Market Responses to
Kentucky’s Michael Clark and GinnyWilson
The authors would like to thank Greg Freedman and D.J.
Wasson for their help Introduction In 1994, the Kentucky General Assembly began a series of three major revisions to the laws governing health insurance in the state. The first revision was designed to provide relief for policyholders facing unaffordable premiums because they were judged by insurance carriers to pose a greater risk of large claims. One feature of the legislation was to limit the ability of insurance companies to charge different premiums based on individuals’ characteristics such as age, gender, and health status. Supporters of the measure generally expected that this change would substantially lower premiums for individuals with a high probability of filing large and numerous claims, and somewhat increase the premiums for those with a low probability of filing large and numerous claims. However, subsequent legislative and executive provisions created an opportunity for policyholders to segment themselves by risk, essentially allowing low-risk individuals to avoid the higher premiums which were intended to offset the lower premiums of high-risk people. This article discusses how the health insurance market has reacted to these changes. Legislative History of Health Inrsuance There are two "pure" types of rating structures that insurance carriers might use to price policies. These are experience rating and community rating. With experience rating, insurance carriers price policies to reflect the expected claims of the policyholder; those with higher expected claims are charged more than those with lower expected claims. With community rating, the expected claims of the entire covered group is estimated and policyholders pay an average premium sufficient to pay total group claims. The basic distinction between these two pricing structures is whether risks are pooled or segmented. Under an experience-rated pricing system for health insurance, the health risks associated with particular policyholders are segmented. This means that those with similarly low expected claims are placed in one category and charged a low price, reflecting their low level of risk, while those with higher expected claims are placed in another category and charged a higher price, reflecting their similar level of risk. Thus, those with different risks are segmented into defined categories, with differing prices attached to each category, according to the level of risk the category represents. The risk categorization of any policyholder is only valid for the time period covered by a specific contract, usually a contract year. At the time the contract is renewed, the risk expectations for each policyholder are re-evaluated, and a policyholder may be assigned to a new category if his or her risk status has changed. The other pure form of insurance pricing, community rating, is based on risk pooling. Under community rating everyone pays a premium closer to the average for the whole group and those who move into a different risk category do not see a parallel increase in premiums. In pure community rating, risks are pooled across all individuals in the group. Under this insurance pricing scheme, those with lower expected claims, who tend to be younger and male, generally subsidize the premiums of those with higher expected claims, who tend to be older or female. Community rating yields a price that is more stable across subsequent periods for all individual policyholders but which is higher for those who would benefit from risk segmentation. Modifications to pure community rating, such as allowing adjustments for age, gender, or other factors, simply restrict the risk pooling to categories of individuals who share specified demographic characteristics. Even with modifications for some demographic factors, however, community rating remains essentially different from experience rating, in that health status factors are not used to adjust community rates. Prior to the initial action by the General Assembly in 1994, insurance carriers in Kentucky had no restrictions on their rating structures. Although rate increases for some carriers were subject to approval by the Kentucky Department of Insurance, the criteria used for approval were that the increases be actuarially justified and that they preserve the financial health of the carrier. Insurance carriers did not have to issue policies to those with high expected claims, waiting periods and pre-existing condition exclusions were set as desired, and most premiums were set through experience rating. These conditions led to a situation in which some policyholders with high expected claims were being denied coverage or were being priced out of their policies. A rate filing from the Department of Insurance showed that from 1980 to 1993, rates increased by 1,600 percent for one group of policyholders. The increase in the medical care Consumer Price Index (a medical care inflation rate) for the same period was 105 percent. Employers and individual purchasers of insurance were complaining about large price increases, and those with employer-provided policies were afraid to change jobs for fear that long waiting periods or pre-existing condition exclusions would leave them exposed to both financial and medical catastrophe. 1994 Legislation In response to these conditions, the 1994 Kentucky General Assembly enacted House Bill 250 (HB 250). HB 250 made many changes in Kentucky’s health care market, but the ones pertinent to this discussion were those relating to insurance carriers’ underwriting rules and rating structures. Kentuckians who purchased insurance in the small-group and individual markets were the prime targets of changes in these rules because they were believed to be the most negatively affected by experience rating and denial of coverage. Individuals purchasing health insurance in large groups were more insulated from the particular impact their own claims might have on the average costs for the whole group. HB 250 required guaranteed issue and guaranteed renewal. Guaranteed issue and guaranteed renewal mean that carriers must provide a policy to all applicants willing to pay the premium. In the absence of these requirements, health insurance carriers could refuse to issue or renew policies to those policyholders believed to represent an unacceptable risk of large claims. HB 250 also mandated a modified community rating structure for policies sold to small groups, individuals, and participants in the Kentucky Health Purchasing Alliance. In setting rates for these policyholders, carriers could consider age, geography, family composition (single, couple, parent plus children, family), richness of benefits, and cost-containment restrictions on choice. Age adjustments were allowed only to the extent that the oldest policyholder in the group could be charged no more than three times the amount charged the youngest, and geographic adjustments were limited to a total of 15 percent. Neither gender nor claims experience could be considered in setting rates. The provisions of HB 250 became effective in July 1995. Around that time, insurance carriers began to withdraw from the Kentucky market. Prior to 1994, only health maintenance organization (HMO) products and individual indemnity products had to be reviewed by the Department of Insurance. Therefore, there is no information on the total number of carriers operating in the state at that time. Following the enactment of HB 250, however, 28 carriers notified the Department of their decision to withdraw from Kentucky. In the period between the enactment of HB 250 (July 1994) and SB 343 (July 1996), three Executive Orders were issued extending coverage issued prior to July 15, 1995, to individuals, small groups, large groups, and trade associations. This extension, which was granted through July 15, 1996, was given, in part, to allow carriers to maintain existing policies until legislative changes anticipated in the 1996 Regular Legislative Session could be evaluated. These extensions were given for individuals, small groups, and small associations at the option of the insured and with no ability for carriers to increase rates. The Executive Order issued for large groups and large associations permitted rate increases, however. The Executive Orders did not prohibit a carrier from terminating all business and withdrawing from the health insurance market. Rather, carriers wishing to remain in the market were required to continue renewing the pre-reform policies. 1996 Legislation As the General Assembly convened in Regular Session in January 1996, withdrawal of insurance carriers from the Kentucky market was forcing many policyholders to find new coverage. Legislative committees heard evidence that some remaining carriers were conducting mass mailings to policyholders warning of restricted choice and higher premiums. Policyholders who believed they had been harmed by the changes were vocal in expressing their displeasure. In response, the 1996 General Assembly passed Senate Bill 343 (SB 343), which allowed gender to be used as an adjustment factor for community rates. Consideration of past or expected future claims experience was still not allowed. Another change in the reforms was that SB 343 exempted certain trade associations from rating restrictions, allowing the associations to provide health insurance that varied by customers’ characteristics, including health status. Another Executive Order was issued on June 5, 1997, again permitting coverage under pre-reform policies to be extended, at the option of the insured. Carriers were permitted, however, to file rate increases on that business. In the interim period following the enactment of SB 343 and the Special Legislative Session on Health Insurance in October 1997, an additional 23 carriers withdrew from the insurance market in Kentucky. No changes were enacted in the Special Legislative Session on health care called by Governor Paul Patton in October 1997. Although it was anticipated that amendments to SB 343 would be made by the 1998 General Assembly, the Patton administration did not issue an additional Executive Order to extend coverage under pre-reform health benefit plans through the end of the Regular Session. Rather, an Executive Order was issued prohibiting the renewal of pre-reform health benefit plans after December 1, 1997. During the time that the freeze was in place, premiums for these policies did not increase. This extension was made to allow carriers time to complete any necessary internal administrative and filing issues with the Department of Insurance. Following the end of the Special Legislative Session, six more carriers notified the Department of their withdrawal from the state insurance market. 1998 Legislation Thus, between 1994 and the end of 1997, a total of 57 carriers ceased selling policies in Kentucky. Many of those that remained reported major losses. In the individual market, only Anthem and Kentucky Kare, the plan created to be a self-insured plan for state government employees, continued to offer policies. To address the loss of carriers to the market, the 1998 General Assembly passed HB 315. This legislation essentially returned the market to experience rating, with a few exceptions. HB 315 stipulated that, for any demographic characteristics except health status, the highest rate charged can be no more than five times the lowest rate. For individuals with the same demographic characteristics, those with higher expected claims because of a health condition can be charged an additional 135 percent of the index rate (the median rate for the covered group for that carrier). HB 315 also designated a list of health conditions, such as juvenile diabetes and leukemia, as high-cost conditions. Because people diagnosed with these high-cost conditions were likely to have higher claims, insurance companies were permitted to charge people with these conditions that did not have prior coverage up to 150 percent of the index rate. For those without one of the high-cost conditions listed, these rules essentially returned the market to the rate spread that existed in 1994, where the highest non-high-risk premium can be 11 times greater than the lowest premium. Those who would have been denied coverage prior to 1994 could purchase coverage at these rates. Once issued, all policies are guaranteed renewable. Premiums Under Risk Segmenting and Risk Pooling In the course of completing the analysis requested by the Department of Insurance and the General Assembly, the authors had access to approximately 1.5 million records covering the claims incurred by participants in Kentucky Kare from June 1995 to August 1997. The data included:
Although the file included data for claims processed from June 1995 to August 1997, there was often a lag of several months between when a claim was incurred and when it was processed. Therefore, the decision was to analyze primarily claims incurred in calendar year 1996. Kentucky Kare paid approximately $117 million for 590,000 claims incurred in 1996. The average payment per claim was $199, while the maximum payment was $190,000. The median payment per claim was $44. Although Kentucky Kare is not representative of all insured people in Kentucky, these data do provide a useful illustration of the relationship between gender, age, and claims.1 This relationship provides the basis for risk segmentation in an experience-rated market. Table 1 shows the breakdown of total claims paid in Kentucky Kare by age and gender. It is apparent that women between the ages of 21 and 60 accounted for a much larger share of total payments than did men of the same ages. The average cost of claims was not substantially different between males and females, except that the average claim for the oldest males was more than double that for females of the same age. The gender difference in total payments is accounted for by the fact that, for ages 21 to 60, women had twice as many claims as men.
To further investigate these differences, we made specific comparisons by gender for those ages 21 and 64. Table 2 shows the covered lives in each age and gender category, along with the total claims paid for that category and the claims paid per covered life. The data indicate that claims paid per covered life for young women were 2.3 times those for young men. This gender ratio declines as age increases, until it is nearly equal for those above age 50.
A comparison within genders for the effect of age shows a large difference as well. For women, the claims paid per covered life for those ages 61 – 64 were 2.6 times that of women ages 21 – 30. For men, the oldest category had claims per covered life 6.7 times that of the youngest category. These data show how the differences in average costs are related to gender and age. These relationships generally hold for sufficiently large numbers of customers, and insurance companies have been able to use the relationships to vary premiums by demographic characteristics. On average, young males are less expense to cover, so their premiums are relatively low. If insurance companies can also determine the health status of an individual through medical histories or testing, they can use that information to alter premiums further. Effects of Community Rating on Premiums Under pure community rating, risk segmenting is not permitted and all insured persons would pay the same rate. For example, Kentucky Kare would have needed to charge an annual premium of approximately $1,600 per covered life to equal their total claims in 1996. This would have created a situation where those with low expected health care costs would have subsidized those with high expected health care costs. Alternatively, under modified community rating, Kentucky Kare might have segmented their customers by gender and age. When doing so, young males would be charged relatively low premiums while older males would be charged higher premiums. The subsidization under modified community rating is less than with pure community rating. Within each risk group, however, healthier persons would still be subsidizing less healthy people. Finally, under experience rating Kentucky Kare could have further segmented young males by health status, eliminating the subsidy across health status. As discussed, the market for health insurance was an experience-rated market prior to the reforms of 1996. With the passage of HB 250 and SB 343, however, insurance carriers were restricted in how they could segment risk, redistributing the cost of health care across demographic groups in the individual and small group markets. Based on rate filings from the period, an actuary retained by the Department of Insurance estimated that, in 1994, individual health insurance rates for young women were 150 percent of those for young men, rates for older men were 450 percent of those for young men, and rates for those with the highest expected claims (and who were still allowed to purchase a policy) were 250 percent of those with the lowest expected claims. This meant that, for those allowed to purchase a policy, the highest rates could be 11 times more than the lowest rates. Given a rate of $68 per month for the youngest, healthiest males and the rating spread above, rates for a comparable policy for others can be calculated, holding all other factors constant. These rates are shown in Section 1 of Table 3. These rates did not reflect coverage for those with very high expected claims, who were denied coverage.
Starting with the actuarial assumptions, and using the age, gender, and health status distributions of those in the Kentucky individual insurance market, the Legislative Research Commission made simulations of the effect of changes in pricing rules on the distribution of rates under HB 250 and SB 343.2 The effect of the demographic factors on the distribution of rates among policyholders under the various pricing schemes is quite clear. Under experience rating, young, healthy males faced a relatively lower premium, while those who were female, older, or had poorer health status faced higher premiums. The effect of changes in the rules of issue and pricing enacted in HB 250 on this distribution of rates is shown in Section 2 of Table 3. Compared to experience rating, the new rules would have required those who were younger and healthier to subsidize the purchase of insurance by those who were older and less healthy. Also, young men would have subsidized the purchase of insurance by young women.
One point to be made is that, all else held constant (a major assumption), the changes contained in HB 250 only addressed the problems of those who could not afford health insurance because of high premiums. Those unable to purchase insurance because of low incomes would not have been helped. To the extent that the young are more likely to be healthy but have low incomes, they would have been less likely to be able to afford health insurance under these rules. SB 343 somewhat softened these effects by allowing greater consideration of age and by allowing gender differences in pricing (Table 3, Section 3). Since health status could not be considered, however, those with good health continued to subsidize those with poor health, and those with low incomes might still be priced out of the market.
HB 315 largely eliminated the subsidization of premiums for those with large expected claims costs. Only to the extent that claims of those with medical conditions not defined as high-cost exceed 135 percent of the median for those of the same age and gender, and those with high-cost conditions exceed 150 percent of the median for those of the same age and gender, will there be cross-subsidization by those with lower expected claims. The provisions established in HB 315 essentially returned the distribution of rates to that existing in the unregulated market of 1994, except that those with high cost conditions now have a price ceiling. Expected Claims and Adverse Selection As shown in the previous section, moving from an experience-rated market to a modified community-rated market redistributes the cost of health care by making premiums for all people the same, regardless of their health risk. This makes health insurance more affordable for the high-risk individuals but also raises the premiums for low-risk individuals. Because low-risk individuals are paying the higher premiums that reflect the risk of the community, they have an incentive to separate themselves from the high-risk individuals. This is an example of "adverse selection," which can have substantial impacts on the market for health insurance. The classic prediction of adverse selection is that the healthiest individuals will separate themselves from the rest of the market by dropping their health insurance. Those remaining in the market will be generally in poorer health than those who dropped out. Therefore, premiums must increase to reflect that expected average costs per person would be higher. This increase would then force more people out of the market and raise premiums even higher. Eventually, this cycle of rising premiums would eliminate the insurance market. This was not likely to be the result in Kentucky, however. The combination of legislative changes and the executive order freezing pre-reform insurance policies gave low-risk individuals options for separating themselves from high-risk individuals without having to drop coverage. Although HB 250 established modified community rating (MCR), the executive order freezing policies in effect prior to the implementation of HB 250 allowed people to keep policies priced under experience rating. For the healthy, the pre-reform policies were likely less expensive than policies rated under MCR, providing them with an incentive to keep their old policies. Unhealthy individuals, however, could potentially receive lower premiums under MCR. In addition, SB 343 exempted associations from the MCR restrictions, allowing them to set premiums based on health status. Individuals purchasing health insurance through an association could be charged a lower premium if they were healthy, or a higher premium if they were unhealthy, than they would pay if they purchased a non-association policy. Under this pricing structure, people who are healthier than average had an incentive to seek out the lower premiums charged by the associations, while those who are in poorer health would prefer to be insured under MCR. Both the freeze of pre-reform policies and the exemption of associations from MCR provided an avenue for the healthiest individuals to pay lower rates than they would under MCR. Theory suggests that those in poorer health, therefore, would have been more likely to obtain health insurance under MCR and those in better health would have been more likely to be covered under a pre-reform plan or an association plan. Since the premium charged under MCR reflects the average risk of those insured under MCR, the healthiest individuals would always be able to find a cheaper premium through an association. Eventually only those in very poor health would continue to purchase coverage under MCR. The premiums they would be charged would reflect their high risk. To determine whether adverse selection did occur in the individual market, we estimated the effect that health status had on whether an individual was covered under a modified community-rated policy (reform policy) or under an experienced-rated policy (pre-reform or association policy) using data from the 1997 Kentucky Health Insurance Survey. This survey interviewed Kentucky households about various aspects of their health insurance. To simplify the analysis, only single coverage policies were considered. Although those purchasing family plans face the same situation, it is difficult to analyze insurance decisions of families because of the interaction of family members with different characteristics. Analysis was also restricted to adults under age 65, because Medicare should provide coverage for those 65 and over. There were 155 individuals under 65 with private single coverage. Of those, 85 were covered under a pre-reform or association plan. That is, they either kept coverage that was in effect prior to HB 250 or they purchased coverage through an association. Table 4 summarizes the data.
Whether an individual purchases an MCR policy or an experienced-rated policy depends on a number of factors. As discussed, the difference in premium is a key determinant when deciding the type of policy to purchase. Under reform policies in Kentucky, premiums were based on gender, age, and geographic location. In addition to these factors, health status could be considered in the pricing of pre-reform and association policies. For the reasons noted above, poor health status is predicted to increase the probability that an individual would have purchased a reform policy. Respondents were asked if their health in general was excellent, very good, good, fair, or poor. Those with fair or poor health were defined as being "unhealthy." It is uncertain whether age would have had an effect on price differences, because under both community rating and experience rating, age could be used as a rating factor. However, MCR restricted carriers’ use of age so that the premium they charged the oldest person could only be three times greater than the premium they charged the youngest person. If this restriction was a binding constraint, then younger individuals might have been able to find lower premiums through experience-rated policies. The effect of gender on premium difference is uncertain. While HB 250 prevented rating based on gender, SB 343 lifted the restriction. In addition, because of the complexity of setting premiums under reforms, some people may not have accurately understood how the different rating mechanisms affect them. Therefore, people who are better informed are expected to make better decisions. Those who are in poor health and better informed are more likely to purchase coverage under MCR. Those who are in good health and better informed are more likely to be covered under an experienced-rated policy. Although it was not possible to determine each respondent’s level of understanding of health reforms, their highest level of education was used as a substitute measure for this understanding. Analysis of the data showed that health status did have a statistically significant effect on the probability that a person was covered by a reform policy. The probability that those with excellent, very good, or good health status would be covered through an experience-rated policy was 59 percent. The probability for those with only fair or poor health was 26 percent. This suggests that adverse selection did exist in the Kentucky market for individual health insurance. The large difference can be partially explained by the fact that people who were previously priced out of the market because of health conditions were able to purchase health insurance at lower prices after reforms. This probably resulted in a surge of unhealthy people purchasing reform plans. Even without an increase of unhealthy people entering the market, there was still evidence of adverse selection. Analysis of only those who had coverage prior to reforms yielded similar results. Healthy people were more likely to be covered under pre-reform or association plans. The results also show that older insured persons were more likely to purchase experience-rated coverage. This finding was somewhat unexpected, because reforms limited the rating spread on age. This limit should have increased the subsidy to older individuals. It may be that older people were more likely to have health insurance prior to the reforms and may have been less willing to change their coverage. Even though they might have found lower premiums by switching plans, they may have been reluctant to drop the coverage they had for some time for a new policy. Males were no more likely to be covered by an experienced-rated policy than females. This result is not surprising because SB 343 allowed gender rating for all policies, which should have reduced any cross-gender subsidy that may have existed under HB 250. Although it was expected that education would improve a person’s choice, it did not have a significant effect. It should be noted that the estimates above reflect a point in time when healthy individuals had two avenues for avoiding MCR policies: pre-reform plans and association plans. Eventually, however, the freeze on pre-reform plans ended. After December 1, 1997, all pre-reform plans were supposed to be converted, upon their renewal, to comply with current law. This provision removed one of the ways healthy people were able to avoid MCR plans and may have slowed the effects of adverse selection. The presence of adverse selection in the Kentucky market could have had major effects on the outcome of health insurance reforms. Adverse selection provides an incentive for the healthiest individuals insured under MCR to seek cheaper coverage through an association, where they are not subsidizing less healthy people. As these people leave the community, those remaining have a higher expected cost per person, and premiums under MCR must rise to reflect these higher costs. As premiums rise, more people will find it beneficial to obtain coverage through associations, further reducing the subsidy to the unhealthy individuals remaining in the community pool. The end result of adverse selection is the Kentucky market’s reverting back to a pure experience-rated market. The market was not allowed to move to this end, however, because passage of HB 315 eliminated most of the restrictions on risk segmenting. Conclusion Adverse selection occurred in the Kentucky insurance market as a market response to price changes. As health insurance premiums for low-risk insured persons increased under modified community rating, these people looked for alternatives to paying the higher rates. One alternative is for people to drop their coverage and self-insure. The exemption of associations from the restrictions against risk-segmenting and the executive order permitting pre-reform policies to be renewed, however, allowed low-risk people to avoid the higher premiums associated with MCR and still maintain insurance coverage. Low-risk people are therefore always better off with a pre-reform or association policy. Eventually this situation would have returned the market to pure experience rating. The legislative changes made in HB 315 largely reduced the potential for adverse selection but did not completely eliminate it. Under HB 315 carriers can once again segment their customers according to health status. But carriers cannot deny coverage to high-risk people, and the premium they charge them is capped. Therefore, the subsidy across health status is reduced but not eliminated. Even though there is a subsidy across health status, it is much more difficult for low-risk people to avoid the subsidy. Because pre-reform policies can no longer be renewed and associations have the same rating restrictions as the rest of the market, low-risk people can only avoid the subsidy by dropping coverage, making it much less likely that adverse selection will occur. Endnotes
This document is a © of CBER, Center for Business and Economic |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||