Potential forms of risk sharing

The purpose of this section is to highlight the essential elements of risk sharing and the choices that can be made. Because risk sharing can be seen as a

3. Forms of risk sharillg

mandatory reinsurance program for the insurers in which the regulator acts as the reinsurer, the essential elements of risk sharing are similar to those of a reinsurance contract (see e.g. Carter, 1979). Let us assume that the period to which the risk sharing applies is one year for all members for whom some risk is shared. Besides this period, Table 3.1 presents four other essential elements (Van Barneveld et a\., 1999b).

Table 3.1 Foul' essential elements of a form of risk sharing (I) The group of members for whom some risk is shared (2) The types of care for which the risk is shared

(3) The extent of the risk that is shared

(4) The price that insurers have to pay to share some risk

3.2.1 The members

If the members for whom some risk is shared are known in advance of the contract year, we can speak of prmpective risk sharing. If these members become known during the contract year, the risk sharing can be called retro-spective. An extreme case is risk sharing for all members. Then the distinction between prospective risk sharing and retrospective risk sharing is irrelevant.

In the case of prospective risk sharing, the regulator may stipulate for which members some risk will be shared, for instance members with a certain medical condition (AIDS patients, cancer patients, transplantation patients) or with high prior costs. This can be called prospective cOlldition-specific risk sharing. Under this form of risk sharing the percentage of members for whom some risk is shared will probably not be the same for all insurers. Another possibility is that an insurer is free to select a fixed percentage of its members whose costs then are (partially) shared. This form of risk sharing will be referred to as risk sharillg for high-risks (RSHR). With risk sharing for high-risks, the fraction of members for whom some risk is shared may be set the same for each insurer or it may vary over the insurers (Van Barneveld et a\., 1996). In the latter case, the percentage would depend, preferably, on the risk that an insurer represents as far as this risk is not reflected in the capitation payments. However, the 54

3.2 Potellliai forms of risk sharing

option of varying percentages of designated members will not be explored in the empirical analysis because the present study focuses on the insurers' incentives for selection and not on the consequences of risk sharing at insurer level.

Moreover it seems likely that the question of how to differentiate the percen-tages of members that insurers are allowed to designate will be difficult to answer.

An alternative for prospective risk sharing can be retrospective risk sharing. As with prospective risk sharing, the regulator may stipulate which members are eligible for risk sharing, for instance members that have or develop a certain (medical) condition in the contract year [(new) AIDS patients, (new) cancer patients, (new) transplantation patients] or members who died during the year.

This can be called a retrospective condition-specific risk sharing. Again the regulator can also leave it up to the insurers to select a fixed percentage of their members for risk sharing. This type of retrospeetive risk sharing will be called risk sharing for high-costs (RSHC). As with risk sharing for high-risks, it is possible to differentiate the percentage of members for whom some risk is shared per insurer which could improve the effects of risk sharing at insurer level. However, for the same reasons as mentioned above with respect to risk sharing for high-risks, this form of risk sharing for high-costs will not be explored in the empirical analysis.

Condition-specific risk sharing, whether prospective or retrospective, requires a list of (medical) conditions that make members eligible for risk sharing. It is likely that such forms of risk sharing create discussions over which (medical) conditions make members eligible for risk sharing (Swartz, 1995). Condition-specific risk sharing could conflict with privacy rights of members and there may be a possibility of manipulation, namely by 'inflating' diagnoses to make members eligible for risk sharing (Swartz, 1995). The extent of the latter problem can be mitigated by using only (medical) conditions with relatively low discretion as well as a good monitoring system".

Risk sharing for high-risks and risk sharing for high-costs can be seen as

12 DisclIssions on condition-specific risk sharing are similar to those on diagnosis-based risk adjustment.

3. Fol'll/s of risk sharing

alternatives for prospective and retrospective condition-specific risk sharing respectively. Under risk sharing for high-risks and for high-costs, it is up to the insurers themselves to decide for which of their members they want to share some risk, so discussions over eligibility can be avoided. The regulator only has to specify the percentage of members that each insurer is allowed to designate for risk sharing. Given a certain average percentage of designated members per insurer, it could be that the group of members for whom some risk is shared under risk sharing for high-risks is similar to the group of members under prospective condition-specific risk sharing. In that case the insurers' incentives for selection as well as for efficiency will be similar. The same holds for risk sharing for high-costs in comparison with retrospective condition-specific risk sharing.

For the reasons mentioned above, it can be concluded that risk sharing for high-risks and risk sharing for high-costs have better administrative feasibility and are less vulnerable to manipulation than condition-specific risk sharing, while the consequences for the insurers' incentives for selection and efficiency can be similar. Therefore condition-specific risk sharing will not be analyzed empirical-ly in the second part of the study.

3.2.2 The types of care

The types of care for which the risk is shared could be set the same for each member that is designated or it could vary between these members (Wallack et a!., 1988). In the latter case, the regulator has to stipulate which types of care can be distinguished in the risk sharing and the insurers have to register their (designated) members' expenditures per type of care. Fm1hermore the regulator will have to determine the price of risk sharing that an insurer has to pay, for each of these different types of care.

If insurers would be allowed to decide themselves for which types of care the risk is shared, prospective risk sharing might simulate - to some extent - the exclusion of pre-existing medical conditions. However, a distinction between several types of care within the specified benefits package does not seem to go well with a flexible description of the benefits package. This is a major element of many competitive health care reforms and should provide insurers with tools 56

3.2 Potential fOf'lllS of risk sharing

for cost-effective substitution of care. A form of risk sharing that makes a distinction between several types of care might stimulate undesirable substitution of care. This could happen if the risk for relatively expensive types of care is shared and the risk for relatively inexpensive types of care is not. Furthermore such a form of risk sharing might be ' gamed' by the insurers if they register health care expenditures of one type of care (for which expenditures are shared) under another type of care (for which expenditures are not shared). In sum risk sharing with a distinction between several types of care within the specified benefits package may provide perverse incentives to insurers, and will be rather difficult to register and to monitor. For these reasons the remainder of this study assumes that the risk sharing applies to all types of care within the specified benefits packageIJ. If the risk for all types of care within the spec-ified benefits package is shared, risk sharing for high-risks might simulate - to a certain extent - the refusals to sell insurance to high-risk applicants for whom an appropriate premium (capitation payment) can not be calculated.

3.2.3 The extent

The extent of risk sharing may be the same for all members that are designated for it or it may vary between these members (Newhouse, 1986). In the latter case it may be the regulator that decides the extent of risk sharing or it can be left to the insurers themselves. Such flexible forms of risk sharing would require that the regulator determines a price for every possible extent of risk sharing. This might be a difficult task. For this reason it is assumed that the extent of risk sharing is the same for all members for whom some risk is shared.

The risk sharing applies to an individual level and not to an aggregate level (Gruenberg et a/., 1986). Risk sharing on an aggregate level could resemble stop-loss reinsurance. This form of risk sharing would primarily concern the aggregate financial result of insurers. Because the present study focuses on tradeoffs between incentives for selection and efficiency and not on the

conse-Il An argument in favor of risk sharing for specific types of care within the specified benefits package is that in some circumstances, insurers have no tools to improve efficiency for certain types of care. For this reason in the Netherlands the sickness funds have been given little financial responsibility for production-independent hospital costs since 1996.

3. Forms of risk sharing

quences of risk sharing at insurer level, this form of risk sharing will not be analyzed empirically. In the present study risk sharing reimbursements will be based on the actual costs of individual members for whom some risk is shared.

Then the risk sharing may apply to all costs of a designated member, may be limited to the member's costs above his normative costs, or may be limited to the costs above a certain threshold. In each of these variants risk sharing may apply to a certain percentage of the cost involved. Of course it is possible to use more than one threshold in combination with different percentages of the costs involved. However, to avoid UlUlecessary complications, the empirical analyses are restricted to only one preset threshold (if any) together with one percentage of the costs involved. If the risk sharing applies to all members and all expendi-tures above a certain threshold are shared, the risk sharing is called outlier risk sharing (ORS). It resembles excess-of-loss reinsurance (Bovbjerg, 1992). If the risk sharing applies to all members and a certain percentage of all expenditures is shared, the risk sharing is called proportional risk sharing (PRS). This form of risk sharing resembles quota-share reinsurance (Bovbjerg, 1992).

3.2.4 The price

The financing of a risk sharing arrangement can take several forms. Risk sharing may be financed externally, internally or via a combination of both.

External financing means that there is some flow of money towards the whole

,

payment system,. for instance from the govermnent. For example Beebe's (1992) outlier pool was assumed to be financed this way. With such a flow of money, risk sharing is not budget-neutral from the regulator's point of view.

However, in this study it is assumed that the regulator requires risk sharing to be budget -neutral at the macro-level. This can be achieved by using an internal financing mechanism i.e. the risk sharing is financed via mandatory contribu-tions from all insurers in the market. This way the risk sharing only shifts (limited) amounts of money from some insurers to others. In practice the price that an insurer has to pay to the regulator can be calculated at the end of the year, when the proportion shared expenditures is known. In the empirical analyses it will be assumed that the nonnative costs, on which the capitation payments are based, are reduced proportionally to finance the risk sharing. A disadvantage of this financing method is that it is not necessarily budget-neutral 58

3.2 Potential forll/s of risk sharing

for each of the risk groups that are distinguished in the capitation formula. This may create some new selection problems. However, for limited extents of risk sharing, there is no reason to assume that a refinement of the financing method would alter the conclusions of this study (see chapter seven).