Dimensions: 6'' x 9'' 115 pages AEI Press  (Washington) Publication Date: October 1999 Hardcover ISBN: 0844741191 Price: $ 39.95 Add to Cart   Examination Copies

Health insurance can help directly with the financial consequences of health risk and, by increasing access to care, can make possible some alleviation of the threats to health. For both of those reasons, a majority of Americans who are not eligible for public insurance voluntarily arrange for private insurance coverage for themselves and their dependents. This book addresses the following questions: How well do private insurance markets function to smooth or pool medical expenses? Do some parts or types of insurance markets work better than others? Should steps be taken to improve market functioning, and might some actions, while helping to hold down medical-care spending or achieve greater equity, simultaneously threaten the appropriate pooling of risk?

Mark Pauly is the Bendheim Professor of Health Care Systems at the Wharton School, University of Pennsylvania, where Bradley Herring is a Ph.D. candidate. This summary is drawn from the book's introductory and concluding chapters.

Risk pooling and risk segmentation are undoubtedly the most complex, confusing, and contentious issues of "market failure" in the analysis of private health insurance markets. In formulating and implementing public policy toward risk pooling and risk segmentation, it is extremely difficult to avoid unintended consequences and to do more good than harm.

The research and analysis we report in this study performs some useful functions. First, it definitively disproves some common misconceptions about how insurance markets function. Second, it clarifies the likelihood that a few possible changes in public policy might be improvements in public policy, making some actions look less likely to provide benefit and increasing the chances that others will help.

The theoretical insight that guides our study is this: if consumers value risk pooling, competitive markets can develop ways to produce that pooling--perhaps not perfectly, but at least to a moderate extent. Conversely, market failures in insurance must have specific causes that can be identified and measured.

The empirical portion of the study uses data from the 1987 National Medical Expenditures Survey (NMES) to examine the relative extent of risk pooling across three different submarkets: large employment-based group insurance, small-group insurance, and individual (nongroup) insurance. It finds that all three markets pool risk to some extent, and that the differences among the three markets are smaller than they are usually thought to be.

Purpose

Private health insurance markets in the United States are subject to a hodgepodge of state and federal regulations, with rules, mandates, and subsidies ebbing and flowing at a moderately rapid rate. The present unsettled state of these markets provides the policy motivation for questions about performance. This book addresses the question of how markets could perform, and do perform, by offering a combination of analysis based on the economic theory of competitive insurance markets and empirical evidence concerning the performance of markets from a period in the recent past when markets were less regulated and less changeable than they are now.

We look primarily at data drawn from the NMES of 1987, a time when those markets were subject to considerably less federal and state regulation than now exist. The NMES is a data-rich household survey of more than 38,000 persons including detailed information for insurance purchases and individual health conditions. We first construct estimates for individual and household risk (expected medical expenses) based on individual age, gender, location, and health status, and then examine the empirical relationships between those estimates of risk and premiums, coverage, and wages within each of the three insurance submarkets.

The study focuses specifically on the question of whether employment-based group insurance, furnished in various settings, is more effective as a mechanism to pool risk than nongroup (individual) insurance. Since the employment-based mechanism is, despite significant erosion, by far the most common way of furnishing private health insurance to consumers in the United States, and since the performance of that market is increasingly challenged as deficient and inadequate by critics on the left and the right, this study will help to suggest whether there are ways to improve public policy and private behavior.

The policy trade-off is between group health insurance provided through the employment setting and nongroup health insurance marketed and purchased in the same way as other consumer insurances, such as homeowner or auto insurance. At present, public policy favors group insurance, providing it with a tax subsidy in excess of $100 billion per year.

What Isn’t Necessarily So

The conventional wisdom in insurance-market analysis assigns the highest marks for risk pooling to large employment-based group insurance, and the worst score to the nongroup individual market. Perhaps that ranking is correct, but our results suggest that the appearances that gave rise to it may be quite deceiving. If we look at the individuals who actually buy nongroup insurance and what they actually pay, that type of insurance appears to be much less effective in segmenting risks than one would assume simply by reviewing what insurers say they try to do.

Considerable valid research, some quite recent, describes nongroup insurers who seek to charge low premiums to low risks and high premiums to high risks, and who consequently avoid selling to high risks and seek out low risks as prime customers. But that research is misleading, because it looks only at one side of the market--what sellers would like to have happen. If we look at buyers' objectives, most buyers (regardless of risk level) try to purchase decent insurance to protect themselves from financial ruin--and to do so at the lowest premium they can find. (A small minority of potential buyers may not share that goal, of course. Low-income households may at some point view accepting charity or caring for bad debt as preferable to paying market-insurance premiums, which they cannot afford, to protect important assets, which they do not have.) Buyers appear to have some success in their quest for a decent deal. Most important, they frequently chose policies with guaranteed renewability built in, even when not required by law.

Our results show that there is much more risk pooling in nongroup insurance markets than is suggested by effective risk rating. Across a random sample of citizens, there is still substantial variation in the premiums paid for nongroup insurance for a given level of coverage--but that variation is by no means proportional to risk. To be specific, persons whose estimated risk level (understood as expected expenses) is twice the average are paying premiums only about 20 to 40 percent higher for a given individual insurance policy. Moreover, premiums do not vary with the presence of high-risk chronic conditions. We also find, in this small sample, no conclusive statistical evidence that ordinary Americans seeking insurance in the nongroup market are differentially deterred from obtaining coverage because they are higher-than-average risks.

This nongroup market is, nevertheless, far from perfect: people at all risk levels are charged, and pay, premiums that are high relative to the benefits they will get back. That high loading arises because this market, like all customized markets (and like its analogues in automobile and homeowner insurance), is very expensive to administer. The correct conclusion to be drawn is that the nongroup market is high in cost, although it may not be very unfair. Everyone overpays, and that discourages insurance purchasing with an approximately even hand.

What should we conclude about the group insurance markets, especially the large-group markets idealized as "corporate socialism?" Here appearances are even less clear, and they are also quite deceptive. If we take the employee premiums actually charged and paid as our measure of what employees pay for group insurance, the picture is one of enormous variation--much more than for nongroup health insurance or for any other insurance. Some people take jobs where they literally pay nothing for insurance, and others pay a lot, although few pay the full per-person cost. These employee-premium payments do not vary with risk, but they do vary over a wide range. The odds that an employee will pay a premium two or three times the average under group insurance are much higher than the odds of similar discrimination in the nongroup market, although of course the absolute value of the average employee premium is much lower than the average nongroup premium for the same coverage.

Economists will object to the preceding discussion because it ignores the bulk of the premium for employment-based coverage, which is nominally paid by the employer. Of course, if the employer does really pay the premium by accepting lower profits or raising prices to consumers, those premiums are spread around, albeit in an uneven (and perhaps illogical) way. But virtually all economists believe that, in some reasonably moderate run, the employer premium in the aggregate comes out of worker wages. Even if that proposition were granted (and not all policymakers would grant it), it leaves two difficult, unanswered questions: If employees sacrifice wages, which employees sacrifice? How much?

Our firm conclusion is that no one really knows how this cost is distributed, and no one knows whether it might be related to risk or to something else that leads to substantial and unpredictable variation across individuals. After all, the problem is not that high-risk individuals pay premiums much above average, but that anyone pays premiums much above average, because of unforeseen events over which the payer has no control. Such created uncertainty is surely possible in the group market.

We explored the question of how employee wages are affected by employer payment of insurance and found evidence, consistent with other recent research, that these reductions in cash wages are not spread equally across all workers, regardless of risk level. Specifically, we found evidence that other workers receive lower raises for seniority over their careers in firms where they obtain insurance than in firms where they do not. Thus, through lower wages, older workers effectively pay more for their insurance. In contrast, wages did not vary with higher risk attributable to female gender or to the presence of long-term chronic conditions. In short, while there is some risk pooling in employment-based group insurance, it is far from complete.

What Probably Is So

This study provides circumstantial evidence for an important conclusion already known from other sources. The serious problem with nongroup insurance is that it is expensive (relative to what the consumer gets) for everyone, regardless of risk level. Our measures of variation in total expenses, including out-of-pocket payments, attest to the lower likelihood of coverage for people without access to employment-based settings--even people with adequate income.

The other certain conclusion is that all insurance pools risk. In 1987, the specter of insurance so accurately risk-rated as to be pointless was not present--as it will not be present in any future setting. The great bulk of variation in actual health expenses and benefits, our results clearly show, is always unpredictable. People who buy insurance and get sick receive transfers after the fact from people who buy insurance and do not get sick. The latter group, unfortunate as it is to have been "taken" by the insurer in a financial sense, would surely not trade places with the former.

The most important policy implication from this discussion is that political energy should be directed toward designing ways to furnish good and cheap nongroup insurance, and should not be diverted into a fixation with risk segmentation. Moreover, group insurance also fails to pool some risks, so it is not an unequivocally preferable alternative even when it is feasible. Indeed, influences other than risk cause so much apparent variation in total premiums for a given, nominal level of coverage that the variation attributable to risk levels is very difficult to detect, much less isolate. The possibility of risk segmentation may be the Achilles' heel of heroic competitive insurance markets, but this particular organism has many other, more immediately threatening points of vulnerability.

Toward Big-Picture Policy

What do our results mean for the formulation of policy toward private health insurance in general and employment-based health insurance in particular? The theoretically important and empirically measured effects of the differential tax subsidy to employment-based health insurance are well known. The subsidy is inequitable, horizontally and vertically. It leads to excessive levels of insurance coverage--and to deficient levels of opportunities for individual choice of coverage, especially in small firms. Our results strongly suggest that the only good thing left to say about the tax subsidy--that it does, despite all these harms, strongly encourage risk pooling--is probably not true. Considerable pooling occurs even without the subsidy. It does not encourage much pooling, and some of the redistribution it might sometimes encourage across risk classes--redistribution from young workers in large firms to middle-aged workers in large firms, but not across firms--is not especially socially desirable.

If improvements are to occur, they must address the major issues surrounding nongroup insurance--high loading costs, lack of persistence in purchasing, and mistaken choices. Some adequate solution to the sales, billings, and collection tasks is needed. There may be no obvious way to solve those problems completely, short of mandated purchase, but improvements may be possible.

Like the nongroup market, the small-group market also needs some work, especially for low-wage groups. Our results do show that high-risk, low-wage employees in small groups are less likely to obtain insurance than others. While "discrimination" is confined to this segment of the market (and is not apparent in either nongroup or large-group markets), it should be of concern. Redirecting the tax subsidy to the nongroup alternative might do the most good here, but so might some imagination directed at ways to lower administrative costs in the small-group market. An improved group-to-individual conversion mechanism to tide people over when the employer drops coverage might help as well.

Setting Priorities

How serious is the problem of risk segmentation in nongroup health insurance in the United States? Our research and that of others suggests that the answer may well be, not as serious as is commonly believed. The risks that do not get pooled, such as age, may not really need to be pooled. While there are possibly some pockets of the kind of risk rating that combines serious illness with high premiums, they are rare. Guaranteed renewability is a common policy feature.

In contrast, other problems with private health insurance are so prominent that they cannot be ignored. High administrative costs discourage some buyers, especially low-income ones. Subsidies are hopelessly mistargeted and ineffective. The availability of free care, in the face of high prices and minimal help, discourages the very people who need insurance from buying it. Policy should focus on where it can do the most good.