AN EVALUATION OF OHIO'S
HEALTH INSURANCE
MARKET REFORMS
 

Wake Forest University School of Medicine
Mark A. Hall, J.D., Principal Investigator
December 1998

I. EXECUTIVE SUMMARY

II. BACKGROUND AND METHODS

This study evaluates how well Ohio's health insurance market reforms have met their objectives and whether they have avoided possible harms and failures. This is part of an intensive case study of seven states that have enacted varying reforms (Colorado, Florida, Iowa, New York, North Carolina, Ohio and Vermont), funded by the Robert Wood Johnson Foundation. This multiple-case study consists primarily of two rounds of structured, in-depth, open-ended interviews, as well as an analysis of documentary and secondary data. The principal reforms we evaluated are contained in the 1993 law (known as House Bill 478) that provides, in the individual and/or small-group markets, for: (1) open enrollment and guaranteed issue, (2) renewability and portability of coverage, (3) rating restrictions, (4) restrictions on underwriting practices such as risk selection and preexisting condition exclusions, (5) a reinsurance mechanism, and (6) sales through purchasing alliances and private associations. This report is intended to inform lawmakers, regulators, insurers, agents, purchasers, and the public policy community whether and how state and federal reforms have achieved their multiple purposes or caused any negative consequences, and how these reforms might be improved. The following is a summary of the major findings.

Almost all of our interview subjects had generally favorable views about the overall impact of Ohio's health insurance market reforms. These are its most successful features:

• The law has not caused market disruptions and has generated few complaints.
• The market remains highly competitive in price, product diversity, and number of carriers.
• Enrollment has increased and prices have held steady.
• Private purchasing cooperatives and associations are popular and self-sustaining.
• Enforcement and implementation has not been difficult or burdensome.

• Open enrollment and guaranteed issue have had only a minuscule impact on coverage of high-risk groups.
• Loose rating rules allow insurers to deter most high-risk subscribers and hamper portability.
• Carriers have found a number of techniques to continue competing on the basis of risk selection, including field underwriting through agents.
• A number of practices of questionable legality are in fairly widespread use, such as list billing and excluding sick members of employer groups.
• Reinsurance has been little used and has not reduced risk pool differentials.

A. Methods

The primary sources of information for this study are various components of the insurance industry. In Ohio, we conducted 21 interviews of 34 people in the spring, summer and fall of 1997, and an additional round of 15 interviews was conducted with 22 of these same people or their designated replacements in the summer of 1998. Represented in this interview pool were five officials with the Ohio Department of Insurance (DOI), six independent agents active in small-group sales, 19 people at eight insurance carriers (four in-state and four out-of-state), and five administrators of two purchasing cooperatives. These were semi-structured interviews based on an interview guide, but the discussions were free-ranging and the coverage of topics varied somewhat among them. Five subjects were interviewed by phone, and 16 were in person. Most interviews were one-on-one lasting 1-2 hours, but some were in groups of 2-4, and five interviews (covering a total of 11 of the interview subjects) were out-of-state in which Ohio issues were addressed only briefly.

We also collected quantitative and documentary information in the form of market activity data, sales literature, and newspaper articles. Finally, we conducted a market testing study to determine the ability of an actual small employer and unhealthy individual to obtain insurance. An employer with three employees contacted 18 agents throughout the state in late 1997 and early 1998 to inquire about the availability of health insurance for the group of three as well as for a group of two plus individual coverage for one unhealthy employee. These multiple sources of information and data were analyzed using both qualitative and quantitative techniques.

This report is organized in two main sections. The first section reviews the history, purpose, and content of these reforms. The second section presents our findings, to evaluate whether these reforms achieved their purposes and avoided potential harms or failures.

Before we begin, a word or two is required about terminology. Health insurance, like any other industry, has a specialized vocabulary with terms of art that sometimes differ from common understandings, and that often are used inconsistently even within the industry, due in part to regulatory differences among the states. For our present purposes, we value simplicity over precision, so we will use a lay vocabulary that glosses over many of the distinctions that are important within the industry. Thus, we use "insurer" to include, generically, both indemnity carriers and health maintenance organizations (HMOs). We use Amanaged care" to refer primarily to HMO plans, including point-of-service (POS), in contrast with Aindemnity," by which we mean both traditional unconstrained fee-for-service as well as more managed forms of indemnity such as preferred provider organizations (PPOs). When we speak of agents, we generally intend to refer to independent agents, which are sometimes called brokers. We use the terms premium, price, and rate interchangeably to refer to how much insurance costs. And, by health insurance, we mean comprehensive major medical, in contrast with more limited or specialized coverages. Other, more technical, terms will be defined later in the context in which they are important.

A final piece of background information is to note how Ohio compares with other states. Ohio is a populous state, with about 10 million people. Demographically and economically, it resembles national averages, with these exceptions (the following figures are from 1992-93 data). A smaller portion of the work force is self-employed (5.4% in Ohio versus 7.5% nationally) and a smaller portion is composed of small firms with 25 or fewer workers (25.1% in Ohio versus 29.3% nationally). The economic base is somewhat more heavily weighted toward manufacturing. Ohio health statistics are generally somewhat more favorable than national averages. The percentage of the population without insurance is somewhat lower than the national average (see below), as is the percentage of small firms (25 and under) without insurance in 1992 (44% in Ohio versus 48% nationally).

B. Content and Purpose of Reform

The major piece of insurance reform legislation in Ohio is House Bill 478, enacted in 1992, effective January 1993. Initially, the law applied only to groups of 2-25, but it was expanded in Senate Bill 150, effective November 1995, to reach groups of 2-50, with certain exceptions. The main exception is for the open-enrollment mechanisms. They extend only to groups up to 25 employees, but they also encompass individual health insurance. More recently, in July 1997, Ohio adopted amendments to bring its law in compliance with the federal law known as the Health Insurance Portability and Accountability Act (HIPAA).

The starting point of reform is to make sure that any willing purchaser has access to insurance and can retain that insurance through subsequent renewal periods. "Guaranteed issue" requires all insurers who participate in the small-group market to accept any applicant year round, and "open enrollment" requires this only during a designated period. An important distinction exists between states that require only designated policies to be issued and those that require guaranteed issue or open enrollment of all policy types marketed by an insurer. Prior to HIPAA, Ohio had only an open-enrollment requirement for designated products. Following HIPAA in 1997, open enrollment was replaced in the small-group market by guaranteed issue of all small-group products. Open enrollment remains in the individual market, however.

Prior to HIPAA, Ohio had several different open-enrollment mechanisms, some of which existed prior to the reform law, and only some of which now exist. The reform law adopted the open-enrollment requirement that applied to commercial indemnity insurers. They were required to take open enrollment all year, but only until each insurer took in .5% of its existing book of business for the relevant portion of the market. The market quota was based separately on the insurers' individual versus small-group market shares, and for this purpose the small-group portion reached only up to groups of 25. The open-enrollment requirement for HMOs was adopted prior to the reform law. It required HMOs to select a single month each year to take open enrollment for up to 1% of their total book of business. Both of these open-enrollment requirements applied only to two state-mandated benefit plans, which differ between indemnity and HMO insurers.

A third open-enrollment requirement, which also existed prior to the reform law, applied to nonprofit mutual insurers such as Blue Cross, but with critical differences. These insurers were required only to receive and evaluate applications from any person or small group during open enrollment; they were not required to accept all applicants. Despite the label "open enrollment," which one regulator called a "ruse," nonprofit indemnity insurers are still allowed free reign in rating and medical underwriting.

Each of these open-enrollment requirements still remains in effect in the individual market. In the small-group market, however, open enrollment was replaced in 1997 with guaranteed issue of all products. Ohio also uses open enrollment as its "alternative mechanism" to comply with HIPAA's requirement of coverage for individuals leaving group coverage and seeking to purchase insurance individually. The focus of this evaluation will be on the open-enrollment system as it existed prior to HIPAA, and on the guaranteed-issue requirement for small groups that was imposed by HIPAA.

The reform law promotes continuity of coverage in three ways. First, insurers are prohibited from refusing to renew insurance except for fraud, nonpayment or similar malfeasance. The second aspect of continuity is to regulate the use of preexisting condition exclusion clauses. Insurers are prohibited from ridering out specific health conditions altogether. They are allowed (for small-group insurance) to place only an initial 12-month preexisting exclusion on any condition manifested within six months before the date of coverage. Third, "portability" or continuity of coverage is promoted by requiring that subscribers, once enrolled, be able to transfer coverage to a new insurer either by changing jobs or changing insurers within the same workplace, without undergoing a new exclusion period, so long as the gap in coverage does not exceed two months.

The second major component of the reforms is to restrict the degree of price variation among subscribers. These rating restrictions both: (1) limit the amount insurers may increase the price for a specific subscriber over time; and (2) compress the range of prices that the insurer can charge across its entire block of business at any given moment in time. Ohio law follows the National Association of Insurance Commissioners' (NAIC) model, which limits year-to-year premium increases for any given group to 15% above the insurer's "trend." Trend is defined as the increase in the insurer's rates for new business. The concept is to allow marketwide cost increases that are driven by technology advances, inflation in the medical sector and the like, but to limit those increases that reflect group-specific health risk. (Trend is keyed to new business rates because this is where insurers are the most competitive.)

The second component of the rating reforms prevents any small-group insurer from varying its prices among subscribers at any point in time more than a defined amount above or below its midpoint for policies with similar benefits and "case characteristics." Limits on both the low end and the high end are necessary in order to keep insurance affordable for higher risks and to lessen an insurer's incentive to engage in cherry picking or cream skimming by offering deep discounts to highly favorable groups. Ohio allows small-group rates to vary +/- 35% to reflect individual health status, which is broader than the 0-25% ranges allowed in most states. This rate variation is in addition to rating factors for age, gender, location, and other allowable "case characteristics," which are unconstrained. As a consequence, considerable rating flexibility remains. Since each allowable rating factor can, in theory, be added to each of the others, at the extreme these restrictions can still allow more than a 10-fold difference in the rates charged two groups at either end of the possible combinations of risk factors. On the other hand, these distant outliers might be very rare.

The third major component of insurance market reforms is an administered reinsurance mechanism that allows each insurer to reinsure any risks that are expected to generate costs exceeding the prices they may charge. Reinsurance encourages insurers to willingly accept all applicants by allowing them to pass their worst risks over to an industry-funded reinsurance pool. The principal funding for the reinsurance entity is from the reinsurance premium paid by the ceding insurer. Ohio follows the NAIC model by allowing insurers to prospectively reinsure either whole groups or high-risk individuals within groups. Ohio has a separate reinsurance pool for individual open-enrollment policies. The premium to reinsure high-risk groups is 150% and for individuals is 500% of the marketwide average for a policy of similar coverage and case characteristics. Any excess losses the reinsurance entity suffers beyond this premium pool are spread back to the insurance market through assessments against participating insurers based on the small-group market share. Participation in the small-group reinsurance system is optional, but the reinsurance system for open-enrollment business is mandatory.

The reforms discussed so far are aimed only at availability, not affordability, meaning they are only designed to offer insurance to any willing purchaser at prices that do not far exceed the market average, not to impose rate regulation or reduce prices across the market. Some other states, but not Ohio, create government-sponsored purchasing cooperatives for the small-group market to help bring down costs by increasing competitive pressures and lowering administrative costs. Instead, Ohio leaves these mechanisms to the private market by promoting and overseeing private purchasing "alliances." In large measure, this is because Cleveland is the home of COSE, the Council of Smaller Enterprises, which is the private purchasing cooperative that is widely regarded as being the most successful in the country. COSE was created in 1973 without any special legislative assistance, but the Ohio reform law attempts to facilitate replication of the COSE model elsewhere in the state. It does so by clarifying the regulatory auspices under which private purchasing alliances function and by providing some minor facilitation, primarily by waiving the 2% premium tax that is imposed on indemnity health insurance for alliances with enrollment of at least 2,500 lives.

C. The Dangers of Reform

These reforms have attracted some critics who warn about possible adverse consequences, and a number of quieter voices that warn against setting hopes too high about their success. The strongest fear is that these reforms could be counterproductive, since they have the potential to increase prices and decrease coverage. These reforms may raise prices because they make insurance most attractive to the highest-risk subscribers by holding prices to less than the policy's actuarial value to them. The excess is built into the premiums paid by all purchasers, which will inevitably drive an undetermined number of lower-risk purchasers out of the market, thus raising the market average even more. This phenomenon is known as "adverse selection" against the market as a whole. This potential exists because the decision to purchase insurance remains voluntary, and existing purchasers are thought to be highly price sensitive.

These reforms also create the potential for administrative and regulatory complexity, circumvention, and strategic manipulation. High-risk individuals might pose as small groups to obtain more favorable rates, or low-risk employers might facilitate the purchase of individual insurance by their workers or might try to artificially aggregate into groups that appear larger than the 50-worker threshold in order to avoid these laws. Insurers might attempt to avoid higher risks through various legitimate or illegitimate strategies, or they might pull entirely out of these regulated market segments. Also, these rules might cause distortions or unlevel parts of the competitive playing field that tend to favor some types of insurers over others.

This outline of the purposes of these reforms and their potential harms and failings points to four central criteria that can be used to evaluate the success of these reforms: the extent to which they promote (1) insurance availability, measured through increased enrollment; (2) affordability, measured through average prices; (3) market competition, measured in a variety of ways; and (4) regulatory administrability, also assessed in a variety of ways. This report organizes its analysis of the empirical evidence by focusing on these four criteria.

Various components of the reforms have importance across each of these categories. For instance, guaranteed issue and open enrollment, which point primarily to availability, also might increase prices or lead to various circumvention techniques that affect administrability. Or, rating restrictions, which affect primarily affordability, might result in less insurance being purchased. Many components of these reforms affect market competition, and some components, such as purchasing cooperatives, affect each of the criteria in equal measures. Therefore, this categorization scheme does not result in a neat pairing of each component and each effect. This is true to the complexity of this regulatory scheme, however, since each component interacts with all the others and with market and social conditions that are independent of these laws. Also, keep in mind as various statistical and descriptive data are presented that it is impossible to know for certain the actual and full impact of these reforms. A host of other economic and social conditions were changing simultaneously and so we will never know what the conditions would have been absent reform, even if we can tell what they are before and after reform. Nevertheless, by following the interwoven threads of information in this complex tapestry, it is possible to draw some solid conclusions about whether these reforms have worked as intended, and, if so, why, and, if not, why not.
 
 

A. Availability

1. Enrollment Generally

Insurance reforms have not had a noticeable effect on the total level of uninsured people in the state, which remains at about 13%, although this is better than the national average of about 18% (Table 1). Considering data that is relevant to the small-group market in particular, there likewise appears to be no dramatic impact one way or the other. The percent of workers with private insurance in firms with fewer than 25 employees has fluctuated following the 1993 reform law and dropped five points since 1992 (Table 2 and Figure 1), but due to sample sizes these fluctuations are not statistically significant at a 95% confidence level. (The 1997 level is significantly lower than 1992 and 1996 at a 90% confidence level, but that statistical standard is generally considered too lenient.) The percent of workers with private insurance has increased significantly for firms sized 25-99 (Table 2 and Figure 1), but the increase is significant at the 95% confidence level only in comparison with 1993, almost three years prior to when the reform law reached the 25-50 employer size range. Also, the 1993/1997 comparison is probably affected by a 1994 change in the survey methodology that tends to increase the estimate of coverage in later years^.1

In the individual market, the number of people with private coverage not sponsored by an employer appears to have dropped sharply in 1994 (Table 1). These figures should be interpreted with caution, however. They reflect a broad residual category of private coverage other than through employment, which might include many types of health insurance besides comprehensive major medical.² Also, changes in the survey methodology could account for much of the drop in 1994, as seen in the national figures in Table 1. It is unlikely the reform law caused this drop since, as discussed below, its requirements are so lenient in the individual market.

In the small-group market, which is the primary focus of this evaluation, it appears from enrollment trends that the market is healthy. Additional confirmation comes from COSE, the purchasing alliance that dominates the market for small-group insurance in the Cleveland area. There, total enrollment has increased 5% between 1994 and 1998 (Table 3), and the number of groups has risen over 10% during the same time period. The experiences of selected insurers reported in Table 3 tends to confirm this picture, although looking at particular insurers is not very revealing since this could simply reflect product design, pricing strategies, or other market factors unique to them.

We can also gauge the status of the small-group market by insurers' and agents' subjective views. No one complained that the reform law has been disruptive, even the more sweeping guaranteed-issue requirement of HIPAA. Insurers commented that, by 1997, they had learned from their experience in other states that this is a logical, incremental step for market reforms to take in their evolution and that Ohio's more gradual implementation of reform allowed them to build confidence in these market rules, in contrast with Kentucky and some other states where reform caused a major upheaval. Most insurers see Ohio's small-group market as offering good opportunities for increased sales. One HMO sees its primary strategic opportunities as small-group, individual, and Medicare insurance, but not much opportunity for new business among larger groups. Another HMO had targeted the small-group market in the early 1990s, thinking it was overlooked at the time. But when the reform law took effect, the HMO began to experience much more competition because the law focused attention on this market segment. A third insurer, with both indemnity and HMO products, also views the small-group market in highly favorable terms. The carrier sees this business as offering favorable medical loss ratios similar to those in the individual market, but with lower marketing costs, more similar to the large-group market where profits are leaner. However, this insurer, like the previous one, viewed the small-group market in this favorable light both before and after the reform law. They do not see the law as having made a major difference, other than by leveling the playing field somewhat.

2. Open Enrollment and Guaranteed Issue

One way to determine direct impact of market reform on insurance availability is to observe the number of people who enrolled through the various open-enrollment mechanisms, for these are subscribers who otherwise could not have found coverage. Similarly, for guaranteed issue, one can inquire how many new subscribers previously would have been declined. This inquiry reveals that open enrollment and guaranteed issue have had very little impact on insurance enrollment.

In 1994, the open-enrollment mechanism that applies to commercial indemnity carriers resulted in issuing coverage to only 44 new groups with a total of 230 employees. We do not know how much total open-enrollment coverage was in effect that year (in contrast with newly-issued coverage). It is clear, however, that the small-group open enrollment process has had only a minute impact on total enrollment or on the level of uninsurance. One of the largest small-group insurers in the state does not have a single open-enrollment group, and another one of the largest small-group insurers has had only one group in three years. The representative for another small-group insurer did not even know there is open enrollment for small-group indemnity. Of the agents we spoke with, four had never sold a single open-enrollment plan, one guessed he sold fewer than two a year, and the agent who was most active said that open enrollment accounted for less than 1% of his business. One subject commented that indemnity open enrollment has "become a joke among carriers." Several of our interview subjects did not accurately understand that there is a separate open-enrollment mechanism for small-group indemnity insurers, thinking instead that the only open-enrollment options were for individual coverage, primarily through HMOs.

These dismal findings are offset to some extent, however, by the fact that Kaiser, and perhaps other HMOs, traditionally have offered open enrollment to small groups as a matter of corporate policy, and so a number of otherwise uninsurable groups have been able to obtain enrollment without invoking the mandated process. Kaiser's policy did not extend to groups of five and under or to point-of-service products, however. Also, two subjects asserted that the demand for open enrollment among groups has been very low because competition among insurers for increased market share has been so intense, resulting in very few groups that are not able to find medically underwritten coverage. The plausibility of this speculation must be compared with alternative explanations below that are more negative.

The open-enrollment mechanism for individuals has had somewhat greater impact, but still only marginal in terms of the overall population and the number of uninsured. Commercial indemnity insurers sold only 32 open-enrollment plans to individuals in 1994. The similar open-enrollment process for the former Blue Cross Blue Shield of Ohio (now called Medical Mutual) had 424 individual policies in effect in 1994, and 717 policies in 1995. The numbers are much larger for HMOs, however. In 1995, HMOs had 7,125 people covered by individual open-enrollment policies. The HMO open enrollment at one large insurer, however, has dropped steadily each year, from 488 in 1993 to 330 in 1996.

Various explanations are available for why open enrollment has had such limited impact. First, there simply may be low demand for these products, even by groups and individuals that traditionally cannot obtain insurance. This is reflected by the statistic that, in 1994, commercial indemnity insurers made 283 offers of individual open-enrollment coverage, but sold only 32 policies. One agent explained that he is usually able to secure better coverage at a lower price elsewhere by placing his high-risk clients with other insurers, or enrolling an individual under a spouse's group coverage.

The unattractiveness of open enrollment might be due to either low benefits or high price, or both. Indemnity open-enrollment policies have several important benefit limitations that diminish their attractiveness, as well as diminish potential eligibility. Small-group enrollees were not eligible unless they had not been able to obtain insurance elsewhere for a year. This waiting period was in addition to the standard preexisting condition exclusion of one year that took effect when coverage began. Coinsurance under even the more generous of the plans was 40%, and the out-of-pocket maximum was $3,000 a year. HMO open-enrollment policies do not have as many of these limitations. Moreover, HMOs have been required to offer open enrollment for many years and therefore the opportunity is better known. Consequently, the vast bulk of open enrollment has been with HMOs.

As for price, rating rules allow indemnity insurers to charge 250% more for open-enrollment plans than they would charge the highest-rated subscriber for an equivalent policy that is medically underwritten. A 250% increase is quite large on its face, much larger than the 150% increase typically allowed by high-risk and reinsurance pools in other states. The increase is even greater in effect, however, due to the impact of the rating bands that apply to medically underwritten coverage. As discussed later, rating rules for medically-underwritten plans allow a +/- 35% variation in small-group rates to reflect individual health characteristics, and the 250% increase was on top of the high end of this band. This does not mean simply that open-enrollment rates were 373% higher than normal underwritten rates. Many insurers choose to issue their normal rates at the low end of the band, in order to allow maximum use of the 70 percentage-point spread. So, effectively, indemnity insurers could rate open enrollment small-group coverage more than seven times higher than standard coverage. (This calculation can be done by starting with a base rating factor of .65 (1 minus 35%), and then increasing the maximum factor of 1.35 by 250%, to 4.725, which is 627% higher than the base.) These possibilities are confirmed as real by our interview sources, one of whom commented that "the [open-enrollment] rates are atrocious. . . . Who can afford $1,000 a month for a family?" Couple this with the fact that many people with serious health problems have lower income because they have difficulty maintaining a good job.

The situation for HMO rates is not as extreme, but rates for HMO open enrollment are also substantially higher than for standard coverage. For HMO open enrollment, rating is done in a traditional fashion based on each carrier's claims experience and actuarial projections for the open-enrollment block as a whole. Because this separate risk pool is not medically underwritten, the actuarially-justified rates are much higher than standard coverage, and HMOs obviously have an incentive not to be among the cheaper offerings. One large insurer reports that its HMO open-enrollment rates are about 250% higher than standard rates. For some HMOs, the open-enrollment rates in 1996 ran as high as $400-500 a month for single coverage and well over $1,000 a month for family coverage (Table 4). At other HMOs, however, the rates were much lower: $200-300 a month for single coverage and $500-600 a month per family. Similar disparities existed in 1997 rates, except that some insurers had increases as much as 40-60%, while others had no increases or substantial decreases. The disparity in these rates may result from the small blocks of business on which claims experience is based. One insurer commented that it had attempted to combine its open-enrollment block with its mandatory-conversion block (i.e., people leaving group plans and converting to individual coverage) to make the rates more "credible," but the DOI disallowed this based on the concern that the conversion rates would end up subsidizing the open-enrollment rates.

These rating disparities were reduced somewhat for small groups in 1997 when the guaranteed-issue requirement took effect for all small-group products. This replaced the open-enrollment process for small groups, and so allows high-risk groups to now purchase coverage within the ordinary +/- 35% rating bands. However, as explained below, this narrower range still allows high-risk groups to be charged twice as much, since insurers can issue standard coverage at the bottom end of the range (1.35 is twice as high as .65). And, as discussed below, many more insurers are using this aggressive rating tactic than before guaranteed issue took effect. Most subjects we interviewed in 1998 explained that the 100% price increase that high-risk groups receive is enough to deter them from buying. About 10% of new applicants are rated at the top end of the bands and so are groups who previously would have been declined. Most of these applicants choose not to purchase the coverage even though it is now available because it is too expensive.

We also observed several indications that agents or insurers may be failing even to offer open enrollment or guaranteed-issue coverage to higher-risk subscribers. There are several possible sources for these failures. Some agents said it is not worth their effort to submit these applications since no one wants to pay those prices. Most agents we spoke to said or implied that they do not actively recruit high-risk clients. It is also notable that representatives from COSE, which dominates the small-group market in Cleveland, knew nothing about the indemnity open-enrollment option for small groups that existed prior to July 1997, nor did they view it as their role to refer applicants to open enrollment who were turned down in normal medical underwriting.

Insurers obviously are not eager to offer this coverage, and the law in Ohio does not require them to be very proactive. Nonprofit mutual insurers such as Blue Cross are required only to receive and evaluate all applications during open enrollment; they are not required to accept all applicants. Other insurers must actually offer open enrollment, but commercial indemnity insurers need not advertise their open-enrollment coverage at all, and HMOs must publish only a specified notice in local newspapers. Neither is required to offer open-enrollment coverage when medically-underwritten coverage is refused. In contrast with the procedures required in other states, insurers in Ohio must respond only if specifically asked by the subscriber or agent. Some insurers fail to comply with even these minimal requirements. In 1998, the DOI issued a consent decree finding that CIGNA had failed to fully comply with the open-enrollment requirements in 1997 and 1998.

Insurers adopt various strategies to keep a low profile with respect to open enrollment. We were told by several agents that most insurers do not pay any commission for this coverage. This appears to be in direct contradiction of the statute, which requires a 5% commission for new open-enrollment business and 4% for renewal. Some confusion may exist by virtue of the fact that the statute applies only to commercial indemnity insurers, and most open enrollment is sold through HMOs. Therefore, agents we spoke to may not have had indemnity open enrollment in mind, but some were adamant that no commissions were paid for any open-enrollment business. This confusion was resolved to some extent when the DOI sent out a stern letter in 1998 reminding insurers of their obligation to pay commissions on all health insurance sold through agents.

However, insurers are still allowed to pay lower commission rates for higher-risk business. In defense of this practice, insurers observe that they are offering open enrollment and guaranteed-issue plans as a public service and are losing money on them, so agents should also make some sacrifice for the public good. Also, a lower commission percentage is justifiable on the grounds that the premium for higher risks is much higher for the same amount of work, and insurers want to avoid creating a reward for higher risks. The DOI allows insurers to create a risk-neutral payment incentive, either by paying a per-case flat fee, or by capping commissions at standard premiums, so there is neither a reward nor a penalty for bringing in higher-risk applicants.

Nevertheless, some insurers maintain bonus arrangements that may encourage agents to avoid higher risks by causing them lose income by writing this business. Some insurers in Ohio pay agents a profitability bonus that is based on the loss ratio for an agent's own book of business with that insurer. This is a very common practice in property and casualty insurance, but is considered suspect or illegal for health insurance in many states. In Ohio, we heard inconsistent views, even among regulators. One regulator said that a bonus for better risks is permissible and only penalties for worse risks are prohibited. Another regulator, however, said he would "come after" anyone paying a bonus based on health status, but he said the DOI has no confirmed reports of this occurring.

One rationale for these bonus incentives is to encourage better "field underwriting." This term refers to a practice of encouraging agents to screen out applicants they know or suspect will not meet underwriting criteria. This practice efficiently avoids unnecessary work for the insurer and agent, and helps to steer subscribers to the plans and carriers that are most likely to offer affordable coverage. This practice also helps to detect when applicants are not being truthful about their risk factors. In part, this practice and these reasons explain why agents refer to themselves as underwriters in their professional certifications and trade association names, even though they do not perform the full underwriting function of insurers in the home office. Insurers give their lead agents field manuals for software programs that contain the general underwriting parameters so agents will know whether it is worth their effort to submit applications. Insurers or general agents also give field agents informal, "pre-screening" opinions by phone for the same purpose. As a result, many applications from uninsurable clients never come in to the home office for a formal decision.

Many interview subjects acknowledged that field underwriting in various forms occurs for health insurance, and none denied it. One knowledgeable source who has worked both as a regulator and in the industry admitted candidly "that's how it's done." Insurers "clean out" their agent rolls of those who are "dumping" bad risks on them, and refuse to sell insurance directly so that agents can perform this screening function for them. This subject explained that insurers tell agents openly that they only want the good business, and they encourage agent loyalty through various perks such as bonuses and free vacations based on the amount of business they produce. One purchasing alliance administrator carefully tracks the loss ratios incurred from each agent to be sure no one is dumping bad risks on the pool. An insurer said it does the same and it demands more documentation from agents it believes are acting this way. A DOI official said the agency is receiving more complaints from agents about being hassled by insurers in this fashion.

In order to gauge the extent and impact of field underwriting, we conducted a market testing study to determine the ability of an actual small employer and unhealthy individual to obtain insurance. A small employer composed of two people in good health and one with juvenile diabetes contacted 18 agents throughout the state to inquire about coverage for both group sizes of three and two as well as individual coverage for the person with diabetes. All of the agents offered coverage to the group of two, and the agents overall were responsive, helpful, and worked hard to find coverage possibilities. One agent even sent information about the new HIPAA law. However, five agents (28%) suggested or recommended that the unhealthy member be excluded from the group's coverage, and two of these (11%) indicated that coverage for the group of three would not be available. The other three of these five reluctant agents suggested the unhealthy member should obtain individual coverage. Twelve of the 18 agents (67%), however, said the unhealthy member would have difficulty finding individual coverage. This is far more than in other states, where only 25% of agents had the same reaction.

Although field underwriting may be legitimate and efficient for most lines of insurance, these practices undermine the functioning of an open-enrollment or guaranteed-issue environment by discouraging agents from taking higher-risk clients. This also creates an uneven playing field by encouraging agents to send higher-risk clients to one insurer rather than another. As one agent explained, AIf everybody has to take everybody and one carrier gives a profitability bonus, man, I'm taking all my good business there so that I can get the bonus." In the words of another, "I think the carriers feel like, well, . . . if there is a piece of garbage out there and if you are on profitability bonus, maybe you won't give it to us." It is worth noting, however, that sometimes this strategy backfires. One agent gave us the following report:

Indemnity insurers also have the potential to manipulate the open-enrollment periods. They are required to maintain open enrollment only so long as they do not reach their maximum new enrollment obligation, which is .5% of their total enrollment in the respective market segment (individual or small-group). In the three-year span of 1994-1997, only a few insurers notified the DOI that they met their open-enrollment quotas: six in 1994, five in 1995, three in 1996, and one in 1997. All of these were for small-group open enrollment, except for one that was for individual open enrollment. When audits were conducted, it was found that a sizeable majority of the groups signed up under open enrollment did not meet the qualification of lacking insurance for the previous 12 months. In 1994, 15 of 23 groups did not qualify, in 1995 and 1996 combined, the ratio was nine of 15, and in 1996 three of four groups did not qualify. This suggests that some insurers may have attempted to avoid their open-enrollment obligation by filling their quota with lower-risk groups early in the year in order to keep from having to enroll groups of much higher risk. On the other hand, this audit figure could simply reflect systematic oversights due to flawed administrative systems or inexperience due to the small number of cases involved each year. Audits of individual open enrollment in 1996 and 1997 found no mistakes in insurers' decisions about who qualifies.

In any event, it is clear that, from whatever cause, the burden of accepting open enrollment is not large for most insurers, and it is unevenly distributed among insurers. For HMOs, the uneven distribution is reflected in part by the premium figures given above, which show that some HMOs have open-enrollment risk pools that are twice as high as those of others. Uneven distribution is also reflected in the enrollment survey conducted in 1995, which found that some HMOs receive thousands of subscribers while others receive almost none (Table 4). However, these figures are not adjusted for each HMO's overall market share.

Low enrollment and uneven burden also exist for indemnity insurers. We analyzed data from 1994 for the top 15 individual and small-group insurers, to determine fulfillment of open-enrollment quotas (Table 5). Among individual indemnity insurers, one company reached 42% of its open-enrollment quota for the year, four insurers reached only 9-12% of their quotas, and the remaining 10 insurers sold no open-enrollment plans. Among small-group indemnity insurers, one reached 61% of its open-enrollment quota, another reached 12%, and a third reached only 1%, with the remaining 12 insurers not selling open-enrollment policies that year.

There is one additional flaw that plagues the open-enrollment mechanism for the individual market. The open-enrollment requirement applies only to insurers who sell true individual policies, and their quota is based on the number of such policies sold. Much of the business in the individual market, however, is coverage that is sold through "group trust" arrangements, which mimic large-group plans but which actually are vehicles to sell to individuals. This technique involves issuing a single master plan to an out-of-state trust created by the insurer, and then having the trust issue certificates of coverage as individuals are enrolled in the trust group. Sometimes the trust is described as an association, which individuals join in order to obtain insurance. One subject described these as "air-breather associations," meaning that the membership qualifications are broad enough that anyone can join. However, they receive insurance only if they meet underwriting criteria. This is a recognized and widespread technique to avoid regulations that apply to individual policies, such as more intensive rate regulation. This arrangement is deterred to some extent by mandated benefits that apply only to group policies. Nevertheless, group trusts account for at least half of the non-employer ("direct pay") coverage sold in Ohio. This practice took hold prior to the open-enrollment law, but avoiding the open-enrollment requirement is another inducement to retain the practice. As a consequence, several of the largest indemnity insurers in the state do not have a single open-enrollment subscriber. However, group trusts must now offer open enrollment under HIPAA to individuals moving from group coverage.

As a consequence of these many problems, most subjects consider the open-enrollment aspects of the Ohio law to be a failure, with the exception of HMO open enrollment for individuals. These mechanisms have now been superseded for small groups by the guaranteed-issue requirements enacted in response to HIPAA. One insurer subject, who conceded he is generally Athe last person to suggest stricter regulation," sees this as an improvement over the prior scheme. Whether guaranteed issue for small groups works more successfully remains to be seen. Some of the flaws that plagued open enrollment may still affect guaranteed issue, such as the ability to use rating bands to maximum effect, and the techniques for encouraging field underwriting. Moreover, the flaws we have identified are still in place with respect to open enrollment for individuals.

3. Purchasing Alliances

Ohio law has had somewhat greater success in promoting small-group insurance availability through purchasing alliances. Cleveland is the home of COSE, the Council of Smaller Enterprises, which is the private purchasing cooperative that is widely regarded as being the most successful in the country. COSE was created in 1973 without any special legislative assistance, but the Ohio reform law attempts to facilitate replication of the COSE model elsewhere in the state. It does so by clarifying the regulatory auspices under which private purchasing cooperatives function and by providing some minor facilitation, primarily by waiving the 2% premium tax that is imposed on indemnity health insurance for alliances with enrollment of at least 2,500 lives.

The success of this law might be measured both by the number of alliances that have been created, and by the size of enrollment. In 1995, just one year after the law took effect, there were 10 officially registered "purchasing alliances," most of which were created in response to the law. One has since ceased operations. Enrollment figures for COSE are reported in Table 3. Enrollment for six of the other registered alliances ranged from 2,200 to 9,500 in 1996, and for most of these this enrollment was obtained after only about a year of operation. The other three were too early in their operations to have significant enrollment yet. In our market testing study of 18 agents, 14 (78%) suggested or mentioned the possibility of purchasing through some type of alliance or association, more than in any other state we studied.

We asked interview subjects how important the law is to the success of these purchasing arrangements. Most said that the law was largely irrelevant: it neither helped nor hindered their operation, although avoiding hindrance is an important achievement. One way in which the law avoids hindrance is by not imposing more demanding rating and underwriting restrictions than exist in the regular small-group market. This is a mistake that several other states have made that has undermined the success of their cooperatives by causing severe adverse selection effects against the coops (i.e., primarily only bad risks leaving the regular market and purchasing through coops). In fact, COSE itself experienced fairly severe adverse selection earlier in its history by imposing underwriting restrictions that were too severe, so it learned by experience that it had to allow its insurers to depart from community rating and to engage in medical underwriting.

Apart from avoiding any disabling requirements, the reform law has had only marginal relevance to the success of alliances in Ohio. This can be seen in the fact that COSE was formed and achieved its success many years before the law took effect. Other alliances were in fact formed in response to the law, but their total enrollment (30,000 in 1995) pales in comparison to COSE's 200,000. However, it is still early in their development. More telling, however, is the fact that a number of other cooperatives exist in Ohio that are not officially registered as purchasing alliances. We refer to these as associations rather than alliances, but operationally they are virtually identical. There are several times as many of these unregistered associations as there are registered alliances. We lack enrollment figures for them, but examples we were given indicate that some have more enrollment than the new alliances, although none has anywhere near the success of COSE. The major way in which the law encourages alliance formation is through the waiver of premium tax, but this helps only if the alliance offers indemnity insurance, whereas most cooperatives offer HMO insurance which is not subject to premium tax. So, the law has helped to publicize the COSE model and to facilitate its replication, but it does not appear to play an essential regulatory role.

One might question whether alliances have improved accessibility, or simply diverted enrollment from the regular market. We found the following limited indications of improvement. First, most alliances offer enrollment down to one-life groups, that is, including the self-employed, in contrast with the rest of the small-group market in which the law reaches down to only two-life groups. Prior to 1997, one significant insurer offered no group coverage for groups with fewer than five employees except through an alliance. Depending on how demanding the alliance is in defining membership, it can be quite easy for anyone to qualify for self-employed membership. One interview subject said that even a person who mows lawns on weekends but has a regular full-time job without insurance can qualify to purchase through his alliance. As a consequence, most COSE groups are quite small. The average group size is about six and has dropped in the past few years, from 6.5 in 1994. About two-thirds of COSE's groups have fewer than 10 members. Importantly, this is the portion of the small-group market that previously has experienced the greatest difficulty in obtaining insurance, since insurers generally are reluctant to enroll the smallest groups due to greater per-unit expenses in sales and administration and due to greater adverse selection problems.

Decreasing group size also appears to be happening outside of alliances. Agents like the following one told us that they are seeing significantly more "micro-sized" groups than before reform.

A second potential indication of COSE's success is the fact that approximately 25% of its new enrollees previously had no insurance, which indicates a possible improvement in accessibility for uninsurable groups. This is not the most convincing interpretation of this statistic, however. A similar percentage of previously-uninsured enrollees is typical in the small-group market generally. Many small employers are startup companies, or they have dropped insurance for a time thinking they do not need it or cannot afford it. Informants at COSE did not believe this 25% figure was significantly better than the small-group market as a whole. If it were, one would expect COSE to have a higher-risk subscriber pool than the market, but most people viewed the COSE risk pool as equivalent to or slightly better than the market. This is because the underwriting practices within COSE and other alliances are no different than in the rest of the market. Insurers we spoke with apply the same underwriting criteria and practices both inside and outside alliances and associations. At COSE, approximately 10-15% of applicants were refused coverage prior to guaranteed issue, which is similar to the denial rate we heard of elsewhere in the market.

When small-group open enrollment was in effect, alliances and associations did not have to participate. Insurers that sold through these arrangements did offer open enrollment, but only to the extent of their business in the regular market. Their alliance and association business was treated as if it were large-group business for purposes of calculating their open-enrollment quota. So, similar to the group trust arrangements discussed above that apply to individual insurance, these purchasing arrangements partially sheltered insurers from small-group open enrollment. Moreover, as noted above, alliance administrators did not see it as their responsibility to refer declined applicants to open-enrolling insurers, nor did insurers necessarily offer this option when they declined coverage. Since COSE operated without insurance agents, there was no one with either an economic, a legal, or a social obligation to assist rejected applicants in learning about open-enrollment options elsewhere. Now that guaranteed issue applies to all small-group products, both inside and outside of associations and alliances, these problems have been resolved.

Both associations and alliances potentially offer affordability advantages that facilitate accessibility. Because we treat affordability as a separate theme in our report, we will delay that discussion until later, but to briefly summarize, we found that insurance through these arrangements is slightly cheaper than in the regular market, usually on the order of 2-3% cheaper, but sometimes the discount is greater. These savings come mainly through administrative efficiencies in marketing and billing, not through bargaining pressure. But the savings are largely absorbed by the fees charged for membership and administration. Therefore, there is no large price advantage.

Alliances might also offer advantages in choice. Unlike employer-based group insurance in the regular small-group market which usually offers employees only a single choice of coverage, alliances might be designed to allow individual employees in small groups to select among a range of insurance options. This has not happened in Ohio, however. Most alliances offer coverage through only a single insurer, although that insurer usually has several or many options. Even where there is a range of choice, however, that choice is exercised by the employer, not the employee.

Nevertheless, it is clear that the alliance model in Ohio, especially COSE, has been quite a bit more successful than in most other places in the country. We inquired extensively into the reasons for this success, either related to or apart from the legal climate. Our observations about the direct impact of the law are noted above. Here, we summarize non-legal factors or indirect factors that explain the success in Ohio, especially with regard to COSE.

It appears that the fact that Ohio coops have been implemented through private initiative rather than under government auspices has contributed to their success. Most insurers and agents reflexively react with hostility, or at least suspicion, to any government intervention in their markets. Private coops therefore begin without having to fight these ideological battles. Moreover, government-sponsored coops usually implement a uniform, statewide system under a fairly rigid set of operating rules. Ohio has had some success in allowing coops to compete with each other in overlapping territories, and in allowing them to find their own way through trial and error by adapting operating procedures to local environments and changing conditions. As one interview subject explained, health care delivery markets are local, employers are local, and so should be insurance purchasing arrangements.

Competing against these concerns is the problem of startup costs. One advantage of government sponsorship is that it provides startup capital and expertise that is important for running a large coop well. A number of business associations have failed or have not flourished because they believed they could simply endorse an insurance product, receive a price break, and allow everything else to happen on its own. Instead, coops are more successful when they operate like benefit departments at very large employers. This requires someone with expertise to shop carefully among insurance offerings, bargain aggressively, and then manage enrollment and billing. At COSE, these lessons were learned the hard way over a number of years, an insight captured in the observation that COSE is a "20-year overnight success." Other coops are still early in their development phase, and they struggle from lack of sufficient professional staff. Some coops obtain their expertise by being run by agents, who collect a portion of the commission paid by the insurer, but this cuts into their potential administrative savings. In contrast, COSE pays no commissions. Its expertise comes in part from a staff paid through the administrative fees collected from employers, and in part by volunteer service from its employer membership.

The COSE combination is hard to achieve without the large base of enrollment that it has. Its success might also be due in part to the market conditions that prevailed when it originated, and to its special relationship with the former Blue Cross and Blue Shield of Ohio. When COSE was formed in 1973, Blue Cross was the only insurer offered, and it was far and away the dominant insurer in the Cleveland market. Blue Cross gave COSE major price concessions, so that very quickly, COSE was able to secure a large enrollment base. Since then, Kaiser has been added as an additional insurer, but in many ways it appears that COSE has been run in cooperation with Blue Cross. Some would say they are virtual partners, and in our interviews we detected that COSE representatives tended to speak of Blue Cross policies and practices as if they were COSE's. We do not mean to suggest a bias against Kaiser or inappropriate favoritism for Blue Cross, only to report that the size and closeness of Blue Cross to COSE is a unique factor that helps to explain its success, especially in the early years. Here again, there is a contrast with government-sponsored coops, which by virtue of their public endorsement find it politically very difficult to limit themselves to a single or dominant insurer because this would appear to constitute an unfair exclusion of competitors. This "sole-source" arrangement is perfectly legitimate, however, when arrived at through private negotiations. Whether coops offer primarily one or multiple insurers also affects issues of competition and choice that are discussed more below.

Finally, it was impressed on us that COSE's success is due in part to the ethic of service and the absence of conflicts of interest among its membership and governing body. We were told, and we have observed, both in Ohio and elsewhere, that many times associations are permeated by various conflicts of interest in which Aeveryone is always looking for a buck." Administrators or sponsors want to use the insurance draw to build membership, in order to support other association activities. Independent agents want to avoid being cut out of commissions. General agents want to chance to earn "override" commissions on a large block of business. None of these legitimate, but non-essential, motivations appear to be at the core of COSE's existence, but replicating this purer, member-service-only structure is difficult to achieve elsewhere.

4. Portability, Preexisting Condition Exclusions, and Whole-Group Coverage

Other successful aspects of the small-group reform laws are the portability and preexisting condition provisions. These are widely popular among our interview subjects and appear to have been implemented without much difficulty. No one complained that the preexisting limits were too short, or too long. Several interview subjects pointed to this provision along with the portability provision as being the most successful and beneficial aspects of the reform law. In the words of one agent:

An agent confirmed this concern with the following example, but used it to make the more positive point that, as a result, there is a stronger incentive for agents to work to get coverage for individuals within groups that previously would have been easily excluded:

On balance, the insurance reforms in Ohio have produced mixed and muted results with respect to the availability criterion. Fluctuations in small-group enrollment are not statistically significant, so it does not appear these laws have produced any increase or decrease in the proportion of small-firm workers with private insurance. Individual enrollment counts have dropped. But these figures are not reliable measures, and there is no basis for concluding the drop is caused by the reform law. Nevertheless, it appears that neither market segment has seriously deteriorated.

The reform law's objective of increasing availability for higher-risk groups and individuals has been hampered by several factors that diminish use of open enrollment or guaranteed issue, which have had only a minute impact. Several of these factors appear to be the result of manipulating, circumventing, or perhaps outright violating the law. This situation has improved to some extent by changes in the law made to comply with HIPAA, but not entirely. Even with these improvements in availability, however, high-risk subscribers face barriers of affordability. The impact of the law can be best summed up by comments from two interview subjects. One, an actuary with a leading insurer, observed that even though he is "the last person to suggest stricter regulation," the law as it stood through 1997 has had almost no impact. Prior to the law, high-risk groups could always find coverage, if only through an HMO. The only difficulty was the cost, and that barrier still remains. An agent made essentially the same point:

B. Affordability

1. Prices

As with accessibility, it is difficult to assess affordability without marketwide, longitudinal data about premiums or rates. However, the information provided by a few insurers (Table 6) indicates that prices for small-group health insurance have held remarkably steady during the first few years of reform, especially compared with the double-digit increases that were common in the late 1980s. At Company 3, which is a large insurer of indemnity products, the average monthly premium per covered life hovered around $100 from 1992 through 1996, but has increased since then (perhaps due to changes in its accounting categories). At Company 1, which offers both indemnity and HMO products, average premiums have dropped from 1994 to 1996, and they have held steady at COSE from 1994 to 1998 (Table 6). Based on comments from agents and interview subjects, it appears that prices held steady marketwide through mid-1997.

Whether this remarkable achievement is due to the reform law is a separate question. Premium increases were similarly subdued across the country over the past few years, in large part due to the shift to managed care. The shift to managed care not only potentially held down medical costs, but also created an intensified competitive dynamic (discussed more below) in which insurers were sacrificing profit in order to gain market share. According to one agent:

Observe, though, that these premium figures are not adjusted for any changes in benefits. Average premiums might hold steady only because benefits are being pared back, such as by increasing deductibles or lowering coverage for prescription drugs. In our interviews, we heard that some of this is occurring as employers try to hold their costs steady by "buying down" coverage, but we also heard that some insurers were increasing benefits.

In any event, it is clear that Ohio's initial reform laws did not drive up average premiums, which is what many opponents feared would happen. Rating restrictions have this potential by increasing costs for healthier subscribers, thereby at the margin driving some from the market, and by lowering costs for higher risks, thereby attracting more into the market. This rate compression, coupled with open enrollment, was expected to increase average prices steadily, but our data show no sign of adverse selection against the market up through mid-1997. Average claims cost per employee did not increase noticeably for the insurers who provided us this data (Table 6). Assuming benefits remained steady (which we do not know necessarily to be true), this indicates the health status of the insurance pool did not decline. Subjective opinions from our interviews confirm this picture.

The picture is a bit cloudier following the mid-1997 changes to comply with HIPAA, particularly guaranteed issue of all small-group products. In our 1998 interviews, a number of subjects noted that insurers increased prices 12-15%, in part due to increased losses stemming from HIPAA. Others, however, said that increases were in the 7-8% range and no greater for small groups than for large groups. They noted that some insurers tried to put through larger increases, but quickly retracted them when not all insurers followed suit. Several subjects said that insurers were blaming their increases on HIPAA, but they felt insurers are just "hiding behind" HIPAA, meaning this is just "cover" for increases that would have been the same regardless of the new law because of deterioration in loss ratios that preceded HIPAA. This is confirmed by the discussion above that HIPAA has not brought in a large number of previously uninsurable groups. Some agents and insurers said there was an "initial influx" of worse risks at the outset of guaranteed issue, but this "flash in the pan" soon "tapered off." Also, several actuaries said they did not load their small-group rates for anticipated adverse selection from HIPAA since the market was too competitive to allow the "luxury" of adding "speculative factors" into their rates. Instead, they felt that any adverse selection that might occur would be impact the loss ratio only 2-3% at most, an amount that could be made up in future rate increases if it materializes.

2. Rating Practices

In addition to possible effects on average prices, rating restrictions might affect the range of prices. Rating bands require price reductions for some groups but price increases for others. We inquired whether the move to rating bands resulted in any "rate shock" in which some subscribers received large increases and decided to drop coverage. In contrast with other states, in Ohio we heard of few complaints to this effect. Some employers complained initially as very low-risk groups were brought up to the bottom of the rating bands, but these complaints subsided fairly quickly, with no evidence that many people dropped coverage. Most complaints came from agents who were confused by the separate rule that limits rate increases for any group to 15% and who failed to realized that this 15% limit is on top of the insurer's underlying trend (that is, its average rate increase plus 15%), or they failed to realize that the law made an exception initially for low-rated groups that had to be raised more than this as a transition into the rating bands.

The low level of complaints appears to be due in large part to the fact that the rate restrictions in Ohio are much looser than in other states. Although some other states started with rating flexibility of +/- 25-35%, many have moved toward modified community rating. Ohio has retained its +/- 35% band. On top of this, rating rules also allow adjustments for various "case characteristics" such as age, gender, group size (that is, charging more per person for smaller than larger groups), and differences in benefits. One insurer subject commented that rating restrictions are a "non-issue" in Ohio from both an HMO and indemnity point of view. A DOI official commented that this is a "field day" for carriers, compared to other states.

Also, most subjects thought the 15% limit on rate increases above trend does not have much effect. If trend is 5-10%, this still allows rate increases for some plans as high as 20-25% each year. The ability to compound these rate increases year after year is limited to some extent by the rating bands. But, as noted below, insurers create the maximum room for movement up by setting their standard rates near the bottom of this spread rather than at the middle, thus allowing an individual group to be moved up almost 100% on top of trend over several years if it experiences high claims. Only a few interview subjects felt that the 15% limit was constraining increases, and no one complained that the 15% restriction is too stringent.

Returning to the rating bands, a range of +/- 35%, on first glance, might appear to mean that high-risk groups could never be charged more than 35% higher than average underwritten plans, but that is not the case. In order to make maximum use of the allowable spread, several actuaries told us that they rate standard applicants near the bottom end of the spread, not the middle, thereby allowing an increase of up to 100% based on health risk (e.g., from .65 to 1.35 is a 108% increase). One agent explains this effect as follows:

Another large insurer used the rating band similarly prior to HIPAA, but with differences. It issued new standard rates 15 percentage points below the midpoint (.85), so that it had room to discount more significantly, down to .65, if the group presented a very good risk profile. When this insurer rated up, it went up only to the midpoint so that it had more room at the top end for groups whose experience deteriorated over time. The consequence was to more readily turn down a group that was of marginal or borderline risk. This actuary told us that, even in a group of 30-40 employees, if there was likely to be one very-high-cost person, such as with cancer or a serious chronic illness, the carrier turned the group down rather than rate it up, prior to HIPAA.

There is a limit, however, to the extent to which insurers can protect themselves at the high end of the bands. If they set their substandard renewal rates too high, then their newly- issued standard rates would not be competitive. One insurer observed that, when the rating rules first took effect, it had to cut some of the rates for bad groups in half to bring their rating spread down to a point where their lowest rates were competitive.

Interviews with agents confirmed that these practices were fairly common across the industry prior to HIPAA. Most said that their clients usually were accepted at standard rates, or rejected outright, and that insurers seldom used the allowance in the rating bands to take marginal cases at substandard rates. One agent in particular expressed frustration that insurers also used the leeway at the low end of the bands to issue quotes at premium rates based on perfect risks but then reverted to standard rates once underwriting was completed: "They come in with a proposal that is too good to be true and in most cases it is. And the only way to qualify for it is that you have to be Jesus Christ yourself and then they would probably look hard at holes in the hands and all that."

The only significant departure we heard about from these underwriting practices was Kaiser, which as a matter of corporate policy has used community rating and year-round open enrollment for small groups with at least five members. Another HMO said it too would prefer doing business this way but could not because of adverse selection. As one agent explained, "The problem is if everybody is allowed to have tough underwriting and one carrier tries to be too liberal, they are the ones that get all the junk." Indeed, we were surprised Kaiser had been able to maintain this practice as long as it did. We were told Kaiser has been able to cope with adverse selection because of superior cost controls and because of the counteracting favorable selection that HMOs experience.

Rating practices have changed considerably following HIPAA's requirement of guaranteed issue. Now, no insurers are using modified community rating, and insurers that were previously using the lower part of the rating bands are now issuing new business across the full range. Also, insurers are issuing standard coverage at or close to the bottom of the bands in order to preserve the maximum range for rating up new business that is undesirable. The result is that groups with very good health risks find it difficult to obtain substantial discounts. Also, many agents complained that it is much more difficult than before to obtain reliable quotes, since quotes are issued at the lowest available rate but are subject to underwriting review of medical history.

Two insurers we spoke to, one an HMO and the other indemnity, shifted in 1997 from adjusted community rating to a complex system of multiple rate tiers that depend on detailed medical underwriting. The HMO was very concerned about having to gear up for the first time to conduct medical underwriting with this level of precision since it does not have the Asavvy" its competitors have developed. The indemnity insurer, which specializes mainly in life insurance, used to purposefully keep a simplified underwriting and rating process since health insurance was not its main line of business. Its health underwriter complained that it had to undertake a major investment at great cost and disruption in order to use the rating sophistication required to continue selling health insurance in a guaranteed-issue environment. Likewise, both Kaiser and Medical Mutual, through COSE, now use rating tiers spread across the full range of the rating band rather than their past practice of adjusted or pure community rating. The Kaiser representative explained this is "very un-Kaiser like" but is necessary in order to avoid becoming the only carrier "of last resort."

Other insurers, ones who were used to more detailed medical underwriting, have changed their practices by adding more rating tiers and issuing new coverage across a broader spectrum. They have also begun to make more aggressive use of the ability to increase renewal rates for particular groups 15% more than the insurer's average increases (trend). This practice of "durational rating" requires closer tracking of each group's actual claims experience. In the past, these insurers attempted to rate small groups more as a block of business in order to keep increases more level, but the use of finer gradations in initial rating undercuts this structure. One might expect groups that receive these greater rate increases to be protected by the portability provisions that allow them to switch insurers, but insurers' ability to increase new-issue rates as much as 100% above standard rates means that these groups are unlikely to find a better deal elsewhere. Several agents noted the irony that these new rating strategies in response to HIPAA make it more difficult for higher-risk groups to switch insurers than was the case before HIPAA, even though HIPAA was meant to promote portability. Others, however, noted these changes are not entirely due to HIPAA but also result from the intensely competitive environment in which insurers are eager to find any price advantage they can, especially for new business with better risks.

Although the rating rules are quite complex and are used strategically by insurers in a variety of ways, it appears that the DOI exercises little scrutiny of the data and actuarial philosophies that insurers use in their rating practices. One regulator said he was unaware that insurers were making more aggressive use of the rating bands. All that insurers are required to file is an annual actuarial certification with their rates that contains the following or similar language: "rate differences due to differences in plan design only reflect benefit differences," and "neither rates nor rating factors associated with the statutory standard and basic plans give recognition to the guaranteed-issue feature of those plans." A DOI official conceded that they are "willing to take almost anything" based on this actuarial certification. An actuary with one insurer explained that certification requires only that the insurer has policies and procedures designed to be in compliance, not that the actuary is certain everything is in compliance. All that the certification assures is that, if a noncomplying case is found, it will be fixed, not that any systematic audit will be performed. Also, the certification provides no detail about the data and assumptions that go into various rating factors, nor even the size of the factors. Here is the actual text of the certification submitted by this insurer:

C. Market Competition

1. Price Competition

As noted above, the insurance market for small employers has been a highly competitive one over the past several years and a number of insurers are targeting this market for sales growth. In 1994, the small-group market was composed of 83 insurers, although only 20 had more than 2,500 lives, and the individual market had 29 insurers, although only nine had more than 1,000 lives. We heard of no reports in Ohio of insurers withdrawing from the small-group market due to the reform law, and almost every multi-state insurer we spoke to identified Ohio in favorable terms. Several subjects said that there are fewer competitors now than a few years ago, but they attribute this to a general trend toward consolidation in the industry as a result of the shift to managed care. As in other states, Ohio employers, both small and large, have moved rapidly toward managed care in recent years. The creation of managed care networks is expensive, requiring much greater investment of capital resources in a local market than does offering indemnity coverage, so insurers have become more selective about which local markets they are active in. Thus, even if the total number of insurers in the state remains high or has even increased due to the upsurge in HMOs, the number actively marketing in particular cities appears to be dropping. Some agents said that they are getting quotes for small groups from only 3-4 insurers compared with 8-10 in the past.

The subjects we spoke to do not attribute these changes to the reform law, however, nor do they view the market as uncompetitive. To the contrary, two subjects said that insurers' eagerness to increase market share made it possible for most high-risk groups to find coverage even without guaranteed issue. Many interview subjects commented on the intensification of price competition in the small-group market in recent years. Agents viewed price differences among insurers as being fairly small and as diminishing, after adjusting for benefit differences. One pricing actuary commented that he is Aconstantly amazed" at how small a price difference it takes for employers to switch insurers; some will switch for as little as $1 per employee per month. As a consequence, prices have remained remarkably steady in the small-group market during the initial years of reform (Table 6), but they began to rise again in 1997 and 1998. For instance, Kaiser reduced its rates 5-10% each year in 1995 and 1996, but has increased rates 10-15% a year since then to make up losses.

We should also consider whether the market is excessively price competitive. Some interview subjects complain of new entrants to the market who offer "low-ball" prices in an attempt to "buy market share," but then follow this in subsequent years with rapid increases. New market entrants are inherently able to offer lower premiums because, all things being equal, newer risk pools are healthier than older ones. This is due to three factors: new subscribers are still subject to preexisting condition exclusions; freshly underwritten groups are uniformly healthier than a pool that contains groups that have renewed over a number of years; and people with health problems are reluctant to switch insurance in the midst of treatment (so even without underwriting, new subscribers tend to be healthier than renewing subscribers). As a consequence, new market entrants are able to quickly gain market share by pricing aggressively. Some insurers try to seize on this advantage by creating new subsidiaries or product offerings to separate their new subscribers from their existing pools. This results in a situation that some subjects viewed as an unhealthy degree of market volatility. Others, naturally, might view this as favorable price and product competition.

Regardless, it is not clear whether this market activity is a consequence of these reform laws. It clearly existed to some degree before these laws, but the laws appear to amplify it to some extent. First, guaranteed renewability ensures that unhealthy subscribers can keep their coverage, thereby magnifying the difference between newly-enrolled and renewal subscriber pools. Second, rating restrictions may limit the extent to which insurers can establish different rates for different risk pools if the benefits are the same, which keeps them from competing directly with new market entrants by offering lower rates for newer business. Third, portability makes it easier for subscribers to switch insurers in order to save a few dollars. One insurer reported that its duration rate (the average time that new subscribers stay with the company) has dropped significantly in recent years, from five years to three years. In 1996, another insurer had 55% of its Ohio groups not renew. This tends to undermine the incentive in managed care plans to invest in preventative health and health maintenance for chronic illness. In other ways, however, the reform law discourages rate volatility. The 15% limit on renewals (above trend) and the +/- 35% rating bands limit the ability of low-balling insurers to recoup severe losses, as illustrated in more detail above.

On balance, most subjects felt that low-balling tactics had lessened over recent years, and almost no one considered volatility to be a major concern. One actuary told us in 1997 that he does not use durational rating at all. Instead, as a marketing strategy, his company issues a three-year guarantee that groups will receive the new-business rates. He found this to be a huge success in attracting new business and in lowering the lapse rate. On the other hand, other insurers have increased the use of durational rating within the 15% allowance, as discussed above.

We attempted to determine the degree of price competitiveness by examining insurers' relative prices. This is a difficult evaluation to make because Ohio does not require insurers to obtain prior approval of rates, and the rates that are quoted vary according to differences in benefit structures, the composition of groups, and health risk factors. Nevertheless, we attempted to make a rough evaluation by looking at insurers' average premium per subscriber for all small-group products and all individual products. We used data reported to the DOI in 1994, which is only the first year of the reform law but is the only year for which data are available to us. We looked at the top 13 small-group insurers and the top 10 individual insurers, in order to exclude aberrations due to insurers that are not actively marketing. We then examined how close each insurer's average premium was to the median figure for this group, and observed whether there was any relationship between relative market price and market share. These figures are displayed in Tables 7 and 8. One important caveat: these premium averages are only approximations of relative price, since they do not account for differences in benefit packages among insurers, or for differences in their respective risk pools (e.g., having relatively younger or older policyholders). Therefore, these figures do not, strictly speaking, compare apples to apples. Nevertheless, they are the best insight we have, and we think they are reasonable approximations.

Examining Tables 7 and 8 reveals several interesting facts. First, there is a surprising spread of over 150% in average premiums, even among the top insurers. In the small-group market, average annual premiums range from $146 to $399, with a median of $188. In the individual market, the spread is $55 to $143, with a median of $100. However, 73% of individual enrollees among these top insurers are with plans priced at or below the median. The same is not true for the small-group market, but if the break point is stretched a bit to capture one large insurer whose average premium is only moderately above the median, then the lower-priced small-group insurers capture 50% of the market among the top firms.

It is debatable whether the reform law is responsible for these competitive conditions. Most interview subjects opined that it is not, since the movement to managed care appears to be happening nationwide and independent of these laws. They viewed the small-group market as simply the next logical place for HMOs to look for sales growth after the large-group market began to become saturated. On the other hand, this movement largely coincides with these laws, and the laws are also nationwide. There are several possible theories about why the law might have helped to precipitate the move to managed care. First, HMOs traditionally have more lenient underwriting and shorter or no preexisting condition exclusions, so these laws helped to level the playing field by requiring other insurers to do business on similar terms. Second, to the extent these laws caused any rate shock effect, they may have provoked subscribers to look around for alternatives more quickly than if prices had continued to climb at a more gradual pace. Nevertheless, only one interview subject, an agent, believed the reform law might be partly responsible for these market conditions. Even HMO subjects believed strongly that the market changes are independent of the law. But, at least it is clear that the law has not subverted these market forces.

2. Purchasing Alliances

One way in which the reform law might have stimulated more competition is through the facilitation of purchasing alliances. We discussed above the extent to which the success of these alliances can be attributed to the law, and we found the connection to be only indirect. But, regardless, here we evaluate whether alliances have promoted competition by offering better value through aggregating purchasing power. For the most part, they have not. Except for Blue Cross/Medical Mutual plans in COSE, alliance and association rates are only about 2-3% lower than market rates, and much of this difference is absorbed by the costs of joining the association or administering the alliance. To the extent that the net rates are lower, in the past this was due in part to favorable selection, not to bargaining power. The favorable selection that some alliances receive tends to even out naturally over time as the risk pool comes to replicate the overall market. Selection effects are also dampened by the rating rules, which restrict the ability of insurers to pass on favorable selection in their rates to alliance members. (We are not entirely clear on the full extent to which rating rules prohibit this. We received unclear or inconsistent answers in interviews. The rules may have changed over time, and may differ between certified and non-certified alliances. Also, the rules may not be thoroughly enforced for reasons discussed below.) Nevertheless, COSE's net prices are sufficiently attractive that it dominates the small-group market in Cleveland. We were told, but did not confirm, that the Medical Mutual rates are 12-14% lower in COSE than outside (although there is some question of how significant Medical Mutual's small-group business is outside of COSE). We do not have exact figures, but it appears that only a small fraction of small-group insurance in the Cleveland area is sold outside of COSE.

Lower net prices in alliances result to some extent from administrative efficiencies in marketing policies and collecting premiums. The greatest administrative savings are potentially achieved if alliances do not pay agent commissions, which range from 5 to 10%, but this is rare. COSE is the only prominent example, and it has recently allowed commissioned agents to sell its policies to larger groups (sized 10 and up). Other alliances find it essential to include agents in order to have access to established small-group accounts and because they lack the size or expertise necessary to do effective marketing themselves. Even at COSE, the savings from not paying agents is minimized by the fact that one of its two carriers, Kaiser, traditionally has not made extensive use of agents outside of COSE. Kaiser also noted that it is unable to take full advantage of the potential administrative savings since much of the billing and processing is contracted out to the insurer that is its principal competitor and would involve exposing sensitive information. COSE, for its part, claims that Kaiser is missing opportunities to capitalize on administrative savings by not adapting its administrative and informational systems to COSE's, and its competitor has. Kaiser replies that its competitor has done so mainly because COSE adopted the competitor's systems in the first place. We judiciously decline to take a position on this debate. Both sides, however, appear to agree that Kaiser's price is only 2-3% lower in COSE than outside, and that this differential is due almost entirely to selection effects.

Why have COSE and other alliances not achieved greater savings through purchasing power and collective negotiation? We found that, unlike the theoretical model, the alliances we spoke to did not engage in an annual competitive bidding process to seek out the best value for their members. For alliances other than COSE, the reason may be size; they have not grown to a critical mass sufficient to warrant a competitive bid process or to seek significant price concessions. Also, in a local market, there may be only a few insurers willing to sell through an alliance, and excluding one or more may not be attractive to members. This is especially the case where a smaller market, such as Akron, is split into two, largely mutually-exclusive, hospital-based networks. An alliance that is seeking broad-based membership would be reluctant to exclude half of its potential members by going with only one network. One large indemnity insurer said it refused to engage in competitive bidding for alliance business because, in its view, alliances need it more than it needs them, since members' primary allegiance is to the insurer and its product, not to the alliance association. In other words, in the view of this large insurer, it brings business to the alliance, not vice versa.

This market dynamic does not hold true at COSE. Nevertheless, COSE also does not engage in competitive bidding. Its approach to negotiating price appears to be driven by institutional culture. COSE has always relied primarily on Blue Cross/Medical Mutual and so has declined to put this insurer in jeopardy of being outbid. Instead, it engages in annual negotiations with Medical Mutual over price with the assumption that some deal will be struck. Once, a few years ago, COSE took seriously the idea of switching insurers when it felt it was not being given a good enough price by the former Blue Cross. But before and since then, a more cooperative relationship has been maintained.

We also inquired why COSE did not foster more competition by including more insurers. We were told that COSE follows a "sole-source" philosophy, with the exception of Kaiser, which was added at a time that the former Blue Cross did not have a developed HMO product line. COSE prefers this because it fears that adding more insurers will simply increase the use of risk selection tactics, which will cause more problems in administration and policing. According to one interview subject, insurers would "eat each other's lunch chasing after the good accounts." This is telling commentary as we consider below whether market reforms have been able to suppress the level of competition through risk selection in the rest of the market. Even in COSE with just two insurers and, prior to 1998, both using modified community rating (that is, no 35% bands), Kaiser complained that Medical Mutual engaged in risk selection tactics in how it structures its rating categories within COSE, and Kaiser continues to be unhappy with the competitive playing field inside COSE.

3. Non-price Competition

These laws were also intended by some proponents to alter the nature of non-price competition. Traditionally, insurers competed primarily based on their ability to select and accurately price risks, and by tailoring their benefit packages to consumer preferences. These laws were expected to greatly reduce the amount of risk selection, and in some states (but not Ohio) reformers attempted to move the market toward more standardized benefits. We evaluated performance in this regard also.

As noted above, competition has intensified in the Ohio health insurance market overall, including the small-group segment, primarily as a result of rapidly increasing penetration of managed care. This has focused competitive pressures on price and on the structure of the insurance plan and the composition of the provider network. Whether these developments are attributable to these reform laws is debated above. Most interview subjects believed they are independent of the law, although we theorize that the law may have stimulated or catalyzed these developments.

What is noticeably absent in the small-group market is competition based on outcomes measures of quality. Naturally, this is relevant only to HMOs since indemnity insurers are not in a position to monitor or influence the quality of care, and one of the selling points of indemnity coverage is that subscribers are free to make their own decisions about which are the best providers. However, given the penetration of HMOs, one might expect at least some competitive focus on quality of care measures. To the contrary, in the vocabulary of most agents, "quality" refers to the richness of the benefits, the size of the provider network, and to how promptly and hassle-free claims are paid, not to the quality of care delivered. We reviewed the sales literature from leading insurers, including HMOs, targeted to the small-group market and found almost no reference to outcome measures of quality such as the HEDIS measures developed by National Committee on Quality Assurance (NCQA). At most, there were passing generic references to the quality of providers in the network. The focus of almost all of the sales literature is on the particulars of the benefit packages, and for HMOs on the composition of the network. Even COSE, with its size and sophistication, is only now beginning to look at outcome measures on a limited basis. In one instance, COSE asked the former Blue Cross and Blue Shield of Ohio to drop one hospital from its network based on outcome and cost information, and COSE is collecting this information from other hospitals. It is noteworthy, however, that this information is being used to formulate the provider network rather than to give to employers or employees to assist their choice of insurance.

Here are representative excerpts from our interviews with agents concerning outcomes measures and the focus of non-price competition:

Q: What about competition based on quality and outcomes? Do you hear about this in the [sales] literature a lot?
A: Yeah, I think what you are seeing [is that with] the IBMs of the world and the more sophisticated buyer, the more you can talk about it. . . . In the small-group market, price is still number one and then we get into the network. I mean, we currently have NCQA data on both the major [insurers] here. I've talked to the small-business people and actually asked them what they thought, and well, [they say] "That's nice," . . . [but] it's going to get down to price, benefits and network.

According to another agent, "a lot of employers who do go to managed care, there is only one thing they care about: if their doctors are on the list. They don't care whether their employees' doctors are on the list or not."

We also found that much of insurers' strategic market positioning is focused on differences in benefit packages. Most sales brochures offer a dizzying array of ways to mix and match various components of coverage such as deductibles, co-payment levels, maximum payouts, and various riders for prescription drug benefits or mental health coverage. Some subjects observed that insurers add or subtract benefits to strategically attract healthier subscribers or discourage unhealthier ones. Adding health club membership is an example of the former that was mentioned by several subjects, and reducing prescription drug coverage and raising deductibles are examples of the latter.

Insurers also still compete to a considerable extent on their medical-underwriting abilities. Despite the purpose of these reforms to minimize medical underwriting, it still occurs to almost the same extent as before. Through the middle of 1997, insurers were still allowed to decline coverage for their primary plans. Even the open-enrollment plans require medical underwriting in order for the insurer to evaluate the extent of risk it is receiving. And, even following guaranteed issue, medical underwriting is still required to employ the +/- 35% allowance for health status in the rating bands, and to determine whether to invoke the reinsurance option described below. Accuracy in these endeavors still determines profitability to a considerable extent. HMOs especially feel that their lack of underwriting experience puts them at a competitive disadvantage.

D. Administrability

1. DOI Oversight and General Compliance

Finally, we address a series of concerns about the administrability of these laws. The Department of Insurance has knowledgeable and dedicated staff. It has done a very effective job in writing and updating each year an extensive consumer information booklet about small-group and individual health insurance laws and offerings in the state. This is the best such information booklet we have seen in the dozen states we have reviewed. The brochure has useful information about the structure of the market, the content of the law, comparative prices, consumer complaint records, market shares, purchasing alliances, and contact numbers. Several interview subjects commented that the department is "consumer friendly" or a "consumer watchdog." Most (six out of nine) thought the department is doing a good job enforcing these laws. When complaints are filed, the department has a reputation for responding quickly. One insurer commented that the DOI is "fair, flexible, and thorough," and it sees the department staff as a resource for compliance questions and for expert advice. DOI staff explained that the department consciously fosters a good relationship with insurers in order to attract more to the state and so enhance competition.

In other respects, however, the department has not been fully effective. It did not initiate much enforcement or monitoring activity with respect to these laws until 1998. One agent said there is "virtually no compliance enforcement by the DOI," and one large insurer noted in 1997 that the department has only done two market conduct exams focused on health insurance in its history (we have not verified this assertion), and these have not been addressed to small-group laws. Although the DOI audits insurers every three years, this is focused on financial solvency issues rather than legal compliance. Its enforcement is mostly reactive, by responding to complaints. Most complaints from consumers are motivated by price increases, which usually are within the bound of the law. The other possible source of complaints that might result in enforcement activity is from insurers themselves as they observe unfair or illegal tactics by their competitors. Although insurer subjects we spoke with sometimes alluded to such behavior by their competitors, we saw no indication they regularly lodged these complaints with the department. One regulator explained that they sometimes receive "what if" inquiries from agents or insurers that suggest illegal or questionable activity is being contemplated, but rarely do they receive any direct reports that such activity is occurring. This regulator felt that these suspicions are not sufficient to justify investigation or further inquiry.

In 1998, the department became more proactive in its enforcement of the health insurance reform laws, possibly in response to press accounts that HIPAA was being circumvented. The department issued show cause orders to a few dozen insurers believed to be in the individual market but that did not meet the deadline for filing HIPAA open-enrollment plans. The department also initiated a wave of market conduct examinations looking at HIPAA compliance. It plans to examine the top 15 insurers by the end of 1998 and another eight insurers in 1999. Also, the department issued a consent decree in 1998 that CIGNA failed to fully comply with requirements in the two previous years.

Despite these compliance concerns, we found no indication of widespread violations of the basic requirements of the reform law. Both agents and insurers we interviewed are very knowledgeable about the basic requirements of the law (except for issues relating to open enrollment for small-group insurers, which has since been superseded.) Consumers are well represented by independent agents who are likely to complain directly to the responsible insurer if they detect noncompliance. Insurers are usually (but not always) responsive since they want to stay in the good favor of their agents. Insurers are also motivated to make sure they are in compliance in order to stay in the good graces of the DOI, which affects their business lives in so many different ways. Most insurers we spoke with have well-staffed regulatory compliance officers to track legal developments and carry out corporate compliance. Insurers send frequent operational instructions and updates to agents with respect to their products and procedures. Thus, although the DOI and the agents' professional associations take only limited steps to publicize the law and determine knowledge and compliance, basic knowledge and compliance appear to us to be fairly accurate within the industry.

2. Border Problems and Fraud

We also inquired into particular enforcement issues that might be especially troubling. One of those relates to field underwriting, which is discussed above. Other areas include list billing, self-insurance, and private associations. These are all concerned with what we refer to as "border-crossing" problems. The potential for these problems arises when one segment of the market is regulated differently than another. This creates possible strategic advantages for low- or high-risk groups or individuals to cross into or out of the market, at either the high-size or low-size ends of the market, thereby unraveling or eroding the market divisions that are necessary to sustain this regulatory structure. We will discuss a variety of specific examples.

List billing. This refers to an insurer that excludes certain members from group coverage or sells individual coverage to members within an employer group, either with or without the employer contributing to the cost. This practice was common prior to the reform law for a variety of reasons. One use of list billing was for employers to purchase insurance for only selected employees by reimbursing them for the cost of individual coverage. This might be done in order to offer insurance only to "key employees" such as managers, or in order to avoid the costs of insuring an employee or family with health problems. Other forms of list billing were done as an accommodation to employees whose employers were not willing to buy insurance for anyone, but who wanted to facilitate their employees' purchase of insurance through payroll deduction.

The reform law in most states prohibits list billing, following the philosophy that employers should treat their employees equally, and out of the pragmatic concern that if it were allowed to continue, employers with low-risk profiles would circumvent the rating rules by purchasing non-group insurance, thus bleeding good risks out of the small-group market. In Ohio, however, there is great uncertainty and difference of opinion about the legality and extent of list billing. Within the industry, six out of 10 subjects said that it occurs with at least some insurers, and the four dissenters seemed more unaware than in definite disagreement. One insurer is running advertisements that pitch individual coverage to groups with fewer than five members, calling them groups of individuals. Recall also the different, but related, technique known as "lasering out" that is described above in which a high-risk employee is excluded from group coverage by arranging for individual open enrollment, usually with an HMO, in order for the rest of the group to qualify for a medically-underwritten policy.

Some subjects believed list billing is perfectly legal as long as the employer does not pay for the coverage and it is truly individual insurance. Some insurers have developed a form for subscribers and employers to certify to this effect. Other subjects believed this practice is illegal or of questionable legality, however. Several agents told us that employers circumvent this arrangement by paying their key employees a bonus to compensate for the cost of the individual insurance. Here is one example from an agent interview:

Associations. Good risks might also leave the small-group market at the high end, if small groups attempted to aggregate artificially into a group larger than 50. This might occur through what are known as private associations, the variety of which are too complex to describe thoroughly. In the past, they have gone under the acronyms of MEWAs or METs. Associations might be used to cross the border at the small end by taking high-risk individuals and presenting them as an employer group. We found no indication that either was happening. However, a related but different problem has been noted earlier, namely the use of out-of-state group trust arrangements that result in circumventing individual market regulations by selling individual insurance through a large-group vehicle. We saw no indication that the same is happening with respect to small groups, although we did note that one large insurer uses a group trust arrangement to sell small-group coverage. Apparently, the DOI regulatory policy treats small groups more firmly in this regard than it does the individual market.

Self-insurance. Yet another border crossing concern is the threat that this law would induce medium-sized groups, those in the 25-50 range, to self-insure. This might occur if a group of good health risk felt it could save money by avoiding the rating bands. It is primarily for this reason that rating rules are not extended to groups any larger than 50. Partial self-insurance can be marketed to smaller groups by selling stop-loss policies with very low attachment points, in the range of $10,000 per employee, which is equivalent to a high deductible catastrophic plan, but is structured to avoid characterizing the plan as ordinary insurance. We found some indication (from three of seven subjects) that this was happening in Ohio. However, it is not clear how pervasive this is for small groups, or whether it is done primarily to avoid mandated benefits and premium taxes rather than the small-group laws.

Employer fraud. Other potential circumvention techniques are not as structurally sophisticated, and they are perpetrated primarily by employers, not insurers. An employer with a sick family member or friend might falsely claim the person as an employee in order to take advantage of open enrollment or rating limitations. Or, an employer who truly employs a person with sickness in the family might try to avoid the cost by "hiding" those lives off of the payroll. Several subjects observed that, if circumvention and fraud are occurring in this market, it is probably of this nature, initiated by employers, rather than the forms initiated by insurers. Insurers and agents are sensitive to this potential and take steps to prevent it by requesting payroll and tax documentation. Nevertheless, there are indications this kind of activity is surprisingly widespread. When COSE became concerned that employer fraud was resulting in adverse selection, it conducted a study of the problem in the early-to-mid 1990s. It found that eligibility problems of both types are widespread and "significant," including sick individuals who are not employees, or hiding young healthy employees who do not want to buy (COSE requires 100% employee participation for groups with fewer than five members.) The COSE audit found that problems of this nature existed in an astonishing 40% of its groups, split approximately evenly between both types of fraud. COSE calculated that, overall, this added 7% to the costs of the premium for the overall pool. In COSE's opinion, the noncompliance rate has dropped substantially since then because of the audits and the resulting sanctions for noncomplying employers, who were refused future membership in COSE. Nevertheless, this rate of noncompliance occurred even though COSE had been conducting standard checks of enrollment and tax records to confirm employee lists and participation.

Agents in their interviews elaborated on additional employer tactics:

A final feature of the law that cuts across several of our categories of discussion is reinsurance. There are two reinsurance mechanisms created by the law that are intended to provide a relief valve for insurers who are forced by guaranteed issue, open enrollment, and/or rating restrictions to accept risks they believe are not adequately covered by the allowable premiums. One reinsurance pool is for small groups that are guaranteed issuance; the second is for indemnity open enrollment, which includes federally-eligible individuals under HIPAA. Insurers can cede to the pool either high-risk groups or individuals within these groups, on payment of a reinsurance premium that for groups is 150% and for individuals is 500% of the market average for the coverage and case characteristics in question. If these premiums are not sufficient to cover payments from the pool, losses are made up through an assessment against all participating insurers, proportionate to their relevant market share. This opportunity to cede bad risks is intended to protect insurers from adverse selection and to reduce their incentive to engage in covert risk selection, thus minimizing many of the gaming and policing problems that might otherwise arise under the law. For small-group reinsurance, insurers can elect to either participate or to opt out. Participation in individual open-enrollment reinsurance is mandatory, except for two insurers which the DOI has determined are large enough to be excused (Anthem and Aetna/U.S. Healthcare).

In the small group pool, 45 insurers have opted to participate, as of 1998. Most of the largest small group insurers have opted out of reinsurance. Their fear is that smaller insurers will use this mechanism more aggressively, thereby forcing the larger ones to pay assessments out of proportion to their use of the reinsurance pool. However, as it turns out, participating insurers have used reinsurance much less than was expected. The small group pool has received only 90 ceded lives through April 1998 from only a dozen or so insurers, and the open-enrollment pool has received only 11 lives. Participating insurers seldom reinsure, for the following reasons. First, the discussion above reveals that insurers are not issuing very much high risk or open-enrollment coverage, and allowable premiums are quite large. The primary use of reinsurance for small groups is when an otherwise healthy group has one or two high-risk individuals. Second, it is difficult and costly to evaluate in borderline cases whether reinsurance is worth the cost. We heard in other states that some insurers found they were reinsuring too readily and therefore were ceding risks for which claims payouts ended up being less than the reinsurance premium. In Ohio, although each of the two pools expected that the loss ratios might eventually be as high as 200%, through the end of 1995 the pools had received about $200,000 in reinsurance premiums, but the small-group pool had paid only $42,642 in claims, for an enviable loss ratio of 21%. The open-enrollment pool had paid no claims as of the end of 1995. Only in 1998 have the pools levied an assessment based on anticipated claims (rather than for administrative costs), and then only to keep loss reserves at a high level, not on account of any deficit in actual claims over premiums.

Some interview subjects observed that reinsurance might have served a useful function even if it was not much used, if, prior to guaranteed issue, it encouraged insurers to take borderline risks in medically-underwritten plans. An insurer might accept a group knowing that one member is sick if the risk can be limited by reinsuring the individual. This can be seen in the fact that, as of 1997, almost all of the lives in the small-group pool were ceded as individuals rather than through group reinsurance. The opportunity to reinsure might also encourage more liberal underwriting even if reinsurance is not used initially since having it as a safety net might induce an insurer to take a group whose future claims might go in either direction. However, we heard from one larger participating insurer that it seldom reinsures since, prior to HIPAA, it had a policy of requiring very clean risks, and it would rather decline a group than to take it and reinsure one of its members, considering the costs and uncertainty of making the reinsurance evaluation.

Following HIPAA, reinsurance may have an enhanced role. Insurers are obviously more concerned about receiving bad risks in a market where all small groups must be accepted. Reinsurance might help to keep some insurers in the market that have smaller blocks of business, and larger insurers may be inclined to reinsure more often. This is suggested by the fact that the number of lives in the small-group pool has doubled between 1995 and 1998, with most of the increase following July 1997 (although we do not know when this increase occurred). Also, two insurers who previously did not participate told us they were considering doing so in response to HIPAA. Moreover, the reinsurance mechanism might serve as a modulating device that regulators can use to tighten or loosen incentives for the industry by adjusting the reinsurance premium. If reinsurance appears too expensive and regulators detect increasing signs of covert risk selection, they can lower the premium to take pressure off insurers and encourage taking more risky groups. Or, if the reinsurance pool is being used excessively and assessments are mounting, the reinsurance premium can be raised. So, the presence of the pool may be beneficial even if it is not presently in great demand. On the other hand, it provides another continuing opportunity for insurers to profit by using underwriting and risk selection techniques rather than competing through more efficiency-enhancing innovations.
 
 

 
 
 
 
 

Table 3: Small-Group Enrollment, Selected Ohio Insurers1
Enrollment
Insurer	1990	1991	1992	1993	1994	1995	1996	1997	1998
Company 1						17,695	20,736	36,563	38,5992
Company 23	71,887	76,832	99,473	121,316	137,963	167,500	189,200
COSE					188,145	192,118	193,328	199,683	197,000

¹Data reflect covered lives and were obtained directly from each insurer. Except where noted, figures are for group size 50 and under.
²Enrollment as of April 1998.
³Figures are for group size 19 and under.
 
 

Table 4: Ohio HMO Open Enrollment
Monthly Price
HMO Name	Single		Family		Open Enrollment, ‘94 (Covered Lives)
	‘96	‘97	‘96	‘97
Advantage Health Plan	($) 232	($)	($) 614	($)	29
Aetna Health Plans	275	351	755	878	101
Choice Care	355	433	835	917	343
Cigna Health Care	541	666	1,297	1,559	246
Community Health Plan of Ohio	161	179	394	438
Day-Med					5
Dayton Area HP		346		959
Emerald HMO		168		478
Family Health Plan	245	277	649	647	31
FHP of Ohio/Pacificare	379	399	1,100	1,158	76
Health First	316	388	949	1,163	219
Health Maintenance Plan	326	553	772	1,252	796
Health Plan of Upper Ohio Valley					120
Health Power	290	323	870	969	15
HMO Health Ohio	379	278	1,023	751	1,027
HomeTown Hospital HP	300	334	856	953	41
Humana					83
InHealth					71
Kaiser Permanente	187	322	557	960	3,453
Licking Memorial Hospital					101
Medical Value Plan		281		844
MetLife					55
Nationwide HMO	334	482	917	965
Paramount Health Care Plan	401	355	1,113	977	48
Personal Physician Care	282	387	790	1,147	19
PruCare	422	542	1,265	1,634	119
SummaCare	315	265	898	729
Super Blue HMO
The Health Plan (Upper OH Valley)	272	343	747	943
Total Health Care Plan	314	314	854	854	1
United Healthcare of Ohio	436	476	986	994	145
Total					7,144

Source: Ohio Department of Insurance
 
 
 

¹See Katherine Swartz, "Changes in the 1995 Current Population Survey and Estimates of Health Insurance Coverage," Inquiry 34:70-79 (1997). However, these survey changes would not affect trends that appear from 1994 onward, and probably would not affect comparisons among states.
 
 

²It is impossible to corroborate this estimate with actual counts of insured individuals, since the Ohio Department of Insurance, unlike many other states, does not collect and compile information about annual sales activity for various market segments. The only year for which adequate marketwide information is available is 1994. In that year, 83 insurers covered 341,139 lives in the 2-25 group market, and 97,474 lives in the 26-50 group market, and 44 insurers covered 172,892 lives in the individual market. These figures do not jibe at all with the figures reported in Tables 1 and 2 (they are much lower). This magnitude of discrepancy is due not only to the different wording and methods of these data sources, but also to the fact that the DOI figures for individual insurance report only policies that are sold as true individual contracts and not policies sold as certificates under the "group trust" arrangements that are described below.