1 Why risk adjustment and risk equalization?

An increasing number of countries permit a periodic consumer choice of insurer in their social health insurance schemes (e.g. Belgium, the Czech Republic, Germany, Israel, the Netherlands, Slovakia, Switzerland and Medicare in the United States). A major goal of social health insurance systemsFootnote 1 is to provide access to affordable coverage. However, a competitive health insurance market tends toward risk-adjusted premiums, which may range from less than €400 to €40,000 or more per year, and risk selection (rejection by insurers of high-risk applicants). Therefore, in all these countries, insurers are not allowed to reject applicants (open enrollment) and one or more of the following strategies have been implemented to make health insurance affordable:

1. 1. Risk-adjusted subsidies based on an enrollee’s risk profile, for example, the risk factors insurers would use in risk-rating premiums. These subsidies are given to insurers (rather than enrollees) to reduce transaction costs and increase transparency. In a transparent competitive market, insurers are forced to use the subsidy to reduce each enrollee’s premium. Giving risk-adjusted subsidies to insurers equalizes the different risks enrollees represent. This way of organizing risk-adjusted subsidies is therefore referred to as risk equalization.

2. 2. Ex-post cost-based compensation.Footnote 2 Insurers are fully or partly compensated by government if an enrollee’s health-care expenses exceed a certain annual threshold. Such compensation will be reflected in premium reductions, particularly for high-risk enrollees.

3. 3. Premium subsidies to enrollees. These subsidies are based on the premium the consumer pays. Those who pay a high premium receive a high subsidy and those who pay a low premium receive a low subsidy (or no subsidy).

4. 4. Premium rate restrictions. All of the above-mentioned countries require premiums to be community-rated, that is, insurers must charge the same premium for the same product to each enrollee, independent of the enrollee’s risk profile.

Risk equalization is the preferred strategy to guarantee affordable coverage in a competitive individual health insurance market (Van de Ven et al., Reference Van de Ven, Van Vliet, Schut and Van Barneveld2000). However, in practice, risk adjustment is still imperfect (see below). Each of the other strategies has substantial drawbacks, resulting in serious trade-offs. Ex-post cost-based compensation reduces insurers’ incentives to operate efficiently (see e.g. Newhouse, Reference Newhouse1996). Premium subsidies to enrollees reduce the incentive for high-risk individuals to shop around for the lowest premium. By lowering the competitive advantage of the most efficient insurers, they also reduce price competition and insurers’ incentives to operate efficiently. To the extent that there is consumer choice of coverage, they may also encourage high-risk individuals to buy more insurance than they would have done in the absence of a subsidy (moral hazard), potentially resulting in welfare loss. Both ex-post cost-based compensation and premium subsidies to enrollees therefore present policymakers with a trade-off between affordability and efficiency.

Community rating also has major drawbacks. Although its goal is to create implicit cross-subsidies from low risks to high risks in the same pool, this pooling creates predictable profits and losses for subgroups in the pool, thereby giving insurers incentives for risk selection, which can have several unfavorable effects (see Box 1). Community rating therefore presents policymakers with a trade-off between affordability and risk selection.

In addition, it is questionable to what extent community rating can be effective in the long term in achieving its goal, because product differentiation (if allowed) may result in indirect premium differentiation.Footnote 3 Such risk segmentation across the product spectrum can be observed in voluntary health insurance markets in Australia, Ireland, the US Medicare Advantage program and South Africa, where premiums must be community rated (Colombo and Tapay, Reference Colombo and Tapay2003; Gale, Reference Gale2005; Armstrong, Reference Armstrong2010; Mcleod and Grobler, Reference McLeod and Grobler2010). In this way, community rating per product results in low premiums for low risks and high premiums for high risks, which undermines community rating.

The importance of good risk adjustment is that if it is sufficiently refined (see below), there is no need for any of the other strategies. The better risk equalization is adjusted for risk, the less severe are the trade-offs the other strategies present.

2 Risk adjustment and risk equalization: state of the art in practice

The quality of the risk equalization system in many countries can be qualified as poor to moderate (see e.g. Van de Ven et al., Reference Van de Ven, Beck, Van de Voorde, Wasem and Zmora2007). The most sophisticated risk-adjustment formulas can be found in the US Medicare program (Pope et al., Reference Pope, Kautter, Ellis, Ash, Ayanian, Iezzoni, Ingber, Levy and Robst2004) and the Netherlands statutory health insurance scheme (Van de Ven et al., Reference Van de Ven, van Vliet and Lamers2004). A crucial question is: how good are these formulas?

The results of an evaluation of the Dutch risk equalization formula in 2007 indicate that the formula provides insufficient compensation for groups defined by health status, prior utilization and prior expenses (Stam and Van de Ven, Reference Stam and Van de Ven2007). These groups can be easily identified by insurers and their size ranges from less than 1% (e.g. people with a hospital admission in three of the last five years) to 40% of the population (people who have had contact with a medical specialist in the last three years). In case of an average premium, the average predictable losses per adult in these subgroups are in the order of hundreds to thousands of Euros per person per year. For example, given an average community-rated premium, the average predictable loss per adult for 21% of the population who report their health status as being fair/poor equals €541 (i.e. 33% of the overall average expenses per person). The results also indicate predictable losses for the groups of insured whose disease is included as a risk adjuster in the equalization formula (e.g. heart problems and cancer). Clearly, not all of these patients fulfill the criteria to be classified as a patient eligible for a higher equalization payment.

So an important finding is that even sophisticated risk adjustment formulas currently in practice are not yet sufficiently refined and – due to community rating – leave ample opportunity for risk selection.

All countries with a risk equalization scheme in their competitive social health insurance market require premiums to be community-rated, with the aim of making health insurance affordable for high risks. As argued above, given the insufficiently refined risk adjustment, this requirement creates incentives for risk segmentation and risk selection. In European countries, particularly Germany and Switzerland (Van de Ven et al., Reference Van de Ven, Beck, Van de Voorde, Wasem and Zmora2007), there is evidence of increasing risk selection. Other countries (e.g. Belgium and the Netherlands) have reduced incentives for risk selection by giving insurers ex-post cost-based compensation. However, this compensation reduces insurers’ incentives for efficiency. Therefore, given the imperfect risk adjustment, there is a trade-off between affordability, risk selection and efficiency. Improving the risk equalization scheme is the only effective strategy to resolve this complicated trade-off.

It is important to emphasize that the prevention of risk selection does not require predictable profits and losses to be reduced to zero. Insurers may incur costs in selecting risks. For example, a bad reputation resulting from selection activities such as preventing patients from accessing the highest quality care can be costly to an insurer. In addition, the information necessary for risk selection may also be associated with costs. If predictable losses and profits after risk equalization are sufficiently small, this may not be a problem, because an insurer has to take into account the costs of selection and (statistical) uncertainty about the net benefit of selection.Footnote 5 Therefore, a ‘perfect’ risk equalization formula is not necessary. It should be refined to such an extent that insurers expect the costs of selection (including the cost of a bad reputation) to outweigh its benefits. By making the risk groups in the equalization more homogeneous, the costs of selection increase while on average its profits fall. But it is still an unanswered question how much ‘imperfection’ is acceptable.

Good risk equalization is essential. Without it, the disadvantages of a competitive insurance market (largely due to risk selection) may outweigh the advantages. However, international experience suggests that in practice the implementation of even the simplest risk equalization scheme is very complex (Van de Ven et al., Reference Van de Ven, Beck, Van de Voorde, Wasem and Zmora2007; Armstrong et al., Reference Armstrong, McLeod, Paolucci and Van de Ven2010).

3 What needs to be done?

Given that the current risk equalization schemes in most countries are far from perfect, the first priority should be further investment in improvement. Investment is needed both in better data and in research and development of better risk adjusters. Since it is still questionable whether in practice a sufficiently refined risk adjustment system can be developed, policy makers should also seriously consider the use of ex-post cost-based compensation and premium subsidies to enrollees, while at the same time reconsidering the use of community rating, which induces incentives for risk selection and may not be effective in the long term. Policymakers must understand that risk selection is not inherent to the competitive insurance market, but is primarily the result of one possible form of regulation (community rating), and that alternative forms of regulation result in other outcomes.