RESEARCH DOSSIER
Long-Term Care Insurance Demand and Family Care Expectations
Abstract. The expectation that relatives will provide help during future dependency is a determinant of long-term care insurance demand, but its empirical effect changes with contract design, public support, family resources, and the opportunity cost of caregiving. This research monograph examines how anticipated family care enters insurance decisions, why such expectations diverge from eventual care, and how insurance purchases change the behavior of adult children. Its source base combines original survey wording, peer-reviewed market analyses, structural and instrumental-variable research, stated-choice experiments, and official international social-insurance and care documentation. The analysis distinguishes contractual reimbursement from cash benefits, perceived from realized family assistance, market purchase from hypothetical preference, and recipient welfare from caregiver employment and public budgets. The evidence supports a conditional synthesis: informal care may displace reimbursable services and lower demand for conventional private coverage, while flexible care-income benefits may complement family assistance; insurance can also affect children's housing and employment without immediately reducing observed informal care hours. No new proprietary theory or priority claim is advanced. The contribution consists of original comparative interpretation, source-critical evaluation, outcome-specific distinctions, and an empirically testable agenda.
Keywords: Long-term care insurance; family care expectations; informal caregiving; insurance demand; private long-term care; Medicaid; public insurance crowd-out; caregiver labor supply; household welfare; insurance purchase; family spillovers; behavioral risk perception; flexible cash benefits; adverse selection; stated preference.
Introduction
Long-term care financing begins with a future state of dependency whose probability, duration, cost, and available caregivers are imperfectly known. The prospective purchaser must anticipate changes in personal functioning, access to relatives, professional care prices, household assets, public protection, and the performance of insurance contracts. The WHO functioning classification treats functioning as situated in physical and social environments. An insurance entitlement is narrower: it depends on the actual policy or statute, assessment, covered expenses or cash triggers, and benefit limitations.
The central question is how beliefs about relatives' future help influence voluntary long-term care insurance ownership. A potential caregiver can be emotionally willing but geographically absent, employed full time, physically unable to assist, or responsible for another dependent person. A promise that appears adequate when a policy is purchased can become infeasible as families and labor markets change. Conversely, a child who was not expected to help may provide substantial assistance when the need arises. An anticipated source of care is therefore a belief about a contingent household resource.
The puzzle of limited voluntary insurance purchase has several documented foundations. Pauly's rational nonpurchase analysis (1990) explains why an informed risk-averse individual may rationally decline cover under particular family and utility assumptions. Brown and Finkelstein's Medicaid crowd-out study (2008) identifies a public-private substitution channel. Brown, Goda, and McGarry's insurance demand survey (2012) finds that perceived care needs, family availability, savings, premium concerns, and insurer trust all matter. Mommaerts's dynamic family analysis (2025) models parent-child decisions, while Wu and colleagues' flexible insurance experiment (2022) examines an income benefit usable during informal care. These sources describe related mechanisms but estimate different outcomes.
The investigation proceeds through original conceptual reconstruction, primary survey measurement, demand economics, comparative family welfare, and methodological audit. Its unit of analysis shifts deliberately among the insured individual, family, insurer, care provider, and public program. A measured insurance purchase rate, a hypothetical product choice, an expected willingness to help, a care hour actually delivered, and a policy claim are different observations. All reported empirical results remain tied to their original populations and dates. This monograph did not estimate microdata regressions or replicate closed-source structural computations.
Historical and Conceptual Foundations
Family support and institutionalization of long-term care finance
Family care remains a major source of help for dependent adults in high-income societies. Yet its relationship to insurance is shaped by public institutions. The OECD care affordability report (2024) documents extensive costs and uneven protection across countries. Its standard need scenarios reveal substantial gaps after public financing, particularly where living alone, severe disability, and low income interact. These are standardized comparative constructions, not individualized insurance quotations.
In the United States, long-term services and supports may be funded through Medicaid subject to eligibility, benefits, and state implementation. The Medicaid home-care programs describe supports delivered in homes and communities. Private long-term care insurance remains a contractual product whose covered care settings, benefit periods, exclusions, inflation options, premiums, and rate-adjustment histories vary. The NAIC consumer guidance makes these contractual dimensions central to informed comparison.
German statutory care insurance, established as a branch of social protection in 1995, provides a different institutional setting. The German care insurance rules describe compulsory coverage, assessed care grades, cash support or professional services, and substantial residual costs borne elsewhere. Singapore's CareShield Life scheme supplies mandatory protection for specified cohorts in the event of severe disability, using cash benefits after defined functional thresholds. Different enrollment rules and benefit types mean that compulsory scheme participation cannot be equated with voluntary purchase.
Historically, the policy-relevant development is the expansion of ways to distribute dependence risk across family, government, insurers and providers. The same household may use all four simultaneously. A comparative study must establish which costs each institution bears before declaring that one arrangement substitutes for another.
Definitions and observable outcomes
Long-term care need refers to enduring assistance with basic or instrumental daily activities and, where relevant, supervision. Informal care is assistance by family, friends or other unpaid or incompletely remunerated helpers. Formal care is organized, usually remunerated service. The relevant boundary is functional: a family member who is paid under a public cash scheme may remain a relative while providing economically compensated care.
Family care expectation refers to anticipated availability, ability and willingness to assist in a future state. It includes uncertainty about task type, intensity, duration, proximity and future circumstances. Actual informal care is realized time, physical effort, coordination, transportation and sometimes reduced wages. Neither the existence of a child nor a positive survey answer measures those realized inputs.
Latent insurance demand is coverage desired under a specified choice set; stated demand is hypothetical choice; revealed demand is actual policy purchase or continuation at feasible prices and underwriting. Lapse and persistence are separate decisions. Contractual claims reflect both eligible need and benefit wording. A purchased policy can have value even without a claim, while a future care recipient may have serious needs that never qualify for contractual payment.
Welfare is broader than insurer revenue or paid-care expenditure. It includes recipient consumption, autonomy, quality and safety; caregiver time, work and health; and fiscal resources. Separating these outcomes is necessary whenever insurance shifts work between relatives and paid providers.
Early economic arguments and empirical checks
Pauly (1990), in his expected-utility treatment of nonpurchase, shows that insurance against long-term care expenses can have low private value under certain care-state utility and family assumptions, even for an individual who dislikes risk. The model directs attention to what an insured benefit actually changes: recipient spending, estate value, or the family's care bargain. Its assumptions make the result conditional rather than a universal prescription.
Sloan and Norton (1997) investigate selection, bequests, and public coverage in national U.S. survey evidence. They report adverse-selection patterns and Medicaid crowd-out among older respondents while finding little support for the bequest and exchange motives they measured. This juxtaposition disciplines theory: a coherent possible motive need not explain observed purchasing behavior. Child availability, bequest preferences, and expectations of caregiving are distinct variables whose associations need not share one sign.
Brown and Finkelstein (2007), in the private-market pricing analysis, find pricing loads and limitations in purchased U.S. contracts, yet argue that supply imperfections alone do not explain the market's small scale. Their later review of market evidence (2009) underscores the importance of demand-side mechanisms. These estimates describe historical contracts and buyer cohorts; they are not current premium or loss-ratio estimates for 2026.
Two substitutions and one complementarity
Consider a task that a child can perform instead of a paid worker. Informal time can substitute for an eligible paid service, reducing expected reimbursable expenses. Under a policy that pays only invoices, expected benefits may fall when families anticipate providing more care. Under a cash policy triggered by need rather than invoices, benefits can finance transportation, adaptive equipment, professional respite, family living expenses or foregone wages. Family assistance can then complement insurance.
The substitution between informal and formal care must therefore be kept separate from the substitution between family support and insurance. A third interaction occurs when formal services preserve the capacity of relatives to continue providing nonprofessional help. Substitution and complementarity may coexist across tasks, duration and severity.
Care quality and access matter. A professional nurse and an adult child may provide different services, and one hour of each is not clinically interchangeable. The economic analysis therefore specifies task equivalence before calculating avoided expenditure or unmet need. Institutional arrangements determine whether a financially compensated family caregiver is recognized for benefit purposes.
Family Care Beliefs and Their Measurement
The original HRS questionnaire
An unusually direct measurement is available in the Health and Retirement Study. The 2022 HRS respondent codebook identifies variable SG097, which asks whether relatives or friends other than a spouse or partner would be willing and able to provide help for a long period if personal care tasks such as eating or dressing were needed. A subsequent question records the relationship of potential helpers.
The wording establishes a precise object: perceived extra-spousal willingness and ability under a hypothetical personal-care condition. It does not ask for a probability of actual future care hours. It excludes the spouse, does not specify hours or duration precisely, and does not validate the named person's own intention. A yes answer can refer to one person, multiple people, relatives or friends. A no answer from a married respondent is compatible with substantial expected spousal support. A survey coefficient assigned to this variable cannot simply be described as an effect of having children.
Even reported category frequencies must be handled carefully. The codebook provides item counts but a population estimate requires applicable sample restrictions, interview status, weights, and treatment of missing responses. This monograph uses the codebook to establish the primary questionnaire instrument and does not claim a newly computed population prevalence.
A richer measurement system would separately record spouse support, adult children, other relatives, friends, geographic proximity, employment capacity, health constraints, number of available helpers, willingness, reciprocity expectations and specific tasks. Prospective observations would then record whether care occurred, at what intensity, and with what economic sacrifices.
Expectations update after family experience
Coe, Skira, and Van Houtven (2015), in their care-experience study, find that a parent's nursing-home use reduces expectations that the respondent's own children will provide informal care; comparable experience with an in-law does not produce the same expectation shift. Nursing-home experience of a parent or in-law is associated with approximately a 0.8-percentage-point increase in long-term care insurance purchase probability in their analysis.
This result is useful precisely because the expectation and purchase responses differ. Observing care can make dependency more salient and reveal the practical difficulty of reliance on relatives. It may also change wealth, trust, attitudes to facilities, and family relations. The source does not permit identifying every channel from a single estimated purchasing effect. An observed change in belief can be larger or more widespread than actual new purchases where affordability or eligibility binds.
Experience is also selected: households exposed to parental care needs may differ in family health histories and financial resources. Designs should distinguish information conveyed by a parent's needs from effects of caring for that parent and from correlated aging risks.
Risk forecasts and insurer trust
Brown, Goda, and McGarry (2012) document multiple factors in their purchase-belief survey: people who anticipate long-term care needs are more likely to buy, while perceived alternative ways to finance or deliver care can reduce purchases. High perceived premiums and distrust in insurers also depress interest. These are observational and reported-reason findings. The article is evidence against accounts invoking only ignorance, only family care, or only price.
Blavet, Chopard, Rapp, and Sicsic (2026), using 12 HRS waves from 1996–2018, examine insurance take-up in an expectations-miscalibration study. Respondents' subjective survival prospects and expected nursing-home entry relate differently to demand depending on how survival forecasts compare with life-table benchmarks. This heterogeneity suggests that a survey question about family assistance can partly reflect broader ability to forecast future aging. Health knowledge, subjective probabilities and preferences must be modeled distinctly.
The shared methodological issue is endogeneity. Someone with insurance may anticipate using paid services and therefore expect less family involvement; someone expecting little help may purchase precisely because of that expectation. Cross-sectional association offers no temporal direction by itself.
Insurance can reshape family expectations
Coe, Goda, and Van Houtven (2023) examine the reverse relationship using changes in state tax treatment as an instrument for private insurance coverage. Their family-spillover findings report no detected reduction in informal-care use during an observation horizon of approximately eight years, while insurance reduces parents' perceived willingness of children to care in future and is associated with less child co-residence and stronger labor-market attachment.
The combination is not contradictory. A household may change planning or residential decisions before an actual care episode; relatively few may experience a qualifying new need during the observed years; the insurance effect may occur through expected future obligations. The null finding for observed care is not proof of zero lifetime displacement.
This study requires attention to instrument validity: state tax changes must influence child outcomes through insurance rather than through unrelated state-policy or economic channels, conditional on the design's controls. Its causal estimand applies to the population whose insurance behavior responds to the particular policy variation, rather than every potential purchaser.
Parent-child expectations as distinct information
A parent and an adult child may attach different probabilities to the same hypothetical future care commitment. One person's willingness is another person's estimate of reliability. Employment, other dependents, marriage, housing, migration and health shocks can change both estimates. Separate reports allow observation of the expectation gap, whereas interviewing only the parent leaves it hidden.
For future research, define the signed difference between a parent's and a child's probability of providing a precisely described care bundle, given comparable time horizons. Test whether a larger discrepancy predicts greater unmet need or insurance lapses after controlling for resources and functional state. This is a proposed empirical comparison and measurement strategy, not an established novel theory.
Family norms may influence statements even when capacity is limited. Relatives sometimes report willingness because obligation is socially expected; later practical limits may require formal help. A valid instrument must distinguish normative obligation, preference for care, and realistic availability.
Insurance Demand Determinants and Contract Design
A demand accounting thought experiment
Let care need in a future state be decomposed into equivalent units of assistance provided by relatives, purchased services, and unmet need. This is an accounting description, not an assumption that different caregivers are perfectly substitutable. Let eligible paid-care expenses depend on purchased services, and let contractual payment depend on functional eligibility and benefit design.
For reimbursement insurance, the payout is limited by qualifying incurred charges, elimination periods, caps and other terms. Greater family assistance can reduce anticipated reimbursable spending. For a qualifying cash-benefit contract, payment can remain available even if relatives provide care and no corresponding formal invoice is generated. The two contract designs create different marginal incentives from the same anticipated care need.
The purchaser's expected value should include recipient consumption and autonomy, premium payments, future care quality, liquidity, bequest preferences, and caregiver opportunity costs if the purchaser accounts for them. The insurer's required premium, in contrast, depends on discounted expected claims, expenses, margins, capital and long-horizon uncertainty. The two values need not coincide.
A stylized example clarifies the mechanism. Suppose a hypothetical person needs ten task-equivalent hours per week; relatives can supply six and four are purchased. If only billed services are eligible, six informal hours cannot generate reimbursement. A cash benefit payable after the stated functional threshold may support care in either setting. Neither the number of hours nor the payment example is a measured population estimate. Clinical equivalence, contract terms, and access to services remain critical qualifications.
Pricing, liquidity and underwriting
The 2007 long-term care market evidence reveals historical market loads and limited coverage, but more comprehensive policies were also offered to some consumers. Those facts do not settle whether an individual could sustain premiums or qualify medically. Realized purchases depend on income, liquid assets, debt, health eligibility, expected price paths, and competing savings needs.
A family can have substantial housing wealth yet face a serious liquidity barrier to premiums. Insurers can also restrict coverage for applicants with existing health conditions, causing realized enrollment to differ from demand under universal eligibility. The NAIC policy comparison advice directs attention to covered facilities, premium history, benefit limits and rate increases. Studies must identify the exact contract version rather than treating all long-term care policies as homogeneous.
Long-duration premium payment risk is especially important because care needs may emerge decades later. A stated preference for insurance in a one-time experiment need not translate into continuation through decades of payments. Insurance demand should therefore be decomposed into initial purchase, coverage persistence, lapse decisions, and effective protection when a claim becomes possible.
Medicaid and the marginal value of private coverage
Brown and Finkelstein (2008) estimate that Medicaid can crowd out private insurance because benefits financed privately may replace spending that a person otherwise could have received through public assistance. The Medicaid implicit-tax estimate reports an illustrative implicit tax of roughly 60–75 percent for a median-wealth individual in their modeled U.S. environment. This number is conditional on their assumptions and contemporary public rules, not a current tax rate or a forecast for any other jurisdiction.
If a household expects family care, Medicaid eligibility and expected paid care expenses jointly determine the residual value of private reimbursement. The channels can interact nonlinearly; adding a family-help coefficient to a public-program coefficient may not capture the joint decision. Social insurance may also complement private coverage by financing services that keep an individual at home.
A valid comparative estimate must code state-specific Medicaid arrangements, assets and income, eligibility rules, public services and private benefit coordination for the appropriate year. Federal program descriptions cannot establish an individual's state-law entitlement.
How cash benefits change demand
Wu, Bateman, Stevens, and Thorp (2022) present an Australian stated-preference choice experiment in which 1,008 pre-retirees allocate wealth among a life annuity, a liquid investment and flexible long-term care income insurance. Around 75 percent chose some long-term care cover in this designed hypothetical opportunity set. Participants regarded the income benefit as complementary to informal help; the paper reports gender-specific patterns among respondents expecting extensive family assistance.
The 75 percent is the experiment's stated selection rate. It is not an observed commercial purchase percentage, a prediction of renewal, or evidence that insurers can profitably supply the proposed benefits at every illustrated price. Its importance is the product mechanism: a qualifying income stream can finance help within the family while protecting household consumption.
Mommaerts (2025) studies parent-child long-term care decisions using a calibrated dynamic model. The availability of informal care reduces modeled insurance demand by seven percentage points. In the article's counterfactuals, equivalent cash benefits for informal care can increase insurance demand and family welfare while reducing Medicaid expenditures. These are estimated model outcomes under defined assumptions, not a randomized legislative trial. The journal issue is 2025, following online publication in November 2024.
Together, the two studies warrant a contract-sensitive interpretation: the expected availability of family help can make traditional expense reimbursement less attractive while increasing the usefulness of benefits spendable in support of family care. They do not establish that every cash policy is superior or that the stated and simulated effects share one parameter.
Bequests, risk preferences and family transfers
The bequest channel has opposing possibilities. Insurance may protect an estate against large care bills, raising the value of cover to a bequest-motivated household. Alternatively, a parent's estate may form part of an implicit exchange within the family, changing preferences over insurance and care. The 1997 study of bequest motives did not find the hypothesized exchange effect in its measured U.S. samples. It would therefore be inaccurate to present strategic bequests as the dominant empirical explanation of low purchasing.
Risk aversion can support coverage while the disutility of paying premiums, uncertainty about benefit eligibility, or the value of retained liquid funds pushes in the opposite direction. Some households can self-insure part of the expected cost; others face correlated risk from a disabled spouse and reduced family earnings. The same actuarial risk can lead to different decisions because the household circumstances and perceived benefits differ.
Trust and understanding of contracts
Long-term care contracts can be hard to evaluate. A buyer must understand whether the policy pays for home care, how claims are assessed, how inflation adjustments work, what happens if premiums rise, and the implications of lapsing. Brown, Goda, and McGarry (2012) found that insurer trust and cost concerns influence demand alongside household substitutes. These relationships may reinforce one another: reliance on children can reduce attention to insurance details, while distrust can encourage a household to rationalize nonpurchase through expected family support. The observed survey cannot establish each sequence causally.
Selection, Welfare, and International Comparisons
Endogenous purchase and adverse selection
Insurance owners and nonowners are selected on wealth, perceived risk, health information, preferences, marketing exposure, and family resources. If a respondent expects little help from children, that person may both purchase insurance and later use more paid services. A naive comparison could falsely attribute all differences in care to insurance rather than initial family circumstances.
The selection findings in elderly samples show why self-selection matters. The market-structure estimates show why supply and demand must be separated. Insurance itself can later alter care choices, but a causal design must distinguish the original purchase selection from induced changes in behavior.
Family characteristics can be effect modifiers and confounders. The number of children, distance to them, their employment and health, and the preference for privacy or independence may affect coverage and care use simultaneously. Household resources should include potential caregiver income as well as the prospective recipient's assets.
Unpaid care has economic costs
The absence of a paid invoice is not evidence of zero resource use. Van Houtven, Coe, and Skira (2013) analyze HRS longitudinal data in their caregiver work study and identify heterogeneous work and wage effects. Their results include reductions in work hours among employed women providing certain forms of help, with distinct effects across sexes and care tasks. They do not establish one universal wage penalty for every caregiver.
Caregiver costs include lost earnings, reduced promotion, foregone retirement contributions, physical strain, stress, travel, and time unavailable for children or other relatives. Some caregiving also brings positive relational value. A welfare analysis should not mechanically equate hours with wage loss, but neither should it ignore opportunity cost. The OECD informal-care indicators report that a majority of caregivers in their country sample are women and emphasize cross-survey differences in definitions.
A policy that pays to support an adult child's reduced work can protect family consumption even if no formal care is purchased. Another policy may finance paid care and allow the child to remain employed. Either can generate welfare benefits invisible in a claims-only analysis.
Family spillovers and incidence
The insurance spillover analysis connects parents' coverage to adult children's co-residence and labor supply. That result broadens the relevant insurance beneficiary beyond the policyholder. Parent and child may value different outcomes; the parent's preferred care arrangement and the child's labor-market opportunities may conflict. Welfare weights and the distribution of costs should be explicit.
Public cost savings do not by themselves establish social gains. Medicaid expenditure can fall because families provide more unpaid care, potentially shifting burdens to low-income caregivers; it can also fall because insurance supports more efficient services. The source of savings and the consequences for recipients and relatives must be reported separately.
A useful accounting presentation keeps distinct columns for recipient consumption, actual care quality and unmet need, relative's hours, relative's market earnings, insurer benefits, premium payments, and public spending. Aggregate welfare is calculated only after specifying which transfers cancel and which real resources are consumed.
Geographic and institutional differences
The OECD affordability comparison shows wide cross-country variation in care costs and remaining household payments. The German partial-benefit system allows statutory cash assistance in home care, but still leaves costs uncovered. Singapore's severe-disability cash insurance uses mandatory pooling and a functional claim trigger: in 2026 the official page states a S$689 monthly payout under its relevant age-and-claim conditions. That amount is cohort- and timing-sensitive and cannot be treated as a universal cash amount for all claims.
U.S. Medicaid home-and-community services vary with program authorities and state implementation. The 2008 public-insurance model was built in an earlier U.S. environment. Germany, Singapore and the U.S. cannot be pooled as equivalent observations of voluntary demand, since the enrollment mechanism, benefit conditions and provider environment differ.
The OECD long-term care workforce evidence documents pressure on paid-care capacity. A reimbursement promise is worth less when eligible services cannot be purchased promptly, whereas a cash benefit can provide financial flexibility but does not itself create trained workers. Family care may be relied upon because of formal-care shortage rather than positive preference for relatives. Geography and service supply should therefore enter demand and welfare models.
Gender and distributional fairness
Older women can face longer dependency horizons; female relatives often bear disproportionate unpaid care responsibilities. Income and family structure affect both exposure and capacity to buy protection. The OECD caregiver statistics apply to the specifically surveyed countries and age groups, with acknowledged definitional differences; they are not evidence of identical effects across populations.
Insurance can reduce burdens on caregivers with high earnings, preserve options for low-wage caregivers, or exclude applicants who cannot sustain premiums. A product aimed at reimbursing expensive formal care may disproportionately benefit households with access to providers, while benefits supporting informal care may be especially valuable where family labor remains available. The actual distribution depends on eligibility, premium subsidies, covered benefits and care quality.
Primary Source and Model Validation
Source hierarchy and access boundaries
The research uses publisher records for peer-reviewed papers, an official HRS codebook for survey wording, official social-program and regulatory information, and OECD/WHO institutional documents. Where open full text was available, its scope and methods were reviewed; where the publisher provided an abstract and bibliographic record behind a paywall, the manuscript confines claims to supported summary results and recognized model features. It does not represent itself as independent replication of paywalled estimates.
A survey instrument proves what respondents were asked. A survey article supports its measured associations. A structural model supports conditional simulations. An instrumental-variable design supports causal interpretation only under its identifying assumptions. A public agency webpage supports institutional rules at the date checked, not historical rules in every state. These distinctions determine evidence strength.
Comparative evidence matrix
| Source | Evidence type | Supported conclusion | Interpretive boundary |
|---|---|---|---|
| Pauly (1990) | Economic theory | Nonpurchase can follow rational choices under defined assumptions | Not a universal behavioral result |
| Sloan and Norton (1997) | Observational U.S. samples | Selection and public crowd-out are relevant | No universal causal coefficient for family care |
| Brown and Finkelstein (2007) | Historical contract pricing analysis | Supply frictions coexist with demand constraints | Not current policy quotes |
| Brown and Finkelstein (2008) | Structural public-private calculation | Medicaid may reduce private benefit value | Rule-, wealth- and model-dependent |
| Brown, Goda, and McGarry (2012) | Survey | Beliefs, price, trust, savings and family help all correlate with purchase | Observational and self-reported |
| Coe, Skira, and Van Houtven (2015) | Longitudinal study | Experience updates future help beliefs and purchase | Heterogeneous observational pathways |
| Mommaerts (2025) | Dynamic model | Informal care lowers modeled demand; cash-benefit counterfactual changes outcomes | Model-based simulation |
| Coe, Goda, and Van Houtven (2023) | Tax-policy instrumental variable | Insurance may change children’s housing and work outcomes | Instrument validity and complier population |
| Wu et al. (2022) | Australian stated choices | Cash insurance can complement expected family care | Not observed market purchasing |
| Blavet et al. (2026) | Panel regression | Forecast calibration shapes associations with take-up | Residual confounding possible |
| HRS codebook (2022) | Primary questionnaire | Precise expected-help wording | No inference from unweighted raw counts |
| OECD (2024; 2025) | Cross-national official analysis | Care burdens and family provision differ | Survey definition and scenario limits |
This table summarizes study-specific evidentiary status. It is not a meta-analysis and does not pool percentages from different populations.
Structural model validation
Mommaerts (2025) reports a seven-percentage-point lower insurance demand from family-care availability in the estimated dynamic model. The counterfactual cash-benefit result depends on preferences, care technology, eligibility and Medicaid treatment in the calibrated model. Sensitivity should be examined over caregiver wage opportunities, formal-care prices, care episode duration, parent-child bargaining, wealth, and expected survival. The publisher refers to replication materials; no code execution or independent recalculation is claimed here.
The meaningful conclusion is the modeled incentive mechanism and its direction under the specified product. The percentage is an estimate belonging to that model, not a benchmark for every country.
Instrumental-variable validation
Coe, Goda, and Van Houtven (2023) identify coverage effects using variation in state long-term care insurance tax treatment. In the tax-policy identification design, the instrument must affect family outcomes through insurance coverage and not unrelated changes in state policy or labor markets. A validation would require examining policy timing, first-stage strength, migration, other simultaneous reforms, subgroup compliance and alternate outcomes. The eight-year horizon also limits lifetime inference. The paper's findings support a carefully bounded causal interpretation rather than a universal prediction of future caregiver hours.
Stated preference and revealed preference
The Australian hypothetical purchase task makes the offered benefit explicit and observes choices under an experimental menu. It is strong evidence that a flexible income-benefit design can appeal to respondents anticipating family assistance. Its approximately 75 percent selection figure cannot be equated with real enrollment: insurer underwriting, annual premiums, disclosure, insurer trust, lapses and competing commercial products may differ from the experiment.
Future validation would compare actual offered prices and take-up, follow renewals and claims, and measure whether cash benefits are spent on formal services, household consumption or compensation for relatives. A benefit's intended flexibility does not by itself establish how money is eventually used.
Repeated measurement and causal modeling
A prospective design would observe a household before purchase, re-interview both prospective care recipient and adult children, record policy offers and premiums, and follow realized care and work outcomes. The causal effect of family-care expectation cannot be read directly from a regression of ownership on stated child availability: unobserved preferences, health information, risk forecasts and financial literacy affect both. Insurance can also alter expectations after purchase.
At minimum, analysts should report population and insurance eligibility denominators, the survey wave, policy ownership and purchase dates, spouse versus child support, functional severity, care hours and task types, care quality, caregiver employment, public program eligibility, regional service supply, mortality and attrition. Death and serious dependency can make later respondents systematically different from the original panel. Missing data should be treated under documented assumptions, including sensitivity analyses for attrition and unobserved confounders.
Source-to-claim validation and legal specificity
Pauly (1990) was checked against the publisher's Journal of Political Economy 98(1) record, pages 153–168. Sloan and Norton (1997) was checked against DOI and institutional citation. Brown and Finkelstein (2007) is the published Journal of Public Economics article, distinct from their earlier working paper. The 2008 American Economic Review article has DOI 10.1257/aer.98.3.1083. Mommaerts's January 2025 print issue follows November 2024 online release. The 2026 Blavet study is published in volume 26, article 3. The bibliographic entries below distinguish these dates.
Legal and program material is jurisdiction- and time-bound. The German public program description does not specify all individual awards; the CareShield Life benefit description is dated to its 2026 parameters; and the Medicaid services information does not replace state benefit criteria. The WHO functional classification describes functioning but does not create a contractual insurance entitlement. No current personal eligibility or financial advice is inferred.
Numerical and welfare accounting controls
The 0.8-percentage-point Coe et al. (2015) estimate refers to study-specific purchase probability, the seven-percentage-point Mommaerts (2025) finding to a modeled insurance-demand effect, and the approximately 75 percent Wu et al. (2022) figure to stated product choice. These have different baselines, designs and denominators; combining them into a single pooled number is invalid. Brown and Finkelstein's 60–75 percent implicit tax is a model-based marginal-value statement, not a literal government levy.
Insurance claim payments are transfers from pooled funds; care labor and supplies are resource uses; informal caregiving may carry opportunity costs; taxes and Medicaid outlays belong to public accounting. Summing transfers with resource costs as independent social burdens would double count. Evaluation must present care outcomes, caregiver work, claims, premiums and public spending in distinct categories before applying any welfare measure.
Integrated Scientific Results and Further Questions
Family expectations influence the value of benefits through contract terms
The first result is conditional and directly grounded: anticipated informal care may reduce demand for a policy reimbursing formal-care expenses, but can increase the appeal of an income policy usable alongside family care. Mommaerts's family insurance analysis and Wu and colleagues' flexible coverage study provide contrasting forms of support. Their study populations and methods differ, yet their benefit mechanisms are logically consistent.
This is an interpretive synthesis of established evidence, not a new named theory. It explains why estimates from reimbursement-based systems cannot be transplanted to cash-benefit systems without considering the actual contract.
Expected caregivers are uncertain resources
The HRS question on future help records a belief about extra-spousal willingness and ability. The parental experience study shows that beliefs can change with observed care experience, and the insurance spillover study shows that insurance can itself affect expectations. A static indicator for number of children cannot represent these processes. Longitudinal, task-specific measurement is necessary for defensible identification.
Research should compare expectations reported by parents and children, track changes in proximity and employment, and observe later actual care. The signed difference between parent and child assessments is a useful measurable disagreement, but no universal relationship with purchase or care failure is established here.
Household welfare spans generations
Insurance may protect a child’s employment, preserve a spouse's capacity, or change housing decisions. The caregiver wage and work evidence shows that unpaid help can have labor-market consequences; the adult-child spillover evidence indicates that ownership can affect family choices even without detected changes in current care use. A claims-only welfare analysis misses these outcomes.
The person purchasing insurance and the persons receiving its indirect benefits need not be the same. A household aggregate can conceal unequal burdens among siblings or partners. Research and policy decisions should specify whose welfare is counted and how intra-family transfers are treated.
Institutions define the outside option
The Medicaid crowd-out analysis demonstrates how public coverage can alter the value of private benefits. Germany's care-grade insurance arrangement and Singapore's cash severe-disability coverage demonstrate other public-private interfaces. The same family-care belief can produce different voluntary purchase incentives under these rules. Country-specific law and actual benefit details belong inside the demand model, not as incidental controls.
Five empirical tests following from the synthesis
First, link matched parent-child expectation reports to later receipt of actual care, specifying task and severity. Second, compare offered cash and reimbursement products at equivalent actuarial value to isolate a benefit-design interaction with expected informal care. Third, test whether shocks to potential caregivers' employment and relocation change actual policy purchasing, after addressing correlated income and need changes. Fourth, distinguish responses in current care hours from future residential and labor-market planning over longer observation periods. Fifth, estimate how public care subsidies, means tests and geographic workforce shortages modify the link between family expectations and demand.
These are researchable questions requiring new matched data, policy variation, or controlled experiments. They are not represented as completed empirical analyses.
Editorial synthesis
No single sign describes the effect of expected family help on insurance demand across all contract types and public systems. Family help can substitute for formal services, while insurance can support family care or free caregivers for paid work. The measured expectation is incomplete and responsive to prior experience and insurance itself. The correct object of analysis is the prospective benefit under stated rules, actual family resources, and defined welfare consequences.
The contribution of this dossier is a source-specific integration of mechanisms, methods and institutional variation. It identifies where an empirical study demonstrates an association, where a model supplies a counterfactual, where an instrument supports a bounded causal inference, and where an official document defines current program rules. It advances no proprietary doctrine or unsupported priority claim.
Conclusion
Long-term care insurance decisions are shaped by what individuals expect their relatives to do when dependency develops, by whether those expectations are feasible, and by what insurance will actually pay in the resulting circumstances. Historical economic theory, observational U.S. studies, a dynamic parent-child model, tax-policy instrumental-variable evidence and an Australian stated-choice experiment jointly show that the family-care relationship varies with the contract and institutional environment.
Expected care and delivered care remain distinct. Insurance can alter expectations and adult children's employment or housing arrangements before the amount of observed informal care changes. The economic burden of dependency includes the recipient's care needs and the caregiver's time and lost opportunities, while public financing and private claims introduce further transfers. These outcomes must be analyzed on their own terms.
The established scholarly conclusion is that demand research requires specified insurance benefits, measured family-care capacity, public program coordination and an explicit intergenerational welfare perspective. General coefficients on child availability, pooled percentages from different designs, and undifferentiated comparisons of national programs offer insufficient grounds for policy claims. This dossier provides a coherent foundation for future empirical study while retaining the evidentiary limits of every source.
References
Pauly (1990). The Rational Nonpurchase of Long-Term-Care Insurance. Journal of Political Economy 98(1), 153–168. https://www.journals.uchicago.edu/doi/10.1086/261673
Sloan, F. A., and Norton, E. C. (1997). Adverse Selection, Bequests, Crowding Out, and Private Demand for Insurance: Evidence from the Long-Term Care Insurance Market. Journal of Risk and Uncertainty 15(3), 201–219. https://doi.org/10.1023/A:1007749008635
Brown, J. R., and Finkelstein, A. (2007). Why Is the Market for Long-Term Care Insurance So Small? Journal of Public Economics 91(10), 1967–1991. https://doi.org/10.1016/j.jpubeco.2007.02.010
Brown, J. R., and Finkelstein, A. (2008). The Interaction of Public and Private Insurance: Medicaid and the Long-Term Care Insurance Market. American Economic Review 98(3), 1083–1102. https://doi.org/10.1257/aer.98.3.1083
Brown, J. R., and Finkelstein, A. (2009). The Private Market for Long-Term Care Insurance in the U.S.: A Review of the Evidence. Journal of Risk and Insurance 76(1), 5–29. https://doi.org/10.1111/j.1539-6975.2009.01286.x
Brown, J. R., Goda, G. S., and McGarry, K. (2012). Long-Term Care Insurance Demand Limited by Beliefs About Needs, Concerns About Insurers, and Care Available From Family. Health Affairs 31(6), 1294–1302. https://doi.org/10.1377/hlthaff.2011.1307
Coe, N. B., Skira, M. M., and Van Houtven, C. H. (2015). Long-Term Care Insurance: Does Experience Matter? Journal of Health Economics 40, 122–131. https://doi.org/10.1016/j.jhealeco.2015.01.001
Mommaerts, C. (2025). Long-Term Care Insurance and the Family. Journal of Political Economy 133(1), 1–52. Published online November 7, 2024. https://www.journals.uchicago.edu/doi/10.1086/732887
Coe, N. B., Goda, G. S., and Van Houtven, C. H. (2023). Family Spillovers and Long-Term Care Insurance. Journal of Health Economics 90, Article 102781. https://doi.org/10.1016/j.jhealeco.2023.102781
Wu, S., Bateman, H., Stevens, R., and Thorp, S. (2022). Flexible Insurance for Long-Term Care: A Study of Stated Preferences. Journal of Risk and Insurance 89(3), 823–858. https://doi.org/10.1111/jori.12379
Van Houtven, C. H., Coe, N. B., and Skira, M. M. (2013). The Effect of Informal Care on Work and Wages. Journal of Health Economics 32(1), 240–252. https://doi.org/10.1016/j.jhealeco.2012.10.006
Blavet, T., Chopard, B., Rapp, T., and Sicsic, J. (2026). Misperception, Self-Reported Probabilities and Long-Term Care Insurance Take-Up in the United States. International Journal of Health Economics and Management 26(1), Article 3. https://doi.org/10.1007/s10754-025-09408-4
OECD. (2024). Is Care Affordable for Older People? OECD Health Policy Studies. OECD Publishing, Paris. DOI: 10.1787/450ea778-en. https://www.oecd.org/en/publications/is-care-affordable-for-older-people_450ea778-en/full-report.html
OECD. (2025). Health at a Glance 2025: Informal Carers. OECD Publishing, Paris, section of Health at a Glance 2025: OECD Indicators. DOI: 10.1787/8f9e3f98-en. https://www.oecd.org/en/publications/health-at-a-glance-2025_8f9e3f98-en/full-report/informal-carers_6f243c4d.html
Health and Retirement Study. (2022). HRS 2022 Final Release: Section G Respondent Codebook, Variable SG097. University of Michigan Institute for Social Research. https://hrs.isr.umich.edu/sites/default/files/meta/2022/core/codebook/h22g_r.htm
World Health Organization. (2001). International Classification of Functioning, Disability and Health (ICF). WHO, official classification overview, accessed November 14, 2026. https://www.who.int/standards/classifications/international-classification-of-functioning-disability-and-health
German Federal Ministry of Health. (n.d.). Long-Term Care Insurance. Official program explanation, accessed November 14, 2026. https://www.bundesgesundheitsministerium.de/en/themen/pflege/online-ratgeber-pflege/long-term-care-insurance/page
Singapore Central Provident Fund Board. (2026). What Is CareShield Life? Official program rules, accessed November 14, 2026. https://www.cpf.gov.sg/service/article/what-is-careshield-life
Centers for Medicare & Medicaid Services. (n.d.). Home & Community Based Services. Medicaid.gov, accessed November 14, 2026. https://www.medicaid.gov/medicaid/home-community-based-services
National Association of Insurance Commissioners. (n.d.). Long-Term Care Insurance: Consumer Guidance. NAIC, accessed November 14, 2026. https://content.naic.org/consumer/long-term-care-insurance.htm
OECD. (2025). Health at a Glance 2025: Long-Term Care Workers. OECD Publishing, Paris, section of Health at a Glance 2025: OECD Indicators. DOI: 10.1787/8f9e3f98-en. https://www.oecd.org/en/publications/health-at-a-glance-2025_8f9e3f98-en/full-report/long-term-care-workers_9c3bdbaf.html