Social Health Insurance

The Concept Of Social Health Insurance

Unlike private health insurance (PHI), ‘social’ health insurance (SHI) is characterized by three distinguishing features:

  • Compulsory membership, at least for the great majority of the population.
  • Community rating, i.e., premiums unrelated to individual risk.
  • Open enrollment, i.e., even if the insurance market is competitively structured, an applicant cannot be denied coverage by an insurer.

The theory of social health insurance has two distinct branches. In the normative branch, the main questions are: (1) What are the efficiency and equity reasons for introducing and maintaining such a compulsory institution in a market economy; (2) How should SHI be designed so as to optimally attain the respective goals, both with respect to the benefits covered and to the way how it is financed; and (3) How should a competitive market for SHI be regulated to achieve its targets? The positive branch explains why SHI exists in most democracies and why certain observable features are characteristic.

Normative Theories Of Social Health Insurance

Possible Efficiency Reasons For Social Health Insurance

According to neoclassical welfare economics, the violation of one or more assumptions of the First Welfare Theorem (Mas-Colell et al., 1995) in a particular market is necessary (but by no means sufficient) (for sufficiency, it must be shown that government can create an institution which is suitable to bring about a better allocation) to justify government intervention in this market on efficiency grounds. In the case of SHI, the assumptions typically alluded to are market transparency (which may be violated in the cases discussed in the Sections Asymmetric information in the health insurance market and Insurance against reclassification risk) and independent preferences (for violations see Sections Altruism and free riding and Externalities from medical care).

Asymmetric Information In The Health Insurance Market

If competitive insurance markets are best described by the assumptions of the Rothschild and Stiglitz (RS) model, then asymmetric information of the ‘hidden information’ type (i.e., absence of transparency) may lead to an inefficient market equilibrium. If the insured has more precise information on his individual risk distribution than the insurer, the only possible RS equilibrium is a separating one (a separating equilibrium is an equilibrium in which each risk type is offered a contract which is not bought by other risk types.) in which only the highest risk types are offered complete coverage at actuarially fair premiums. Lower risks obtain more favorable terms but are rationed in terms of coverage. They would prefer to have more but this would make their contract attractive to unfavorable risks. Relative to such an equilibrium, SHI which forces all individuals into a pooling contract with partial coverage can achieve a Pareto improvement: high risks are made better off because they pay lower premiums for the mandated part of their coverage, whereas low risks benefit from improved total (social plus private) coverage (Newhouse, 1996).

However, several objections can be raised against this defense of SHI: first, it is unclear to what extent asymmetric information on health risks is really a problem as medical examinations can be used and are used to determine the risk of an insured. Second, the assumptions of the RS model are highly unrealistic: firms can neither cross-subsidize one insurance plan by another nor anticipate their competitors’ reaction to their own market entry. If these possibilities are taken into account, then the inefficiency of the competitive market equilibrium vanishes and so does the justification of SHI.

Insurance Against Reclassification Risk

In contrast to the adverse selection by Rothschild and Stiglitz (1976) model mentioned in Section Asymmetric information in the health insurance market, private health insurers charge different premiums according to health status. As this may change over time in an unpredictable way, individuals face the risk of uncertain premiums (‘reclassification risk’). As before, the existence of a problem does not per se justify government intervention. Indeed, private insurers may cover this risk in two ways. First, they can offer ‘guaranteed renewable contracts’ that provide a premium guarantee to individuals in exchange for a prepayment (Pauly et al., 1995). Second, Cochrane (1995) proposed to insure premium risk by a separate insurance that pays an indemnity to individuals who become a high risk (‘premium insurance’). However, both of these market solutions suffer from problems. Guaranteed renewable contracts lock consumers in, which means that they will be unable to switch to another insurer in case they are dissatisfied with the service. (Thus, the case for government intervention ultimately rests on the assumption of nontransparent product quality.) Premium insurance contracts are likely to be incomplete because it is difficult to define the risk type with sufficient precision. Thus, SHI with community rating may be the only appropriate solution for the problem of reclassification risk.

Altruism And Free Riding

Altruistic rich members of society may be willing to subsidize the provision of healthcare to the poor if they are more interested in the health than in the subjective well-being of the poor – a case of interdependent preferences. Private charity is not suitable to achieve an efficient allocation as donations to the poor, whether in cash or in kind, have a public-good characteristic because they increase the utility not only of the donor but also of other altruistic members of society. The free rider problem of potential donors could be solved either by a tax-financed National Health Service or a specific system with free healthcare for the poor (such as Medicaid) or an SHI with compulsory membership and contributions according to the ability to pay.

Externalities From Medical Care

Besides the ‘psychological’ externalities described in Section Altruism and free riding, there are physical externalities involved in some types of medical care, in particular in the treatment and isolation of patients with contagious diseases as well as in vaccination services. However, given the limited extent of infectious diseases, it is questionable if these effects are a sufficient rationale for SHI or if there is a weaker interference in free markets such as the subsidization of vaccines.

Optimal Taxation When Health And Income Are Correlated

A related justification of SHI is derived from the theory of optimal taxation. If abilities cannot be observed by tax authorities, the extent to which income taxation can be used for redistribution from the high skilled to the low skilled is limited because the high skilled can always ‘mimic’ the low skilled by reducing their labor supply. However, if there is a negative correlation between ability and the risk of illness, a mandatory SHI with community rating implicitly redistributes between the ability groups in the desired fashion and thus improves social welfare. It must be emphasized, however, that this justification departs from Paretian welfare economics by postulating a specific redistributive goal.

Social Health Insurance And Equity

A further and perhaps the most compelling justification, also known as the ‘principle of solidarity’ relates to the achievement of equality of opportunity: people differ in their health risk already at birth, and some indicators of risk are readily observable. Moreover, with the rapid progress of genetic diagnostics and the spread of tests during pregnancy, the ability to measure individual health risks of newborns will become more and more pronounced. In PHI, these differences in risk immediately translate into differences in premiums so that those persons who are endowed by nature with a lower stock of ‘health capital,’ and are thus already disadvantaged, have to pay a higher price for the same coverage on top of this. Behind the veil of ignorance, one would desire at least an equalization of the monetary costs of illness.

There are in principle two ways to achieve solidarity in health insurance. First, PHI premiums can be subsidized for those who would have to pay excessive contributions. The transfer could be on a current basis or a lump sum, equal to the estimated present value of future excess premiums over the whole expected lifespan of beneficiaries. Both have the important advantage of permitting full competition in PHI (or SHI), including insurers acquiring information about true risk. Besides means testing and the need to define a benchmark contract to determine the amount of the subsidy, the second variant has the disadvantage of shifting the risk of longevity to beneficiaries. The second alternative is a compulsory SHI scheme with open enrollment and community rating that prevents differences in health risk from being translated into differences in contributions but, if combined with a competitive structure, induces cream skimming and therefore requires risk adjustment schemes (RAS; see Section Competition in Social Health Insurance) as a secondary neutralizing regulation.

The Design Of The Benefit Package Of Social Health Insurance

The efficiency reasons given above for the existence of SHI with compulsory membership can be convincing only if the design of the SHI contract is in some sense ‘optimal’ from the point of view of some ‘representative’ consumer. The most important design feature is the depth of coverage or, more precisely, the use of copayment provisions in SHI design. What are the main reasons justifying deviations from full coverage?

Administrative Costs

Copayment provisions can be called for to keep administrative costs low such as costs of handling claims. For this reason, and assuming expected utility maximization on the part of consumers, it is optimal to exclude partially or entirely expenditures on healthcare items that occur frequently but in limited amounts such as minor medications. More specifically, if administrative costs are proportional to the expected volume of health expenditures, a feature of the optimal insurance contract is a fixed deductible, which serves to equalize marginal utility of disposable income in all insured states of the world. Only in the absence of administrative costs would the optimal deductible be zero. However, in some SHI systems such as the German one, doctors and hospitals are paid directly by the sickness funds and the payment is only weakly related to the volume of services provided so that the handling of individual claims from the insured person by the sickness fund is not necessary and thus this reason is irrelevant.

Noninsurable Loss

Illness typically involves not only monetary costs but also nonmonetary losses such as pain and suffering. Optimal health insurance equalizes marginal utility of wealth in all states of nature but this is not equivalent to full coverage if there are complementarities between nonmonetary and monetary losses. In particular, if marginal utility of wealth is lower in case of illness than in good health (e.g., due to reduced ability to enjoy expensive types of consumption), optimal health insurance does not fully reimburse the monetary loss. Although some papers find that marginal utility of consumption is higher in case of sickness, the bulk of the evidence points in the opposite direction, thus supporting the use of copayments for this reason.

Ex Ante Moral Hazard

If the insurer cannot observe preventive effort on the part of the insured, a high degree of coverage reduces the incentive for prevention. Hence, there is a trade-off between risk spreading through insurance and maintaining incentives to keep the risk of illness low. This trade-off leads to a premium function which is convex in the degree of coverage, such that full coverage should be particularly expensive. In SHI such a premium function is nowhere observed, although it could be easily administered because consumers cannot circumvent the convex schedule by purchasing many insurance contracts with limited coverage and low premiums. One reason may be that this type of moral hazard is small due to the nonmonetary costs of illness. Moreover, empirical evidence suggests that people with health insurance live healthier lifestyles, although this may be due to a selection effect. Finally, a system of taxes on harmful and subsidies on healthy consumption goods may be a better alternative (Arnott and Stiglitz, 1986).

A different reasoning applies for prevention through medical services, whose costs can themselves be included in an insurance contract although their occurrence is not a random event. Ellis and Manning (2007) showed that it is efficient to include at least partial reimbursement for preventive services in SHI coverage to align privately optimal demand for these services with the social optimum, which considers the effect of prevention on the premium. In particular, the coverage rate for preventive services should be higher the lower the coinsurance is on treatment (so insurance for treatment and prevention are complementary) and the more risk-averse consumers are.

Ex Post Moral Hazard

If the insurer could observe the health status of the insured, the optimal type of health insurance would provide indemnity payments, i.e., the insurance payment would not depend on the insured’s healthcare expenditure. With asymmetric information, however, linking reimbursement to expenditure is inevitable. Still, copayment provisions are needed to fend against overconsumption of medical care. The optimal copayment rate is higher the more price elastic is the demand for the particular type of medical services. Empirical evidence, for example, from the research and development health insurance study, shows that there is a small, albeit statistically significant, price elasticity of demand for most medical services.

Competition In Social Health Insurance

In an unregulated PHI market, high risks pay higher premiums for the same level of coverage than do low risks. Community rating in SHI prevents this, making ‘cream skimming’ attractive, which in turn runs counter to the aim of open enrollment.

Risk selection can take different forms. Health insurers perform direct risk selection if they influence directly who signs a contract. For example, insurers may ‘lose’ the contract form handed in by a person deemed expensive. Individuals who can be expected to be profitable for the insurer can be encouraged to sign a contract by offering them supplementary services at a discount or, in the extreme case, outright payments.

Indirect risk selection, however, consists in designing benefit packages or by contracting with service providers who are attractive for low risks but unattractive for high risks. In particular, insurers may design their benefit package to attract low but not high risk. An example is a contract with a deductible. This is more appealing for low than for high risks as they face a lower probability of becoming ill and, therefore, of having to pay the deductible. The same reasoning applies to the design of the benefit package in general. For instance, an insurer who covers only few services for patients suffering from diabetes can expect these high risks to prefer another insurer. A straightforward counterstrategy is to impose a maximum deductible and a minimum benefit package. This may not be sufficient, however, because insurers can still try to attract low risks by writing policies with ample coverage of athletic medicine and well-baby care. If these benefits are included in the mandatory package, they will also have to be financed by high risks who have no interest in them (Kifmann, 2002). It may therefore be necessary to specify a maximum benefit package as well.

There are a number of options for complementary regulation designed to limit risk selection. The first is a central fund running an RAS, which pays to the insurer the difference between the expected healthcare cost of the insured and of the average of the respective population. Ex ante equalization of expected healthcare expenditures has the crucial advantage of preserving incentives for cost control but is restricted by the availability of data needed to determine payments from the fund. An alternative are cost-reimbursement schemes. These can be based on total cost, costs by service type, and individual healthcare expenditure. In the latter case, the individuals whose healthcare expenditure is reimbursed can be determined prospectively or retrospectively. Various functional forms of reimbursement can be employed. For example, regulation may prescribe a high-risk pool, in which expenditures of the x% most expensive insured (which are identified on the basis of past experience) are covered by the central fund (Van Barneveld et al., 1996). By contrast, Kifmann and Lorenz (2011) found with data of a Swiss health insurer risk selection can most effectively be prevented if costs are reimbursed only up to a limit.

Income-Related Versus Flat Contributions

A second feature of SHI which is under scrutiny is the base on which contributions are levied, where the choice is:

  1. between income-related and flat contributions, and, in the first case;
  2. on which types of income should be included:
  3. only earnings, or
  4. income from all sources,

and whether or not an income ceiling shall apply.

Presently, all countries with an SHI system except Switzerland levy contributions only on the basis of labor income. (In the Netherlands, the employer’s share of 50% is levied on labor earnings, whereas the employees pay a flat fee.) Historically, this was an application of the principle of equivalence between contributions and benefits as long as the majority of benefits consisted of income replacement in times of sickness. In the meantime, with the rise in scientific medicine throughout the twentieth century the percentage of income replacement in total health insurance expenditures has dropped to the single digits so that this justification is no longer valid.

Nowadays, health insurance benefits are virtually the same for all members (except for differences in risk discussed in Section Social Health Insurance and Equity), and thus income-related contributions constitute pure income redistribution between high and low earners. A possible justification would be the principle of ability to pay. But this is very imperfectly measured if only labor income is taken into account, in particular with the declining share of labor earnings in national income in recent years. Moreover, wage-related contributions imply a significant additional tax on labor, and therefore distort the labor-leisure choice as well as the decision whether to work in the official sector or in the shadow economy. These important disadvantages have to be weighed against a single argument in favor of this particular contribution base, viz. the low costs of collecting the contributions at the source, i.e., as a payroll tax from the employer.

Besides the efficiency reasons just mentioned, there are additional arguments for uncoupling health insurance contributions from income. In particular, the decision making on copayments and other features of the benefit package is distorted if contributions are differentiated according to income: low-income persons have an incentive to opt for ‘too much’ insurance coverage whereas the opposite is true for high-income voters.

The only European country with flat contributions is Switzerland, whereas the Netherlands have a mixed system in which the employer’s contribution is wage related and the employee’s contribution is flat. However, even in Switzerland, effective contributions are not completely independent of income from the point of view of the insured, because low-income households receive a so-called ‘premium subsidy.’ This subsidy varies from canton to canton and covers that part of the total premium of household members which exceeds x% of household income (where x is usually between 7 and 10). Thus effectively, the total contribution amounts to x% of income up to an income ceiling that depends on the number and age of household members and can be calculated as total contribution divided by x. Thus, this system has the same effects like an income-dependent contribution in which total income from all sources is taken into account.

It is sometimes argued that the volume of premium subsidies, which have to be financed from general taxation, will place an enormous stress on the government budget and will therefore be a constant matter of political debate.

Therefore, a better and politically more stable method for compensating the losers from a transition to flat contributions could be a general reform of the tax transfer system, in which social assistance transfers and child allowances are increased and the income tax schedule is appropriately changed (higher initial tax exemption and higher marginal tax rates) so that the pure income redistribution, which is now implicit in the system of health insurance contributions, is performed within the general budget but now with a broader tax base.

Positive Theory: The Political Economy Of Social Health Insurance

Having discussed normative theories of the design of SHI, it is important to assess what can be expected from political decisions in democracies. In particular, the theory should explain:

  1. the apparent lack of generosity of the benefit package of SHI, characterized by the massive use of nonprice rationing methods and the simultaneous presence of private financing of supplementary healthcare services; and
  2. the widespread phenomenon of financing SHI with income-dependent contributions (or taxes) and no risk-rated premiums, in which there is a twofold redistribution from the low to the high risks and from high to low earners.

As to the first point, several models have been put forward (e.g., by Breyer, 1995; Gouveia, 1997), which show that under plausible assumptions on preferences:, the majority of voters will support a two-tier system in which citizens can top up their SHI coverage with PHI (for an opposite result, cf. Hindriks and De Donder, 2003). This is a typical ‘ends-againstthe-middle’ result because the groups of voters who are in favor of a small SHI system are members of the lowest and highest income brackets, whereas middleand high-income earners end up buying supplementary coverage.

To the second point, Kifmann (2005) showed that there can be majority support for a system with income-dependent contributions if the choice of regime is taken at the ‘constitutional stage,’ i.e., before the individuals know their health risk, whereas the details of SHI are decided at a later stage after risk types have been revealed. Then even a high earner can vote in favor of a redistributive SHI if the alternative is private insurance with risk-rated premiums and no insurance against a deteriorating risk type over time. This is particularly likely if individuals are sufficiently risk averse and the premiums in a system with risk rating sufficiently dispersed, whereas income inequality in the society should not be too extreme to make the implicit income redistribution too expensive for the high earners. This may explain why in many countries with SHI, contributions are levied only on labor incomes and even on those only up to a ceiling to limit the volume of income redistribution. However, separating ‘pure’ income redistribution from SHI through flat premiums may be more efficient, but not politically feasible because in such a system the political support for SHI with a generous benefit package, which comes only from the high-risk group, may be too small.


Part of the material used in this survey is adapted from Zweifel and Breyer (2006). Valuable comments from the Editors and Mathias Kifmann are gratefully acknowledged.


  1. Arnott, R. and Stiglitz, J. E. (1986). Moral hazard and optimal commodity taxation. Journal of Public Economics 29, 1–24.
  2. Breyer, F. (1995). The political economy of rationing in social health insurance. Journal of Population Economics 8, 137–148.
  3. Cochrane, J. (1995). Time-consistent health insurance. Journal of Political Economy 103, 445–473.
  4. Ellis, R. P. and Manning, W. G. (2007). Optimal health insurance for prevention and treatment. Journal of Health Economics 26, 1128–1150.
  5. Gouveia, M. (1997). Majority rule and the public provision of a private good. Public Choice 93, 221–244.
  6. Hindriks, J. and De Donder, P. (2003). The politics of redistributive social insurance. Journal of Public Economics 87, 2639–2660.
  7. Kifmann, M. (2002). Community rating in health insurance and different benefit packages. Journal of Health Economics 21, 719–737.
  8. Kifmann, M. (2005). Health insurance in a democracy: Why is it public and why are premiums income related? Public Choice 124, 283–308.
  9. Kifmann, M. and Lorenz, N. (2011). Optimal cost reimbursement of health insurers to reduce risk selection. Health Economics 20, 532–552.
  10. Mas-Colell, A., Whinston, M. D. and Green, J. R. (1995). Microeconomic theory. New York, Oxford: Oxford University Press.
  11. Newhouse, J. P. (1996). Reimbursing health plans and health providers: Efficiency in production versus selection. Journal of Economic Literature 34, 1236–1263.
  12. Pauly, M. V., Kunreuther, H. and Hirth, R. (1995). Guaranteed renewability in insurance. Journal of Risk and Uncertainty 10, 143–156.
  13. Rothschild, M. and Stiglitz, J. (1976). Equilibrium in competitive insurance markets: An essay on the economics of imperfect information. Quarterly Journal of Economics 90, 629–650.
  14. Van Barneveld, E. M., van Vliet, R. C. J. A. and van de Ven, W. P. M. M. (1996). Mandatory high-risk pooling: An approach to reducing incentives for cream skimming. Inquiry 33, 133–143.
  15. Zweifel, P. and Breyer, F. (2006). The economics of social health insurance. In Jones, A. M. (ed.) The elgar companion to health economics, pp. 126–136. Cheltenham: Edward Elgar.
  16. Zweifel, P., Breyer, F. and Kifmann, M. (2009). Health economics, 2nd ed. New York: Springer.


Sample Selection Bias in Health Econometric Models
State Insurance Mandates in the USA