbidity, prior utilization and related indicators can predict expenditure; the economic problem arises when insurers can select or repel those characteristics while prices remain socially regulated. The distinction between cost prediction and fiscal implementation is explicit in Kautter and colleagues' HHS-HCC model analysis (2014), which treats model construction and transfer calculation as separate components of the Affordable Care Act methodology.
The institutions examined have different histories and coverage populations. The Netherlands uses a statutory base-insurance arrangement funded partly through a Health Insurance Fund; the Dutch risk-equalization explanation describes community rating and acceptance rules. Germany allocates pooled statutory contributions among sickness funds under the morbidity-oriented risk structure adjustment framework, summarized by the Federal Health Ministry overview. Switzerland compensates insurers within cantonal risk frameworks, as explained in the Swiss federal explanation. Ireland maintains a Risk Equalisation Fund for its community-rated private health insurance market, whereas Australia's private hospital insurance equalization rules operate within a distinct supplementary insurance environment.
These programs have separate benefit definitions and population denominators. The price of a Dutch statutory base-insurance contract, the allocation to a German statutory sickness fund, and an Australian private insurer's net equalization position are not interchangeable economic magnitudes. A rigorous comparative account must identify whether the unit is a person-year, member-month, contract, standardized benefit claim, state-market pool, canton or national statutory allocation. It must also specify the benefit year and the point in the settlement cycle at which a reported amount becomes final.
Risk equalization affects financial distribution, but its policy rationale is more specific: it seeks to make covering a person with high predictable needs financially compatible with the market's acceptance and solidarity rules. Where predictable marginal profits persist, selection can remain profitable even while aggregate transfers reach substantial amounts. The scope of that proposition is established in selection incentives research by van Kleef, Eijkenaar and van Vliet (2020).
Economic Logic of Risk Equalization
Consider insurer i with N_i covered persons, per-person premium revenue p_i, an expected risk-adjusted transfer t_i, expected covered medical costs c_i and nonmedical expenses a_i. Its expected underwriting result before investment income and taxes is M_i = N_i(p_i + t_i - c_i - a_i). The regulator may separately impose minimum service standards, required benefits, premium categories, network rules and restrictions on refusing enrollment. Expected profitability for a subgroup g is then determined by p_g + t_g - E[c|g] - a_g, evaluated with the appropriate price, benefit and accounting bases.
This representation reveals two forms of insurer selection. Direct selection changes the composition of members through marketing, product choice, renewal or access mechanisms. Indirect selection changes the attractiveness of particular service lines, waiting times, provider networks or treatment pathways to groups with predictable needs. The study measuring payment-system efficiency by Layton, Ellis, McGuire and van Kleef (2017) distinguishes demand-side sorting distortions from supply-side benefit design distortions. Efficient risk compensation should be evaluated against both channels.
Equalizing predicted costs is also related to premium competition. If insurers receive compensation for exogenous expected morbidity, a plan offering efficient clinical management can retain part of the benefits of lower realized cost, depending on the design of ex post settlement. But adding a variable linked to actual spending may remove some financial incentive for prevention or efficient contracting. The efficiency-based payment weights analysis by Layton, McGuire and van Kleef (2018) derives the policy relevance of service-level incentives rather than fitting statistical predictions alone.
Selection incentives depend on residual costs, consumer preferences and switching. Layton's imperfect adjustment and sorting analysis (2017) shows that a transfer system may alter which residual costs matter to pricing and can either improve or worsen selection outcomes depending on how demand and predicted risk correlate. The inference is that a system with better overall fit can still have problematic subgroup margins and competitive distortions.
The objective must therefore be declared at the level of outcomes: access for high-risk members, sustainable regulated prices, incentives for care and prevention, administrative credibility and feasible financing. One observed transfer total cannot independently establish success on these dimensions.
Formal Transfer Accounting and Budget Neutrality
Let r_i denote a standardized expected reimbursable expenditure per person for insurer i within a properly defined pool, w_i=N_i/Σ_j N_j its enrollment weight, and r̄=Σ_j w_jr_j the market-weighted predicted mean. A deliberately simplified risk-neutral transfer rule is
T_i = N_i × k × (r_i − r̄),
where k is a common positive scaling factor expressed in units consistent with the monetary risk score. Summing across the insurers in that pool gives Σ_i T_i = k[Σ_i N_ir_i − r̄Σ_i N_i] = 0. This identity is mathematical: it follows from the choice of the weighted mean. It establishes equal charges and payments for the defined pool, subject to the book-keeping convention that positive T means a receipt.
Budget neutrality can also be enforced by centering a more complicated raw formula: define a_i as the initial raw allocation and set T_i = a_i − (N_i/N)a_total, with N=ΣN_i and a_total=Σa_i. Then ΣT_i=0 whatever the underlying risk function. This simple centering step is often useful for understanding the accounting identity; it does not replicate the full policy's adjustments for actuarial value, geographic costs, age factors, induced demand, premium scaling or risk-pool segmentation.
Several distinct neutrality claims require precise treatment. Within-pool zero sum states that transfer charges finance transfer payments within a specified accounting period and policy scope. Fiscal neutrality to the central government refers to whether government contributes additional resources, a different question. Economic neutrality means no effect on prices, care or incentives, which generally does not follow from the transfer identity. Intertemporal neutrality is another distinct question: a fund can balance on an accrual basis yet carry receivables, interest, timing differences or failed collections.
In the United States the 2027 federal payment rule retains the HHS state payment transfer formula and its high-cost treatment. It would be a mistake to replace that legally specified expression with this introductory equation. The U.S. formula also distinguishes a plan's relative expected risk from its allowable rating revenue and combines additional factors. The Swiss risk compensation framework explicitly describes the payment system as zero-sum, but its cantonal grouping and PCG adjustments differ from a nationwide single-pool formulation.
The counterfactual for any transfer evaluation is the market organization with the same prices and enrollment restrictions but a different equalization rule. The observed sum of transfers is chiefly an accounting property; changes in coverage accessibility, plan composition and care quality are substantive empirical questions.
A Reproducible Three-Insurer Example
The following three-insurer example is independently derived for this monograph and is hypothetical. It uses a single period and pool, equal benefit scope, equal standardized premiums and no geographic or actuarial-value adjustment. Insurer A enrolls 10,000 people with predicted annual medical expenditure of $1,000 each; B enrolls 6,000 with $1,600 each; C enrolls 4,000 with $2,800 each.
| Insurer | Members | Predicted cost per member | Predicted total cost | Transfer per member | Net transfer |
|---|---|---|---|---|---|
| A | 10,000 | $1,000 | $10,000,000 | −$540 | −$5,400,000 |
| B | 6,000 | $1,600 | $9,600,000 | +$60 | +$360,000 |
| C | 4,000 | $2,800 | $11,200,000 | +$1,260 | +$5,040,000 |
| Total | 20,000 | Weighted mean $1,540 | $30,800,000 | — | $0 |
The weighted mean is ($10.0m + $9.6m + $11.2m)/20,000 = $1,540 per person. Set k=1 in the simplified rule. A pays (1,000−1,540)×10,000 = −$5,400,000; B receives (1,600−1,540)×6,000 = $360,000; and C receives (2,800−1,540)×4,000 = $5,040,000. Their net cash flow sums exactly to zero, and all arithmetic can be reproduced from the displayed quantities.
If the uniform stylized premium available for medical costs is $1,540, then premium plus transfer exactly matches each plan's modeled expected covered costs: A has $15.4m−$5.4m=$10m; B has $9.24m+$0.36m=$9.6m; C has $6.16m+$5.04m=$11.2m. This equality is a consequence of the assumptions and deliberately excludes administration, capital returns and random cost variation. It says nothing about actual profitability after enrollment dynamics.
A first sensitivity test increases C's predicted cost from $2,800 to $3,200 with member counts held fixed. The market mean rises by $80 to $1,620. Transfers become A −$6.2m, B −$0.12m and C +$6.32m. Once again the total is zero. The example illustrates that a higher risk score for one plan changes every participant's allocation through the endogenous pool average, even when another plan's clinical population is unchanged.
A second sensitivity replaces prospective modeled expenditure with actual realized expenditure. Under that rule an unexpectedly expensive event raises the transfer, partially insulating the plan; but it also decreases the marginal financial payoff to cost containment. This distinction motivates separating forward-looking compensation from retrospective risk sharing. Such counterfactuals provide directional theoretical insight, not forecasts of any actual U.S. or European plan.
Ex Ante Models, Ex Post Adjustments, and Risk Corridors
An ex ante risk-equalization scheme allocates resources using information fixed before, or administratively defined for, the relevant contractual risk period. It may use prior-year diagnostic codes, drug groups, historical service use, age and demographic variables. The goal is to compensate predictable differences while leaving a meaningful residual to be managed by the insurer. The Dutch risk equalization settlement cycle explicitly distinguishes ex ante determinations, spring recalculation, preliminary settlements and a final calculation up to four years after the equalization year.
Ex post adjustment can take several forms. A reconciliation may correct the number of enrolled members or improve the data underlying a prior-period predicted risk. An explicit risk corridor may share aggregate gains and losses. A high-cost pool may reimburse some spending exceeding a threshold. These mechanisms need not have the same incentive properties. An enrollment correction changes exposure measurement; retrospective reimbursement changes the insurer's financial exposure to the marginal claim.
The high-cost payment alternatives study by Layton and McGuire (2017) demonstrates analytically that appropriately modified risk adjustment and transfer formulas can perform functions similar to reinsurance, while identifying important differences among designs. The U.S. high-cost risk pool rule maintains the $1 million threshold and 60 percent coinsurance arrangement for qualifying costs above the threshold in its particular market program. This provision is supplementary to the state-level transfer formula and should not be mistaken for a universal percentage of each plan's costs.
Germany's reformed risk structure adjustment combines morbidity-based allocation with a high-cost risk pool and an explicit prevention allowance. The German institution reports an 80 percent reimbursement rate for expenditure beyond the specified, annually adjusted high-cost threshold. The different thresholds and funding bases in the United States and Germany preclude a simple comparison of nominal percentages as if they represented identical insurance protection.
For interpretation, define α as the fraction of incremental qualifying cost reimbursed by an ex post mechanism. For a marginal eligible dollar of actual expenditure, the plan bears approximately (1−α) dollars before additional incentives and adjustments. This partial derivative is a stylized local calculation and is valid only within the defined reimbursement region. It does not state the entire plan's exposure, nor the effectiveness of utilization oversight.
In every system, the policy question is how much unforecastable financial risk the insurer should retain and what share of savings from genuine quality improvement should remain with the entity responsible for providing or purchasing care. The answer depends on whether clinical risk is manageable, whether the insurer faces capital constraints, and whether a patient's access to essential care is threatened by residual losses.
Predictable Profits, Residual Selection, and Market Sorting
Let the group's expected margin after risk equalization be μ_g=E[p_g+t_g−c_g−a_g|g]. A regulator concerned about indirect exclusion can study the distribution of μ_g across identifiable groups under the published benefits and premium schedule. Equalizing a risk score at the whole-population level need not force μ_g to zero for clinically meaningful subpopulations. The Dutch residual selection study of van Kleef et al. (2020) finds meaningful incentives despite a sophisticated system; the clinical subgroup method and studied institutional context matter to the inference.
The time horizon also matters. multi-contract-period analysis by Withagen-Koster, van Kleef and Eijkenaar (2023) follows identified Dutch groups through multiple years using linked survey and administrative information. The authors report persistent average losses for many chronically ill groups and persistent profitability for healthy groups. Single-year average compensations can therefore understate the value to an insurer of attracting a low-risk individual who is likely to remain enrolled.
Even groups with zero mean residual margin may differ in financial attractiveness. Oskam, van Kleef and Douven (2024) show in heteroscedastic residual spending research that unequal residual variance can sustain selection incentives through uncertainty and safety-margin costs. A group with greater variance or catastrophic downside may be less attractive to a capital-constrained plan than another group with an identical mean, particularly under constrained community-rated pricing.
The study healthy-group overpayment analysis by van Kleef, van Vliet and Oskam (2024) uses Dutch model-year 2021 data to identify healthy groups with measurable positive margins. Reported mean overpayments across identified groups range from €38 to €167 per person; those estimates refer to defined datasets and a specific model version. Their proposed methodological remedy is constrained regression, not a universal adjustment coefficient transferrable to unrelated programs.
Risk-selection analysis should therefore use a vector of diagnostic outputs: subgroup mean residuals, distribution of losses, persistence over time, heteroscedasticity, correlation with switching, patterns of benefit use and plan exit, and whether providers or plans can identify the relevant risk before enrollment. A transfer scheme can have an impressive conventional predictive R² while retaining exactly the selection incentives that regulation was designed to address.
A further causal distinction is essential. Observed overrepresentation of a subgroup within one insurer may reflect preference, network location, switching costs, price and reputation. It is not by itself proof of intentional risk selection. The evidentiary burden is to identify the channel by which a predictable profit opportunity changes insurer actions or consumer access, ideally with quasi-experimental or experimental variation.
Quality, Prevention, and Service-Line Incentives
An insurer's optimal behavior depends on the marginal cost of improving patient health and the marginal adjustment to its transfer. Suppose a preventive intervention reduces expected future covered cost by ΔC>0, costs K, and reduces future transfer revenue by ΔT>0 because the model registers lower future utilization or morbidity. Its simplified net return to the insurer is ΔC−K−ΔT, measured on a consistent time horizon and after enrollment attrition. The benefit to the insured person and the broader health system can exceed this private return, and the causal effect of the intervention must be identified independently.
For example, if an intervention costs $40 per enrolled person, reduces future expected treatment expenditures by $120 and reduces next-year adjustment payments by $50, the insurer's simplified gain is $30 per person. If that adjustment loss were $90 instead, the gain would be a $10 loss despite a $120 decline in future care costs. These numbers are hypothetical and serve to expose the marginal incentive, not to demonstrate empirical prevention effectiveness.
Risk adjustment variables derived from diagnoses or prior utilization may create tensions between precise expenditure prediction and efficient service delivery. Layton, McGuire and van Kleef's payment-weight efficiency analysis (2018) examines how assigning weights in light of actual service allocation can counter the incentive to distort particular services. Statistical fit and welfare-grounded service incentives need not rank payment models in the same order.
There is also the question of clinical access. If an insurer can reduce visits to specialists who serve disproportionately undercompensated patients, its cost-cutting may be privately profitable while undermining covered benefit delivery. A comprehensive assessment should therefore track appointment availability, meaningful service access, chronic disease management, patient-reported outcomes, appeals, and potential spillovers across insurers and provider markets.
German policy explicitly includes a prevention allowance in the statutory sickness-fund reform, demonstrating that institutional designers recognize prevention as an independent incentive domain. In 2026 the U.S. CMS 2026 benefit parameters fact sheet describes adding a PrEP-related factor to address incentives that might otherwise limit access to preventive medication. Such measures are relevant illustrations of design intent. Their presence in a rule does not itself establish measured quality gains.
An actuarially defensible framework for evaluation connects predictable group profitability to observed service outcomes. That requires preserving the difference between a diagnosis used as a legitimate predictor of unavoidable need and a utilization variable that may partly reflect discretionary decisions by a plan or provider.
Coding Intensity, Auditing, and Information Governance
A diagnosis-based risk model depends on recording clinical conditions. This creates the possibility that a higher risk score reflects changes in underlying illness, completeness of ascertainment, administrative coding practice, or strategic coding intensity. Accurate documentation is necessary for fair compensation, while unsubstantiated additions corrupt the relationship between morbidity and payment.
Geruso and Layton's Medicare upcoding investigation (2020) provides empirical evidence of large coding differences between private Medicare plans and the U.S. fee-for-service setting. Their estimates concern Medicare Advantage and the specific measurement methods of that study, not direct measured effects in the Affordable Care Act individual-market scheme, the Netherlands or Germany. The portable lesson is about mechanism and audit design: when payment responds to clinical coding, a credible verification process is required.
The HHS 2027 final rule specifies continuing risk adjustment data validation provisions and relevant plan liability calculations. Data validation must check not only a provider's reported code but the clinical documentation and eligibility of the coding event under the program's exact version. Audit sampling, error measurement, correction timing, appeals and financial penalties should be explicit.
There is an analytical tension between individual-level sensitivity and public auditability. Insurers need reliable data for risk prediction, while patients need confidentiality and fair use of health information. A high-dimensional model may predict well but obscure which variables generate payment or selection incentives. Administrative reproducibility requires archived model coefficients, risk-factor classifications, code crosswalks, enrollment-period rules, changes in diagnostic coding standards and independent access to aggregate validation results.
When a model revision reduces unexplained subgroup spending, it is still necessary to test whether the improvement reflects a genuinely better prediction or a more intense administrative incentive to document codes. A defensible review examines pre- and post-revision trends in coding frequencies, utilization and outcomes, and comparisons with variables or settings less directly exposed to coding incentives.
The science of risk equalization thus includes information provenance. The same financial transfer may have a mathematically valid formula and a data input whose accuracy is disputed. Traceability of source records is a substantive condition for trusting the allocation.
United States: Affordable Care Act Market Transfers
The United States uses risk adjustment for specified individual and small-group commercial health insurance market pools under the Affordable Care Act. It is not identical to the separate Medicare Advantage risk-adjusted federal payment arrangements, nor to the Medicaid managed care payment rules. The original HHS-HCC design paper describes models using demographic and diagnosis indicators for adult, child and infant groups, with plan liability and cost-sharing differences central to the estimation design.
The operative transfer mechanism uses a plan's relative risk and rating characteristics within a defined state market risk pool. The May 2026 final payment rule confirms the existing state transfer formula was maintained for the 2027 benefit year, together with revised coefficients and the high-cost risk pool. HHS announced that it would administer the program in every state and the District of Columbia for that benefit year. Model-specific adult, child and infant calculations and the interpretation of geographic and metal-level rating terms prevent the use of a single population-wide pure morbidity average as a literal statutory formula.
The official 2025 transfer summary report, released June 30, 2026, records 557 issuer participants, down from 592 in 2024. It reports that the absolute value of state transfers in the 2025 benefit year, excluding the high-cost risk pool and subject to the report's specified pool conventions, was about 9.0 percent of total premium compared with 8.9 percent in 2024. The published individual noncatastrophic and small-group percentages depend on separate pooling denominators and must not be summed or treated as the share of national health spending.
Those percentages describe the scale of redistribution, not a direct change in health outcomes. A transfer amount may rise because composition changes, premiums change, risk classification changes or enrollment migrates between plans; it does not isolate the causal effect of the policy on access. Smaller or newer insurers may experience forecast errors and cash-flow constraints as well as different underlying risk mixes.
The method also uses audit and data validation procedures. A properly specified research design would relate issuer-level net transfers to member months, premiums, actuarial value, standardized risk scores, realized eligible claims, data validation adjustments, and market exits by state and year. Without those denominator and program-version controls, a correlation between high transfers and high spending merely restates the purpose of risk adjustment.
The U.S. case illustrates a regulated marketplace transfer with strong within-state pool accounting and a separate high-cost component. Its institutional significance resides in converting standardized expected risk into a charge or payment while preserving financial accountability for plan-specific performance within the policy's governing formula.
Netherlands: Statutory Health Insurance Fund Contributions
The Dutch statutory base health insurance market pairs broad enrollment rights with risk equalization. The Zorginstituut system description explains that insurers must offer the basic plan and cannot charge a higher basic premium to a member on the basis of illness. The Health Insurance Fund finances risk-based contributions to insurers, combining a specific public financing institution with insurer competition.
For 2026, the official annual equalization page states that the ministry establishes the macro performance amount and annual equalization regulation while Zorginstituut applies detailed policy rules and distributes contributions according to expected risk. Risk factors include demographics, geographic characteristics and information on health care or pharmaceuticals, with category-specific normative amounts. The administrative process is revised as enrollment and other information become known, with a final multiyear settlement cycle.
The Dutch policy shares the objective of reducing selection with the U.S. system, but its finance cannot be interpreted as an identical zero-sum issuer-to-issuer charge on a state marketplace. The contribution from the Health Insurance Fund is a component of a broader financing package. Household premiums, tax- or income-linked financing and the fund's macro budget must be included when describing total redistribution. An observed annual contribution is an allocation from a pooled fund rather than simply a single counterparty insurer's transfer.
Empirical research provides a particularly rich test bed. The sophisticated-adjustment selection study of van Kleef et al. (2020), the multiyear profitability study of Withagen-Koster et al. (2023), and the healthy-enrollee margin analysis of van Kleef et al. (2024) examine residual incentives using differing designs and data sources. Together they show why apparently sophisticated risk adjustment can coexist with predictable profitability differences.
The Dutch case emphasizes that even a carefully managed annual equalization budget cannot eliminate every strategic incentive. The research problem is to distinguish risk factors that justifiably compensate unavoidable need from variables whose inclusion weakens insurer responsibility for appropriate care. The published theory and evidence do not establish that a single coefficient adjustment will solve every subgroup residual, particularly when residual variance and contract persistence differ.
Germany: Morbidity-Oriented Sickness-Fund Allocations
Germany's statutory health insurance allocates pooled financing among competing sickness funds through the morbidity-oriented risk structure adjustment system, commonly called Morbi-RSA. The German federal health ministry account describes the reforms operating from 2021: coverage of the full disease spectrum, a regional component, a high-cost risk pool, removal of an earlier disability-pension risk factor, an exclusion approach for unusually fast-growing morbidity groups, and a prevention allowance.
The legal and operational context differs from both the Dutch model and U.S. marketplace transfers. A sickness fund's revenue from the common statutory allocation depends on demographic and morbidity mix, with specified additional adjustments and relevant pension or eligibility categories; the insured's statutory contribution and the insurer's discretionary supplemental contribution are governed by other financing rules. A positive or negative net allocation under this framework must be read with the German program's own financing denominators.
The Federal Office for Social Security's 2024 annual settlement analysis, dated January 12, 2026, provides technical definitions and aggregate tables. It documents the 2021 reform components, separate annual settlement indicators and longitudinal individual-level model statistics. For the 2024 settlement, its table reports an R² of 55.9 percent for specified standardized expenditures, within its stated data exclusions and methodological limits. This is a goodness-of-fit statistic for that particular risk-adjustment exercise; it is neither an insurer-wide measure of fairness nor a comparable R² from the United States or the Netherlands.
The German high-cost pool reimburses 80 percent of qualifying expense above an annually adjusted threshold according to the federal reform overview. This is a clear instance of explicit risk sharing complementing morbidity adjustment. The prevention allowance and exclusion of anomalous morbidity-group growth also demonstrate how regulators can embed counter-incentives to gaming in an otherwise predictive allocation system.
Interpretation should examine whether a reported settlement is ex ante, preliminary or final; whether results include sickness benefit components; how disease groups were revised; and what share of costs are standardized in the numerator. Aggregate annual reports support descriptive comparisons within their own series, while causal statements about the impact of reforms require separately identified counterfactual evidence.
Switzerland: Cantonal Zero-Sum Risk Compensation
Swiss mandatory health insurance uses standardized premium rules within specified cantonal or premium-region arrangements. The Federal Office of Public Health's Swiss risk compensation description identifies age, sex, prior hospital or nursing-home stay, and pharmaceutical cost groups based on prior dispensed medicines as risk indicators. Insurers with relatively fewer high-risk members make payments to insurers with more high-risk members.
The official explanation specifically states that total charges equal total payments, giving a clear example of a zero-sum transfer institution. It also distinguishes cantonal charge/contribution rates from nationwide pharmaceutical cost group adjustments. A study that aggregates all Swiss insurers without observing canton boundaries can therefore conceal differences in local risk composition and the scheme's explicit grouping decisions.
This architecture illustrates why national risk equalization and regional solidarity are separate policy choices. If equalization is computed within cantons, cross-canton differences in risk or service prices are not necessarily shared through the mechanism. If a national system compensates an insurer for regional characteristics, the redistributive reach is broader. Neither form can be labeled more equitable without a declared objective, premium system and population denominator.
The Swiss approach relies substantially on indicators observable in a prior period, an arrangement that can reduce immediate reimbursement of realized current spending while still making drug prescribing and administrative records important inputs. Independent evaluation should check how the introduction or revision of pharmaceutical cost groups changes predictive accuracy, chronic-illness residuals and incentives to steer members.
The Swiss case supplies a useful contrast with the U.S. state-market framework and the Dutch fund-contribution model: all are motivated by risk selection, but the fiscal object and geography of balancing materially differ.
Ireland: Community-Rated Private Insurance Credits
Ireland's risk equalization is associated with its community-rated private health insurance market. The Health Insurance Authority reporting portal publishes risk equalization assumptions, overcompensation reports and scheme materials. The governing financial arrangement uses a Risk Equalisation Fund and credits linked to characteristics of insured persons and certain utilization or high-cost claims.
The Health Insurance Amendment Act 2025 revises specified credits and high-cost parameters to apply from April 1, 2026. The statutory text is especially important because a changing annual credit or levy changes the financial incentives even where the broad policy rationale remains constant. A study combining pre- and post-April observations without an effective-date variable would make its estimated average incentive difficult to interpret.
Unlike the simplified symmetrical zero-sum insurer formula, the Irish arrangement includes fund receipts through stamp duties and credit disbursements under the statutory rules. Annual receipts and payments may differ due to timing, balances, and the institutional budget. Budget neutrality, if asserted, must specify whether it means the long-run fund design, the annual cash account or the actual net payment of an individual participating insurer.
The HIA's public program documents provide a basis for comparing compensation for age and hospitalization against insurer premium strategies and population risk profiles. Evaluating whether overcompensation or residual selection occurs requires the scheme's exact credited benefits, tax and duty bases, annual settlement data and relevant market-contract segments.
The Irish case is particularly useful because it explicitly links actuarial solidarity and the regulatory governance of competing private insurers. Its institutions should be classified according to the coverage market they govern, rather than collapsed into a generalized national health financing indicator.
Australia: Private Health Insurance Risk Equalization
Australia operates risk equalization for its private health insurance sector through a statutory pool subject to private health insurance legislation. The official 2025 Risk Equalisation Policy Rules set the policy framework. The Department of Health's 2025 rules renewal notice describes replacing the earlier policy rules through a substantially similar 2025 instrument.
The Australian Prudential Regulation Authority's 2024–25 equalization statistics release, published March 31, 2026, provides annual insurer-level net results based on regulatory returns. The dataset is valuable as a transparent administrative basis for tracing which funds pay into or receive from the system. It also requires care: a private hospitalization insurer's result in an Australian risk equalization system has a different benefit and enrollment base from a German sickness fund's statutory allocation.
The significance of Australia's data architecture is that the redistribution can be observed at the insurer level in a regulator-published annual series. This allows descriptive work linking net equalization positions to fund scale, private hospital membership and product mix, but an insurer's net contribution is not evidence of inefficient claims management or favorable risk selection by itself.
Identifying behavioral effects would require an external source of changes in risk equalization parameters, a credible comparison group or a carefully articulated structural model. Annual movement in equalization totals can arise from enrollment changes, benefit growth, population aging or revised rules. The presence of annual statistics provides observability rather than causal identification.
Australia therefore broadens the comparison from mandatory base coverage markets to a substantial private insurance sector, reinforcing the need for a coverage-specific definition of the research denominator.
Comparative Interpretation and Institutional Boundaries
A comparison of six national approaches can be organized around four dimensions: who is insured, how a risk score or risk proxy is calculated, where the net flow is funded and balanced, and which decisions remain commercially sensitive to expected medical spending.
| Jurisdiction | Covered institution or market | Financial architecture | Primary design feature |
|---|---|---|---|
| United States | Specified ACA individual and small-group pools | State market-level charges and payments; related high-cost pool | Risk and rating terms, HHS model and federal data validation |
| Netherlands | Statutory basic health insurers | Health Insurance Fund risk-based contributions | Ex ante normative amounts and multiyear reconciliation |
| Germany | Statutory sickness funds | Common statutory fund allocations | Morbidity and region adjustments, high-cost risk pool, prevention allowance |
| Switzerland | Mandatory insurers within cantonal grouping | Explicit zero-sum insurer compensation | Prior hospital/nursing-home and pharmaceutical cost groups |
| Ireland | Community-rated private insurers | Risk Equalisation Fund credits and levy receipts | Annual statutory credits and overcompensation scrutiny |
| Australia | Private health insurance funds | Statutory risk equalization pool | Regulator-published annual fund-specific net outcomes |
These descriptions are grounded in their respective government sources: the U.S. payment notice, Dutch 2026 regulation summary, German RSA description, Swiss program specification, Irish regulator records and Australian statistical release. The table captures institutional features at the research cutoff; it does not impose an identical score formula on all countries.
Three comparative findings follow. First, a universal word such as equalization hides at least three fiscal designs: transfers among plans, contributions from a pooled national financing institution, and statutory credits financed by levies or taxes. Second, risk-adjustment design cannot be separated from the premium and enrollment rules that create selection incentives in the first place. Third, measured predictive performance, redistribution volume and a regulated insurer's ability to offer good care are distinct objects requiring distinct evidence.
The external validity of an empirical result depends on the environment. Dutch subgroup profitability estimates concern statutory basic health insurance and a specified year of model coefficients. Medicare coding results concern a different U.S. risk-adjusted payment institution. Australian net transfer data concern the private insurance sector. Treating those studies as identical would generate false international precision.
The basis for comparison is therefore a matched concept, not a matched dollar figure: the expected marginal financial gain or loss from enrolling and appropriately treating an otherwise comparable person, under the actual local contract and benefit rules.
Identification, Validation, and Sensitivity Analysis
A rigorous empirical evaluation can distinguish six classes of output. Predictive fit measures how well a model forecasts observed covered cost on the assessment sample. Group calibration compares predicted and realized average costs for predeclared subpopulations. Selection exposure measures predictable net margins after the applicable transfer and premium rules. Incentive outcomes measure plan responses in benefit design, service delivery, coding and marketing. Financial stability measures volatility, capital sensitivity and the timing of settlement. Welfare analysis assesses whether enrollees receive accessible care at defensible total cost.
Data should be indexed by benefit year t, insurer i, product or market m and subgroup g. A minimal record includes exposure in member-months, contract eligibility, standardized covered claims, risk scores or diagnostic input categories, transfer receipts and charges, premiums or contribution revenue, benefits, cost-sharing tier, high-cost adjustments and contemporaneous changes in program rules. Each data item should retain its legal and actuarial definition. The denominator for an issuer's transfer-to-premium ratio differs from the denominator for a per-member predictive error.
Model validation requires out-of-sample assessment and transparent specification of the target variable. For example, if y_igt denotes actual covered spending and ŷ_igt a predicted amount, the group prediction error is E[ŷ−y|g]. Overall mean error may be zero by calibration while group-level errors persist. Individual R² weights high-cost observations and is sensitive to the covered-cost basis. A more direct incentive diagnostic is the expected residual profit including permissible premiums and transfers, calculated for groups that an insurer can plausibly identify.
The residual variance study highlights a limitation of examining means only. The economic efficiency framework and welfare-grounded adjustment weights show why evaluation should also examine service-line benefits and the actual decisions insurers make. If the policy question concerns insurer exclusion of patients with a particular chronic condition, both group margin and related service availability are relevant.
Causal inference can exploit changes in risk-factor inclusion or coefficients, phased implementation, regional boundaries or threshold discontinuities if exogeneity and pretrend assumptions are plausible. A difference-in-differences estimate requires parallel counterfactual trends; a regression discontinuity requires no manipulation around the threshold; an instrumental variable requires an exclusion restriction. These conditions cannot be inferred from official administrative datasets alone. A published policy reform is a source of possible variation, not automatic identification.
Predeclared sensitivity analyses should vary the accounting horizon, claimant truncation, insurer fixed effects, demographic and diagnostic definitions, region, risk model version, premium treatment, high-cost exclusion, timing of settlement, outliers and observation weights. Analysts should report both subgroup mean residuals and tail outcomes, and explain whether an apparent improvement reflects fewer high-cost episodes, a change in coding or a changed transfer coefficient.
This monograph conducts documentary verification and reproducible arithmetic; it does not possess a cross-country person-level dataset and does not report a newly estimated causal coefficient. That separation protects the interpretation of its comparative results.
Governance, Fairness, and International Policy Implications
Risk equalization is often defended as an instrument of solidarity, but the practical allocation of resources is an institutional choice that must be monitored over time. A design that removes a predictable loss for a chronically ill subgroup may improve a plan's willingness to cover and serve that group. The same design may require higher charges on insurers with less expensive populations, changing premium competition and entry incentives.
The risk equalization goals article explains why the normative goal of risk equalization shapes what counts as an improvement. A narrow forecast objective rewards statistical accuracy, whereas a selection objective emphasizes predictable financial advantages across identifiable groups. An efficiency objective additionally evaluates incentives to provide appropriate service and manage avoidable cost. These objectives can align but are not mathematically identical.
Governance should therefore allocate responsibilities for defining policy objectives, approving factors, publishing coefficient versions, auditing data quality, assessing adverse subgroup outcomes, evaluating private insurer behavior and providing a method for correcting settlement errors. Public accountability is strengthened when authorities publish both the governing rules and enough aggregate outcomes to test whether redistribution is proceeding as intended. The Australian insurer-level statistics, German settlement tables and Irish scheme reports illustrate different official forms of administrative transparency.
A further issue is risk tolerance and capital. Two groups with identical expected residual claims may require different liquidity and capital buffers when their claim distributions have different tails. The variance and residual risk analysis establishes the relevance of this distinction. A regulator can consider how reinsurance or a high-cost pool affects those differential burdens, provided its design does not reward clinically unnecessary spending.
The overall comparative implication is that an appropriate transfer mechanism should be tested against foreseeable strategic responses rather than evaluated only after payouts. Questions about who is attractive to enroll, which services remain attractive to provide, how codes are generated and how predictable group margins change are central scientific observations.
A single universal adjustment algorithm would have to ignore fundamental differences in benefit entitlements, fiscal institutions and national legal rules. The more defensible international approach is a common analytical structure for comparing local systems, combined with transparent legal and empirical conditions for transferring research conclusions between them.
Conclusion
Risk equalization alters the distribution of financial risk among health insurers and consequently the expected returns to enrolling people with different health needs. The mathematical budget-neutrality identity gives a precise foundation for understanding charges and payments within a specified pool. It establishes an accounting balance and leaves the substantive outcomes to market conduct, risk classification, benefit regulation and empirical observation.
The comparative evidence supports five integrated conclusions. First, identifying the funding entity and transfer perimeter is essential: insurer-to-insurer payments, statutory fund allocations and levy-financed credits represent distinct fiscal designs. Second, risk prediction is one input into equalization; predictable subgroup margins, residual variability and multi-year enrollment determine whether selection opportunities remain. Third, insurer incentives to improve care and prevent illness depend on how much of the resulting savings the plan can retain after future transfer adjustments. Fourth, diagnostic coding and administrative audit practices are integral to payment accuracy. Fifth, the evaluation of success requires the joint inspection of financial balance, access, service quality, competitive conduct and data integrity.
The independently derived three-insurer example demonstrates both the zero-sum mechanism and the consequences of changing a single plan's predicted risk. It is an accounting illustration, not an estimate of any existing national scheme. The documented Dutch studies show residual subgroup incentives; German, Swiss, Irish, Australian and U.S. institutions show distinct solutions to the funding and risk-sharing problem. None of those separate findings establishes an internationally invariant optimal coefficient.
The practical scientific outcome is a source-grounded framework for interpreting insurance risk-equalization transfers across their proper jurisdictions and causal settings. Continued investigation can estimate how specific reforms change enrollment and quality with versioned administrative data and credible counterfactual identification. Such questions remain empirical work for future studies rather than claims of novelty or new theoretical doctrine in the present monograph.
Research Method and Evidence Limitations
The research was conducted as a critical narrative and comparative documentary review through November 14, 2026, prioritizing government rulemaking, competent supervisory agencies, peer-reviewed empirical work and published bibliographic records. The primary sources were inspected at their discoverable official or publisher locations; the U.S. 2025 program report was consulted through the official PDF's indexed text, while visual screenshot retrieval of that document was unavailable in the research environment. The German 2024 annual report's PDF front page was independently inspected as an image. Some journal arguments were assessed from accessible publisher abstracts and public full-text versions or indexing metadata, not from a uniform exhaustive full-text audit.
The illustrative algebra was derived independently and checked for zero sum and arithmetic conservation under two internally stated risk compositions. No personal medical information, claims database, insurer-level unpublished transfers, or original econometric estimate was used. Cross-country tables describe institutions and should not be read as rankings. Numeric figures are labeled with the report year and pool conventions wherever included.
The principal limitations concern institutional heterogeneity, the timing of official finalization, observable versus unobservable selection, reimbursement versus actual total benefit cost, and the fact that correlation in official administrative aggregates cannot identify behavioral effects. Changes enacted after the research cutoff require an updated rule and data review. All factual research links in the chapters are linked to the evidence ledger and listed in the complete bibliography.
No independent external peer review was requested or represented. The work establishes a comparative analytical synthesis, uses existing published economic and actuarial theory, and introduces no separate original theory, doctrine or conceptual school.
References
U.S. Department of Health and Human Services. 2026. “Patient Protection and Affordable Care Act, HHS Notice of Benefit and Payment Parameters for 2027; and Basic Health Program.” Federal Register 91 (97), May 20, 2026. Federal Register document 2026-10050. Official or publisher source
Centers for Medicare & Medicaid Services. 2026. Summary Report on Individual and Small Group Market Risk Adjustment Transfers for the 2025 Benefit Year. June 30, 2026. Official or publisher source
Kautter, John, Gregory C. Pope, Melvin Ingber, Sara Freeman, Lindsey Patterson, Michael Cohen, and Patricia Keenan. 2014. “The HHS-HCC Risk Adjustment Model for Individual and Small Group Markets under the Affordable Care Act.” Medicare & Medicaid Research Review 4 (3). doi:10.5600/mmrr2014-004-03-a03. Official or publisher source
Zorginstituut Nederland. 2026. “Zvw 2026.” Official information on the 2026 risk equalization contribution under the Health Insurance Act. Official or publisher source
Zorginstituut Nederland. n.d. “Wat is risicoverevening?” Accessed November 14, 2026. Official or publisher source
Bundesministerium für Gesundheit. n.d. “Risikostrukturausgleich (RSA).” Accessed November 14, 2026. Official or publisher source
Bundesamt für Soziale Sicherung. 2026. Risikostrukturausgleich: Ergebnisse des Jahresausgleichs 2024. Report dated January 12, 2026. Official or publisher source
Federal Office of Public Health (Switzerland). n.d. “Health Insurance: Risk Compensation.” Accessed November 14, 2026. Official or publisher source
Health Insurance Authority (Ireland). n.d. “Risk Equalisation.” Publications and scheme reports, accessed November 14, 2026. Official or publisher source
Ireland. 2025. Health Insurance (Amendment) Act 2025, No. 14 of 2025. Electronic Irish Statute Book. Official or publisher source
Australian Prudential Regulation Authority. 2026. “Private Health Insurance Risk Equalisation Statistics.” March 31, 2026. Official or publisher source
Australian Government. 2025. Private Health Insurance (Risk Equalisation Policy) Rules 2025, F2025L00895, made August 4, 2025. Federal Register of Legislation. Official or publisher source
van de Ven, Wynand, Gerrit Hamstra, Richard van Kleef, Mieke Reuser, and Piet Stam. 2023. “The Goal of Risk Equalization in Regulated Competitive Health Insurance Markets.” European Journal of Health Economics 24: 111–123. doi:10.1007/s10198-022-01457-7. Official or publisher source
van Kleef, Richard C., Frank Eijkenaar, and René C. J. A. van Vliet. 2020. “Selection Incentives for Health Insurers in the Presence of Sophisticated Risk Adjustment.” Medical Care Research and Review 77 (6): 584–595. doi:10.1177/1077558719825982. Official or publisher source
Withagen-Koster, Anja A., Richard C. van Kleef, and Frank Eijkenaar. 2023. “Predictable Profits and Losses in a Health Insurance Market with Risk Equalization: A Multiple-Contract Period Perspective.” Health Policy 131: 104763. doi:10.1016/j.healthpol.2023.104763. Official or publisher source
Layton, Timothy J., Randall P. Ellis, Thomas G. McGuire, and Richard van Kleef. 2017. “Measuring Efficiency of Health Plan Payment Systems in Managed Competition Health Insurance Markets.” Journal of Health Economics 56: 237–255. doi:10.1016/j.jhealeco.2017.05.004. Official or publisher source
Layton, Timothy J., Thomas G. McGuire, and Richard C. van Kleef. 2018. “Deriving Risk Adjustment Payment Weights to Maximize Efficiency of Health Insurance Markets.” Journal of Health Economics 61: 93–110. doi:10.1016/j.jhealeco.2018.07.001. Official or publisher source
Layton, Timothy J. 2017. “Imperfect Risk Adjustment, Risk Preferences, and Sorting in Competitive Health Insurance Markets.” Journal of Health Economics 56: 259–280. doi:10.1016/j.jhealeco.2017.04.004. Official or publisher source
Layton, Timothy J., and Thomas G. McGuire. 2017. “Marketplace Plan Payment Options for Dealing with High-Cost Enrollees.” American Journal of Health Economics 3 (2). doi:10.1162/AJHE_a_00071. Official or publisher source
Geruso, Michael, and Timothy Layton. 2020. “Upcoding: Evidence from Medicare on Squishy Risk Adjustment.” Journal of Political Economy 128 (3). doi:10.1086/704756. Official or publisher source
van Kleef, Richard C., René C. J. A. van Vliet, and Michel Oskam. 2024. “Risk Adjustment in Health Insurance Markets: Do Not Overlook the ‘Real’ Healthy.” Medical Care 62 (11): 767–772. doi:10.1097/MLR.0000000000001955. Official or publisher source
Oskam, Michel, Richard C. van Kleef, and Rudy Douven. 2024. “Heteroscedasticity of Residual Spending After Risk Equalization: A Potential Source of Selection Incentives in Health Insurance Markets with Premium Regulation.” European Journal of Health Economics 25 (3): 379–396. doi:10.1007/s10198-023-01592-9. Official or publisher source
Centers for Medicare & Medicaid Services. 2025. “HHS Notice of Benefit and Payment Parameters for 2026 Final Rule.” CMS fact sheet. Official or publisher source
Australian Government Department of Health, Disability and Ageing. 2025. “PHI 67/25 Private Health Insurance – Remade Private Health Insurance Rules.” August 7, 2025. Official or publisher source