Center for Studying Health System Change

Providing Insights that Contribute to Better Health Policy


Insurance Coverage & Costs Costs The Uninsured Private Coverage Employer Sponsored Individual Public Coverage Medicare Medicaid and SCHIP Access to Care Quality & Care Delivery Health Care Markets Issue Briefs Data Bulletins Research Briefs Policy Analyses Community Reports Journal Articles Other Publications Surveys Site Visits Design and Methods Data Files

Printable Version Session One - Overview Panel: The State of the Debate

Moderator: Paul Ginsburg

Panel Members:
Mark Pauly, University of Pennsylvania
Jack Hadley, The Urban Institute and HSC
Rob Cunningham, Health Affairs

Thomas R. Hefty, Blue Cross/Blue Shield United of Wisconsin
John Bertko, Humana

Mark Pauly: Thank you, Paul, and I’m delighted to be here as well. The individual or non-group insurance market, I was first asked to look at it by the organization formerly known as the Alpha Center about four years ago, to look at that market, and among other things, it proves something I always tell my graduate students, that almost anything is interesting if you look at it carefully enough.

Cosmically speaking, of course, the number of people insured--I don’t know this for a fact, but I’m speculating--in the individual insurance market is probably about equal to the number of people who buy pet insurance. But as Paul has already suggested, not only is it interesting in and of itself, but because the possibility of offering individuals tax credits to help them buy insurance in private markets, among which individual market would probably be prominent, because that’s now--I don’t know about traction, but at least it’s made it onto the political radar screen. That’s what we’re here.

And, again, in the list of your mission, should you choose to accept it, Len Nichols and I were both asked by John Iglehart first to see whether we could agree, and then if we could agree, to write a paper that would give the background for discussion of the subject of this conference, the individual insurance market, also known as the sacrificial victim. So what I plan to do here in the new few minutes is just give you the highlights of our paper.

Again, one message that I would convey--or two messages, actually. First, I found it a pleasure to work with Len on this, and we’ve agreed to disagree a lot over the years. In this case, we were able to agree to agree. And, second, it was also a pleasure to work with the wonderful community tracking survey data that was able to attach some numbers to things we always knew must be true. So you’ll see some of that.

Personally, as a researcher, as well as a person interested in policy, I think there are some surprising and also maybe some confirming things that came up in this exercise.

So here’s what I’m going to say quickly. I’ve already said a little bit about why the topic is controversial, and at least we tried to come up with what analysts could agree on about the individual insurance market. It’s probably too much to expect political types to agree. I’m always surprised at how much passion surrounds an issue like this. It seems to me to be just a fairly technical issue. But I suppose when the game is winning and losing elections, you attach passions to the strangest things.

Then, secondly, what analysts disagree about, well, we are going to talk about facts that analysts more or less agree about, but they disagree about facts and they disagree about values, and the hard part often is even for the authors to kind of separate those two. I will try as hard as I can to stick to facts until the end, when I’ll throw in my two cents’ worth about what I really think about individual insurance markets. And then there are policy disputes that follow from these markets.

Well, here is our quick summary of what we think the facts are, and I added, although it didn’t come out very well--it almost looks like a computer’s frown or something. But even the facts, it’ll be sort of "yes, but." And the most you can say about these facts is not that they’re absolutely true but that they’re very likely to be true. But at least these are the things we thought were reasonably uncontroversial.

First of all, just a description of the individual insurance market. I’ll show you a table on this in a moment and comment on it. But it’s small relative to the size of the health care sector, to even the total private sector, where the great bulk of Americans get their health insurance as part of their compensation for their job not in this market. But as Paul mentioned, some do get it in this market. But it would probably be used by subsidized uninsured, and it’s also shrinking and I’ll say a little bit about that.

The second factoid that I think there’s pretty much general agreement on is that, other things equal, if you buy the same insurance policy in the individual market as in the group market, much less in the public sector through Medicare or Medicaid, the portion of the premium that will have to go for administrative expenses, the main differential is not claims payment but it’s selling and billing expenses, is going to be higher for individual insurance. It’s a kind of custom-designed product, and it costs more than an off-the-rack suit.

The third is that although--I was going to say "yes, but." There are some people in some circumstances that may actually find a better deal in the individual insurance market than what at least somebody might guess they were being charged in the form of lower wages in the group insurance market, maybe; at least there’s arguments about that.

Thirdly, the risk of adverse selection is probably greater than in group insurance, especially where there’s regulated rating, whether it’s community rating, which requires insurers to ignore information about differences in risk levels that they can observe. Whether or not the problem is so serious if insurers were allowed to use all the information at hand is, I think, more controversial, but at least where they’re not--which happens fairly frequently--that’s a problem.

Our judgment is that the reforms, which generally have taken the form of reducing the extent of risk rating and the extent to which insurers have discretion over who they will insure in the individual market, has almost surely not increased the total number of people with insurance coverage. So if the ball that you want to keep your eye on is how do we get more Americans to have health insurance, you can do a lot of good things potentially with reform of the individual insurance market, but so far nobody has figured out how to hit that ball with this particular bat.

What you may be able to do, although even this turns out to be controversial, is given a total number of uninsured people with rating reforms like community rating, you may be able to move out of the uninsured, the high-risk people, and into the uninsured, the low-risk people, but we’re not even sure about that.

And then, of course, this is the fundamental message from economics, if only we needed to look at it again. All policy interventions produce tradeoffs. There’s no intrinsic good or evil in the individual insurance market. It’s just a matter of what tradeoffs you like and what values you place.

So here’s the kind of data that I mentioned, and I draw your attention to the last column. I think that’s the most interesting set of numbers. There was a decline in the percentage of--in the number of people enrolled in the individual insurance market between 1996 and 2001 of 11.5 percent. Of that, about 80 percent was attributable to the fact that the number of potential candidates to buy insurance in this market fell, so that fell by 8.6 percent, in part because of growth in employment-based insurance and in part because of growth in CHIP and Medicaid. But there was about 20 percent of it that was due to the lower likelihood of taking this insurance, given that you were stuck in that market. And so not exactly covered with glory in terms of performance in the last half-decade or so.


ere’s what analysts disagree about. That was just an illustration of factoid number one. And I think number one on this list is the most controversial. In the individual insurance market, some people face tough choices, and we hope you kind of know what that means.

I’ll say a little bit about it. But people actually do have quite different opinions of how tough is tough, and they also have quite different opinions, even if they agree on it, of how many people face those tough choices. So we tried to bring some information to bear, and actually I think this is helpful.

We at least were able to explain why it’s confusing because it kind of depends on what measure you use of tough, and maybe say something--and here we’re winging it a bit and out on a limb--about the answer to that question.

Then another thing that analysts disagree about, you get less discussion about this, I think, in the general public policy area, although you do among some of my friends, which is those people who attach value to imagining that people can get the insurance policy they want, not the policy that the boss’ benefits manager, who’s probably his son-in-law, anyway, picked for them, attach value to the individual insurance market. It, of course, doesn’t give you as many choices as you’d have if you were a federal employee, but compared to working at Gus & Otto’s Garage, even if Gus and Otto offer you group insurance, you’re going to have more choices in the individual insurance markets.

Then there do exist high-risk pools to pick up some of those people who have tough choices, and we comment on their effectiveness, or at least the debate about their effectiveness.

And then, as I’ve already mentioned, there’s controversy about both facts and values in terms of evaluating the effects of regulation of market performance.

So the number facing tough choices is tough to count, and we kind of--at least this is what you might be interested in counting. The truly uninsurable, we use the folk literature number of 1 percent of the population. I think that was quoted in testimony to Congress about 30 years ago, and it has never been updated. But we do bring to bear some data from a large insurer on what proportion of applicants they turn down, and it was about 4 percent. And if you assume that some of those people by scrambling would have got coverage, in the range of 1 to 3 percent is probably about where you would end up, my guess.

Then there are people who are insurable but they have important exclusions. If you have hay fever and you write that down on your application form or your doctor snitches on you, the insurer will probably not cover that. That can still mean that insurance is valuable to you for all the other things that have yet to happen, but it does exclude things.

But the most serious issue, probably, of tough-choice people are the people who are insurable but at very high prices, either an absolute sense or relative to income. And we observe people who don’t--and I need to say, and I think we show in the data, that there are people who don’t buy insurance even though they can afford it, and there are some people who buy insurance that it doesn’t look like they can afford. So it’s not just a matter of affordability, but it’s certainly a matter of consternation and confusion.

And then the main--one of the main facts we’ve found is that how bad you think things actually are for people who are high risk depends on how you measure high risk. So that’s what I’m going to talk about now. This is sort of the main research thing. Just give me a few minutes to talk about it because it’s exciting to researchers.

It turns out that what you would conclude about how easy it is, at least in terms of revealed behavior, not to mention all of the sleepless nights and the telephone calls and so forth, but in terms of what people finally are able to do, is it easier or harder for high-risk people to get individual insurance than low-risk people to get individual insurance? It’s hard for both groups to get individual insurance because it’s expensive insurance because of that high administrative cost. But it’s differentially difficult. It depends on how you measure health risk, and there we have three candidates there: self-reported health status, presence of chronic conditions, and age. And we say a little bit in the paper about individual versus family concept, but I won’t talk about that here.

I hope this is semi-legible to people in the back. What is worth paying attention to is the differentiation in the last two columns which talk about the non-group market--the first two columns talk about the group market--between the percentage of people, given income as measured by the band around poverty line, who would be candidates for non-group insurance who actually get non-group insurance. So running all the way down to the bottom, the Evil Knievel’s of health insurance there, the people who have incomes greater than 4 percent of the poverty line, but they’re still about--it depends on how you count it--about a fifth of the uninsured.

You can see that the percentage of people with a chronic condition at 60 percent who get this insurance is actually higher than the percentage of people in households where somebody has no chronic condition, which is about 40 percent. And you see the same qualitative difference, although, of course, the percentage falls as income falls. So at least if chronic conditions you measure, the people with chronic conditions actually are more likely to end up with this insurance, whatever else is going on, than the people without chronic conditions in their household. And this is a summary of that.

The ESI take-up, the employment-based insurance take-up is clearly higher than the individual take-up. At least personally I think that. This is an opinion. But I think that’s because of lower loading, but there could be other things going on, including corporate socialism and other favorite topics of Uwe Reinhardt.

Income is clearly important as well. No matter what, the higher your income, the more likely you are to have insurance.

People with chronic conditions, we think, would probably expect to see riders and higher premiums in the individual insurance market; nevertheless, they are more likely to end up with insurance than people in households without chronic conditions. And I guess that’s the punch line down there at the bottom.

Now, this shows, though--that’s one set of facts. You want a different set of facts to testify for the other side? Well, those exist, too, and it depends if you measure health status by this excellent, very good, good, fair, poor. Again, drawing your attention to the final two columns, you can see--I’ll just pick the rich uninsured: 33 percent of them, if they are in poor or fair health, only one-third of them end up with non-group insurance; whereas, if they’re in excellent on down to good health, more than half of them end up with non-group insurance. And, again, you get the same differentiation as income falls.

So it all depends on how you measure things, and this is in the nature of you open one door--you close one door, you open another door. What is the puzzle here? The puzzle is in some ways to reconcile the two findings that I just presented. How can we explain why people in poor health are less likely to have health insurance than people in good health? Well, one possibility is, of course, that insurers discriminate based on that. But then why didn’t they discriminate at least effectively based on chronic conditions?

The other possibility is--which is, actually, I think, the reason why we care about the uninsured. Not having insurance makes you sick, so maybe the causation goes the other way.

This looks at age, and unequivocally, age is an indicator of risk. And, again, you get the phenomenon closer to the chronic condition story, that holding income constant, the older people are, the higher risk they are, the more they’d have to pay in the individual market for sure, the more likely they are to buy insurance in that market. So you can see the conclusion there, which basically says what I just said.

We also did a multivariate model, and the message here is the computer told us what we could see in tables, which must mean it’s true. And so our overall conclusion about health status in the non-group insurance market is that the non-group market works passably well, even for high risks--I’ll take about one more minute here, Paul, if I can--particularly for--this is where we’re most out on a limb, so I’m sorry I have to talk so fast--particularly for the 40 percent of households who are not income constrained. And we hazard the guess that perhaps 80 percent or a lot of non-group households would have access to what we call actuarially acceptable premiums.

There’s the 40 percent who can afford it, and then there is 42 percent, about two-thirds of the remaining 60 percent, who would at least be charged premiums which, from the point of view of insurance, is about the best the insurer can do. And our data on applicants to the unnamed company found that 78 percent of them were offered insurance at premiums that were at most 25 percent more than the standard rate. This was my editorial comment, not Len’s. I think it’s government failure here. We fail to help out low-income people, not necessarily market failure. So that’s probably the most controversial conclusion, and I’ve put quotations around "works" because that’s what everybody will fuss about.

Then here’s one last factoid that we thought was interesting. There may be another kind of market failure in the non-group market in that the average expenses of people who don’t buy insurance are lower than the average expenses of people who do buy insurance. Well, that’s no surprise. That’s moral hazard. But they’re so much lower when you look at the data that it’s hard to believe it could be moral hazard alone. So that there must be something about the people who remain uninsured, particularly strong if you look at the people who are listed in fair or poor health. Either they’re not in such bad health as they seem, or they’re very stoic when it comes to seeking care. But the problem then with the market is, if the insurer bases their premiums on the expenses of the people who do buy insurance, they’ll end up offering something that looks like a terrible deal to the people in good health. So although the belief, mistaken belief in immortality probably explains why a lot of uninsured are uninsured, it may be also that when they actually look at the facts, the real facts about themselves, it is a pretty bad deal.

Other major disagreements, I’ve already commented on the feasibility of tailoring packages. The effects of high-risk pool, I think on balance they do good things, but they don’t do enough good things, and the values of effects of market reforms. The remaining policy disputes, I’ll just quickly summarize here. How well, when push comes to shove, does the non-group market work now? Here’s my two cents’ worth: Not as bad as you might have expected, but not as good as God would want it to. But we may not be able to improve it very easily. And how relevant is that to a world where we might do something dramatic, can only hope to do something dramatic--something dramatic about the uninsured. And if it was to offer fairly generous tax credits, the hypothesis that we discuss--of course, can’t prove because we haven’t been there yet--it is that that would change the way the non-group market functions. Certainly it should change to increase the--or likely it would change to increase the generosity of policies. So I’ve objected to trying to draw inferences about what the non-group market might be like in the nirvana based on what it’s like now on that hypothesis. And probably--and I can say more about this if somebody wants to fuss about it. My prediction is that if you subsidize this insurance so it sells itself, it will cause insurers to have to spend less on selling the stuff themselves and less on screening out the high risks.

Would the group market fall apart if the non-group market became heavily subsidized? The short answer to that is it depends on how you subsidize the non-group market, but not obviously true. But that’s what people fuss about. And are there any other policy interventions that are better? Is this the best of all possible worlds or is this the best of all thinkable worlds?

I think those are my comments.


Paul Ginsburg: Jack Hadley?

Jack Hadley: Thank you, Paul, and Tad and Roland, for getting me up on the screen. I never could have done it myself, I know for sure.

The title of my presentation is "Health and the Cost of Non-Group Insurance." It’s based on work, on continuing work I’m doing with Jim Reschovsky at the Center for Health System Change.

One of the major concerns that people have about using tax credits as a policy option for expanding insurance coverage via the non-group market is that their value may be limited by certain structural characteristics of the non-group market. Mark touched on all of these in his presentation. Our presentation focuses on the first two: medical underwriting, primarily, and to a smaller extent, the issue of less generous benefits.

There are two basic policy questions that I’ll try to address in this presentation. One is: Will the tax credits be generous enough to make non-group insurance affordable for the target population, that is, uninsured people who don’t have access to either employer-sponsored insurance or public insurance? And, secondly, how many people would benefit from tax credits and what are the characteristics of those who don’t?

The main contribution of our analysis--and I hope it’s value-added--is that we adjust for the statistical problem of selection bias. And I’ll explain what I mean by that in just a moment. After making the adjustment, we project non-group premiums to the target population in our data set, and then assign them tax credit values using two well-known tax credit proposals. And then we’ll compare the proposals in terms of the amount of subsidy they provide, their implications for affordability, that is, the percentage of income that people would have to pay, both before and after a tax credit. And at the end I’ll show you some information comparing the benefits of non-group insurance and employer-sponsored insurance.

This next slide gives you a very quick summary of the data we used from the community tracking study, and one of the things you’ll probably come away with at the end of the presentation is that, even though we use the same data that Mark and Len used, we don’t actually go the same way and come up with somewhat different results, although at the end of the day, I think we wind up pretty much in the same place.

So let me talk about selection bias. Basically it refers to the possibility that people covered by non-group insurance are not representative of the uninsured and that they’re healthier, possibly, in both observable and unobservable ways, and, therefore, it’s making inferences from the cost of non-group insurance based on what people with coverage--who actually have that coverage have to the uninsured is biased. We call it selection bias because it works through two mechanisms. One is the self-selection of healthier and richer people into the non-group market. The other is based on insurers’ offering and pricing decisions that possibly screen out sicker and poorer people.

With our data we can actually look at one element of this, the health characteristics, observable health characteristics of the people based on the kind of insurance coverage they have. And one difference you can see right off the bat between our analysis and Mark’s and Len’s is that we use bigger numbers. So I think you can probably see them a little more easily.

The first two rows measure adults’ self-reported health status. The bottom two rows measure a couple of family health characteristics. And I think it’s pretty clear as you compare non-group to employer-sponsored to uninsured that, in fact, people with non-group coverage really are--appear to be much healthier, both in terms of the proportion who say their health is fair or poor and the proportion who have a child in fair or poor health. So without going into any details about our statistical analysis, let me just say that what we do is estimate a relationship between premiums charged for the non-group policies in our data set as a function of personal and family health characteristics, also controlling for age and family structure, costs of medical care in the community, several measures of state insurance regulation, and the presence of high-risk pools.

To adjust for the effects of selection bias, we use what’s known in the trade as a Heckman selection model, which is named after James Heckman, who won the Nobel Prize in economics a couple of years ago. So it’s a method that has some pedigree, although there are people who still think it’s all hocus-pocus.

This slide summarizes the main results of our analysis, and let me explain the columns first, and then I’ll talk about the rows.

Each of the columns reports percentage differences in premiums predicted by our model. The first column incorporates the Heckman selection adjustment. The second column, the one labeled "unadjusted," does not.

Now, for the rows, the first five rows measure the health of the policy holder relative to a person who’s in excellent health with no chronic conditions, the reference group. And then each subsequent entry shows you--predicts how much higher the premium would be as the policy holder’s health deteriorates.

The last three rows measure family health characteristics, and there the entries indicate, again, how much higher or lower the premium would be if you cover a spouse who has chronic conditions, if one of the people covered is a female of child-bearing age, or if you cover a child who’s in fair or poor health.

Let me start with the unadjusted column. The main story there is that the numbers are all small. Some are negative. None are significantly different from zero. So the implication from the unadjusted approach is that there is little or no medical underwriting in the non-group market.

However, if you use the selection-adjusted estimates, they suggest a very different story, namely, that as health deteriorates, premiums do show a pretty sharp upward increase. In fact, for a policy holder who’s in very good or good health but has two or more chronic conditions or someone who’s in fair or poor health, our model estimates the premium would be about 40 percent higher.

Now, for those of you who are number-challenged and like to look at pictures instead, this slide shows essentially the same thing. It plots the premium that we predict for a family of four, varying only the health of the policy holder and the age of the adults. And what you see is that the two unadjusted lines are essentially flat or show no clear relationship between premium and health status, while the ones that use the Heckman selection adjustment, or selection correction, the green and the red lines, show a clear upward increase in premium.

So what we do next then is take the selection-adjusted premiums, and we project them onto the uninsured families that are in our data set, and then we use the parameters of two tax credit proposals to assign tax credits to the people as well. And the two proposals we use are the one that was put forward by the Bush administration and the Senate bipartisan REACH proposal.

What we find is that both proposals would provide credits to a large number of people in the target population. The REACH proposal would provide some credit for about 95 percent of the population, the Bush proposal about 85 percent of the population. The REACH proposal is more generous, but it’s also more costly, and you can, you know, sort of do what you want with that, whether you like it or not.

The average subsidy, that is, the percentage of the premium that will be covered by the credit, is about almost 70 percent under the REACH proposal and about 46 percent under the Bush proposal.

This next slide shows the distribution of people by the amount of subsidy that they would receive, and, again, under both proposals substantial proportions of people would receive large subsidies of 50 percent or more, 46 percent under the Bush proposal, 73 percent under the REACH proposal.

Well, what about affordability? How much do people have to pay relative to their incomes? In this slide, we see that without a tax credit--and this may be one of the areas where we disagree a little bit with what Mark and Len conclude--a relatively small proportion of people, only about 30 percent, would pay less than 8 percent of their income, and about 45 percent would pay more than 16 percent of their income to buy non-group policies in the target population.

However, with the tax credits, those numbers change pretty dramatically, and under Bush, more than half, 55 percent, would fall into the, quote, affordable range, and under the REACH proposal, about two-thirds. So this is the glass is half-full or the glass is about two-thirds-full picture. What about the other 17 to 23 percent for whom non-group insurance would be still pretty expensive? And I’ll speed things up a little bit here. Here we look at the characteristics of the people who are tax credit winners and losers under the REACH proposal, which is the more generous of the two, and what we conclude is that those who don’t do well or don’t benefit very much from the tax credit are, in fact, much sicker, and from this slide they also tend to be much older and poorer.

The right-hand column indicates that almost 70 percent of those who find policies unaffordable are in the--have incomes below poverty.

I said I’d tell you a little bit about benefits. When you look at the cost-sharing structure of non-group policies compared to employer-sponsored policies, what we see is that, in fact, the structures look pretty similar, that most policies have copays of less than $20; most have coinsurance of less than 20 percent; and the great majority have deductibles of less than $500.

However, this information doesn’t tell you anything about exclusions or limits or services that are not covered. An indirect way of looking at that is to take a look at people’s out-of-pocket expenses, depending on the kind of insurance they have. And we do that here focusing on single-person families and comparing their out-of-pocket expenses over the year. And what we see is that if you have non-group insurance, your out-of-pocket spending is from one and a half to two times larger than if you have employer-sponsored insurance. And given that, not surprisingly, when we asked people directly how satisfied are you with the amount your plan pays relative to the benefits, we see that those with non-group insurance are much less satisfied.

So, to summarize, as I indicated, I think we actually do wind up at pretty much the same place that Mark and Len did, which is to say that tax credits have the potential to help many people, 46 to 73 percent receive large subsidies, 55 to two-thirds would find that the cost of insurance would fall to less than 8 percent of their income.

On the flip side, however, the tax credits based only on income don’t help everyone, and our estimate of those who, what we would say, face impossible choices is roughly 17 to 23 percent; tough choices, about an equal proportion, 16 to 22 percent; and those people tend to be sicker, older, and poorer. And it does appear to be the case that non-group benefits are less generous.

So, to sum up, what are the implications of this? I guess this is where the rest of the conference will be going today. I think no big surprises here. Varying credits with age would probably help since health is correlated with age. Increasing the size of the credit for poor people would help. Expanding eligibility for public insurance as a complementary policy would help. And possibly expanding subsidized high-risk pools could be an asset as well.

Thank you.


Rob Cunningham: Thank you. Good morning, everybody. Can you hear me? Is it off?

Paul Ginsburg: I think it’s on.

Rob Cunningham: This program and the package of papers that’s--no mike? What do we need?

Rob Cunningham: In today’s meeting and in the package of papers that Health Affairs is publishing today, there’s a high focus on a discrete subject, the non-group market, which is complex and difficult, and you’re going to be asked to really have a workout on looking at the details. This piece of the program is sort of a stretching exercise where you can stop doing that for a minute and prepare yourself, exhale for the next round of high-focus presentations. I want to talk a little bit about the contextual factors Mark mentioned, that there’s a lot of passion about the non-group market that, when you look at the non-group market, you say why do people get passionate about this.

What’s happened over here? Uh-oh. I’m denied access to my slides.

Rob Cunningham: Okay. I can’t hardly see it. There. I have to put on a different pair of glasses to do this. I can’t see the arrow.

Robert Cunningham: The passion comes sometimes from confusing contextual factors with the issues that arise and the difficulties that arise in the non-group market per se. And I think we’re better off if we can recognize and identify the contextual factors that sometimes put a load on our concerns about the questions of the non-group market. So I want to try to identify a few of them in a very non-researchy kind of way to help us be aware of the things that may make it more difficult than it has to be to understand what we want to get.

The first is that the non-group market is not the only arena in which tax-based subsidies have to operate. Let me summarize briefly, and we’ll also look at the non-group market is not necessarily being offered as a replacement for employment-sponsored insurance, although there are some voices that occasionally suggest that it is and that subsidies for the non-group market, insuring of the non-group market are a first step towards ultimately replacing the employment-based system, which we have a lot of discontents with. And I think sometimes that’s a reason that people get passionate about this a little bit prematurely.

The third point is the ideological conflict between public and private solutions to the uninsured, which always looms large but we may need to look at perhaps more creatively than we have and, again, not load up our deliberations about the problems of the non-group market and what to do about them with larger questions that in some ways are misapplied.

The Bush administration proposal essentially, which set the agenda to some extent, although it’s not live in Congress right now, makes the non-group the exclusive arena for subsidies or virtually the exclusive arena. The primary argument here, one is to foster choice, but the other is that it’s unfair to provide a subsidy to the employment-based market because it already enjoys a large subsidy as it is in its double-dipping, in effect.

A number of tax credit proposals address both the non-group and employment-sponsored insurance. The REACH proposal that Jack mentioned is one of them. The way they cope with the inequity and the double-dipping factor is by having a smaller subsidy--I think it’s 25 percent in the REACH proposal--that’s available to income-eligible workers to purchase at the workplace.

The Granger-Wynn proposal that came out this past summer has a similar notion, a lesser subsidy to employment-based coverage, so that you’re covering both.

These are administratively complex. The REACH proposal is huge. That’s partly because they don’t attempt to prevent crowd-out effects. They think that it’s legitimate to give these subsidies for non-group and group market and for people who already have coverage. It’s not designed exclusively for the uninsured. So it represents tax relief essentially for low- and moderate-income workers. But it is possible to use tax-based subsidies for employment-based coverage, and we know that the majority of the uninsured are in small employment. It’s not simple to visualize how these subsidies would flow. Employers who don’t offer would have to somehow realize and be moved by the fact that their employees had a little more money to deal with cost sharing if they had non-generous policies or something like that. But it would take a couple of steps, but shoring up the non-group market with subsidies is also not a slam-dunk, so shoring up the small-group market with subsidies is something that we needn’t rule out and we needn’t think of the non-group market as an option that excludes the possibility of tax-based subsidies for a small-group market.

Let me add that those proposals have considerable support, and there’s been a number of--well, the Chamber of Commerce, the National Association of Health Underwriters have expressed support for the Granger-Wynn proposal. Health Care Leadership Council also is a supporter. Stuart Butler, who’s a long-time advocate of tax credits, criticized the Bush proposal for not including the employment-based market. So these are not sort of left-field ideas.

The second point stretches a little bit further out of the immediate context. The notion that the non-group market is going to eventually--should eventually replace employment-based insurance is--everyone is free to argue it. There are, however, some sort of truisms that have been thrown around increasingly, I think, in the past few years that are misleading in this respect, and I want to address a couple of them because I happen to have come across this subject with some specialized knowledge of my own from studying the history of Blue Cross and Blue Shield and the origins of the health insurance system.

We have in particular the myth of the accidental system. We hear this a lot of the time, that employment-based insurance is the unintended consequence of wartime wage and price policies that led to the tax preference for employment-based coverage, and we wouldn’t really have this stuff if it hadn’t been for those policies. This is simply not correct, historically. I haven’t done exhaustive research on the subject. I’ve been getting a lot of calls about this since my paper on which this presentation is based appeared in the Health Affairs website a month or so ago.

In fact, the tax preference that came along during the war didn’t really get settled legally until, I think, in 1949, a Supreme Court decision, and then on into the ’50s some further legal clarification.

In the meantime, as of 1952, there was about 70 million people who had hospital coverage. This is about half the market at that point. It was about 50/50 between Blue Cross and commercial insurers, and at the Senate Labor and Human Welfare Committee--I think it was called them--hearing on this subject, Blue Cross and commercial insurance representatives estimated that employers contributed only 10 to 12 percent of workplace situations on the Blue Cross side, and the data was poorer on the commercial side, but it was a less than 20 percent number of employers where contributions were made, and these contributions were often small.

Initially, health insurance was sold on a check-off basis, retail to employees. They were given the option to take it. The paycheck was a collection point. The employer was not a payer in any sense and was not eligible to receive the tax benefits in all of the first-generation coverage up to the point--I mean, the market is made. Later on tax policy becomes more important in cementing it and completing the maturation of the market, but it’s not the reason we got it. The reason we got it is because it was a real good product and everybody wanted it, and we were in an expansion economy and it didn’t take a tax break to convince an employee that it was worth paying $15 a month to make sure you didn’t get bankrupted by a hospitalization. So this was a good product per se, and that’s why it sold.

There’s a couple of other things that we tend to say when we say that the employer-sponsored insurance is on the verge of death, that it’s declining, that it doesn’t work. I just want to point out that in the numbers that we published in a paper by John Olihan (ph) several months ago, the data showed that there was a growth of, I think, about 16 million in the number of people covered in the workplace between 1994 and 2000. This is not a dying system.

I also want to point out we have the picture of the Mom and Pop place of employment as a non-viable setting for administering and managing insurance. Forty percent of the uninsured are in workplaces of 100 or more; 60 percent in workplaces of 25 or more. So these are not necessarily very small employers where a lot of people are.

One more point, which is the erosion in employer-sponsored coverage. There’s a lot on the retiree side. If you imagine what a Medicare drug benefit would do to retiree health costs, you could see stemming erosion at that point.

Last, just on the subject of Medicaid, we had a couple of years ago a bunch of strange bedfellow events around Washington where the idea was that a bipartisan initiative on the uninsured would be achieved by you scratch my back, I’ll scratch yours, I won’t oppose your tax credit proposal if you won’t oppose my Medicaid expansion. This was based on Stuart Altman’s principle that everybody’s second--the status quo is everybody’s second choice.

The idea of significant expansions of Medicaid went out the window with the tax cut and the declining economy. We can’t really make that deal anymore. I think it was kind of simplistic to begin with. But if we think about the pressures that exist on Medicaid now--and I’m trying to get at the ideological polarization around public and private programs--any way of providing relief to Medicaid stressed in multiple directions by long-term care, mental health, et cetera, in addition to the two million increase that’s associated with the recession, any kind of relief that we can provide for that program to prevent cutbacks in it for those who are interested in shoring up the public programs is going to be helpful to Medicaid. So if we can reduce the attrition in employment-based insurance or in the non-group market, if we can reduce the number of uninsured with these kinds of subsidies, then Medicaid benefits, essentially, and can continue to be strengthened. And I think that may be a more creative and more organic way to think about it.

Mental health parity would do the same thing. It would destress--a Medicare drug benefit would do the same thing. It would destress Medicaid, help stabilize Medicaid.

This is a schematic. There are no real numbers. But if you imagine the Medicaid population at the bottom of the income distribution, employment-based or non-group-based, private insurance at the time, the uninsured are in the middle. The idea is both to push up on the bottom and pull down on the top and narrow that gap, and then a lot of further solutions become more achievable. So now you have to go back to dealing with the non-group market.

That’s the end of the stretching exercise.


Paul Ginsburg: We’ll hear from Tom Hefty. Will you turn his microphone on?

Tom Hefty: I think we’re on. Thank you. And I’m going to give you a perspective, or at least one perspective from the private sector, more importantly, a perspective from the Midwest, and particularly Wisconsin.

Wisconsin is a laboratory of social experiments. We train Secretaries of Health and Human Services.


Tom Hefty: And we’re working on the next one. And in terms of the insurance marketplace, Wisconsin has a generous Medicaid program and SCHIP expansion that covers relatively few people, only about 10 percent of the population versus 15 percent nationally. But we have the lowest uninsured rate, again, in the latest census in the country in an unregulated private market. We have standard group regulations and an individual market that is basically unregulated in terms of rates and guaranteed coverage. And that gets us an uninsured rate one-half the national average and very high levels of satisfaction in terms of consumers. In the latest study, three of the four top-rated HMOs in the country are located in Wisconsin, which may explain why we train Secretaries along the way.

The question of what do we agree or disagree on in this, it’s a question I asked a couple years ago. What’s the price elasticity of demand for health insurance and how does it vary by segment? Because you really have two questions: What’s the supply side? How do you get a competitive supply in the market, under what terms? And then what’s the price elasticity of demand?

I think from my perspective in the Midwest, much of this debate ignores the substantial regional variations in market success. The Midwest in general has very high rates of insurance. It has regulation that encourages market solutions. It has low rates of Medicaid and, I would add, low rates of Medicare reimbursement. And yet we do well in the private market.

In ten states in the Midwest, you have more private purchased insurance than you have Medicaid or government programs. You could argue that that’s a tradition in terms of the agricultural economy, but I think it ignores what’s going on when you break the market down and that elasticity down by segment. You have young people. You have baby boomers and early retirees in particular have fallen out of the market, and then you have farmers and the current version of entrepreneurs falling out of the group market.

If you look at cities, you may say Iowa and Wisconsin are just unique and different, but if you look at San Francisco, if you look at a metro area, San Francisco actually has more individual coverage, individually purchased coverage than public-sponsored programs. So you have that change in the economy that’s driving the purchase of individual insurance, and much like the early ’90s, you have group rates going up, employers dropping coverage, and very rapid growth. And at least our experience is the individual market in the last two years has exploded. We are doubling sales every year in the individual market in a state--this is just Wisconsin--in a state that has very high rates of group insurance coverage. And so this is an important debate.

I’d also add that crowd-out is an issue. If you look at states with high rates of uninsured people, I would say look at your Medicaid program and then look at social expectations of consumers. Any family who thinks that they can get public health coverage if they really get sick ends up choosing not to pay for health insurance and then opting into the program in the emergency room. That’s the experience in Wisconsin, and when you look at the states with high rates of uninsured, you say it’s bad public policy, not a bad market.

I always enjoy talking at Washington policy debates because coming from Wisconsin, sort of the home of welfare reform, and as well as high rates of health insurance coverage, it’s always bright researchers discussing elegant solutions to simple problems. And health coverage is relatively simple. You need consumers to find a policy that matches their needs with intelligent market regulations so they’re treated fairly, and then work at subsidies for the groups where the private market does not work.

If you look at what’s going on in the market today, not only is individual coverage growing rapidly in the last two years, but the trend towards defined contribution is going to accelerate that. If you look at a 30 percent employee contribution, a 25-year-old can buy cheaper individual coverage on his or her own purchase decision and get the coverage that they want.

I’d also add that the Midwest uniquely has a number of associations and cooperatives that buy group insurance. About 10 percent of the farmers in Wisconsin get coverage through a farm or a dairy cooperative. That does work. And when you look at the statistical reports, in some states that shows up as group insurance; in other states it shows up as individual insurance. It’s why some of the statistics may not lead to good policy decisions.

In Iowa, for example, the Farm Bureau sells individual policies. In Wisconsin and other states, the Farm Bureau sells a group policy. And people respond to questions differently based on what they have, which is purely a historical accident in the marketplace.

I’d add there are other vehicles out there. If you think of the very young in terms of schools and the aging baby boomers, namely, the AARP, you have association plans out there that reduce the cost of administration. And that is another change that’s going on very rapidly. The ability to market direct and particularly using the Internet addresses what has been the legitimate issue in individual insurance, the high administrative and selling costs. Those are coming down rapidly.

We actually discount our sales on the Internet by 3 percent because it’s cheaper. Think of no-load mutual funds. And I’d tell you in at least our experience--and I ran our data through June of this year in terms of numbers--our individual insurance block averages an age of 43, and our group insurance block is an average age of 44. So it’s not exactly a major demographic shift. But in terms of product design and that elasticity of demand, our average single premium in the group market, employer-sponsored market, is $280. Our average premium in the individual purchase block is $172. That’s individuals making choices in terms of deductibles and coverage. That’s not high-pressure sales. We, in fact, do not use brokers in the individual market. We sell only direct or through the Internet.

And you say consumers are making intelligent choices, and I think when you look at the data by region and by segment, you come to a better conclusion. That’s probably where I’d disagree with some of the earlier speakers. If you average the country together, your data comes out in a fashion that is not very helpful in terms of public policy. Or put another way, when the third Wisconsinite sits in the Secretary’s office, we think some of the Midwest solutions on individual insurance may be helpful.

Thank you.


Paul Ginsburg: John?

John Bertko: Okay. I’m going to give a somewhat similar Midwest perspective, but based on where I live in California also. I’m a unique person that our company covers primarily that middle part of the country, in contrast to Tom’s company, which is essentially one state. We might be called super-regional, and we cover most of the Midwest, down through the Southeast, and Southwest.

So, first off, I’d like to say that the paper by Mark Pauly and Len Nichols is a really good contribution. I’ve also read the other papers, which are quite good. I’d like to maybe add a few more factoids and maybe, again, like Tom did, some modest corrections here from people who do the real work out in the boonies.

The first thing I’d like to say is that I think you can interpret the glass here of individual health insurance as perhaps being half-full. We’re a new player to the market. We have been offering individual insurance now for about six months. It’s been about a year and a half pick-up on it, and we see big opportunities. So although I think the data that you pulled upon to show the market was shrinking in the late ’90s during the boom period is true, the data, like most academic studies, is now obsolete. So, you know, Tom’s comment here that the market is growing is absolutely the case, and, in fact, the intersection of the small end of the small-group market and the individual market, at least in our perspective, shows that the individual will grow, and probably that small end of small-group is going to shrink.

To confirm what Tom has said here, if you looked at non-benefit-adjusted type of premiums and you saw some numbers in the $2,000-per-year range in several of the papers, or you would see them, that’s about the same cost on small-group, in that small-group end of the market, and in my quick assessment, for about the same types of benefits, the $1,000 deductible PPO type plans. And so those two markets are lying almost on top of each other.

Another comment that I’d like to make, though--and this was, I think, not stated in any of the papers--when you compare individual health insurance with at least small-group insurance, you’re comparing apples and oranges. They’re the same thing. They’re both fruit, but they’re two different products that you’re buying. Small-group insurance--and, in fact, any of the large-group type of products--are one-year-term products. That is, you’re buying a coverage for a year period. There is no extension. After that, premiums are rated up depending on what happens.

On the individual side, our terminology is that we have to not only pay for the term for that first period, but also active health reserves, because we are guaranteeing renewability here. The average period that we have a policy hold is about four to five years, and in addition to trim, people have their underwriting selection wear off. And so I agree with what several of the speakers have said in terms of the moment someone enters and passes through the underwriting screens, they are, of course, much healthier. That healthiness wears off, and we’ve got to actually incorporate that. So the premium really comes in two pieces. It’s a piece that’s term and it’s a piece that is, in other words, an option to buy coverage without underwriting for the next four or five, even ten years.

A couple of other comments here. One of the other papers noted that while there continue to be real and significant admin cost differences, the brokers in this area do play an important role. I did a little calculation that said if you had a broker instead of an HR department with a 50-employee group, it’s equivalent to about 3.5 percent of premium. And the broker serves the role of a half-time FTE. So on and off--I mean, those brokers really do serve reasonable purposes. There are diseconomies of scale, but like Tom has said here, many of us, and our company included, are attempting to push through what I’ll call structural reforms on commissions, on use of Internet, on real-time discussions with underwriting and fulfillment that will, in fact, reduce admin costs and thus make these things a little bit less expensive.

Another modest issue is just that the Pauly-Nichols paper has a factoid, I guess, about one company, and I would just tell you that that’s--I would confirm it on the basis of our experience. People do come in, about 50 percent of them, as I have said, would go through clean and get a policy issued. And those are people seeking insurance.

I would remark, though, that I believe--and I’m not sure--that this company probably had an agent distribution system. And what the agents do on that is they screen some of the folks in there that otherwise wouldn’t be able to pass. They just say don’t bother to apply. And I’ll tell you that because we have both an agent-based system and a direct system, and the people coming through the direct system, we get about 10 percent more people who are declined.

Which brings you to the point about how many people are truly uninsurable. Roughly the numbers there, perhaps 1 percent is correct on a population basis as a whole. Then you need to look at it for those seeking insurance. And people seek insurance because they need it. The number that you would then decline in this market is probably in that 10 to 15 to 20 percent range, depending on the channel that’s used.

I think, Paul, that’s probably the extent of my main comments.


Paul Ginsburg: Thank you, John.

Since the Pauly-Nichols paper is kind of the main inspiration for the whole issue, I’d like to give Mark a chance to react to some of the comments made.

Mark Pauly: Well, I just wanted to offer a few further thoughts on what I think is the most difficult tradeoff here, and it is this half-full or half-empty. The individual market works well for most people who use it. Whether it’s 80 percent or 70 percent or 60 percent kind of depends on what you mean by works well, but not for--but leads to high premiums and turndowns for some minority of applicants.

There are two things I think I wanted to say about that. First, I think the right way to think about it is in cost-effectiveness terms rather than absolute terms. Here’s what I mean: I think most people’s reaction is, well, if I was worrying about the uninsured, I’d rather have a sick poor person get insurance than have a not-sick, not-poor person get insurance. And everybody I think would feel that way. The issue, though, is that the size of the credit that you would need to offer, or subsidy or something, to get the sick poor person insured may be four or ten or some substantial multiple of what you would have to pay to get the non-sick, non-poor person insured.

And as the father of three 20-somethings who are not sick and not poor but it’s an eternal struggle to keep them insured, I’m personally concerned about the consequences for their health, even if it’s good at the moment, of them becoming uninsured. And, actually, I think of the enormous funds being spent on research and rehash of data on what it means to be insured versus uninsured is neglectful in not paying attention to that population as far as I have seen so far.

The other comment I wanted to make has to do with affordability, and this is also by way of a commercial of a paper that I’ve written that will be an NBER working paper soon, so it won’t appear in Health Affairs, I guess. But what it says is there’s two definitions of affordability. The most common one is the one Jack talked about. Look at how high are these premiums relative to your income. It’s with Kate Bundorf, I should say, so she could take some of the blame.

The other definition, though, is what we called a behavioral definition, and the idea is you would say that a person can afford something if most people like them already buy it. And if you use that definition, you get a difference of, for example, in particular, the effect of age. We give a higher credit as people get older. As I get older, I think that’s a good idea. But from a policy point of view, as we’ve shown, although it’s somewhat tempered by low income, older people actually do end up, other things equal, more likely to get insurance. So maybe you don’t need to give them a higher credit if you used it on the basis of their behavior rather than on the basis of the alternative.

Again, I’m not trying to be doctrinaire about this. I’m in no position to do so. But I think that keeping in mind that there are these tradeoffs between how you evaluate a lot of not-sick-at-the-moment people moving from uninsured to insured compared to a few sick-at-the-moment people, and how you think about these two alternative meanings of affordability I think can--well, I don’t know if they help to clarify, but they make judgment more discriminating among various policy options.

Paul Ginsburg: Thank you, Mark.

Please join me in thanking the first panel that got us off to a great start.


Previous Section
Next Section


Back to Top
Site Last Updated: 9/15/2014             Privacy Policy
The Center for Studying Health System Change Ceased operation on Dec. 31, 2013.