Preconditions for risk-selection via SI

The first precondition is that insurers have to have incentives to perform risk-selection in BI. This implies that health insurers must bear financial risk for the provision of BI, and some risk types must be more attractive than others. Traditionally, BI was provided by non-competing administrative carriers that were fully compensated for the medical costs of their enrollees (Cutler, Reference Cutler2002). However, during the 1990s these BI carriers in several countries were exposed to competition and financial risk. In order to preserve universal financial access, these changes were typically accompanied by premium rate restrictions, open enrolment, and risk-adjusted compensation payments. As far as the risk-adjusted compensation payments are not sufficient to compensate for predictable losses on high-risk individuals, health insurers in these countries face incentives to select risks (van de Ven et al., Reference Van de Ven2003). The second precondition for the use of SI as a selection device in BI is the presence of a link between the purchasing decisions for the two types of insurance. This link must exist in order for health insurers to be able to use SI to influence the consumers' decision to buy BI. For instance, if consumers are legally obliged – or have a strong preference – to obtain SI from the same carrier that provides BI, selective underwriting by insurers of applicants for SI may undermine open enrolment requirements in BI.

Determinants of risk-selection via SI

If both preconditions are present, the probability that SI will be used for risk-selection in BI critically depends on the strength of the incentives for risk-selection in BI and the strength of the links between SI and BI. We distinguish the following determinants of the strength of the incentives for risk-selection in BI. The first crucial determinant is the quality of the risk-equalization scheme. The more accurately insurers are compensated for each risk type, the weaker the incentives for risk-selection are. A second determinant is the level of financial risk for the health insurer. The more the insurer is at risk for the financial results of selling BI, the stronger the incentives for risk-selection. A third determining factor for the incentives faced by insurers is the level of price competition. Strong price competition will force insurers to exploit the available opportunities for risk-selection. By contrast, weak competition leaves room for cross-subsidization, which may reduce incentives for risk-selection, particularly if BI is carried out by non-profit insurers pursuing social goals (Douven and Schut, Reference Douven and Schut2006).

Next, we discern the following determinants of the strength of the links between SI and BI. First, the strength depends on the type of link. In practice, three types can be observed: (1) regulatory or formal links, (2) insurer-established links or tying provisions, and (3) consumer-preferred links.

The strongest link is constituted by a legal requirement that SI and BI have to be sold as a joint product by the same health insurer. Next, insurers may be able to enforce joint purchase by means of tie-in sale provisions in SI policies. For instance, SI contracts may include provisions that the contract will be terminated or a surcharge will be required if the subscriber switches to another BI carrier. Since these tie-in provisions are not legally required, they are likely to constitute a weaker link than a formally enforced one. If the government forbids tie-in provisions or even requires that BI and SI must be sold by different legal entities, insurers may still be able to establish a link between the two. For instance, insurers could establish such a link by joining the same holding company and using the same brand name for SI and BI products. Finally, even in the absence of any formally or insurer-established link, consumers may have strong preferences for a joint purchase of SI and BI. One-stop shopping may be attractive because it lowers search and transaction costs and because it may facilitate the coordination of basic and supplementary benefits.

A second determinant of the strength of the link between SI and BI is the extent to which health insurers are free to set the terms of the contract, the enrolment rules, and the types of benefits covered by SI. Health insurers can effectively use SI for risk-selection in BI by means of selective underwriting, by premium discounts and by benefits design targeted at favourable risk groups (with respect to BI). Selective underwriting can be based on health history questionnaires to SI applicants. By including questions that are particularly relevant to assess the applicant's risk for BI, insurers can subsequently decide to deny SI coverage or calculate a high surcharge to compensate for the expected loss on BI. If allowed, insurers may also decide not to renew the SI contract of enrollees who are unfavourable risks in BI (selective disenrolment).

The design of specific SI-packages is another strategy that insurers may adopt to differentiate between low risks and high risks (e.g. early cancer diagnosis is more likely to be demanded by healthy individuals, whereas cancer therapy by sick people). In the same way, insurers may advertise SI to certain (profitable) risk categories, e.g. by using specific distribution channels (internet, fitness clubs etc.), offering high rebates for deductibles, informing unprofitable enrolees about their right to change insurer and providing bonuses to agents who are successful in getting rid of the most expensive cases by shunting them off to competitors.

Insurers may also attract favourable risks by offering SI premiums below actuarially fair levels. By using health history questionnaires, insurers can determine which applicants are likely to be profitable in BI and they may use these expected profits to offer these applicants a premium discount. Cross-subsidizing SI contracts by profits on BI contracts may be more attractive than lowering the community-rated BI premium. This is particularly the case for new entrants who may be able to use SI for attracting only favourable risks and therefore do not have to recover any losses in the BI market (Kifman, Reference Kifman2005).

Finally, the strength of the link between SI and BI is also determined by the importance of SI for consumers. If SI comprises only a small fraction of individuals' health care expenses or if only a small proportion of the population purchases SI, then the link between both types of insurance is likely to be weak. Hence, both the share of SI in health care financing and the share of the population covered by SI are likely to be positively related to the probability that SI will be used for risk-selection in BI.

Main features of SI

In Belgium, three types of SI can be distinguished, two of which can be used as a tool for risk-selection in the BI market. First, almost all local BI providers offer mandatory supplementary coverage (MSI). It is stipulated in the statutes of these local insurers that SI is mandatory for all their members. Moreover, according to the law all insured are legally obliged to enroll with the same local insurer for both BI and MSI. Second, BI providers offer voluntary supplementary coverage (VSI) on top of BI and MSI (primarily extra hospital services). Again, insured who want to buy VSI are obliged to enroll with the same BI/MSI provider.

From 1994 until 2002, SI expenditures as a share of BI expenditures remained constant at 2% (Table 1), indicating that SI and BI expenditures grew at the same rate. Currently, MSI covers costs related mainly to hospitalization, health care abroad, transportation of the sick, logopaedics, orthodontics, alternative medicine, and home care. VSI expands the coverage of hospital costs. The benefits package is regulated by the law, which obliges every BI provider to offer at least one item in the MSI. In reality, all insurers offer dozens of additional benefits ranging from health promotion to all kinds of advantages such as family allowances in case of childbirth, cheap holidays for children and pre-marriage savings.

The level of concentration of the SI market is low, the market share of the four largest insurers being only about 19.4%. Premium rate restrictions (e.g. community-rated premiums) are absent in both SI markets. In case of financial problems, the Control Office can oblige insurers to take some measures. There is no formal open enrolment in the MSI market. Nevertheless, insurers do not deny access in practice. SI providers are not allowed to exclude new enrollees because of their age or health status. SI contracts can be terminated yearly, subject to three months' notice by the insured.

Incentives for risk-selection

The risk-equalization scheme in Belgium is based on demographic and socio-economic information. There are also some morbidity-related risk-adjusters, but they are not based on diagnostic information, meaning that BI providers can easily identify highly profitable risk groups (Schokkaert and Van de Voorde, Reference Schokkaert and Van de Voorde2003). Despite the rather poor risk-equalization scheme, incentives for risk-selection are substantially mitigated by the limited financial risk for BI providers (7.5% of gains and losses). In addition, the out-of-pocket BI premium is almost negligible, so price competition is hardly feasible. Finally, in absence of competition, BI carriers have no strong drive to select risks, since they are non-profit entities pursuing social goals. Due to the limited financial risk and the absence of price competition, BI providers have relatively weak incentives for risk-selection.

Usefulness of SI as selection device

The law stipulates that the insured must obtain BI and MSI from different entities of the same insurer. These entities must be financially independent and cannot cross-subsidize each other. In practice, however, it is difficult to attribute marketing and administrative costs to the different entities, since the same insurer offers SI and BI. Typically, BI and SI are marketed as joint products. For insurers, the strong formal links make SI a potentially important tool to attract customers.

For Belgian insurers the most straightforward strategies to select favourable risks in SI markets are product and premium differentiation. Premiums are differentiated according to risk categories, mainly age, family composition, and employee/self-employed status. The benefits package is regulated but the freedom to differentiate is sufficiently large to attract specific risk groups. Under the Belgian law, comparative or misleading advertising is forbidden. In addition, the law forbids that benefits are granted with the purpose of inciting switching. In practice, these regulatory restrictions on selective marketing and benefits design are not very effective (see below).

Finally, selective underwriting, particularly in the forms of waiting times and exclusion of pre-existing medical conditions, is allowed and used in the VSI market. Despite the fact that SI offers insurers a potentially effective tool for risk-selection, the usefulness of SI for this purpose is restricted due to the limited importance of SI in health care financing. This together with the weak incentives for risk-selection makes it unlikely that SI is a frequently used tool for risk-selection in BI.

Evidence of risk-selection via SI

Nevertheless, there are indications that SI is increasingly used for favourable risk-selection. Although evidence is largely anecdotal, selective advertising has increased during recent years. Certain insurers selectively promote new insurance products related to sports (such as a reduction of registration fee for the sports club or discounts for children participating in a sports camp). In the leading Belgian newspaper De Standaard (Tegenbos, Reference Tegenbos2005), CEOs of three major insurers' associations (with a combined market share of 77%) accused two other insurers' associations (with a combined market share of 21%) of attracting young and healthy enrollees by offering them advantageous SI contracts. Since 2000, the number of items included in the SI has steadily increased, with substantial differences between the different funds. According to the three CEOs, this risk-selection undermines access, and they appealed to the government to make these selection activities impossible by imposing a uniform standardized benefits package and by prohibiting any marketing activities involving a comparison of SI contracts. Evidence from a survey among VSI members of the Christian insurers showed that specific benefits are particularly attractive to specific risk groups (Christian Mutualities, 2003). Not only enrolment in VSI substantially differed between various age and socio-economic groups, but an analysis of the survey data also revealed a positive relation between characteristics such as education, income, and marital status and the probability of having hospital insurance.

Main features of SI

Two types of supplementary insurance can be distinguished in Germany: MSI and VSI.

MSI is exclusively offered by (most) BI providers. On top of the standardized mandatory basic benefits package (MBI), German insurers are allowed to offer extra benefits, up to a maximum of 5% of BI expenditures (Buchner and Wasem, Reference Buchner and Wasem2003). MSI benefits include services such as spa treatments and hospice treatment. Supplementary benefits are determined in the by-laws of the individual insurer and are mandatory for all subscribers of that insurer (no opting-out). Open enrolment for BI also applies to MSI. Moreover, insurers must charge a single contribution rate for both basic and supplementary benefits. However, in contrast to BI, expenses for MSI are not included in the risk-equalization scheme.

German citizens can also buy supplementary insurance from competing for-profit carriers on a voluntary basis (Wasem et al., Reference Wasem, Greß, Okma, Saltman, Busse and Figueras2004). VSI mainly covers costs regarding upgraded hospital accommodation, dental care, alternative medicine, glasses, and co-payments. The VSI market is unregulated: no premium rate restrictions (e.g. community-rated premiums), no open enrolment requirements, and no standardized benefits. VSI providers are allowed to calculate risk-rated premiums and to exclude pre-existing conditions from coverage. Usually, applicants 55 years or older receive no contract at all. BI carriers are not allowed to offer VSI. VSI comprises only 2% of BI expenditures and 11% of the German population. The market share of the four largest VSI insurers is about 45% (Table 1).

Incentives for risk-selection

The risk-equalization scheme in Germany does not effectively neutralize incentives for risk-selection (Buchner and Wasem, Reference Buchner and Wasem2003). This and the high level of financial risk for health insurers, accountable for 97% of gains and losses, create strong incentives for risk-selection. In addition, the high level of competition due to the presence of about 250 BI carriers, and the absence of entry barriers further reinforce the incentives for risk-selection. For instance, new BI carriers (e.g. Betriebskrankenkassen, BKKs) grew rapidly after entering the market because they were successful in attracting favourable risks by offering low contribution rates (Tamm et al.,Reference Tamm, Tauchmann, Wasem and Greß2006).

Usefulness of SI as a selection device

BI and MSI are tightly linked, since they are offered and marketed as joint products by a single provider at a single price. If subscribers want to opt out, they can only switch to another BI provider that offers a different package of supplementary benefits. The strong link between MSI and BI makes MSI a potentially useful tool to attract favourable risks. Despite this, the usefulness of MSI for risk-selection in BI is restricted by the open enrolment requirement, the absence of a separate premium, and the limited scope of the supplementary benefits that can be included. In fact, the only way that MSI can be used for risk-selection is by the design of the benefits package.

Until 2004, VSI and BI were completely separated by law. Then, the government decided to allow BI providers to act as agents for VSI providers. Most BI providers now cooperate with one VSI provider and offer premium discounts to their subscribers. Moreover, as part of this cooperation, some VSI providers waive their right to reject applicants. However, none of them has waived its right to calculate risk-related premiums. If subscribers switch to another BI provider, in most cases they lose their VSI discount. Opportunities for BI providers to use this link as a tool for risk-selection are rather limited, since BI and VSI providers are also not allowed to cross-subsidize each other and must be financially independent from each other.

Evidence of risk-selection via SI

Nevertheless, there is some evidence that BI carriers use the composition of MSI benefits to attract low risks. A comparison of supplementary benefits for the three main types of BI providers shows substantial differences (Nuscheler and Knaus, Reference Nuscheler and Knaus2005). Whereas most of the traditional regional BI carriers (AOKs) offer benefits that are attractive to the high risks – such as chiro therapy (91%), cancer therapy (64%), homeopathic medicine (70%) – only a small minority of the fast growing and lower-priced BKK funds offer these benefits (14% offers cancer therapy, 26% homeopathic medicine, and 33% chiro therapy). By contrast, BKKs more often than AOKs offer benefits that are attractive to favourable risk groups, such as health checkups (25% vs. 9%) and cancer screening (25% vs. 0%). Given the weak and only recently established link between BI and VSI providers, risk-selection via VSI does not seem to be an issue yet. However, many of VSI contracts offer rebates that may be lost if enrollees switch BI carrier. Andersen and Grabka (Reference Andersen, Grabka, Göpffarth, Greß, Jacobs and Wasem2006) conclude that attractive SI packages had an impact on switching. However, the risk profiles of switchers are unknown. MSI and VSI may become a more important tool for risk-selection in the near future, if the basic benefit package is further reduced. However, in contrast to the trend of a gradual expansion of the role of SI in health care financing, the 2006 German health care reform plan includes a transfer of a number MSI benefits (palliative care and some spa treatments) and some VSI benefits (acupuncture) to the basic health insurance package.

Main features of SI

In Israel, SI can be bought voluntarily and on an individual basis from the same insurer providing BI, a commercial insurer, or both. During the period 1995–2002, the share in the population buying SI from a BI provider rapidly increased from 35% to 66%, while the share buying SI from a commercial insurer increased from 16% to 24% (the share buying both types of SI grew from 5% to 20%). Since only SI provided by BI carriers can be used as a tool for risk-selection in BI, we focus our analysis on this type of SI. Currently, SI covers costs of some surgical interventions in Israel and abroad, dental care, preventive screening, alternative medicine, co-payments in BI (especially for drugs), and IVF. In the period 1999–2002, SI expenditures as a share of BI expenditures grew from 3.3% to 5.3% (Table 1). In Israel, the four largest BI/SI providers together hold 91% of the SI market share. There is an open enrolment requirement and insurers may charge age-related premiums for SI coverage after government approval. Law does not determine the composition of benefits package and the contractual conditions. Nevertheless, the extent of product differentiation is still moderate (Brammli-Greenberg and Gross, Reference Brammli-Greenberg and Gross2003).

Incentives for risk-selection

The quality of the risk-equalization scheme in Israel is quite poor (only age-related subsidies), therefore BI providers can easily identify risk groups that are highly unprofitable, such as chronically ill people (Shmueli et al., Reference Shmueli, Chernivkovsky and Zmora2003). Since BI suppliers are fully at risk and are not allowed to charge a premium for BI in addition to the risk-adjusted capitation payments they receive from the government, the incentives for risk-selection are strong. These incentives are mitigated, however, by the weakly competitive structure of the BI market. Traditionally the BI market is dominated by a few non-profit insurers, while legal entry barriers effectively prohibit any potential competition. Switching rates are low (about 1% per year) (Laske-Aldershof et al., Reference Laske-Aldershof, Schut, Beck, Greß, Shmueli and Van de Voorde2004).

Usefulness of SI as a selection device

BI providers are allowed to offer SI only to their own enrollees. Alternatively, BI insured may choose to buy SI coverage from commercial insurers, but they cannot obtain SI from another BI provider. BI providers exercise the double function of SI and BI suppliers in a regulated context. The law requires that BI/SI providers keep a separate financial administration for SI and BI. Cross-subsidization between BI and SI is forbidden and providers are not allowed to sell other types of insurance. Despite the strong link between BI and SI, the usefulness of SI as a selection device is restricted by open enrolment and by the limited freedom for insurers to differentiate premiums. The limited role of SI in health care financing and the low switching rates also reduce the usefulness of SI as a selection device.

Evidence of risk-selection via SI

Currently, there is no evidence of insurers using SI for risk-selection in BI. Although the share of SI expenditures on BI expenditures is still small (5%), it steadily increased during the last decade. If this trend continues, SI is likely to become a more useful tool for risk-selection, particularly if the risk-equalization scheme is not improved.

Main features of SI

In the Netherlands, SI is exclusively voluntary. More than 90% of BI insured buy SI. SI can be bought on an individual basis or via group contracts. Currently, SI covers almost all dental care for adults, alternative medicine, maternal home care, physical therapy, psychotherapy, anticonceptives, and IVF. In the last decade, the number of services excluded from the basic benefits package and covered by SI has steadily increased. The same applies to SI expenditures as a share of BI spending, which grew from 2% in 1994 to 6.5% in 2003 (Table 1). The Dutch SI market is not regulated by the government. There is no open enrolment requirement and there are no restrictions on premium rate setting and the benefits package. Although risk-rating is allowed, insurers still charge predominantly community-rated premiums (Schut et al., Reference Schut, Laske-Aldershof and de Bruijn2004). The most likely explanation for this is that SI is traditionally offered as a by-product of BI. Since almost all insured buy SI and BI from the same insurer and switching rates were low until 2006, there was hardly any competitive pressure to differentiate SI premiums (Laske-Aldershof et al., Reference Laske-Aldershof, Schut, Beck, Greß, Shmueli and Van de Voorde2004; Douven and Schut, Reference Douven and Schut2006).

Incentives for risk-selection

The risk-equalization scheme in the Netherlands is the most sophisticated among the five countries. Nevertheless, BI providers can still easily identify risk groups that are highly unprofitable relative to the community-rated premium that BI providers have to charge (Prinsze et al., Reference Prinsze, van de Ven, de Bruijn and Schut2005). Since BI carriers are accountable for about 50% of gains and losses, insurers can substantially benefit from risk-selection. Prior to 2006, incentives for risk-selection were mitigated by the weak competition among non-profit BI providers due to the limited propensity of consumers to switch to another insurer (Laske-Aldershof et al., Reference Laske-Aldershof, Schut, Beck, Greß, Shmueli and Van de Voorde2004; Douven and Schut, Reference Douven and Schut2006). Since 1 January 2006, the Dutch health insurance system has been profoundly reformed by the introduction of a new MBI scheme. The former distinction between social health insurance (MBI) for low/middle income groups and private health insurance (VBI) for high-income groups has been abolished. Under the new BI scheme, all enrollees can switch plans at annual open enrolment periods (two months). BI premiums have to be community-rated and a risk-equalization scheme applies as before. Dutch citizens had to choose a new contract for BI and SI. Anticipating that many customers would consider switching, insurers engaged in a price war (Douven and Schut, Reference Douven and Schut2006). Indeed, the switching rate increased dramatically, from about 3% to 18% (NZa i.o., Reference NZa2006). Although switching rates in 2006 are likely to be high because of the radical change in choice setting, price competition is expected to remain strong in the future. Strong price competition substantially increases the incentives for risk-selection in the new BI scheme.

Usefulness of SI as a selection device

In the Netherlands, there is no regulation that requires insurers to link SI with BI coverage. On the contrary, both types of insurance have to be offered by different juridical entities. Cross-subsidization between BI and SI providers is forbidden and the providers must be financially independent. Despite the separation of providers, SI and BI have always been sold as a joint product under the same brand name (Schut et al., Reference Schut, Laske-Aldershof and de Bruijn2004). SI and BI providers typically belong to the same holding company, making marketing and administrative costs difficult to ascribe to the different entities. Prior to 2006, most SI contracts had a clause that the contract would be automatically terminated once the insured would switch to another BI provider. Although termination clauses are forbidden under the new BI scheme, this prohibition does not effectively preclude tie-in sales. For instance, several insurers made it clear that they would charge higher SI premiums to enrollees choosing other BI providers and some insurers offer SI only to applicants that obtain BI from the same company. In practice, almost all consumers still buy SI and BI from the same company (NZa i.o., Reference NZa2006).

Since SI providers are free to set premiums, determine the benefits package, and apply medical underwriting, SI can be effectively used for risk-selection in BI. If consumer mobility remains high after 2006, this would make SI a particularly powerful tool to discriminate between different risk groups.

Other straightforward strategies would be to use health questionnaires to identify favourable and unfavourable risk groups that apply for SI. Such strategies, however, are effective only if a substantial proportion of the insured is willing to switch (Schut and Hassink, Reference Schut and Hassink2002).

Given the links between BI and SI and the absence of legal restrictions on underwriting, product differentiation, and premium setting, SI is a moderately useful tool for risk-selection.

Evidence of risk-selection via SI

Before 2006, both the incentives for risk-selection and the usefulness of SI as a selection device were limited. Survey results indicate that less than 1% of the applicants were refused, despite the fact that insurers used health questionnaires for SI applicants (Bruijn et al., Reference Bruijn, Prinsze, Schut and van de Ven2005; Laske-Aldershof and Schut, Reference Laske-Aldershof and Schut2005). Since 2006, price competition and consumer mobility increased, which made SI a powerful selection device. However, to accommodate the transfer to the new BI scheme, all insurers promised to accept all applicants for SI without medical underwriting. Recent investigations show that in 2006 only a few insurers used health questionnaires for SI applicants and, except for extensive dental coverage, no applicants were refused (Bruijn and Schut, Reference Bruijn and Schut2006; NZa i.o., Reference NZa2006). Since the promise to accept all SI applicants only holds for 2006, the probability that SI will be used as selection device is likely to increase in the future. A possible counteracting factor may be the damaging effect of such behaviour on an insurer's reputation. For several years already, the National Patient Federation (NPCF), the Ministry of Health, and the Dutch Health Authority (NZa) have monitored the underwriting practices of SI providers and they publish the results of their investigations (Laske-Aldershof and Schut,Reference Bruijn, Prinsze, Schut and van de Ven2005; Bruijn et al., Reference Bruijn, Prinsze, Schut and van de Ven2005; NZa i.o., Reference NZa2006).

Main features of SI

In Switzerland, SI is exclusively voluntary and covers dental care, sick-leave payments, alternative medicine, upgraded hospital accommodation, access to all physicians and hospitals all over the country, cross-border care, and transportation costs for accidents in the mountains. Prior to the health care reform of 1996, SI accounted for almost 30% of BI expenditures. Due to the introduction of a comprehensive BI scheme, however, the share of SI on BI expenditures decreased to about 20% (Table 1). Nevertheless, the popularity of SI has grown, with the share of BI insured holding SI coverage increasing from 62% in 1997 to 71% in 2003. Government intervention in the SI market is limited. There are no open enrolment requirements, standardized benefits, and premium rate restrictions. Three types of SI carriers can be distinguished: independent SI providers, daughter companies of BI carriers, and integrated BI/SI providers. Consumers are free to choose among these three SI carriers. The regulatory regimes that apply to daughter companies and integrated providers are quite different. Integrated insurers are restricted to the exclusive provision of health coverage by law, while daughter companies may offer all kinds of insurance coverage besides health insurance. Combined with the more liberal supervisory regime, this explains the growing popularity of daughter companies at the expense of integrated ones.

Incentives for risk-selection

The risk-equalization scheme in Switzerland is based only on age, gender, and region, implying that BI providers can easily identify risk groups that are highly (un)profitable. Moreover, BI carriers are fully at risk and there is no reinsurance or state aid in case of losses. Therefore, insurers have to charge sufficiently high premiums to survive. Since insurers face substantial competition, the incentives for risk-selection are strong.

Usefulness of SI as a selection device

Formally, basic and supplementary benefits are strictly separated. It is legally forbidden to link rebates in SI to BI, i.e. SI carriers are not allowed to cover the medical expenses that fall under voluntary deductibles in BI. In terms of accounting requirements, SI providers are obliged to keep their financial administration separated from BI providers. Cross-subsidization between BI and SI providers and tie-in sale provisions in SI contracts are also forbidden.

Despite the legal separation between BI and SI carriers, most SI carriers are daughter companies of or integrated with BI carriers, which makes it difficult to establish whether marketing and administration costs refer to BI or SI carriers. Moreover, despite the fact that tying-in is forbidden, when people switch BI provider they usually terminate their contract with the related SI carrier. This likely reflects a consumer preference for a joint purchase of BI and SI. Since BI and SI benefits overlap to some extent, consumers may not want to figure out which plan is responsible to cover the costs of care once they fall ill. Moreover, insurers try to prevent consumers from a separate purchase of BI and SI ‘by all kind of tricks, for instance by taking away the premium discount for families or by surcharges on the premium for extra administrative expenses' (Beobachter Kompakt, Reference Beobachter2006). Beck (Reference Beck2004) also found that having SI significantly reduces the probability to switch. Hence, despite the separation between BI and SI, most consumers purchase both products from the same company. Given the important role of SI in health care financing and the possibility of selective underwriting, selective advertising, and product and premium differentiation, the joint purchase of SI and BI makes SI a potentially useful tool for risk-selection in the BI market.

Evidence of risk-selection via SI

Beck (Reference Beck2006) shows that the impact of risk-selection on premiums increased from 1997 to 2006 by a factor of 12 and can explain a substantial part of the variation in premiums. Although empirical research is hampered by the fact that risk-selection is a hidden activity by insurers, there is substantial evidence suggesting that risk-selection via SI in the BI market is becoming increasingly important (Colombo, Reference Colombo2001; Beck, Reference Beck2004). Each year several insurers launch new SI products that are particularly attractive to healthy customers. In addition, SI premiums are increasingly risk-rated and new SI products for upgraded hospital accommodation are targeted at specific age groups (e.g. SI policies were launched under the label “Hospital 20” and “Hospital 30”). These new SI products are attractive for young people as long as they pass the mandatory health check.