Congressional Testimony
March 9, 2004
Statement of Len M. Nichols, Ph.D.1
Vice President, Center for Studying Health System Change
Before the U.S. House of Representatives
Committee on Ways and Means
Health Subcommittee
Hearing on the Uninsured
Madame Chair, Representative Stark and members of the Subcommittee, I am honored to have been invited to testify before you today on a topic of such importance to our nation, facts about those who live without health insurance. My name is Len M. Nichols and I am an economist and the vice president of the Center for Studying Health System Change (HSC). HSC is an independent, nonpartisan health policy research organization that is principally funded by The Robert Wood Johnson Foundation and is affiliated with Mathematica Policy Research. We conduct nationally representative surveys of households and physicians, site visits to monitor ongoing changes in the local health systems of 12 U.S. communities, and we monitor secondary data and general health system trends. Our goal is to provide members of Congress and other policy makers with unique insights on developments in health care markets and their impacts on people. Our various research and communication activities may be found at www.hschange.org.
I am also a member of the Policy Advisory and Research Review Committees of the Economic Research Initiative on the Uninsured (ERIU), a project of The Robert Wood Johnson Foundation that convened a group of health and labor economists to sort out what we do and do not know about the uninsured in our country. The ultimate goal was to inform policy makers who may consider specific policy responses. The project was directed by Catherine McLaughlin, a professor of economics at the University of Michigan. I was a co-author of a chapter in a recently published book that grew out of this project, Health Policy and the Uninsured (Urban Institute Press, 2004). My chapter was titled, "Why Are So Many Americans Uninsured?"
My testimony today is organized around a theme called "Myths About the Uninsured." This theme was also the one used at a recent press briefing, which Mark Pauly—professor of economics and health care systems at the Wharton School of the University of Pennsylvania—and I did together to report on the research contained in the ERIU book. Dr. Pauly and I took turns clarifying the research pertinent to each myth, and we both essentially agreed with what the other said. Dr. Pauly has kindly allowed me to use some of his logic and words in my written testimony. I take sole responsibility for any remaining errors or ambiguity, however. In this testimony I have combined and rephrased some of the myths we used that day, and I have added one more that grows out of the spirit of the research but is wholly my contribution to your deliberations. The 10 myths about the uninsured my written testimony will highlight are:
Below I explain why economists think all these myths are misleading to an important degree.
Myth #1: We know how many people are uninsured. Forty-four million is the "official" number from the most recent Current Population Survey, but the truth could be (and is) on either side. The CPS asks: did you have health insurance at any time in the 12 months ending two months ago? Penn State Professor Pamela Farley Short’s chapter clarifies the overwhelming evidence that many respondents answer the CPS insurance questions incorrectly. Even if answered perfectly, this concept omits quite a large number of people who lack insurance for a period shorter than 12 months or the interval in which they lacked insurance did not match the particular window asked about. So the truth is that far more than 44 million are uninsured for a period shorter than 12 months in a given year.
On the other hand, other surveys make clear that the 44 million number overstates by as much as a factor of two the people who were uninsured for all of the prior 12 months. The Census Bureau’s Survey of Income and Program Participation, HSC’s Community Tracking Household Survey, and AHRQ’s Medical Expenditure Panel Survey, as well as the Urban Institute’s National Survey of America’s Families, all have probed survey respondents for years and said, now, are you really sure that you didn’t have any insurance for that time period?
The subtle lesson here is to pay attention to time frame. The longer the period of time, the smaller the number of people who are always without health insurance and the larger the number of people who are without insurance for some of the relevant time period.
Perhaps the most important thing to establish from a policy perspective is not the precise number, as long as we are confident that the number of uninsured for an entire year is in the tens of millions, and researchers are confident of this. The most important analytic measurement may be the time trend in the percentage of non-elderly Americans who are uninsured, which has recently been quite adverse. Trends are more reliably calculated, assuming that the same kinds of respondent errors and measurement imperfections are present each year, which is a reasonable assumption.
Myth #2: The uninsured are all alike. This is manifestly false. The uninsured tend to be somewhat lower-income and in somewhat poorer health, but because there are so many of them and because they do span various dimensions of American life, there are many who are young and healthy but there are many who are not; there are many who are reasonably well off, including a sizable fraction above the median income. And then, as is also important to note, there is a sizable fraction below the poverty line who are also sick and in a very bad way. The message of this diversity for policy design in a world of public budget constraints is that you probably want to be careful and clever in making limited funds go as far as they can toward expanding coverage. Of course, policies that are target efficient are also more complex. In addition, there are inherent trade-offs in choosing a target population, for example, in extending lower cost coverage to a larger number of relatively healthy uninsured vs. extending higher cost overage to a smaller number who are likely to have more health risks. Value judgments are unavoidable when making actual policy choices in this case.
Myth #3: Coverage is Coverage is Coverage. Designs of insurance policies really do matter. Insurance is not insurance. Insurance differs in terms of the kind of financial protection it offers, in the potential for improvement in health it offers, and the humanity of the treatment when you contact the healthcare system. To put it slightly differently, imagine a policy that gave every American as much insurance as $100 could buy. Every American would then have insurance, we’d have zero uninsured, but we wouldn’t really be in that much better of a situation than we are now.
But the punch line is that the head counts of coverage are not enough, that the actuarial value2 of insurance may vary, and even given the same number of dollars spent on insurance, the consequences of insurance may be different, depending on the form that insurance takes. Furthermore, the harm of not having insurance may vary with the length of time coverage is lost, as well as with nature of the people without coverage.
Moreover, the kind of insurance that people get depends very strongly on where they get it. If they work for a large Fortune-500 firm whose benefits department is run by professionals, they will get very good and well-designed coverage. If they get it from Gus and Otto’s Garage, and neither Gus nor Otto was trained as an actuary, it may not be such great coverage. And if they get it in the individual market, it depends on how good the consumers are at searching through the wide range of possibilities available to find the best buys out there compared to other less satisfying policies that are also available and may be easier to find.
Myth #4: Health insurance would improve the health of all the uninsured. This is among the more complicated and emotional disputes in health policy analysis. I will clarify how the literature may be correctly interpreted on what is accepted as proven now, and take some care to distinguish this from what we would like to know and from what we might think policy should do in the face of real-world imperfect knowledge.
Helen Levy and David Meltzer, both professors at the University of Chicago, were asked to review the literature to assess this question: "Does health insurance really affect health status?" They were rightly concerned that standards of proof about causation in this area have often been lower than they should have been in many published papers, even in many prestigious journals over the years. And they chose to use a standard of proof that is quite high, but is nonetheless becoming increasingly common in the social sciences, that causation is not likely to be appropriately inferred unless there has been an adequate natural experiment or a true experiment in which a representative sample of people are assigned to have or not have insurance for the duration of the experiment. This standard of proof for causation has become more widely shared as researchers have realized that there may be important but unobservable differences in people that make different choices about things like insurance, diet, exercise and education. If we merely observe what people do, it is hard to be sure what caused and what merely reflected health outcomes. For example, if some people (for whatever reason) have a low value for their health, it is likely that they will not obtain health insurance but also will not take steps (like preventive care and better health habits) that are known to affect health. We can easily observe the association of lack of insurance and low health, but it will be their low demand for health that causes the poor health, not lack of insurance per se.
Now, this standard of proof has rarely been met in the research literature, but when it has, the bulk of the evidence suggests that health insurance does indeed have positive effects on the health of certain populations, and indeed, those most often at the center of a policy debate: the poor, the elderly, the truly sick and children. What has not been proven by this standard is that universal coverage would improve the health of all of the uninsured, and this leads economists to the following three inferences. (1) Because we do not have an unbiased measure of the effect of health insurance on health in general, we cannot say with certainty that more public subsidies for health insurance for the general population would improve health status more than would an increase in the capacity of public health centers or public hospitals, better education about diet and exercise, or a more equal income distribution for that matter; (2) Understanding more about the complicated pathways that different types of people traverse from coverage to health status through health services, and indeed, health insurance and health education, would help us make far better calibrated recommendations to policymakers; (3) There are many reasons to support universal coverage, but the analytic case for the short-run positive health effects is not the strongest one, at least for the higher income and basically healthy uninsured who comprise roughly 40 percent of the uninsured today.
Another element of this generalized myth is that universal coverage would eliminate poor health status among vulnerable populations. Despite considerable policy attention and focus, rather large disparities in health care outcomes among different population subgroups persist in our country. At least part—and perhaps a very large part—of the reason lies in differential access to health insurance. Harold Pollack and Karl Kronebusch, from the Universities of Chicago and Yale, respectively, have written a chapter that focuses on access to health insurance by six subgroups that are often considered vulnerable for one or more reasons. The groups are the low-income population, children, racial and ethnic minorities, people living with chronic conditions, the near-elderly, and people suffering from psychiatric and substance use disorders.
Each group raises distinct concerns for public policy, health insurance and the healthcare delivery system. Pollack and Kronebusch conclude there are four basic reasons vulnerable populations often lack health insurance: (1) they have medical and social needs that hinder their access to good jobs and to private health insurance markets; (2) they have general economic disadvantages, including lower incomes, which impede their ability to pay for health insurance when it is available and less access to jobs with employer-sponsored insurance, which makes it cheaper; (3) they sometimes face discrimination based on race, ethnicity or language; and (4) they sometimes suffer from impaired decision-making and rather imperfect proxy decision-making. And unfortunately, many people in vulnerable populations face multiple barriers at the same time.
As an example of troubling disparities, taken from AHRQ’s recent healthcare disparities report,3 black women have lower rates than white women of cancer screening and higher rates of diagnosis in late stage and consequently higher death rates. These death rates apparently persist even after controlling for education and income. They also appear to persist after controlling for insurance. This suggests that insurance alone cannot solve the problems faced by vulnerable populations. Pollack and Kronebusch wrote: "The data provide ample warning that one should not oversell the possibilities of improving health status and individual well-being through expanded health coverage. Expanded coverage is unlikely to eliminate the high rates of death and illness that arise from multiple causes and require multifaceted interventions." In other words, insurance will help these populations and reduce gaps,4 but eliminating the disparities gap will require multiple policy changes.
Myth #5: Individuals without insurance choose to be so. In some general sense this is true. No law prohibits people from buying insurance, and most could buy individual insurance, although if you are a very high-risk person you might find the price quoted to exceed what you expect to get back in benefits, and a small fraction of people are outright denied access to insurance at any price. But, more generally, if we think of realistic choice or reasonable choice for low-income people or for people at high levels of risk, if they don’t have insurance now, obtaining insurance voluntarily without further subsidies is probably not a realistic option.
We also know—especially from some of the studies described in the chapter that Linda Blumberg of the Urban Institute and I wrote —that job matching is not perfect and there are some people who probably want insurance who can only find a job in firms that do not offer insurance. Now, they do not want it so much they are willing to pay whatever it may take in the non-group market, but they do want insurance and can not get it. There are also some other people who would rather have higher wages than health insurance but can only find a job in a firm that offers health insurance to them along with an acceptable wage. The out-of-pocket premium required of them may even be low enough to induce them to take-up this employer offer, but maybe not, and thus this low relative demand—or willingness to pay—for health insurance may be the core reason roughly 20% of workers do not accept their employer’s offer.
Myth #6: U.S. employers spend $400 billion a year for workers’ health care. This issue reveals how differently economists think from most people. Imagine that somebody could wave a magic wand and end $400 billion of employer payments for health insurance. First, the definition of "pay" in economics is not who writes a check, but the definition is wrapped up in the question, would employers then get to keep $400 billion more of profits that they could distribute to stockholders on to increase compensation of their senior executives, or to do whatever they wanted to do with it?
And the answer that economics gives—well summarized in a couple of chapters in the ERIU volume—is no. One way to think about why the answer is no is to think about why employers offer health insurance. Now maybe some of them do it out of the goodness of their heart, and some of them do it because they think insurance makes employees healthier and therefore more productive, and under certain circumstances there may be a business case for doing that. But most employers, at least if you locked them in a room and asked them, "Why are you doing this if you whine and complain about it all the time, why don’t you just stop offering health insurance?" And their answer is, "Well, we need to offer health benefits to be competitive in the market for workers, to be able to attract and retain high-quality workers," which is another way of saying they offer health insurance to obtain a given quality of worker for less total compensation outlay than they would have to expend in the absence of health insurance.
And so the punch line is that if somehow employers were not allowed to spend $400 billion on health insurance, then in order to attract the workers that they were formerly attracting with this benefit, they would have to use money or some other benefit that could well eat up or even exceed all of the savings. So that’s at least one way to think of why economists are out of step with the rest of the world. Our theoretical logic—and some careful empirical work—tells us that (most) employers actually do not pay for health insurance (and by the way, then, health insurance costs are not what makes U.S. products noncompetitive internationally). Economists believe that ultimately most workers end up paying for health insurance in the form of lower wages.
This argument also works in reverse, which may be more germane for the current situation. Imagine that employers are mandated to provide health insurance, as has been passed in some states and introduced at the federal level from time to time. Who’s going to actually end up paying for that? Well, the story is just the same as above but in reverse. Initially of course employers will do most of the complaining about it, as they have, and threaten to lay off workers, but that will, at least over time, soften the labor market, cause raises to be smaller than they otherwise would have been, and sooner or later, the bulk of workers will end up paying for the health insurance that policy makers gave them with the best of intentions. They’ll end up paying for it themselves through reduced wages and fewer jobs unless they receive a subsidy. Of course, if they receive a generous subsidy or their employer does, that subsidy will ultimately go to workers.
Myth # 7: The decision to remain uninsured has no effect on anyone else. An overarching feature of modern labor markets is worker heterogeneity; we all differ in many important dimensions, including our preferences for health insurance arrangements. One consequence of heterogeneity is that different kinds of compensation packages may exist in equilibrium, some with a broad array of health insurance choices attached, some with one health insurance option embedded, and some with only cash wages to entice a prospective employee to give up their leisure time. Michael Chernew and Richard Hirth of the University of Michigan focus their critical review essay on the connections between decisions made by different people in the nexus of labor and health insurance markets. This myth was chosen to highlight the reality that some workers’ willingness to work at jobs without health insurance—while this may be a minority of workers today—has important consequences for the rest of us.
First and foremost, it means employers have a choice about whether to offer health insurance, and they will make this decision largely based on the preferences, expectations and productivity of the dominant type of worker they need to produce their products and services, as well as on their own unique costs of delivering health insurance to their workforce. For example, higher-wage workers are likely to be willing to pay more for health insurance in the form of reduced wages, and so employers of highly productive high-wage workers are more likely to offer than are employers who can get by with mostly lower-wage workers. This effect is amplified by our current tax subsidy for premiums nominally paid by the employer, a subsidy that works out to be roughly proportional to the marginal income tax rate of the worker. It is also amplified for large firm employers of high wage workers, since they have the lowest costs of providing health insurance, for they can take advantage of various economies of scale.
But worker heterogeneity also means that local labor market conditions can significantly affect offer rates, since firms offer only when they must to compete for the workers they want, and we do observe offer rates differ by as much as 20 percentage points across the United States. This variation in offer rates also affects ultimate coverage rates, of course. Differential offer rates and employer-sponsored insurance (ESI) coverage rates also affect the contours of the coverage problem faced by policy makers. For example, states with high offer rates find it cheaper and easier to be more generous with Medicaid and SCHIP eligibility—Minnesota and Wisconsin come to mind—than do states with very low employer offer rates, like Arkansas and Mississippi.
Myth #8: Workers used to be afraid to switch jobs because of health insurance, and HIPAA fixed that. "Job lock" is the shorthand term economists applied to the phenomenon of workers remaining with less productive jobs than they could get because they fear losing health insurance if they were to switch. This was originally investigated with some vigor in the early 1990s during the debates over the Clinton Health Security Act, for it was argued that if the aggregate amount of lost productivity was large enough, there could be a very large hitherto uncounted gain to universal coverage, and thus the net cost to society might be much lower than simple budgetary cost estimates.
Since then, much research was done, and HIPAA was passed, which among other things, was designed to make the portability of insurance more real and reduce job lock. Jonathan Gruber of MIT and Bridget Madrian of the University of Pennsylvania reviewed the complex research evidence and concluded that the studies with the most defensible methods do indeed find some pre-HIPAA job-lock, though the welfare cost from this job lock is essentially impossible to quantify. This means economists cannot tell, at the moment, if additional policy interventions are justified.
Gruber and Madrian also highlight two broad reasons to believe that many workers are still reluctant to switch jobs for health insurance-related reasons, even after HIPAA: They stem from Myth # 3, coverage is coverage is coverage. First, workers could have more generous coverage on their current job than HIPAA requires, in terms of pre-existing condition waiting periods, actuarial value or access to preferred providers. Second, insurance in the individual market costs more per dollar of coverage, so that higher wages—exactly equal to what the previous employer "paid" toward health insurance, for example—may not be able to make one whole. Thus, workers are often reluctant to leave a job with health insurance for a job that might pay higher wages but does not have health insurance attached. The cost advantages of group purchase are large.
Myth #9: Economists don’t know anything about why people are uninsured. Sometimes it seems that a normal person might listen to economists argue among themselves or read a whole book devoted to methodological flaws in prior work and reasonably conclude that economists actually think we know exactly nothing, that nothing has been satisfactorily proved, and we therefore need millions of dollars and years more to study and argue before we will be able to say anything at all that is useful to policymakers. This is not the case, and this idea is so important, I will devote the last two "myths" to embellishing the point. There are three things I think most economists actually do believe about the lack of insurance coverage. And I think the chapter by Linda Blumberg and myself make these fairly clear, even, and maybe especially, to non-economists.
Myth #10: The combined research evidence supports doing nothing to address the problems of the uninsured today. Economists and health policy analysts cannot tell you—as a scientific matter—that you should implement new subsidies and other policies designed to reduce the number of the uninsured. We can—when we’re at our best—articulate and help you see the tradeoffs involved, but only you who have been entrusted with the power of our people can decide if the opportunity cost is worth it, i.e., which competing priorities will and should get less attention and fewer resources. For let there be no doubt, if you really want to make a serious dent in the uninsured problem, you’re going to have to be willing to claim and redirect a considerable amount of public resources.
But at the same time, a politically neutral observer might reasonably conclude, from the decades we have been discussing this issue as a nation even while the number and percentage of uninsured keeps trending upward, that the case for doing something substantial about the uninsured must be widely perceived to be weak. I believe this is the wrong conclusion to draw from the evidence I’ve reported on today, as well as from the empirical work my colleagues at HSC and others around the nation have done these last few years.5 Perhaps the best evidence of the value of health insurance is not in statistics or econometrics, however, but rather lies in the fact that all the health policy analysts I know—and I know quite a few around the country—actively seek out and keep health insurance at all times, even when self-employed, and they even buy it for their recalcitrant adult children when the latter emerge from college feeling immortal but also stunned at the rental price of nice apartments in our great cities these days.
The choice is less funny for two working parents who make say $7.50 an hour each—that’s more than $2 above the minimum wage—and if they work full time as most do, they therefore earn $30,000 a year. Their children would in most but not all states be eligible for SCHIP, but you can know they would not likely be offered health insurance at their jobs, and they make far more than Medicaid income cutoffs in the vast majority of states in our country. They are also not very likely to feel like they can afford to spend a third or more of their gross income on family health insurance in the non-group market. To add one final touch of realism, you may assume they are healthy today.
Are we willing to require them to obtain health insurance? If they do get sick, they will most likely access health resources that will impose costs on the rest of us in various ways, and a requirement to purchase then would be responsive to the so called "free rider" justification for universal coverage. But of course they cannot afford it, so we would also have to subsidize their purchase of it, or impose an inequitable burden upon them. At the same time, they are healthy now, so the nation would be partially buying for them true insurance with no necessary immediate health benefit, that is, we would be buying protection from risk, a risk of potentially devastating financial, emotional and health consequences of unforeseen health problems which could strike any of us this very afternoon. The question comes down to, are we willing as a society to pay to protect these parents from living with this risk that we all pay to avoid for ourselves, and to protect us all from living with their free-rider risk? These are the ultimate questions that only you and your colleagues can answer.
I devoutly wish it were otherwise, but we economists cannot tell you with certainty the best particular way to expand health insurance coverage,6 but I can say the case for some kind of significant coverage expansion seems strong to many health economists and health policy researchers today. The prudent strategy in the event you do move in that direction would be to monitor the outcomes quite closely and be prepared to alter details of the program or change course altogether if credible evidence warrants it. We at the Center for Studying Health System Change and in the economics and health services research professions more generally will undertake to try and keep you well informed.
I would now be glad to answer any questions my testimony today might have provoked.
1 I am grateful to Paul Ginsburg, Mark Pauly, Hanns Kuttner, Mary Harrington, Chrissie Julianno, and Catherine McLaughlin for comments on a previous draft. I remain solely responsible for all opinions expressed herein and any remaining errors or omissions. Contact: lnichols@hschange.org.
2 Actuarial value can be thought of as the percentage of expected health-related costs for an average risk person that the policy is designed to cover. It is thus a measure of generosity of a health insurance policy.
3 http://qualitytools.ahrq.gov/disparitiesReport/download_report.aspx
4 Hargraves, L. and J. Hadley. "The Contribution of Insurance Coverage and Community Resources to reducing Racial and Ethnic Disparities in Access to Health Care," Health Services Research 38:3 (June 2003).
5 B. Strunk and P. Cunningham. "Treading Water: Americans’ Access to Needed Medical Care," Tracking Report No. 1. Center for Studying Health System Change. March 2002. http://www.hschange.org/CONTENT/421; Care Without Coverage: Too Little, Too Late. Institute of Medicine, National Academy Press, May 2002; J. Hadley. "Sicker and Poorer—The Consequences of Being Uninsured: A Review of the Research on the Relationship between Health Insurance, Medical Care Use, Health, Work, and Income," Medical Care Research and Review Supplement to Vol. 60, No. 2 (June 2003).
6 For a range of coverage proposals developed by thinkers with many different perspectives, see the Covering America web page at www.esresearch.org. This Robert Wood Johnson Foundation project was directed by Jack Meyer of the Economic and Social Research Institute.