Among developed countries, the United States is unique in having a high proportion of persons who have no private or public health insurance coverage. In 1998, 43.5 million persons were uninsured, representing 16.1 percent of the U.S. population (Bennefield 1998). Further, many persons with coverage were considered to be "underinsured" to the extent that they had insufficient coverage to protect them well against the expenditure risk of a major adverse health event (Short and Banthin 1995).

Unlike some other forms of insurance, health insurance is on the public agenda because it is not considered to be a matter of purely private consumption. Various public good aspects of health care, including simple altruism, create a shared sense of concern when fellow citizens do not have access to needed personal health care services. Therefore, the individual purchase of health insurance coverage is considered a public issue whereas, at least in comparative terms, purchase of other coverages, such as life or homeowners= insurance, is not.

Why Are There So Many Persons Without Such Insurance?

In view of the benefits of health insurance and the nearly $100 billion annual federal subsidy provided to employer-sponsored coverage (CBO 1999), why are there so many individuals without any form of health insurance? There are several reasons why the personal benefits of health insurance coverage are more than offset by the premium for such coverage. The first is use of medical underwriting to counter adverse selection. Private insurance markets tend to segregate risks into homogenous pools for purposes of pricing and coverage. For health insurance, this is done through techniques of medical underwriting that identify higher risks (e.g., older persons, those with chronic, preexisting conditions) and either increase rates, deny coverage, or limit coverage. If an insurer did not medically underwrite, it would attract mostly higher risks and drive away the lower risks needed to maintain a stable and affordable risk pool. Fortunately, these problems are greatly mitigated for most private insurance by the fact that employers purchase on behalf of a heterogenous pool of workers. Employer sponsorship, however, leads to a second problem: moral hazard. Widespread, subsidized insurance encourages the provision of excess care or necessary care at excess cost, which drives up the cost of insurance, thereby deterring its purchase by employers. This effect is amplified by mandated benefits, which have been shown to reduce demand for coverage (Gabel and Jensen 1989 ; Jensen and Gabel 1992 ; Sloan and Conover 1998).

Third, where insurance is not available in the workplace, its cost is even higher in the individual market due to lack of scale efficiencies. The insurance premium reflects both the future outlay anticipated by the insurer for medical expenditures, and a loading factor. The loading reflects administrative expense and a risk surcharge required by the insurer to accept risk. Both components increase on a per-insured basis as the size of the insured unit decreases (Hall 1994). Individual health insurance typically carries a load of 30-40 percent, compared with 10 percent or less for large groups.

Finally, the availability of public insurance in the form of Medicaid has probably "crowded out" some demand for private health insurance coverage. Households eschew private coverage when they are healthy, or not as sick as they might become, reasoning that public coverage will be available in the event that catastrophic illness occurs. The same is true for free services available through private hospital emergency rooms.

Reform Strategies

Federal efforts to address these gaps in coverage occurred primarily in the mid-1960s. Medicare extended coverage to virtually all seniors and many disabled individuals and Medicaid now covers about half of those below poverty. In the late 1980s, Congress enacted expansions designed to considerably increase the number of pregnant women, infants and children covered by Medicaid. However, various proposals dating back to 1912 to establish a system of universal coverage have failed repeatedly at the national level, the most recent being the demise of the Clinton reform plan in 1994. In the absence of a federal policy affecting all demographic and income groups, much of the impetus for increasing health insurance coverage in the past decade has come from individual states.

Because the federal government provides 50% or more of Medicaid funding, states have used their Medicaid programs as the primary vehicle for expanding health insurance coverage. For instance, in the late 1980s and early 1990s, federal relaxation of Medicaid eligibility rules has culminated in a doubling of the number of pregnant women covered by Medicaid and an increase in eligibility for children of nearly 50 percent compared to 1987 (Cutler and Gruber 1997). Today every state must cover all pregnant women and children up to age six with incomes up to 133 percent of poverty and by the year 2002, all remaining children under 18 with incomes below poverty will be Medicaid-eligible. Ironically, however, the near-quadrupling of Medicaid expenditures between 1987 and 1998 (CBO 1999) has been accompanied by a steady increase in the number of uninsured. There are at least twice as many and possibly four times as many uninsured now as were reported 20 years ago (CBO 1981).

Affordability of Coverage. One of the earliest state efforts to make coverage more affordable were state high-risk pools for the medically uninsurable (typically defined as having been rejected by 1 or more insurers) that began in the mid-1970s. These allowed those denied coverage by the private market to obtain coverage at capped rates --typically 150 to 175% of standard rates--with subsequent losses covered through a mix of general revenues and assessments on insurers (Communicating for Agriculture, Inc. 1993). Most states with such pools retained them even after adopting small group and individual market reforms that came later. By the late 1980s there was a recognition that these pools only address a small segment of the uninsured population and that the uninsured also were by no means limited to the poor or even near-poor that Medicaid expansion could plausibly covered. Given that three quarters of the uninsured were either workers or dependents of workers (CRS 1988), states began focusing on ways to encourage more employers--especially small employers--to offer health insurance coverage.

One approach that a few states attempted was to offer subsidies to entice more small employers to offer health coverage. These took the form of tax credits for both employers and, in a handful of states, for individuals. A far more widespread initiative was bare-bones coverage that permitted small employers to obtain lower cost coverage by exempting selected plans from mandated benefits. States mandate that insurance cover certain benefits such as mental health, substance abuse, or alternative practitioners, which they think should be included but that private purchasers often decline to purchase. States had been adding benefit mandates since the 1960's at such a rate that there were nearly 700 by 1988. Some observers see this as a response to market failure caused by adverse selection, while others see this as a response to political pressures from various stakeholders. Critics claimed that, by driving up health spending, such mandates made coverage less affordable: they were estimated to be responsible for as much as one quarter of the uninsured population (Goodman and Musgrave 1988).

Medical savings accounts (MSAs) are an alternative vehicle for effectively bypassing the high costs of mandates. They couple high deductible catastrophic coverage with a pool of funds that effectively allow individuals to cover their front-end medical costs with tax-free dollars. Because people use their own dollars for such care (and could benefit from any dollars that remain in their account unspent), MSAs are viewed as a way to avert the moral hazard created by insurance while at the same time ensuring the availability of funds to limit the risks involved in such large front-end deductibles. Purchasing cooperatives are another approach to making coverage more affordable: by pooling small employers (and in some cases individuals such as the self-employed) into a single pool, states hope to provide them with some of the same economies of scale and purchasing power that large employers historically have enjoyed. There are many different ways in which such health alliances or purchasing cooperatives could be (and are) established and run (see Hall 1999b for details).

Small Group and Individual Market Reforms. The limited success of early efforts to subsidize coverage, coupled with the perception that private insurance markets were beginning to "unravel," leading to increasing numbers of uninsured, prompted many states to begin exploring ways in which to reform private insurance markets as a strategy for expanding coverage. States' options for private insurance are more limited, due to a federal law known as ERISA, which prevents states from regulating in any manner health benefits that employers self-fund rather than purchase. Self-funding is common, especially by larger employers, and may have become more prevalent on account of ERISA. However, states are allowed to regulate the coverage, pricing and underwriting of insurance that is voluntarily purchased, either by employers or by individuals.

Health insurance reforms have multiple purposes, but they all tend to converge on a central aim: to create a marketplace dynamic that will both promote health insurance coverage and restrain health insurance costs (Hall 1999b). State initiatives used a combination of carrots and sticks to steer private insurance markets into more socially acceptable directions. Whereas experience rating or surcharging based on health status, and selective underwriting are deemed to be acceptable practices in some lines of insurance (e.g., life, auto liability), there is widespread concern about their use in marketing health insurance. At their core, many of these reforms represent efforts to overcome a central dilemma: that 1 percent of the population generates 30 percent of all health spending, while 50 percent of Americans account for only 3 percent of spending (Blumberg and Nichols 1996). In a system of voluntary coverage, it is not always easy to keep those who are healthy in the same pool as those who are sick since they could obtain coverage far less expensively by finding a pool that avoids these higher-risk individuals. But those in the unlucky 1 percent have average expenditures that exceed $140,000 in a single yearBfar beyond the means of all but the most well-to-do to handle. Thus, the chief objective of many reform efforts is to overcome the tendency towards risk segmentation in voluntary health insurance markets by encouraging the pooling of risks in a variety of ways. States began such reforms in the small group market in 1990, but by 1992 these had also spread in some states to the individual health insurance market. Overall, these various reforms had three components: access to coverage, rating restrictions, and risk pooling.

Availability of Coverage. Reform begins with ensuring that individuals have access to insurance and can retain it over time. This requires that insurers offer open enrollment (also called guaranteed issue) to all applicants, regardless of their health status. Guaranteed renewability further ensures that subscribers will not have their coverage canceled due to poor health. Limitations on pre-existing condition exclusions prohibit carriers from excluding anyone due to a particular health problem, but such limitations must be structured carefully to avoid the adverse selection that might otherwise occur in a voluntary market. If individuals knew that they could obtain coverage at any time regardless of health, they might opportunistically wait to purchase their coverage only when sick and drop it during periods of good health. Therefore, even under reform, waiting periods are permitted before coverage begins (generally 6 to 12 months) and insurers can only limit benefits for a pre-existing condition for a reasonable period of time (also typically 6 to 12 months). However, to protect individuals from having to be subjected to a new exclusion period each time they change jobs or health plans, portability requirements are needed to allow individuals to continue their coverage through a new group or individual plan. Such portability rules are to be distinguished from continuation-of-coverage requirements such as the federal requirements enacted in the Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA). These require employers with 20 or more workers to continue coverage for up to three years for those losing coverage. COBRA rules ensure continuation of coverage through the same employer health plan through which a qualified individual was covered at a premium not to exceed 102 percent of the group rate. In contrast, portability rules permit workers and dependents to move from one plan to another without undergoing an additional exclusion period. Collectively, these various reforms attempt to replicate for the individual and small group markets the natural pooling of heterogenous risks that occurs in larger employer groups. By suppressing medical underwriting, these reforms also seek to force insurers to compete based on their ability to promote efficiency in the delivery of medical care rather than on their ability to identify and segregate different risks (Hall 1994).

Rate Compression. While all of the foregoing guarantees access to coverage regardless of health status, it does not guarantee an affordable price. If insurers can experience-rate their premiums, proponents of rate compression argue that it will do little good to give high-risk people theoretical access to coverage as most could not afford it. Rate compression does not seek to lower premiums on average, but to reduce the variation across groups and individuals. One way to make coverage affordable for high risks is to charge everyone the same price regardless of health (although location and family size may still be used in setting rates): this is known as pure community rating. But there are both efficiency and equity reasons to question a pure community rate. On efficiency grounds, individuals may lose some incentive to engage in health prevention or to economize on treatment costs if part of the cost of failure to do so is borne by everyone else. On equity grounds, community rates may be objectionable insofar as they result in a transfer from young, relatively low-paid workers to older, typically much higher-paid workers. Thus, most states have adopted the less drastic approach of using rating bands that require premiums to stay within a certain range (e.g., + 25%) of the average premium for a given age or gender group. If the state allows no rate variation for individual health status but still allows adjustment for demographic characteristics, this is known as adjusted or modified community rating. Under either version, insurers can set rates based on selected enrollee characteristics that permit premiums to vary somewhat with predictable health expenditures, but not nearly as much as a pure experience-rated premium would allow.

Possible Unintended Consequences of Reform

In theory, these reforms are designed to increase the proportion of the population with health insurance. In practice, the interplay of these reforms with the complex structure of the private market and existing public programs and regulations may have resulted in no increase in coverage, or even a decrease. For one, as noted above, public insurance may crowd out private coverage. Subsidies or other initiatives to lower health insurance premiums may fail to reduce them sufficiently to cause meaningful increases in demand for private coverage, or may target too little of the relevant population. Similarly, risk pooling alone may not reduce the price of coverage to high-risk individuals to a sufficient extent. Worse still, restrictions on medical underwriting may lead to increased premiums for persons in low-risk categories, causing them to drop coverage, or causing insurers to withdraw from the market, particularly those who are not prepared to compete based on managing the costs of care. At the worst extreme, if too many low risks drop out and community rates rise sharply, medium and even higher risks may also drop out, leading to the additional premium escalation. Conceivably, this process could culminate in a "death spiral" that causes the complete collapse of the market (Rothschild and Stiglitz 1976), or the process could stabilize at a new equilibrium, one where many fewer people are covered, but including more of those with serious health problems who need coverage the most.

The relative weight of these competing arguments cannot be settled on the basis of a priori reasoning. Rather, the "proof of the pudding" is empirical evidence about whether or not these reforms actually increase coverage, as they are designed to do, or, conversely, whether the adverse consequences dominate and so the reforms have no or even a counterproductive effect. Various studies have been conducted using different data sources and methodologies to assess the impacts of health insurance reforms. Most studies assess the probability that an individual has some form of health insurance, but also measure related outcomes, such as the probability of insured people having private versus public coverage, the probability of privately insured people having group versus individual coverage, and the probability of an employer offering coverage. Other studies look at more subtle outcomes, assessed in a qualitative manner, relating to how these reforms alter the competitive dynamic and the interplay between private and public sectors.

The analysis that follows is based primarily on two such studies conducted by these authors. Two of us (Sloan and Conover 1998) conducted a quantitative, multivariate analysis of pooled Current Population Survey (CPS) data for adults for the years 1989-94. The CPS is a large, nationally representative survey conducted by the Bureau of the Census, which collects information each year about the nature and source of insurance coverage. State reforms were merged with CPS data to analyze the effects of these specific reforms and other determinants of the probability of coverage. The second study, conducted by Hall and colleagues, is an in-depth qualitative analysis of private market reforms in seven illustrative states (Hall 1999a). It consists of over 150 expert interviews in 1997 and 1998, with agents, insurers, and regulators. This article also reviews and summarizes more than 30 other quantitative studies. These studies either used CPS data from different years, other household-level data, or data from a separate series of surveys taken of employers. These evaluations from different vantage points and often with different analytical methods, reach some consistent conclusions, but they also differ in important respects.

Empirical Evidence

In general, these reforms have had very limited effects, either positive or negative. One must strain hard to see any impact because of the small fraction of people directly affected, the failure to address fundamental problems that led to lack of coverage in the first place, and the multiple factors that are occurring simultaneously. For the small scale effects that can be detected, some are positive--in the sense of achieving greater numbers of persons with health insurance--some are negative, and still others are ambiguous or subject to interpretation.

Medicaid Expansion. Overall, judged in terms of its effect on the probability of having some form of insurance, the most influential reform by far is Medicaid (Sloan and Conover 1998). The news for Medicaid expansion is not entirely good, however, due to evidence of crowding out private insurance. In states with a "medically needy" option, which allows recipients to subtract medical expenditures from income for purposes of determining Medicaid eligibility, people are both less likely to have private insurance and, for those with such coverage, are less likely to have purchased individual insurance. Although being pregnant (which potentially makes one eligible for Medicaid) has no effect on the probability of having any coverage, pregnancy does reduce the probability of having private insurance and individual private insurance in particular. This suggests that the increase in Medicaid was offset by a reduction in the number of privately insured pregnant women (Sloan and Conover 1998). These findings are consistent with other studies showing that at least 17 percent and possibly as much as half of the coverage of new Medicaid eligibles covered through recent expansions was offset by corresponding declines in private coverage (Holahan 1997, based on Dubay and Kenney, 1997 and Cutler and Gruber 1997).

Affordability. In contrast to the large effects resulting from Medicaid expansion, high-risk pools for the uninsurable do not have statistically significant effects on the probability of having insurance. (Sloan and Conover 1998). This is a much less studied reform, but our finding is consistent with other studies showing that such pools do not affect the likelihood of small firm insurance offers. This is also consistent with common sense: in most states with pools, coverage is measured in thousands whereas the number of uninsured often is measured in hundreds of thousands or millions. We also found some evidence of "crowd-out" as risk pools were associated with a lower likelihood of having group coverage.

For benefit mandates, the results of the Sloan and Conover study imply that removing 11 mandates (the sample mean) would increase the proportion of covered adults by 4%. Given that 18% of the sample was uninsured, this implies that between one-fifth and one-quarter of the uninsured problem for adults can be attributed to benefit mandates, a figure consistent with estimates by others (Goodman and Musgrave 1988 ; Jensen and Gabel 1992). With a single exception, other studies that have attempted to isolate the effect of some of the most expensive mandates, such as alcoholism or drug abuse treatment, generally have found no effects on coverage or, less frequently, negative effects (Table 1). Although much less studied, health insurance premium rate regulation also has been found to have no effects, or--as theory would predict--negative effects on coverage.

Although mandates reduce the likelihood of having coverage, selective overriding of mandates through bare-bones plans does not succeed in increasing the chances of being insured. Most other studies have found these plans have no effect on group coverage. Although a few studies have found that bare-bones plans slightly increase the fraction of small employers offering coverage, we found that they resulted in a statistically significant drop in employees who actually have group coverage overall. Similarly, employer tax credits and other means of subsidizing coverage have had no detectable impact on the probability of coverage in part because a) employers often were not aware of these programs; b) some efforts were just geographically-limited demonstration initiatives, and c) in other cases, they were time-limited subsidies that employers evidently found insufficiently enticing. Some studies did show a slight increase in employer offers of coverage, but on the whole, the response to such incentives was far lower than most reformers had hoped.

Reformers' attempts to achieve greater efficiencies in the small group market have met with only slightly more success. There is some evidence that public purchasing cooperatives have had a slight positive effect in increasing both the availability of plans among small employers and the overall odds of having group or private coverage. However, because they have achieved only minuscule enrollments--typically single-digit market penetration rates into the groups eligible for coverage--these alliances produce no measurable effect on the probability of being insured. The views of industry participants are instructive. Agents are hostile to these cooperatives because they fear their success will drive them out of business. Also, both agents and insurers have an innate antipathy toward government sponsored organizations. Many cooperatives have been operated in a way that makes them magnets for higher risks, which discourages insurers from participating and thereby minimizes the possibility of competitive bidding. Finally, cooperatives have found that it is difficult to achieve any substantial administrative efficiencies since they duplicate many of the functions performed by insurers and agents, and cooperatives usually require that insurance be selected individually by each employee rather than by an employer on behalf of a group.

Small Group and Individual Market Reforms. The numerous provisions in either small group reform or individual market reform designed to expand access to coverage have produced no detectable impact on coverage--positive or negative. Changes in underwriting rules such as guaranteed issue, guaranteed renewability, pre-existing conditions waiting periods and portability all had no detectable effects on the probability of being insured, the probability of having private coverage or the probability of having group coverage. Others who have attempted to decompose the effects of these various reforms at the same level of detail we did have reached the identical conclusion. Furthermore, studies that have examined whether these provisions affected small employer decisions to offer coverage similarly have found no detectable effects. Likewise, in the small group market, a number of quantitative studies have found that open enrollment and community rating have no (or almost no) effect on the probability of having insurance. This is true for different versions of these laws (e.g., loose versus strict rating restrictions). The only discernible impact has been in the individual market, where we found that modified community rating has a negative effect on private coverage, although no detectable effect on the odds of being insured. Interestingly, among those with insurance, non-group community rating increases the share of people with public as opposed to private insurance, suggesting a measure of "reverse crowd out" in which some people who are driven out of the private market due to increased premiums are able to join public programs.

Interviews with industry sources give these explanations for the lack of more positive impact. Medical underwriting was having much less impact in the small group market than was previously imagined. Therefore, most employer groups who wanted insurance before were able to get it, and most of those who previously were priced out still aren't willing to buy. Also, insurers are still able to engage in some degree of risk selection in how they structure their benefits, market their policies, and use the allowable degree of rating flexibility.

Industry sources also explain why there has not been a greater negative impact, as many in the industry predicted. They observe that the greatest potential for harm exists in the individual market, because of the much greater potential for adverse selection. This confirms our observations noted above and the accounts from states such as New York and Washington where reforms to the individual market have created much more difficulty than those in the group market. In the small group market, even pure community rating has not been severely disruptive because there are a sufficient number of insurers willing to compete on these terms. Indeed, price competition has been more intense for small groups following reform. This is due in part to the movement to managed care, in which insurers were aggressively vying for the market share necessary to justify capital investments in forming provider networks. But this movement itself may have been precipitated by the reform laws, which were intended in part to force insurers to compete on terms more consistent with how HMOs were accustomed and equipped to do business. Indeed, some studies have documented that even though small group reform did not have its intended effects on coverage, it did change market structure by accelerating a trend towards HMOs (Buchmueller and DiNardo 1998). Conversely, the impact and effectiveness of small group reform tended to be greater in areas of high managed care concentration (Buchmueller and Jensen 1997).

Reform Overall. One possible explanation of the generally dismal results on various market reforms is that by focusing on the trees, we missed the forest. Even if no single feature of state reforms had an effect we could detect, it is possible that collectively they did. Several studies have attempted to aggregate small group reforms into various clusters that can be characterized on a continuum from "stringent" to "weak" reform. The results are not encouraging. While these studies on average indicate that small group reform has succeeded in slightly increasing the propensity of small employers to offer coverage, they also collectively indicate that if anything, the odds of an individual worker having group coverage decline under small group reform. While paradoxical, this seeming inconsistency exactly mirrors what is now happening nationally in the small group market. A larger fraction of small firms are offering coverage than a few years ago, but many employers are shifting more of the premium costs to employees. As a result, the "take-up" rates by employees have declined by more than enough to offset the increase in employers offering insurance. The main reasons that an employee would not sign up for health insurance coverage with an employer are: (1) the premium contribution required of the employee was deemed to be too high; and (2) the employee had access to coverage from another source, such as through a spouse's employer. The net result has been declining group coverage for small firm workers (Ginsburg, Gabel, and Hunt 1998).

But the "bottom line" of all of these reforms is not what happened to group or private coverage; it is whether they succeeded in reducing the number of uninsured. That, after all, was a principal motivation for reforms. Most of these studies of reform "packages" find no detectable effect. Two such studies show that small group reform on average, and stringent reform in particular, have succeeded in lowering the risk of being uninsured among small firm workers. However, the inconsistency of findings makes it difficult to make a strong case in favor of reform from the evidence at hand.

Interviews with insurers and agents help to understand why small group reform apparently has had such a limited impact (Hall 1999a). They explain that most insurance sales are to people who already have coverage. For those already in the market, there is a high degree of price-sensitivity. Even slight differences will drive people to switch or choose different insurers, an effect that is facilitated by portability or continuity of coverage reforms. To this extent, market reforms have precipitated intense price competition, helping to keep people in the market, but they have not drawn large numbers of previously uninsured people into the market. Moreover, low-cost policies with fewer benefits have not sold well even to those who are purchasing, for two reasons. Subscribers have become accustomed to, and continue to demand, comprehensive coverage. Insurance agents, who mostly work on commission, are reluctant to argue otherwise, and insurers are reluctant to embrace policies with standardized benefits that do not match the structure of their current portfolios or their computerized claims and actuarial systems.

In contrast to the weak or inconsistent findings regarding small group reform, all of the evidence from the few available aggregate studies indicates that individual market reform has increased the odds of being uninsured. The negative effect of community rating for the individual market is confirmed by Hall's study. He documents: large numbers of insurers dropping out of the market, insurers declining to offer indemnity coverage, premium increases, covert risk selection, and circumvention of market and regulatory borders. However, there is no evidence of market collapse or anything approaching a "death spiral." Even under stringent community rating, the individual (non-group) markets in most states (with Kentucky a notable possible exception) have been able to reach a new equilibrium with higher prices and fewer covered, but covering those who are older and sicker. Lack of evidence of a death spiral has been confirmed by other studies and appears to be one instance in which, at least at the moment, the worst fears from theory have been soundly trumped by empirical evidence.

Conclusions

The few indications of either significant success or significant harms may be due to difficulties in measuring these complex and sometimes subtle phenomena. Some studies suffer from small sample sizes at a state level, but those based on the Current Population Survey have ample statistical power. Also, some studies include only a limited number of variables and so are unable to detect or control for complex interactive and confounding factors, but Sloan and Conover included a full slate of reforms and many demographic, income, and disability variables, which had statistically significant effects on coverage in anticipated directions. More troubling is the possibility that state reforms may be endogenous, meaning that states with high levels of uninsured populations may have been more likely to implement these reforms. If so, the effects of reforms may be masked in these studies. With so many reforms and states, it is difficult to fully test or correct for this possibility, but the various statistical techniques used in different studies suggest that endogeneity is not a large problem, or they have found similar effects despite some endogeneity.

Finally, it is entirely possible that it is still too early to know the full impact of these reforms. As one actuary explained in 1998, the market reforms are only "four years young" and even if there is only a "slow trickle" of better risks leaving the market, this can still result in an adverse selection death spiral over a span of 10 years. Other observers note that market reforms were, by happy coincidence, enacted at the low point of cyclical underwriting profitability and during an intense battle among managed care firms for market share. Now that insurers are consolidating and making up for past losses, there will be nothing to shelter vulnerable market segments from the full brunt of inevitable cost increases. Still others are hopeful that a new generation of reforms, focused on particular market segments and designed with greater sophistication, will achieve more pronounced successes. These include subsidies for uninsured children, and private associations for individuals or small groups.

What does seem clear is that it is hard to construct a reform that has any meaningful effect without making things worse. If subsidies are too small, they have no effect. But if they are large enough to have an effect, they result in "crowd-out," that is, people dropping private coverage to move to subsidized programs. Similarly, reforms in the private market at best hold enrollment steady, but also make insurance more available for higher risks. If it is too attractive for higher risks, however, adverse selection sets in and total enrollment drops. These dilemmas are intractable in a system: (1) that maintains a border between public and private insurance; (2) that has skewed risk distribution; and (3) in which the purchase of insurance is voluntary. Perhaps with more empirical study, it will be possible to determine whether future reforms can overcome these daunting odds.

Word count: 5649 (excludes references) 6393 (with references)

References

Bennefield, Robert L. 1998. Health Insurance Coverage: 1997, P60-202. U.S. Department of Commerce, Washington D.C..

Blumberg, Linda J., and Len M. Nichols. 1996. First do no harm: Developing health insurance market reform packages. Health Affairs 15, no. 3: 35-53.

Buchmueller, Thomas C., and Gail A. Jensen. 1997. Small group reform in a competitive managed care market: the case of California, 1993 to 1995. Inquiry 34, no. 3: 249-63.

Buchmueller, Thomas C. 1997. Managed competition in California's small-group insurance market. Health Affairs 16, no. 2: 218-28.

Buchmueller, Thomas C., and Gail A. Jensen. 1997. Small group reform in a competitive managed care market: the case of California, 1993 to 1995. Inquiry 34, no. 3: 249-63.

Buchmueller, Thomas, and John DiNardo. 1999. Did community rating induce an adverse selection death spiral? Evidence from New York, Pennsylvania, and Connecticut, National Bureau of Economic Research, Cambridge, MA.

California Department of Corporations. 1994. Small employer group reforms: Monitoring of standard employee risk rates: Report to Legislative Assembly. Sacramento: California Department of Corporations.

Communicating for Agriculture, Inc. 1993. Comprehensive Health Insurance for High-risk Individuals: A State-by-state Analysis, Communicating for Agriculture, Inc., Bloomington, MN.

Congressional Budget Office (CBO). 1999. The Economic and Budget Outlook: Fiscal Years 2000-2009, The Congress of the United States Congressional Budget Office, Washington, D.C..

------. 1981. The Impact of PSROs on Health-Care Costs: Update of CBOs 1979 Evaluation, U.S. Congressional Budget Office, Washington, DC.

------. 1999. Maintaining Budgetary Discipline: Spending and Revenue Options, U.S. Congress, Congressional Budget Office, Washington, DC.

U.S. Congress, Library of Congress, Congressional Research Service (CRS). 1988. Health Insurance and the Uninsured: Background Data and Analysis, U.S. Government Printing Office, Washington, DC.

Cutler, David, and Jonathan Gruber. 1997. Medicaid and private insurance: Evidence and implications. Health Affairs 16, no. 1: 94-200.

Dubay, Lisa. 1997. Did Medicaid expansions for pregnant women crowd out private coverage? Health Affairs 16, no. 1: 185-93.

Dyckman, Z., and J. Burnette. 1992. Programs to improve health insurance access for small business--what works and what doesn't., Report to the Small Business Administration. Columbia, MD: Center for Heatlh Policy Studies.

Families USA . 1993. No sale: the failure of barebones insurance, Washington, D.C..

General Accounting Office. 1992. Access to health insurance: state efforts to assist small businesses, Pub. no. GAO/HRD-92-90. Washington, D.C..

Gruber, Jonathan. 1994. State-Mandated Benefits and Employer-Provided Health Insurance. Journal of Public Economics 55: 433-64.

Holahan, John. 1997. Crowding out? How big a problem? Health Affairs 16, no. 1: 204-6.

Gabel, Jon, and Gail Jensen. 1989. The Price of State Mandated Benefits, Health Insurance Association of America, Washington, DC.

Ginsburg, Paul B., Jon R. Gabel, and Kelly A. Hunt. 1998. Tracking small-firm coverage, 1989-1996. Health Affairs 17, no. 1: 167-71.

Goodman, John C., and Gerald L. Musgrave. 1988. Freedom of Choice in Health Insurance, NCPA Policy Report No. 134. The National Center for Policy Analysis.

Hall, Mark A. 1999a. An Evaluation of Health Insurance Market Reforms, Wake Forest University School of Law, Winston Salem, NC.

------. 1999b. Health Insurance Market Reforms: Introduction and Background Analysis. Wake Forest University School of Law, Winston Salem, NC.

------. 1994. Reforming Private Health Insurance. Washington, DC: AEI Press.

Institute for Health Policy Solutions. 1995. State Experiences with Community Rating and Related Reforms.

Jensen, Gail A. 1993. Regulating the Content of Health Plans. in American Health Policy: Critical Issues for Reform. editor Robert B. Helms, 167-93. Washington, DC: The AEI Press.

Jensen, Gail A., and Jon R. Gabel. 1992. State Mandated Benefits and the Small Firm's Decision to Offer Insurance. J Regul Econ 4: 379-404.

Jensen, Gail A., and Michael Morrisey. Managed care and the small group market. presented at Managed Care and Changing Health Care Markets, 379-404.

Jensen, Gail A., and M. Morrisey. 1989. State insurance regulation and the decision to self-fund.

Jensen, Gail A., and M. A. Morrisey. 1997. The effects of state inititatives in the small group insurance market.

Jensen, GailA., M. Morrisey, and R. Morlock. The effects of state initiatives in the small group insurance market, (Revised paper originally presented at The Robert Wood Johnson Foundation conference, "The Rapidly Changing Insurance Market: Policy and Market Forces," Washington D.C., 22 March 1995)..

Litow, Mark E. 1994. The impact of guaranteed issue and community rating in the State of New York, Milliman and Robertson, Inc., Brookfield, WI.

Lipson, Debra J., and Jeanne De Sa. 1995. The Health Insurance Plan of California: First Year Results of a Purchasing Cooperative, Alpha Center, Washington, DC.

Marsteller, Jill A., Len M. Nichols, Adam Badawi, Bethany Kessler, Stephen Zuckerman, and Shruti Rajan. 1998. Variations in the Uninsured: State and County Level Analyses, The Urban Institute, Washington, D.C..

Minnesota Department of Commerce. 1995. Study of small employer health insurance reform. Minneapolis: MDC.

Morrisey, Michael A., and Gail A. Jensen. 1996. State Small-Group Insurance Reform. in Health Policy, Federalism, and the American States. Editors Robert F. Rich, and William D. White, 71-95. Washington, DC: The Urban Institute Press.

------. 1997. Switching to managed care in the small employer market. Inquiry 34: 237-48.

Nichols, Len M., Linda J. Blumberg, Gregory P. Acs, Cori E. Uccello, and Jill A. Marsteller. 1997. "Small Employers: Their Diversity and Health Insurance." The Urban Institute, http://www.urban.org/health/smemployers.htm.

Oliver, Thomas R., and Emery B. Dowell. 1994. Small-employer health alliance in California. Health Affairs 13, no. 1: 350-351.

Rothschild, Michael, and Joseph E. Stiglitz. 1976. Equilibrium in competitive insurance markets: an essay on the economics of imperfect information. Quarterly Journal of Economics 4: 630-649.

Short, Pamela F., and Jessica S. Banthin. 1995. New Estimates of the Underinsured Younger than 65 Years. Journal of the American Medical Association 274, no. 16: 1302-6.

Sloan, Frank A., and Christopher J. Conover. 1998. Effects of state reforms on health insurance coverage of adults. Inquiry 35.

Thorpe, Kenneth E., Ann Hendricks, Deborah Garnick, Karen Donelan, and Joseph P. Newhouse. 1992. Reducing the Number of Uninsured by Subsidizing Employment-Based Health Insurance: Results from a Pilot Study. Journal of the American Medical Association 267, no. 7: 945-48.

Uccello, Cori E. 1996. Firms' health insurance decisions: The relative effects of firm characteristics and state insurance regulations, The Urban Institute, Washington D.C..

Zuckerman, Stephen, and Shruti Rajan. 1999. An alternative approach to measuring the effects of insurance market reforms. Inquiry 36: 44-56.

¹ Frank A. Sloan is the J. Alexander McMahon Professor of Health Policy and Management and Professor of Economics at Duke University. He directs the Center for Health Policy, Law and Management. Christopher J. Conover is Assistant Research Professor of Public Policy Studies at the Terry Sanford Institute of Public Policy and Senior Fellow in the Center for Health Policy, Law and Management. Mark A. Hall is Professor of Law and Public Health at Wake Forest University Schools of Law and Medicine and an Associate in Management at the Babcock School of Management.