رتبه بندی جامعه و بازار برای بیمه سلامتی غیر گروهی خصوصی

Prior research on adverse selection in health insurance markets has found only mixed evidence for adverse selection in group settings. We examine the impact of state community rating regulations enacted in the 1990s, which greatly limited insurers' ability to risk rate premiums, to determine if adverse selection is more evident in non-group insurance markets. Using data from large, national surveys we find evidence of a shift to a less healthy pool of non-group enrollees as a consequence of community rating. Community rating made healthy people 20 to 60% less likely to be insured by non-group health insurance; in addition, we found evidence that young and healthy people were 20 to 30% more likely to be uninsured as a result of community rating. We also find evidence that individuals in poor health were 35 to 50% more likely to be insured in the non-group market, but only limited evidence suggesting that persons in poor health were less likely to be uninsured. Our results are further supported by findings suggesting that non-group enrollees were sicker as a result of the community rating laws. Lastly, we find evidence suggesting that HMO penetration in the non-group market increased disproportionately in states that implemented community rating relative to states that did not.

The phenomenon of adverse selection in health insurance markets has been noted since at least Arrow (1963), yet only recently have economists made some headway in empirically demonstrating adverse selection. Insurance provided within large groups is unlikely to be affected by significant adverse selection, but as the group size shrinks there is more potential for individual enrollees to influence overall health care expenditures. Hence adverse selection becomes incrementally more likely, though still not assured, in small groups. It is therefore not surprising that work on the subject has been mixed: some studies have found little evidence of adverse selection in small group markets (see for example Buchmueller and DiNardo, 2002), while others have found some evidence of adverse selection (see Simon, 2005 and Monheit Schone, 2003). As the group size diminishes to one the presence of adverse selection becomes less ambiguous. However, another vein of research has highlighted the importance of multiple forms of heterogeneity that cloud the ability to identify adverse selection in insurance markets more generally (see Chiappori et al., 2006). Finkelstein and McGarry (2006) examine long-term care insurance markets and determine that purchasers are heterogeneous in the type of private information they have: some are high risk and thus purchase more insurance and use more services while others have a strong taste for insurance and thus purchase more insurance but use less services. Fang et al. (2008) study the Medigap insurance market and find that cognitive ability dominates risk preference and other characteristics as a factor explaining the decision to purchase coverage. Our goal is to identify and understand the nature of adverse selection in non-group health insurance markets. Specifically, we examine how state community rating regulations combined with guaranteed issue laws, which eliminate insurers' ability to experience rate premiums based on characteristics associated with health care expenditure risk such as age, gender, or health status, affected the purchase of non-group insurance by different risk groups and how the composition of the risk pool changed as a result of the regulations. Our research differs from most prior work because we estimate the impact of community rating on both ends of the risk distribution: the healthy and the sick. We examine whether the regulations served to induce some people to purchase non-group coverage and other people to drop non-group coverage, and whether we observe reciprocal effects on uninsurance. We find that community rating was associated with a 20–60% reduction in the likelihood of being covered by non-group health insurance though no significant increase in the probability of being uninsured among healthier individuals overall. However, we do find that among young healthy unmarried men community rating laws are associated with both a decrease in non-group coverage and a 20–30% increase in uninsurance. In contrast, non-group coverage increased 35–50% among unhealthier individuals, though we did not find consistent evidence that uninsurance decreased among those in poor health. The combined impact of these results are further supported by examining the impact of community rating on health status characteristics and health utilization of persons with non-group insurance before and after community rating for a subset of states, which suggests that the enrollees as a group were sicker after community rating laws were enacted. This evidence is consistent with the hypothesis that community rating worsened the extent of adverse selection in the non-group market. However, perhaps tempering the adverse selection effect, we also find evidence that HMO penetration in the non-group market increased disproportionately in states that implemented community rating relative to states that did not. 2. Community rating in the non-group health insurance market During the early- and mid-1990s states were active with regulatory efforts aimed at improving the perceived inequities and inefficiencies in the small and non-group health insurance markets. While small group regulatory efforts were more common, in many cases the small group regulations were done in tandem with similar regulations implemented in the non-group health insurance market. Of the numerous regulations enacted by states in the non-group market during the 1990s2 the regulatory regime that is most likely to have an unambiguous effect on health insurance markets is community rating combined with a guaranteed issue requirement. Community rating requires insurance carriers to charge the same premiums for all plan participants regardless of age, gender, health status, or other factors.3 By restricting the ability of insurers to charge differential premiums by risk, states intended to create a market in which those in poor health would not be “discriminated” against in the form of higher premiums. Table 1 lists the eight states that enacted non-group community rating laws and guaranteed issue for all non-group health insurance products during the 1990s.4 Note that of the states that enacted community rating, most enacted modified community rating laws, which allows limited premium variation by specified demographic characteristics (typically age) or region. Although modified community rating is likely to diminish somewhat the potential for adverse selection, the absence of underwriting for health status represents a qualitatively different regulatory regime in relation to non-community rated states. Similarly, the bite of the policy is greatly augmented by the guaranteed issue requirement because it further restricts the insurer's ability to limit coverage to those deemed a profitable risk. The table also indicates availability of data for residents of the states in our two data sources, the Survey of Income and Program Participation (SIPP) and the National Health Interview Survey (NHIS).

Our results suggest that community rating of the non-group health insurance market was associated with a significant change in the risk composition of the non-group market. Using data from large, national surveys we found compelling evidence of adverse selection in the non-group health insurance market: community rating made healthy people less likely to be insured by non-group health insurance. We found less consistent evidence that healthy people were more likely to be uninsured as a result of community rating, suggesting that alternatives likely in the form of group health insurance are available for most healthy individuals. Increases in uninsurance were evident, however, for young healthy males, who apparently are more willing to self-insure or have more difficulty finding jobs with group insurance. We also found evidence that the unhealthy were more likely to be insured through non-group policies yet only limited evidence suggesting that the unhealthy were less likely to be uninsured. The lack of evidence suggesting a consistent reduction in uninsurance associated with community rating, even for older unhealthier individuals, would appear to suggest that the policy's primary goal was not achieved. Our non-group take-up results are further supported by examining the impact of community rating on the health status characteristics and health utilization of persons with non-group insurance before and after community rating in a subset of states, which suggested that enrollees as a group were sicker as a result of the community rating laws. We also found evidence that HMO penetration in the non-group market increased disproportionately in states that implemented community rating relative to states that did not implement community rating. It is difficult to rule out that increasing HMO penetration might have played a part in bringing about community rating, but if the goal of non-HMO incumbent insurers was to reduce HMO penetration our evidence suggests that the aim was not achieved. Given the lack of information regarding the pricing of non-group insurance products, it is difficult to estimate the deadweight loss associated with community rating. We can, however, gain some sense of the implicit transfer from the healthy to sick in the presence of community rating by using the 1997 Medical Expenditure Panel Survey (MEPS) data set to estimate actuarially fair premiums for each individual in the SIPP sample to approximate a non-community rated environment. To generate an estimate of the premium under community rating we calculate the weighted average of predicted health care expenditures for the sample of SIPP respondents who purchased non-group policies under a community rated regulatory regime.19 The estimated community rated annual premium from this procedure is $2275. The implicit redistribution from the remaining healthy individuals in the non-group market to the sick non-group policy holders is associated with healthy paying roughly $1040 or 84% more per year than the efficient premium for the policies. The amount represents a sizeable implicit transfer to the unhealthy non-group enrollees who are paying less than fair premiums for their policies.