[Joint House and Senate Hearing, 108 Congress]
[From the U.S. Government Publishing Office]



                                                        S. Hrg. 108-766

                    ``EXPANDING CONSUMER CHOICE AND
                ADDRESSING `ADVERSE SELECTION' CONCERNS
                         IN HEALTH INSURANCE''

=======================================================================

                                HEARING

                               before the

                        JOINT ECONOMIC COMMITTEE
                     CONGRESS OF THE UNITED STATES

                      ONE HUNDRED EIGHTH CONGRESS

                             SECOND SESSION

                               __________

                           SEPTEMBER 22, 2004

                               __________

          Printed for the use of the Joint Economic Committee



                    U.S. GOVERNMENT PRINTING OFFICE
97-228                      WASHINGTON : 2005
_____________________________________________________________________________
For Sale by the Superintendent of Documents, U.S. Government Printing Office
Internet: bookstore.gpo.gov  Phone: toll free (866) 512-1800; (202) 512ï¿½091800  
Fax: (202) 512ï¿½092250 Mail: Stop SSOP, Washington, DC 20402ï¿½090001

                        JOINT ECONOMIC COMMITTEE

    [Created pursuant to Sec. 5(a) of Public Law 304, 79th Congress]

SENATE                               HOUSE OF REPRESENTATIVES
Robert F. Bennett, Utah, Chairman    Jim Saxton, New Jersey, Vice 
Sam Brownback, Kansas                    Chairman
Jeff Sessions, Alabama               Paul Ryan, Wisconsin
John Sununu, New Hampshire           Jennifer Dunn, Washington
Lamar Alexander, Tennessee           Phil English, Pennsylvania
Susan Collins, Maine                 Ron Paul, Texas
Jack Reed, Rhode Island              Pete Stark, California
Edward M. Kennedy, Massachusetts     Carolyn B. Maloney, New York
Paul S. Sarbanes, Maryland           Melvin L. Watt, North Carolina
Jeff Bingaman, New Mexico            Baron P. Hill, Indiana
                    Paul A. Yost, Executive Director
                Wendell Primus, Minority Staff Director


                            C O N T E N T S

                      Opening Statement of Members

Senator Robert F. Bennett, Chairman, U.S. Senator from Utah......     1
Representative Paul Ryan, U.S. Representative from Wisconsin.....     3
Senator Jack Reed, U.S. Senator from Rhode Island................    14

                               Witnesses

Statement of Mark V. Pauly, Ph.D., Professor of Health Care 
  Systems, Insurance and Risk Management, and Business and Public 
  Policy in the Wharton School, and Professor of Economics in the 
  College of Arts and Sciences, University of Pennsylvania, 
  Philadelphia, PA...............................................     5
Statement of James H. Cardon, Ph.D., Associate Professor of 
  Economics, Department of Economics, Brigham Young University, 
  Provo, Utah....................................................     8
Statement of Jeffrey M. Closs, President and CEO, BENU, Inc., San 
  Mateo, CA......................................................    10
Statement of Linda J. Blumberg, Ph.D., Senior Research Associate, 
  The Urban Institute, Washington, DC............................    12

                       Submissions for the Record

Prepared statement of Senator Robert F. Bennett..................    29
Prepared statement of Senator Jack Reed..........................    30
Prepared statement of Mark V. Pauly, Ph.D., Professor of Health 
  Care Systems, Insurance and Risk Management, and Business and 
  Public Policy in the Wharton School, and Professor of Economics 
  in the College of Arts and Sciences, University of Pennsylvania    30
Prepared statement of James H. Cardon, Ph.D., Associate Professor 
  of Economics, Department of Economics, Brigham Young University    39
Prepared statement of Jeffrey M. Closs, President and CEO, BENU, 
  Inc............................................................    42
Prepared statement of Linda J. Blumberg, Ph.D., Senior Research 
  Associate, The Urban Institute.................................    49

 
                    ``EXPANDING CONSUMER CHOICE AND
                     ADDRESSING `ADVERSE SELECTION'
                     CONCERNS IN HEALTH INSURANCE''

                              ----------                              


                     WEDNESDAY, SEPTEMBER 22, 2004

             Congress of the United States,
                          Joint Economic Committee,
                                                     Washington, DC
    The Committee met at 10 a.m. in room 628 of the Dirksen 
Senate Office Building, the Honorable Robert F. Bennett, 
chairman of the Joint Economic Committee, presiding.
    Senators present: Senators Bennett and Reed.
    Representatives present: Representative Ryan.
    Staff present: Tom Miller, Leah Uhlmann, Nancy Marano, Mike 
Ashton, Colleen Healy, Wendell Primus, John McInerney.

OPENING STATEMENT OF SENATOR ROBERT F. BENNETT, CHAIRMAN, U.S. 
                       SENATOR FROM UTAH

    Chairman Bennett. The Committee will come to order. I 
welcome you all to our hearing on Consumer Choice in Health 
Insurance.
    The whole issue of health care, what to do with it, how to 
pay for it, where to take it in the future, seems to be front 
and center in the Presidential debate, so I think it is 
appropriate that we continue our series of hearings on the 
whole health care issue.
    As those who have followed this Committee will know, we are 
not attempting to try to fashion any particular health care 
solution at the moment. What we're attempting to do is to lay 
out a record of the various challenges with respect to health 
care, with an attitude of a clean sheet of paper, that is, not 
what do we need to do to fix the present system, but what 
should we do, if we were starting from scratch, had no 
constraints, and could create an ideal system? What components 
should that ideal system really have?
    That's been the overarching attitude of this Committee 
since we started this series of hearings, and I think we are 
well prepared with a variety of witnesses today, to continue to 
look at it in that fashion.
    Now, many consumers would like to have greater choice and 
control of their health care and health insurance coverage. 
They know, from their experience with other types of goods and 
services, that choice and competition that matches their 
different tastes and preferences end up providing the best 
value and the greatest opportunity of choice.
    Well, recent efforts to increase consumer choice in health 
care have included: Providing multiple health plan options with 
employer-sponsored coverage, and that's the kind of thing we 
see in the Government program; offering new consumer-driven 
health care arrangements such as Health Savings Accounts--now, 
we're seeing some experimentation with that now--reforming 
Medicare financing to strengthen private plan alternatives. 
That was originally called Medicare Plus Choice, back in 1997, 
but the name was changed to Medicare Advantage in last year's 
Medicare Modernization Act.
    Then there is trying to level the playing field for those 
purchasers who select individual insurance market products, 
rather than employer group insurance coverage, trying to make 
that option less expensive.
    Well, some of these initiatives have advanced further than 
the others, and, in part, that's because various plans to 
increase consumer choice in health insurance often face 
criticism that they will trigger a host of purported dangers. 
The one we hear the most about is called adverse selection.
    That term tends to be loosely defined, but widely used, in 
health care debates. Its most accurate definition is when 
consumers know significant private information about their 
expected expenses that insurers do not know.
    The assumption is that insurance buyers who know that their 
risks are greater than average will want to purchase more 
insurance than that which is based on average risk. You know 
you're going to be sicker than the company knows, you're going 
to want to buy more coverage than the company otherwise would 
sell you, in order to take advantage of your increased 
knowledge. That's adverse selection.
    Buyers who expect their risk to be lower than average, will 
prefer less insurance coverage.
    Now, this simple description of adverse selection projects 
that insurance premiums for the original coverage offered will 
increase more than otherwise, because low risks either switch 
to other types of insurance, or, in the extreme, drop coverage 
altogether. The end result is presumed to trigger a death 
spiral of rising claims costs and fewer paying customers to 
finance them, under the initial insurance policy, and, in the 
worst case scenario, the death spiral extends to the overall 
health insurance market, which can break down completely.
    With today's hearing, we hope to examine that conclusion to 
see how valid it really is. It's important, because those who 
believe that more consumer choice in health insurance is just 
another risky scheme will attempt to handicap it, if not 
prohibit it altogether, through such policy measures as 
community ratings, standardized benefits, coverage mandates, 
and preferential subsidies.
    Today's hearing will examine whether or not employers and 
insurers can offer better choices to consumers in practice, 
without producing the sort of adverse consequences that I have 
just described in theory.
    We want to determine what really happens in insurance 
markets, in the pooling and pricing of risks, and sort out real 
problems from imagined ones. It appears that there are natural 
limits on the scope and scale of potential adverse selection 
problems.
    Employers and insurers seem to manage remaining ones rather 
effectively in most cases. Nevertheless, there may be policy 
opportunities to improve access to the care and coverage that 
consumers value the most.
    Now, health insurance coverage, of course, as we have tried 
to stress over and over through these hearings, is just a means 
to an end. The real objective should be better health. Better 
health means, ultimately, lower acute care costs.
    We also, of course, want better outcomes from medical 
treatment when the acute care is necessary, but improving the 
value of insurance that's available to a diverse population of 
consumers, is, of course, an important part of that process. 
Increasing choices, rather than reducing them, seems to be a 
fundamental starting point for upgrading the status quo.
    [The prepared statement of Senator Bennett appears in the 
Submissions for the Record on page 29]
    Chairman Bennett. Now, we usually limit opening statements 
to the Chairman and the Ranking Member. Mr. Stark is unable to 
be with us this morning, and Mr. Ryan has come over, and since 
we want to be ecumenical about Senate and House--this is, after 
all, a Joint Committee--why, we will allow Mr. Ryan to speak on 
behalf of the House in Mr. Stark's absence. That might send 
such a chill down Mr. Stark's spine that he'll rearrange his 
schedule to be here next time. I do not in any sense mean to 
criticize his absence. I know he has a very pressing conflict 
this morning, and we will miss him, because Mr. Stark adds a 
flavor to these hearings that is very valuable. Mr. Ryan, 
you're speaking for the People's House.

        OPENING STATEMENT OF REPRESENTATIVE PAUL RYAN, 
               U.S. REPRESENTATIVE FROM WISCONSIN

    Representative Ryan. And for Mr. Stark.
    [Laughter.]
    Chairman Bennett. I wouldn't go that far.
    Representative Ryan. I'll be brief, and thank you very 
much, Chairman, for your indulgence. I actually enjoy my other 
Committee--Ways and Means--with Mr. Stark, where we have had 
great conversations about this.
    I'll be very brief. This is a very timely hearing. The 
whole adverse selection issue is really coming back up because 
we now have Health Savings Accounts [HSA] that are coming out 
in the marketplace.
    We've had two amendments in the House, just in the past 
week, trying to prejudge this issue before any data comes out, 
as the Federal Employee Health Benefit Plan is now opening up 
its open season to a new HSA product. Those amendments were 
defeated, but, nevertheless, there are many who already want to 
make the conclusions that HSAs or consumer-directed plans 
encourage adverse selection.
    I would argue that it's just too soon to tell, but the 
early data we're seeing from various sources--the companies 
selling HSAs, the clearinghouse websites like e-
healthinsurance, is showing, from a preliminary data 
standpoint, that the opposite is happening.
    I would like to hear from the experts, what they are 
seeing, so, to me, this is a very timely hearing. I would argue 
that you can't make conclusions, now that we've just got new 
products out there that are just taking place in the 
marketplace. This is something that we need to watch very 
closely, the data, so that as each product is being sold, it is 
incorporated so that we can track this very well.
    I think the ability to track this data is much better than 
it was a few years ago, so while we see a fight here, 
politically or ideologically, on Capitol Hill, over this issue, 
I think it's reasonable to conclude that neither side knows 
what the answer is.
    That's why I think this is a very timely hearing. The end 
of the story is, each constituent of ours, whether they work 
for a big company, a small company, or are on their own, is 
probably facing double-digit health insurance cost increases, 
and answers are needed.
    We have some answers that are being deployed in the 
marketplace. There are more things we're proposing, so I think 
this is a very timely hearing, and I look forward to the 
witness testimony. Thank you for your indulgence, Mr. Chairman.
    Chairman Bennett. Thank you very much. I think we have a 
panel of witnesses across the spectrum of experience, 
background, and opinion, that can help us get a balanced view 
of this.
    We welcome Dr. Mark Pauly, who is one of the nation's 
leading health economists from the Wharton School at the 
University of Pennsylvania, who has written extensively on the 
operations of health insurance markets and how public policy 
can shape them.
    Dr. James Cardon of Brigham Young University, has examined 
whether consumers have private information advantages over 
insurers, information that could trigger adverse selection and 
distort health insurance markets. His most recent research 
focuses on the effects of the new consumer-driven health care 
choices like Health Savings Accounts.
    Jeffrey Closs is President of BENU, a company that provides 
benefits choice services to mid-size employers. BENU uses risk 
adjustment tools to encourage insurers to compete more 
vigorously for a portion of an employer's business and to 
provide more meaningful choices for covered employees.
    Then Linda Blumberg brings some direct governmental 
experience to the table. She is with The Urban Institute, but 
has been an advisor to both HHS and OMB, and has studied issues 
of risk selection and risk segmentation in voluntary insurance 
markets, particularly those involving small employers and 
individual consumers.
    I welcome all four of you and appreciate your willingness 
to come and share your expertise with us. We will hear from you 
in the order in which I have introduced you, which means Dr. 
Pauly, we start with you.

  STATEMENT OF MARK V. PAULY, PH.D., PROFESSOR OF HEALTH CARE 
             SYSTEMS, BUSINESS AND PUBLIC POLICY, 
 INSURANCE AND RISK MANAGEMENT, AND ECONOMICS, WHARTON SCHOOL, 
     UNIVERSITY OF PENNSYLVANIA, PHILADELPHIA, PENNSYLVANIA

    Dr. Pauly. I thank you, Mr. Chairman and Members of the 
Committee, for the opportunity to testify on adverse selection 
in health insurance, and related issues.
    The world in which we live is one in which health spending 
risk varies before the fact, in the sense that different 
consumers reasonably expect to collect different amounts in 
benefits from a given policy, because they expect in the 
future, to get sick with different frequencies and severities.
    Insurers can identify and measure some characteristics that 
they know predict above- or below-average benefits, 
characteristics such as age, location, and the presence of 
chronic conditions. Insurance markets can still function in 
such a world, but either premiums or purchases will be 
different for different people, and this makes life 
complicated.
    What will happen depends crucially on whether insurers have 
and can use the same information that predicts benefits as 
consumers can use. If everyone has the same information, and 
the information does predict different risk levels, then 
insurance theory tells us that insurers will choose to charge 
below-average premiums to the lower risks, and above-average 
premiums to the higher risks. Someone who has four times the 
expected benefit from a given policy, will be charged about 
four times the premium.
    At those premiums, insurers will be equally eager to sell 
to low and high risks. In insurance theory, this situation of 
proportional risk rating will be stable, and probably will be 
one in which low risks are no more or less likely to buy 
insurance than high risks.
    A few very high risks with low incomes may find that 
premiums are so high and expenses are so near certain, that 
they are just as well off not buying insurance and paying those 
expenses directly, when and if they can, but that's the 
exception.
    Insurance markets, the same theory tells us, will be very 
different if insurers do not have equal information that buyers 
have, or if insurers are not allowed to use the information 
they do have in setting premiums and bidding for business.
    In the extreme case in which insurers cannot distinguish 
among risks or are not permitted to do so, they will be forced 
to charge the same premium to everyone who buys insurance, but 
if insurance purchasers know their risk levels, their 
willingness to buy insurance at this premium will vary.
    Higher risks will be very enthusiastic about buying, since 
they can, on average, collect in benefits, more than they pay 
in premiums, but low risks may decide not to buy insurance at 
all, because it looks like a bad deal to them, or may at least 
seek to buy less generous coverage than the high risks desire.
    This situation of community rating will be one in which the 
low risks are less likely to buy insurance than under risk 
rating. In the limiting case in which the low risks bail out 
altogether, the so-called ``death spiral,'' the premium 
insurers will end up charging to the high risks who remain, 
will be the same ones they would have charged under risk 
rating, and the effect of community rating will only be to 
drive all of the low risks from the insurance market, with 
resulting adverse effects on their access to care and financial 
stability.
    It is in this sense that community rating can be 
inefficient, compared to risk rating, since it can make the low 
risks worse off and not make the high risks better off. In the 
less extreme case in which some low risks might continue to 
buy, the high risks could be better off, but the low risks will 
still be worse off than they would have been under risk rating.
    There will still be inefficiency, compared to the ideal, 
because the low risks will choose less coverage than they would 
have chosen, if they had faced premiums reflective of the true 
cost of their coverage.
    Whether there will be cream-skimming, in which insurers are 
more eager to sell to low risks than to high, depends on 
whether the adverse selection is essential in the sense of 
being caused by insurer inability to tell risk apart, or 
inessential, caused by regulation or policies which forbid 
insurers from using information they have, to set lower 
premiums for lower risks.
    In the case of essential adverse selection, as in the case 
of risk rating, there should be no cream-skimming, because all 
potential purchasers look equally profitable to insurers. 
Insurers might want to cover only the low risk, but they cannot 
tell who is who.
    In the less extreme case, the regulation-required community 
rating, insurers will try to avoid selling to high risks they 
can identify, on whom, as a group, they are sure to lose money, 
and there will be cream-skimming.
    For these kinds of reasons, some insurance analysts think 
risk rating is better than community rating, but many 
policymakers and some other analysts do not look at it this 
way. They do note the downside of community rating in terms of 
squeezing out the low risks, even to the extent of a death 
spiral, but policymakers also find much not to like in risk 
rating, precisely because the higher premiums for high risks, 
may bite into their ability to consume other necessities of 
life, if they have low income, and sometimes because observing 
higher income, high risk paying more than higher income, low 
risks, still looks unfair, especially compared to a 
policymaker's dream world in which everybody pays a low 
premium.
    That this is impossible in a world of competitive, but 
unsubsidized insurance, only margins dampens their ardor.
    The most obvious way to deal with these problems is to use 
regulation and to require insurers to charge similar premiums 
or limited premiums for high risk, and forbid low risks from 
buying less generous policies, then require insurers to sell 
policies to high risks they know will be causing losses, and 
when there is enough political nerve, forbid insurers and the 
low risks from dropping out, by mandating insurance purchasing.
    Measures short of this draconian one, can still lead to bad 
adverse selection type outcomes, especially when community 
rating forces insurers to ignore information they have. Then 
when insurers respond to community rating regulations with 
cream-skimming, one needs to write yet more regulations to 
require open enrollment and guaranteed issue.
    To avoid the death spiral, we move to a regulatory spiral. 
As with other kinds of health care regulation, how bad or good 
the regulatory outcome will be, seems to vary in practice 
across states, depending on characteristics of their potential 
insureds and the form and administration of the rules.
    In some states, such rules have seriously curtailed the 
size of insurance markets, while in others, the main effect is 
only discontent among the low risks and the insurers who would 
like to sell to them. Still, on balance, community rating seems 
to increase the number of uninsured.
    The main novel point I want to make here is that recent 
research suggests that, both in theory and in practice, there 
are ways alternative to regulation to get closer to what 
policymakers want, or should want, when risk rating and adverse 
selection are possible.
    The three kinds of solutions to which I want to draw your 
attention are: No. 1, guaranteed renewability at uniform 
premiums; No. 2, group insurance; No. 3, high risk pools. I'll 
speak most about the first.
    The great majority of people who are high risk today, were 
not sicker than average at all times in their lives. Data tells 
us what common sense and even our bones in the morning tell us, 
that even people who are in excellent health, have higher 
medical expenses, on average, as they age, and some pick up 
chronic conditions.
    The age effect on increasing risk is perfectly predictable. 
What is not predictable is the random onset of a chronic 
condition that makes a person high risk, not only initially, 
but for some time to come.
    Most medical expenses for people under 65 are not related 
to chronic conditions; they come from the bolt-from-the-blue 
event of an accident, a stroke, or something that cannot be 
predicted well in advance, and this is precisely the kind of 
low-probability, high-cost event for which insurance works 
well.
    But some events are predictable in advance, and then 
someone who contracts a chronic condition, gets a double 
financial insult. Not only will they have to pay a lot for 
their care in the year in which they are diagnosed, but they 
will probably--they may have to pay higher premiums in the 
future, and if insurance is perfectly risk rated, they are 
subject to the risk of becoming a high risk.
    A protection against that in the form of guaranteed 
renewability exists and was quite common, even in the absence 
of regulatory rules. Specifically, renewability at class-
average rates, requires insurers to charge the same premium to 
people, regardless of their experience or their risk, when that 
has changed after the sale of insurance. It basically means the 
insurers ignore new information about the level of risk, and 
this has the power to protect people against premium risk.
    This feature is not free, of course. Policies that contain 
it, must have high initial premiums, front-loading, than would 
premiums for which the insurer retained the right to increase 
premiums for people who contracted a chronic illness, but it is 
easy to see why rational, foresighted people, would prefer the 
slightly more expensive mature policy.
    Federal law now requires guaranteed renewability, but our 
research on data in a period when it was not required by 
Federal law and not required in many State laws, produces the 
result that there is very strong empirical evidence consistent 
with guaranteed renewability, in that the premiums charged in 
the individual insurance market, which is most subject to risk 
rating, are much higher for lower risks than for higher risks 
than would be consistent with proportional rates.
    Chairman Bennett. Dr. Pauly, can I ask you to summarize?
    Dr. Pauly. Yes.
    Chairman Bennett. This is fascinating stuff, and your whole 
statement will appear in the record.
    Dr. Pauly. Thank you.
    Group insurance can avoid adverse selection, fundamentally 
in two ways: The employer will limit the range of choices, 
which can limit the possibilities for adverse selection, but 
more fundamentally, in most group insurance, the employee who 
chooses the lower cost premium policy, or who chooses to have 
no insurance, will not recapture in cash, all of the money that 
has been saved by making that choice. In that way, people are 
induced not to engage in adverse selection, even if they are 
low risk.
    My conclusion about the current functioning of insurance 
markets, is that adverse selection is not, in general, a severe 
problem, nor is its mirror image, risk rating, which causes 
high risks not to have coverage. To the extent that there are 
problems for high risks, my suggestion is that an appropriately 
run plan of guaranteed renewability of high risk pool, can 
solve that problem.
    My fundamental amateur judgment on policy is that, of all 
of the things that are wrong with America's health care and 
health insurance markets--and there are many--these various 
risk segmentation issues so prominent in insurance theory and 
much policy discussion are a distraction. Not that there is no 
problem there, but compared to other issues like getting 
subsidies to people of all risk levels to help them afford 
insurance, if they are low-income, or making health care, if we 
can, cheaper and just as good, this seems to me to be a very 
low priority item.
    [The prepared statement of Mark V. Pauly appears in the 
Submissions for the Record on page 30.]
    Chairman Bennett. Thank you very much.
    Dr. Cardon.

  STATEMENT OF JAMES H. CARDON, PH.D.; ASSOCIATE PROFESSOR OF 
        ECONOMICS, BRIGHAM YOUNG UNIVERSITY, PROVO, UTAH

    Dr. Cardon. Mr. Chairman and Mr. Ryan, it is good to be 
here today.
    Insurance is desirable because it exchanges a fixed premium 
for a reduction in risk. Adverse selection is caused not by 
imperfect information about future expenditures, but by 
asymmetric information. Buyers and sellers of insurance may 
have private information about those risks.
    There is potential for adverse selection, anytime either 
buyers or sellers have significant informational advantages. 
Akerlof first illustrated the problems of private information 
advantages in the used car market, the market for lemons.
    Seller's private information about car quality leads to an 
unraveling of the market in what is sometimes called a death 
spiral. Rothschild and Stiglitz extended the argument to the 
case of insurance, and identified a simple market solution of 
this information problem that effectively identifies and 
separates, high and low risks.
    In this separating outcome, risk types are fully revealed, 
and the only deviation from the world of symmetric information 
is that the low risk types are forced to accept lower levels of 
coverage.
    It is a mistake to conclude that the separating outcome 
defeats the purpose of insurance, since health care expenditure 
are unpredictable, even given detailed information about 
demographics and medical conditions.
    As used and useful as this model is, there is something of 
a divergence between the theory and its application to real 
markets, and this has led to widespread misinterpretation of 
the statistical evidence.
    There is a crucial difference between selection based on 
private information and selection based on public information, 
such as demographics and income. Theoretical models that lead 
to adverse selection are concerned with private informational 
advantages, and public information poses no problem for 
markets.
    In a paper published in 2001, Igal Hendel and I built a 
statistical model to test for the presence and importance of 
asymmetric information in health care markets. The question is 
whether there is evidence of private information that can 
produce adverse selection.
    The test that we used is based on the link between 
insurance choices and subsequent consumption of health care. 
Intuitively, this test is based on whether this link can be 
explained by mutually observable variables, or whether private 
information plays a significant role.
    We found that the link between health insurance choices and 
health care consumption, is mostly explained by income and 
other demographics. As is normally the case, expenditures do 
vary predictably with income and these demographics, but most 
of the variation in expenditures is purely random and 
unpredictable.
    Our research shows no evidence of private information 
leading to adverse selection in the health insurance market. 
The life insurance market is similar in many respects to the 
health insurance market. Cawley and Philipson find that the 
data are inconsistent with private information on the consumer 
side. Instead, the authors suggest that the insurers in this 
market may have the informational advantage.
    Similar results have been found in the auto insurance 
markets. The papers cited here should cast some doubt about the 
presence of adverse selection. A failure to find evidence of 
informational advantages leading to adverse selection in a 
given market, does not, of course, mean that it cannot or that 
it does not occur; rather, it means that the problems that do 
exist are swamped by other factors, or that the problem has 
been managed by consumers or insurers in some other way.
    Many cases of so-called adverse selection are due to 
deliberate neglect of available information.
    One commonly made argument against Health Savings Accounts 
has been that they will lead to either increased risk 
segmentation and a separating outcome, or to the premium death 
spiral in which exit of the healthy from comprehensive plans 
raises premiums to the point that the market for such insurance 
collapses.
    At a common sense level, I believe that concerns that HSAs 
will distort markets, are greatly exaggerated. There is 
evidence that informational advantages are often either small 
or two-sided, with both buyers and sellers having private 
information.
    As far as risk segmentation is concerned, HSAs are similar 
to existing high-deductible or other plans with high levels of 
cost-sharing, and benefits managers know how to manage 
enrollment among a variety of plans by adjusting premiums and 
plan benefits.
    After all the analysis, it's markets that will provide the 
final test. If HSAs work, then they will become popular. If 
they do not work, then they will disappear.
    Traditional plans will continue to be available, and 
decisions are usually biased against new products. If firms 
find that HSAs are not a good match for their employees, they 
will drop them.
    HSAs will likely become a useful alternative to less 
comprehensive insurance or managed care, and they are worth a 
try. Thank you.
    [The prepared statement of James H. Cardon appears in the 
Submissions for the Record on page 39.]
    Chairman Bennett. Thank you very much.
    Mr. Closs.

 STATEMENT OF JEFFREY M. CLOSS, PRESIDENT AND CEO, BENU, INC., 
                     SAN MATEO, CALIFORNIA

    Mr. Closs. Good morning, Mr. Chairman and Members of the 
Committee. I'm pleased to be here today.
    Our Company, BENU, offers a meaningful choice of health 
insurers to employees of small and mid-sized companies. We're 
able to offer this expanded choice by reallocating premium to 
health insurers to compensate for adverse selection.
    We currently operate in the states of Oregon and 
Washington, with Kaiser Permanente Group Health Cooperative and 
Cigna Health Care, and beginning January 1st, BENU will be 
available in the Washington, DC region, with Kaiser Permanente 
and Cigna Health Care.
    Consumer choice of health insurers doesn't exist for small 
and mid-sized companies. Why? Adverse risk selection.
    Let me give you an example: A marketing executive for Group 
Health told us of their very successful program to treat 
diabetics. He described their sophisticated prescription 
system, which flags new insulin prescriptions, which then 
prompts a nurse to call the diabetic and offer education on 
monitoring and controlling blood sugar, as well as to schedule 
appointments with the dietician to review nutritional needs, 
and more testing for additional diseases.
    I was impressed with the comprehensiveness and 
effectiveness of this care. But when I asked, why not encourage 
all diabetics to join Group Health, he responded, ``We'd love 
to, but we can't. We'd go out of business.''
    The problem is, the premium payment will not cover the cost 
of treating a diabetic, no matter how efficient the care is. 
So, why are premiums insufficient?
    Because employers pay health insurers, based on an average 
cost payment. Health insurers charge the same rate for all 
potential enrollees within a company.
    The problem is that individual members have vastly 
different expected costs, depending on factors such as age, 
gender, and most importantly, the level of chronic illness they 
have.
    We know that a person with diabetes will, on average, cost 
many times that of a 20-year old, yet the health plan that 
enrolls either of these, receives the same payment. On a pure 
financial basis, whom would the insurer rather enroll? 
Obviously, the healthy.
    It's a shame that Group Health has a disincentive to 
promote their award-winning diabetic care program. how do we 
correct for this?
    BENU solves this by increasing the payment to the insurer 
for the more costly diabetic member, while lowering the payment 
for the less costly healthy member, what we call risk-adjusted 
payments.
    Employers continue to pay BENU, average cost payments, but 
what's transformational is that BENU pays insurers risk-
adjusted payments. Insurers now get a fair payment for both the 
sick, as well as the healthy.
    The large chart over in the corner demonstrates the amount 
of premium that we actually move between carriers by employer 
group, which we'll probably touch on later. [Chart appears in 
the Submissions for the Record on page 49.] The results are 
profound:
    Health insurers now have an incentive in enrolling the 
chronically ill, as well as the healthy. They won't fear 
financial losses, if they enroll a disproportionate share of 
the diabetics.
    Employers can now offer a choice of truly competing 
insurers, and provide a fixed subsidy for the lowest cost, most 
efficient plan.
    Employees can choose to buy up to more expensive plans and 
pay the difference. This creates savings for the employer and 
controls health care inflation.
    In fact, BENU has saved its customers an average of 15 
percent on the total employer premium cost. In addition, 
consumers are significantly more satisfied.
    Our customers tell us that their employees love the broad 
choice. Consumers can now choose from a comprehensive closed 
network HMO from Kaiser Group Health, an open access PPO from 
Cigna Health Care, or a consumer-directed plan with a Health 
Savings Account.
    Consumers are making decisions based on their own 
individual financial and health needs, but, more importantly, 
efficiency is brought to a health care system badly in need of 
it. Insurers compete for the right reasons, providing high 
quality care to keep their members. If they don't, employees 
can easily move to another insurer that will satisfy their 
needs.
    While free-market forces drive needed efficiency, the 
social aspect, where the chronically ill receive care at prices 
they can afford, is maintained.
    The truth is, we expect our health insurance carriers to be 
more than just insurance. We expect them to be a good service 
plan, taking good care of the healthy and chronically ill, and 
we expect them to be part social program, spreading the cost of 
the health care evenly among all participants.
    Not surprisingly, with the way insurers are being paid, 
they're having a hard time being either. Mr. Chairman and 
Members of the Committee, what's wrong with the current system 
is not how we fund health care, but how we pay insurers.
    We fund health care by charging everyone the same premium 
for the same plan, no matter how sick they are, but instead of 
paying the insurer the average cost premium, we should adjust 
payments to insurers, based on the chronic illness of those 
enrolled.
    BENU enables small and mid-sized employers to offer a 
competitive choice of insurers, delivery systems, health plans, 
and prices to their employees, and reallocates premium to 
insurers to compensate for the adverse selection that 
inevitably occurs. The result is a competitive consumer market 
that lowers costs, satisfies employees, and motivates insurers 
to provide value to the chronically ill. Thank you for your 
time today.
    [The prepared statement of Jeffrey M. Closs appears in the 
Submissions for the Record on page 42.]
    Chairman Bennett. Thank you very much.
    Dr. Blumberg.

    STATEMENT OF LINDA J. BLUMBERG, PH.D.; SENIOR RESEARCH 
         ASSOCIATE, THE URBAN INSTITUTE, WASHINGTON, DC

    Dr. Blumberg. Thank you, Mr. Chairman and distinguished 
Members of the Committee, for inviting me to share my views on 
adverse selection and health----
    Chairman Bennett. Could you pull the microphone a little 
closer to you?
    Dr. Blumberg [continuing]. For inviting me to share my 
views on adverse selection in health insurance, and its 
implications, when expanding consumer choice in private health 
insurance markets. I applaud the Committee for taking the time 
to carefully consider these issues, which are of paramount 
importance to individuals' access to health care coverage and 
medical services.
    In order to understand health insurance markets, there is 
one overarching fact that must be understood: The distribution 
of health expenditures is highly skewed, meaning that a small 
fraction of individuals account for a large share of total 
health expenditures.
    Because of this fact, the gains to insurers of excluding 
high-cost people, swamp any possible savings from efficiently 
managing care for enrollees. The incentives for insurers to 
avoid high-cost, high-risk enrollees are, therefore, 
tremendous.
    Greater risk segmentation of the market means setting 
individuals' health insurance premiums to more closely reflect 
each individual's expected health care costs. Conversely, 
greater risk pooling implies increasing the extent to which 
individuals with different expected health care spending 
levels, are brought together when determining premiums.
    Providing new insurance options is one way, intentionally 
or not, that the extent of risk segmentation can be increased. 
Reforms that increase risk segmentation, are appealing to some 
because they promise, and sometimes deliver, lower premiums for 
currently healthy persons, and because the majority of people 
are healthy.
    However, gains from segmenting healthy groups can occur 
only if premium costs for the unhealthy are increased, or if 
the unhealthy are excluded from the market to a greater extent 
than is true today.
    Examples of proposed and already implemented reforms that 
will increase risk segmentation in private markets, are Health 
Savings Accounts, tax deductions for the premiums of high-
deductible policies associated with HSAs in the private non-
group market, association health plans, and tax credits for the 
purchase of non-group insurance policies.
    While risk segmentation increases the cost of coverage for 
the unhealthy, the isolated instances where states have forced 
greater risk pooling, have not been successful, either. Efforts 
at pooling have been limited to a small population base, and 
have been foiled by individuals and groups that opt out of our 
voluntary private insurance markets.
    Addressing the problem will require subsidization of the 
costs associated with high-cost individuals, with the financing 
source being independent of enrollment in health insurance--
ideally, all taxpayers.
    In this way, the unhealthy could be protected from bearing 
the tremendous costs of their own care, while there would be 
little to no disincentive for the healthy to give up coverage.
    Three examples of policies that would move closer to such a 
paradigm are: (1) Dramatically increasing funding for state 
high-risk pools, and making the coverage both more 
comprehensive and easier to access; (2) Having the Federal 
Government take on a roll as public reinsurer, particularly for 
the private, non-group market and for modest-sized employers; 
and (3) A more comprehensive strategy would allow groups to 
continue to purchase insurance in existing markets under 
existing insurance rules, while each state provides new 
structured insurance purchasing pools. Through these new pools, 
employers and individuals could enroll in private health 
insurance plans at premiums that reflect the average cost of 
all insured persons in that state.
    For the following reasons, introducing greater choice 
within existing insurance pools, will not solve the problems I 
described. In fact, doing so will likely exacerbate them, even 
given the best available risk adjustment mechanisms.
    First, it is not sufficient to spread risks only within a 
particular insurance pool. Second, benefit package design is an 
effective tool for segmenting insurance pools by health care 
risk. Offering less than comprehensive insurance will tend to 
attract healthier enrollees.
    Third, in private markets where differences in actuarial 
value of plans can be quite large, and where people have the 
opportunity to opt in or out of the market, risk adjustment 
becomes substantially more difficult. Risk adjustment has been 
used in the Medicare Program, and is universally considered to 
be inadequate in that experience. Finally, it is not even clear 
that employers will have a strong incentive to want to risk-
adjust across plans.
    Although most employers want to look out for the well being 
of all their workers, they face incentives to keep health care 
premiums down, while keeping their highest paid workers 
satisfied. HSAs may provide employers with an effective tool 
for responding to these incentives, but place a greater share 
of the health care financing burden, directly on the sick, 
while higher paid employees can be compensated via the tax 
subsidy.
    Further segmentation of risk will not improve the social 
welfare in the United States. Addressing the health care needs 
of all Americans and protecting access to needed services for 
our most vulnerable populations, those with serious health 
problems and those with modest incomes, will require broad-
based subsidization of both those with high medical costs and 
income-related protection for those unable to afford even an 
average-priced insurance policy. Thank you very much.
    [The prepared statement of Linda J. Blumberg appears in the 
Submissions for the Record on page 49.]
    Chairman Bennett. Thank you. We've been joined by Senator 
Reed. Senator, do you have an opening statement? We'd be happy 
to hear from you before we start the questioning.

OPENING STATEMENT OF SENATOR JACK REED, U.S SENATOR FROM RHODE 
                             ISLAND

    Senator Reed. Thank you very much, Mr. Chairman. Just 
briefly, this is a very important topic, given the fact that we 
have legislatively committed to Health Savings Accounts for 
over 10 years, with a price tag of about $70 billion, so I 
think we have to ask the question, are we getting our money's 
worth in terms of broader coverage and more efficient coverage.
    The issue of adverse selection is a critical component to 
answering that question of whether or not these Health Savings 
Accounts are literally allowing healthy individuals to 
accumulate, through the tax system, wealth, while not serving 
the needs of lower-income Americans and particularly very ill 
Americans.
    Now, I think that's at the heart of this issue, and I 
commend the Chairman for raising the issue and for bringing 
together a panel of experts to do this.
    Mr. Chairman, I think that sort of summarizes where we are, 
and I'd be happy to claim my time in questioning at the 
conclusion of your questioning.
    [The prepared statement of Senator Reed appears in the 
Submissions for the Record on page 30].
    Chairman Bennett. Thank you very much.
    Well, you've heard each other on the panel, and what I like 
to do in these hearings, because they are not legislative 
hearings, is move more to an attitude of a panel discussion 
than a direct question-and-answer session. Now, we stay in the 
framework, in that each Member is allowed to conduct the 
discussion, if you will, but having heard the range of opinions 
here, I'd like to get a little interaction going.
    Dr. Blumberg is fairly firm in her conclusion about Health 
Savings Accounts and how dangerous they are. I don't want to 
put words in your mouth. You didn't use the word, 
``dangerous,'' but I get that sense from your testimony.
    Dr. Cardon, you've done a lot of research on this. Can we 
have an exchange between the two of you, and then the others 
get into it, as to where you are on this one?
    Dr. Cardon. OK.
    Chairman Bennett. Bring the microphone closer to you, as 
well.
    Dr. Cardon. Well, I don't think we know how these things 
are--I think Mr. Ryan said what I believe. We're not sure how 
these things are going to play out.
    I don't think they are that dangerous, because I am 
skeptical about the true adverse selection. I guess we've had 
the range of definitions of adverse selection here.
    The way I have defined it, I don't think there's a lot of 
it. It's driven by private information, and not by things that 
can be observable, including things that would be used in a 
risk rating system.
    I think that there's some--I just don't think the dangers 
that are suggested with these things, are as real as might be 
suggested.
    Chairman Bennett. Dr. Pauly, I was interested in your 
concluding remark that this whole debate is something of a 
diversion from the real structural problems that face our 
health care system.
    I have the same feeling. I think some of the truly 
structural problems that we face, are being ignored in much of 
the debate, and at this series of hearings, we're trying to get 
at some of those problems.
    Do you have a comment here on how important is the issue of 
adverse selection, and how valuable is the question of--is the 
opportunity of consumer choice? Is consumer choice a 
distraction?
    Dr. Pauly. All right, actually I think that Dr. Blumberg 
and I both agree with you, that the more fundamental questions 
are ones that involve helping people afford insurance and 
making insurance for the average person work well.
    In terms of--let me make two comments: One is whether 
Health Savings Accounts create adverse selection that we should 
be terribly worried about, and then what should we be worried 
about?
    First of all, if low coverage policies do attract low 
risks, so do aggressive HMO plans. In a way, they are certainly 
no worse, and we've been able to deal with HMOs and tolerate 
them.
    More importantly, if the HSA plan is offered in an 
individual setting, as I've already pointed out, when the 
individual insurance is already risk-rated, the HSA insurance 
would also be risk-rated, and if the insurer knows what it 
seems to know, which is as much as the person knows, there 
would be no adverse selection that would go on there. There 
might still be a choice by healthier people to take the HSA, 
but they would pay--their doing that would not cause the high-
risk to pay anything more, because there is no cross-
subsidization.
    In the group setting, of course, it depends on what 
employers do. I agree with Dr. Blumberg, at least, I think, 
implicitly, that it would be possible for a foolish employer to 
set up an HSA option and then set the reward for choosing the 
HSA in such a way to create adverse selection.
    I think that with the kind of devices that Mr. Closs talked 
about, or some other less formal devices, that insurers, 
employers, and benefits consultants know about, it's possible 
to control the extent of adverse selection, essentially by not 
offering too large a reward to the low risks for the plan that 
they choose, so that not to make it excessively attractive.
    Then I guess the final way to look at this is, in a worst-
case scenario where all the other plans disappeared and only 
the HSA plan was left, if that's a decent plan, which I think 
it is for most Americans--it may not be the best of all 
possible worlds, but it's not the end of the world, either--
but, more generally, of course, the main adverse consequence of 
adverse selection is that it would wipe out choice.
    I guess, to address that specific question, I think there 
is enormous evidence that Americans differ substantially, not 
only in terms of how they want their health care and what kind 
of health care they want, but how they want it controlled.
    Some people want to control it themselves by paying out of 
pocket; other people are happy to have a managed care plan say 
no for them; still other people are happy to spend more money, 
because that's what it takes to be able to have full insurance 
coverage and whatever you want.
    Of course, all of us would like cheap insurance that puts 
no restrictions on us, but, this side of the grave, that's not 
something we're going to see.
    I think that offering those options to people who seem to 
have very different preferences in terms of how they want to 
see their health care financed and ultimately controlled, is 
part of the genius, in a way, of the American system of 
allowing a pluralistic arrangement where there are different 
strokes for different folks, makes the most sense.
    Chairman Bennett. Your description of what we want is the 
subject of Robert Samuelson's op ed piece in this morning's 
Washington Post, when he says Americans want full control of 
choice of their doctors, full access to all services at all 
times, and very low premiums.
    Dr. Pauly. Even grownups are still teenagers when it comes 
to health insurance.
    [Laughter.]
    Chairman Bennett. It's come to that. Mr. Closs, react to 
Dr. Blumberg about your experience, and then, Dr. Blumberg, 
react to Mr. Closs's experience in the marketplace. We've got 
the two professors, and now you're the practitioner here.
    Mr. Closs. You know, Dr. Pauly actually set it up great for 
us, in that everybody has a different need, and if we try to 
dictate what each individual employee needs, and if it's an 
HMO, well, we were there once before, right, when we had 
everybody into an HMO, and that didn't work out so well.
    I think we're potentially setting up the same thing, which 
is that if everybody moves into a consumer-directed plan, high-
deductible HSA, by force, you're going to end up with the same 
situation.
    The fact of the matter is, everybody has different needs. 
Not everybody wants to be engaged in the retail purchase of 
health care. They don't want to go negotiate with their doctor 
about how much that visit is going to cost. Some people prefer 
that closed system HMO where everything is taken care of.
    I think, really, our mission is to provide all of those 
choices in a competitive marketplace, and let free-market 
forces decide which of those plans provides the most value and 
the best efficiency at the right price for each individual.
    I think that's kind of one of our fundamental business 
premises, that choice is good; employees want choice; they want 
to determine how much they're going to spend; and they want to 
determine how much coverage they get for that.
    In terms of adverse selection, I guess I have a couple of 
comments: One is, there's a lot of talk about what do we do to 
prevent it? One of the things that happens is, I think, if we 
get so focused on preventing adverse selection, what ultimately 
happens is, we lose choice.
    We lose a differentiation among products and insurers in 
the marketplace. Why? Because the way you minimize or mitigate 
adverse selection is to make everything the same, right? That's 
counter to what we're trying to do. We want to have 
differentiated choice for the individual.
    If risk selection happens, adverse selection happens, kind 
of our philosophy is, then compensate for it. Don't spend so 
much time trying to prevent it, because of all the bad things 
that come with the prevention of adverse selection, but when it 
happens, the key is that the carriers need to get compensated 
appropriately for the risks that they get.
    If they do, they now have an incentive to take care of that 
member. I think that's been the imbalance in this system, the 
connection of those two.
    Chairman Bennett. Do you buy that, Dr. Blumberg?
    Dr. Blumberg. What I agree with is that you expend a lot of 
resources to try to avoid adverse selection, and that's an 
efficiency loss; that's a lot of wasted time and effort.
    I agree that insurers will always be better than any 
analyst I can think of who attempts to stop insurers from 
pursuing enrollment of a lower-cost risk group, using risk 
adjustment, regulations, or other techniques.
    They are always going to be better than us at it. It's 
their job; they do it all the time. That's why I'm concerned 
about the direction of the policies that have either been 
implemented recently or are being seriously considered. Because 
although we have very limited experience with HSAs, as everyone 
has acknowledged, and we don't know exactly how bad the risk 
segmentation is going to be, all of these kinds of policies, 
the HSAs, the tax credits, the further deductions for high-
deductible policies, the association health plans, will all 
have a tendency to move our market to a more segmented risk 
scenario.
    With that, the proposals are not joined by other proposals 
that would compensate for this greater extent of segmentation 
that we're probably making. These proposals, therefore, do not 
acknowledge that there's no way we're going to be able to stop 
it from occurring, regardless of our different predictions of 
what the magnitude is going to be.
    What I'm suggesting is that we look at proposals that are 
specifically designed to assist those individuals who are most 
vulnerable in these markets, and those are the people who incur 
high-cost illnesses and individuals who are of low and modest 
incomes.
    To that extent, I agree with what Mr. Closs is saying.
    Chairman Bennett. Mr. Ryan.
    Representative Ryan. This is very interesting. I want to 
ask for a reaction on something that's happening right now, 
which is--I mentioned this in my opening statement, the FEHBP, 
the Federal Employee Health Benefit Plan. I'd like to ask each 
of your predictions on what you think is going to happen.
    Now, this is the largest health insurance pool in the 
country with nine million Federal workers and their families. 
What OPM just did, the Office of Personnel Management, who is 
in charge of putting this together, this year--I think the open 
season starts in December--Federal employees will be able to 
choose HSAs.
    To avoid these risks and adverse selection issues, they 
basically made the premium virtually identical to the other 
common low-
deductible premiums, with all the other plans.
    There has been an adjustment made by OPM, consciously, to 
try and avoid this, based on a big premium differential, so 
that's moving forward. We'll see this happening now, and from 
what we've been seeing from other testimony, is that the 
Federal Employee Health Benefit Plan is sort of a trend-setter 
in the large company marketplace.
    We're already seeing a lot of early data coming in, that 
small and medium employers and the individual market, are 
really going toward HSAs. The adverse selection data that's 
coming in is splotchy. It's showing that sort of the opposite 
is occurring, but, again, it's too early to see.
    What's your impression and your belief and your prediction 
on what this new policy will have? There are nine million 
people in this large pool, now having access to these plans. 
Will adverse selection occur, and, because of the premium 
adjustment, do you think that the bottom is going to fall out 
and the healthy and the wealthy employees in the Federal 
Government are going to flock to these and you'll have death 
spirals in the rest of the FEHBP? I'd just be curious to have 
everybody's reaction on that, and if you have studied this.
    Dr. Blumberg. Well, it's an interesting issue, given that 
the structure, as you said, is that they're having the same 
premium being charged. I'm assuming that whatever extra 
contribution that the individual is making or the Federal 
Government is making into that plan, is going to go into the 
account, as opposed to the premium, so it still will be more 
attractive to individuals who have low expected health care 
costs. This is because they could use those dollars, not just 
for potential savings for retirement, which a lot of young 
people aren't that interested in doing, anyway, but also they 
can use it for discretionary types of medical care that aren't 
covered by traditional insurance or even by the high-deductible 
plans; things like eye glasses, cosmetic surgery and 
nonprescription drugs, those kinds of things. I think it still 
will be more attractive to the low cost for those reasons, and 
the premium savings.
    Will they be able to prevent the death spiral? You know, 
FEHBP has some very clear experience with death spirals in the 
context of their Blue Cross-Blue Shield plan. Originally, they 
had two options, High Option and Standard Option Blue Cross-
Blue Shield, with deductibles that were not that different, but 
there was some savings from taking the higher deductible plan.
    The selection that occurred across those two plans, drove 
the premiums apart to the point where over time--and it did 
take time--the High Option Plan was no longer sustainable and 
was dropped a couple of years ago from FEHBP.
    Do I have great confidence that they are going to be able 
to avoid segmentation? No, I don't have great confidence. Does 
it take quite a long time for death spirals to occur? It does 
take years. I don't think you're going to see the impact of it 
over a year or 2 years, but you'll start to see segmentation of 
the pools, if we can collect appropriate data. I don't know 
what OPM is doing to track what's going to be going on. It 
would be great if they were collecting data on the health 
status and the sociodemographics and the wage levels and the 
family status of the people that are enrolling in these 
different plans, so that we can really do a good assessment. 
But because I don't have great confidence in our ability to 
risk-adjust or to have a strong incentive to risk-adjust in the 
long run, I still have concerns.
    Dr. Pauly. Well, my empirical answer to almost any question 
these days is no more than 15 percent, so I guess I would use 
that here. I think the number of Federal employees that are 
likely to enroll in this plan, would be no more than 15 
percent.
    I do think, though, it may well be, as I said in my 
remarks, that a high-deductible plan is the efficient plan, is 
the desirable plan for healthier people to use, compared to 
people with chronic conditions. In the best of all possible 
worlds, we'd like to have people have the efficient plan.
    The way to--the question, though, is, will there be adverse 
consequences for the people who are at higher levels of risk? 
My answer to that is, in a sense, that's up to the Federal 
Government in terms of what premiums it wants to charge for the 
high-option plans, relative to the low-option plans.
    There's no law of nature that requires it to allow the 
premiums for those plans to rise, relative to the HSA or to 
what already exists in fairly aggressive managed care plans. 
That can be adjusted, and if you say, as you might be thinking, 
``Well, what about total health care costs,'' then, if we don't 
penalize the high risks when the low risks are, in a sense, 
getting more than they should? The answer is, at least in 
economic theory, that you can adjust the money wages so that 
your total compensation budget stays the same.
    The general philosophy--Mr. Ryan, you said the Federal 
employee plan is a trend-setter for plans around the country. 
There is one enormous difference: It offers many, many more 
plans than almost any private firm does. To some extent, 
perhaps Dr. Blumberg is right, it has more of a problem with 
the potential for selection, given the wide variety of plans.
    But even so, with appropriate anticipation of what kind of 
risks will enroll in which plan, it's definitely possible to 
set the premium differentials so that people end up where they 
should be. As I said, the kind of last resort to kind of make 
everything work out right, is to adjust the money wages, so 
that the total compensation cost stays within whatever target 
you set.
    Mr. Closs. A couple of comments: One, I suspect that the 
risk selection will be a little higher in an HSA type plan. My 
personal opinion is that if people are sick and they know they 
are going to consume services, most of those services are going 
to happen on the other side of the deductible, all right?
    So, they are going to use their cash, they're going to go 
past the deductible, and when they see that, they're most 
likely to pick a more comprehensive plan and not be exposed to 
that big out-of-pocket difference.
    One would think that if people do have that information as 
they're purchasing, they may, in fact, end up with a more 
comprehensive plan and the more healthy will end up in a 
consumer-directed plan.
    Again, our view is that that's fine, because we're going to 
compensate the carriers appropriately for that different risk, 
such that they are paid right.
    What I want to do is talk just a little bit about the 
transition, and, as you said, how that moves into the large 
employer market, because our world is employer-sponsored.
    Large employers, when they start to introduce a consumer-
directed plan, they have less issues than the employer segment 
that we deal with, in large part because they are all self-
funded. You know, the employer is the insurer, they have their 
own risk pool, so the risk pool is intact, no matter what 
people choose.
    What changes is when you go into a smaller employer 
environment or mid-sized employer, and they buy insurance. If 
you have two insurance companies in that environment, that's 
where the dynamic occurs where the insurers won't share that 
business.
    I think that's what we're trying to fix, is the fact that 
now those employers are really forced to make one choice, and 
that choice isn't going to fit everybody. If they go to that 
consumer-directed plan, it may be good for some, but it's not 
going to be good for all.
    Really what we see is employers forced into a position 
today in this small and mid-sized segment, where they have to 
buy a single insurer, and generally they get one product. I 
mean, in our markets that we're in now, 80 percent of people 
have no choice. Only 20 percent have a choice of carriers.
    I think it's important to understand the dynamic that 
occurs in an insurance environment, to figure out how we can 
make sure that we create a competitive environment where they 
can get all those plans and the death spiral won't happen.
    Chairman Bennett. Did you say 80 percent have no choice, or 
80 percent have a choice?
    Mr. Closs. Eighty percent of employees in the States of 
Oregon and Washington, have one carrier and one plan.
    Chairman Bennett. So they have no choice?
    Mr. Closs. They have no choice.
    Chairman Bennett. I see.
    Representative Ryan. My time is up.
    Chairman Bennett. Senator Reed.
    Senator Reed. Well, thank you very much, Mr. Chairman, and 
thank you to the panel for their insights.
    I was struck by Dr. Cardon's conclusion that at this point, 
it's hard to tell what the effect of HSAs are in terms of their 
impact on the health care system. That raises to me, the 
question of, since we're spending $70 billion already in the 
next 10 years, and the President is calling for $40 billion 
more, are we getting what we're paying for, particularly with 
respect to what I believe are the policy objectives, or should 
be the policy objectives, which are to broaden coverage and to 
reduce the premiums to make it more affordable, which are 
obviously related issues.
    Dr. Pauly, do you think Health Savings Accounts are 
contributing in a meaningful way to increasing coverage and 
reducing the cost of premiums to the average person?
    Dr. Pauly. Well, of course, I need to say, not yet. Nobody 
knows, but the potential--the primary potential advantages of 
Health Savings Accounts is, as I mentioned in my remarks, that 
by having people pay more out of pocket, they will be more 
prudent in their choice of health care.
    That, of course, you could say, for someone who chooses an 
HSA, that's kind of the bargain they make. They agree to have 
less financial protection and more of a chance of getting a big 
bill that wiped out the account, but in return, on average, 
they'll end up being more frugal and spending less.
    I think, you could just say, ``Well, just let individuals 
choose HSAs, those who like them,'' but I think that from a 
general public policy point of view, I think there's pretty 
strong evidence that when the middle class chooses expensive 
health care, the premiums rise for everybody, and it's those 
rising premiums engendered by excessive insurance coverage and 
excessive health care purchases by people like me, that are 
causing health insurance to be un-
affordable to lower income people.
    If you ask me what's likely to be the greatest benefit of 
HSAs or any other cost containment feature, managed care or 
HMOs, I would include as, you know, the Gold Dust Twins of cost 
containment, and it is--the people who buy those plans can make 
their own decisions, but the reason why the rest of us have an 
interest in what those people do, who are generally not poor or 
near poor, is because there is this spillover or trickle down 
effect.
    Then there has been some intriguing data from the HSA 
experience, which suggests that a surprisingly large fraction 
of people who bought those plans, were people who had formerly 
been uninsured. I don't have an easy explanation for that, but 
these days, I'm happy to hear good news, so I count that as 
good news.
    Senator Reed. But let me return----
    Dr. Pauly. But I think the main point is that their primary 
advantage, I think, is cost containment.
    Senator Reed. Cost containment, so the other objectives, 
which would be essentially expansion of coverage, is probably 
less--not well served by HSAs?
    Dr. Pauly. No. I think that's right, and that's well served 
by tax credits to help low-income people afford insurance.
    Senator Reed. That gets to the point of tax advantages or 
the true nature of the tax system to relatively high-income 
people who have tax liabilities.
    Dr. Pauly. Yes.
    Senator Reed. Do you know, offhand, how many Americans 
don't pay taxes because they work, but they don't get to the 
point where they're paying income taxes?
    I mean, my impression is, as I go through Rhode Island, 
that there are lots of people working 40 and 50 hours a week at 
very low-income jobs, who don't have health care, and would not 
be enticed to an HSA, because simply their tax liability is 
minimal. In fact, they pay more in payroll taxes than they do 
in any type of income tax.
    Dr. Pauly. Well, I think that's right.
    Chairman Bennett. Twenty percent of working Americans do 
not pay income tax.
    Senator Reed. That would translate into a lot of people who 
would not essentially be enticed into the HSA model, because it 
doesn't make any real economic sense to them.
    Dr. Pauly. I think that's right. Any plan whose advantage 
is tax deductibility, obviously only matters to people who have 
an income level at which that's relevant, or a tax exclusion.
    In a way, what HSAs do is kind of level the playing field 
for those who do pay taxes, so that they will not be biased 
toward choosing excessively generous coverage and might decide 
to choose a plan that makes them somewhat more frugal, which 
may provide benefits to others. I guess the implication is, if 
you want to help the bottom 50 percent, you have to go to 
refundable tax credits.
    Senator Reed. But the other aspect here is--and I guess I'm 
not talking with any quantitative data backing me up, other 
than my own sense of things--is that if you feel healthy, your 
sense of how much health care you need and how well you're 
going to tolerate the future, increases.
    To a relatively healthy person, these HSAs are appealing, I 
would think. They've made a conscious decision that they don't 
need all the bells and whistles on a big plan, and so they say, 
``I'm a healthy person, I've got good genes,'' and it seems to 
me that this adverse selection issue is based not on any 
economic model, but just on sort of common sense or a common 
tendency of, if I'm healthy, I can get a tax break; I don't 
think I really need insurance. That might even account for 
those people who didn't have insurance before, and now since 
they have a tax break, they buy insurance.
    It goes to this issue, I think, of what kind of people 
you're going to find in these HSA arrangements.
    Dr. Pauly. Well, it is an effort to even things up. Now, if 
you're not healthy and you buy a very comprehensive plan, you 
get a big tax break, and so we're trying to be fair to the 
healthy and wealthy, as well as to the unhealthy and not 
wealthy, but I think, in terms of a factual matter, the 
question of who will choose HSAs and what their risk level will 
be, is more complicated than you think.
    Of course, it depends on what the alternative plan would be 
that they might choose. If the alternative plan is actually the 
kind of plan I have, which has a smaller deductible, a $500 
deductible, but then copayment and out-of-pocket limit which is 
higher than the typical HSA deductible, if I were a high-risk 
person, I'd actually be better off in HSA/MSA plan than I would 
be in my current plan. They're actually compared--of course, 
compared to insurance that pays 100 cents on the dollar for 
everything, which nobody has anymore, HSAs do look more 
attractive to low risks, but compared to the alternative kinds 
of insurance that most people have, including what's in the 
Federal employees' plan, for the really high-risk people, the 
sort of stop-loss feature of most HSA plans, actually may mean 
they're better off than with the other.
    Senator Reed. Thank you, Dr. Pauly. Thank you for your 
response. Dr. Cardon, and then I'll ask Dr. Blumberg. I'll do 
the academicians first.
    Dr. Cardon. Same question?
    Senator Reed. Any comment you have based on this.
    Dr. Cardon. One comment I would have is, so, you're feeling 
good today and you think you have good genes. Well, just how 
sure are you? I think that there's a behavioral thing here that 
probably explains Dr. Pauly's 15-percent number.
    People are uncomfortable with deductible plans, I mean, to 
some degree. I think that there is always going to be a 
tendency to sort of err on the side of caution and go toward 
the comprehensive plan, and I think that limits a lot of the 
selection that otherwise might occur.
    There are other things, too. I mean, if I feel healthy, but 
I have two kids, how healthy are they? I'm not basing my 
selections, just on my own health, but on the sort of average 
health in a family, and that sort of muddies it up quite a bit.
    Senator Reed. Thank you. Dr. Blumberg, your comments?
    Dr. Blumberg. Thanks. First, with regard to objectives of 
our spending, I'd just like to say that in times when we have 
scarce resources, and we seem to always have scarce resources, 
I'd much rather see that our subsidies were being directed to 
those who are least likely to have health insurance in the 
absence of those subsidies.
    I would suggest, and I think it would be hard to dispute, 
that the healthy and wealthy are not the ones that are least 
likely to have health insurance, in the absence of subsidies. 
While we do have some other subsidies that are upside down in 
our system, I would say that it doesn't make a lot of sense in 
terms of trying to expand health insurance coverage, to direct 
more subsidies to that well-off segment of the market.
    In addition, I have relatively great skepticism about the 
potential for savings or the cost containment effect of HSAs. 
The reason I'm skeptical is because the bulk of dollars that 
are spent will still be in excess of that deductible.
    Roughly 80 percent of the dollars in the health care 
system, will occur over those deductibles, even when you're 
talking about a $2,000 deductible. The share of expenses that 
you have at the bottom, that you're saying that people are 
going to spend out-of-pocket through the HSAs, is a small share 
of expenses.
    Even if you're able to conserve a bit on that small share, 
what is the real extent of the savings going to be to the 
system? I would suggest that it would be relatively small.
    I agree that time will tell, but I do feel like we're 
directing our energies in the wrong direction and away from the 
most vulnerable populations.
    Senator Reed. Let me ask a followup question: My 
understanding--and you can clarify it if it's wrong--is that 
someone can have an HSA account, but as soon as they reach the 
age of Medicare, they can go into the Medicare system.
    I would assume, since there is a certain correlation 
between health and age, that the real costs come at the point 
where we're actually picking up the tab through the Medicare 
system on many of the costs, so that these savings, these cost 
containment issues, as you point out, Dr. Blumberg, with a 
healthy 35 or 45 or 55-year old, are not the places where the 
real, the major cost are incurred in the system. Is that fair?
    Dr. Blumberg. Well, I would not go quite there, because 
there are sizable costs associated with those under 65, and the 
distribution of expenditures there is just as skewed as you see 
among the elderly.
    We do have very high-cost individuals who have not reached 
Medicare age, whose needs need to be addressed. But, clearly, 
cost containment is an issue, not just for the elderly, but the 
non-elderly, as well.
    Senator Reed. Thank you. Mr. Closs, I was very impressed 
with the ingenuity and the logic of your business plan, which 
raises the question, why don't insurance companies do this 
themselves? Why do they need sort of an intermediary like 
yourself?
    Mr. Closs. Great question. I think there are a couple of 
reasons: The first one is antitrust, because, you know, one of 
the things, when we get two insurers together, if we're going 
to move premium dollars around, right, we want to make sure 
that that is absolutely protected, such that particularly when 
it comes to pricing, that we want those insurers to compete 
aggressively for the employee in that employer group, such that 
we want each carrier to present prices based on that group, to 
try to get as many members as they can, again, trying to drive 
competition in there.
    Two carriers together can't really coordinate that up-front 
effort toward the sales process and the pricing process. It's 
important to have an intermediary in there that protects that 
information, and we act as essentially a market-maker to keep 
that intact and to promote competition.
    More importantly, after the fact, after people enroll, 
making sure that each carrier gets the appropriate amount of 
dollars. Neither carrier is going to trust each other to 
reallocate money to one another, so it's important to have that 
person behind the scenes, after the fact, to make sure that the 
premium dollars get to the right carrier.
    Senator Reed. Thank you. Recognizing that the legitimate 
goal of an insurance company is to make profits for their 
shareholders or for their members, if it's a mutual 
organization, is there a distinction in the profitability of 
these Health Savings Accounts? I mean, have we seen that data 
anyplace?
    Mr. Closs. I'm not aware that there's any data yet that 
would demonstrate that. I think it's too early.
    Senator Reed. That's a quantitative issue that we haven't 
seen. I guess the other question I would have to the panel and 
to Mr. Closs, is--and this, I think, goes to the question we've 
been addressing all afternoon. What is the chief factor in 
changing the profitability of a Health Savings Account? Is it 
essentially making sure you pick healthy people, or is it 
something else?
    Mr. Closs. Well, I would argue that in a risk-adjusted 
environment, it doesn't matter whether it's a healthy or sick 
person. If the carrier has an incentive to pick the healthy in 
the non-risk-
adjusted environment, of course, there are going to be excess 
profits, potentially generated from taking the healthy.
    But if you're actually putting an environment in place 
where they're competing and they know they're going to get paid 
appropriately for the risk, it takes that dynamic out, so if 
they get a sicker person in that HSA high-deductible plan, 
they're going to get more revenue that goes with that to take 
care of that person.
    Senator Reed. OK, Dr. Blumberg.
    Dr. Blumberg. I would just want to put a caution on what 
Mr. Closs said, in that I would agree with him in a perfectly 
risk-
adjusted environment, but our technologies are not perfect for 
risk adjustment. In fact, the best available that's being used 
now, I believe, by your corporation and also by the Medicare 
Program, can get us to roughly half the theoretically 
explainable portion of the variation in expenditures.
    Yes, it's true that the profitability of selecting risks 
goes away in a perfectly risk-adjusted environment, but nobody 
has that at this time.
    Dr. Pauly. May I make a comment on the virtues of 
imperfection, which, of course, is the old maxim that you don't 
want to make the perfect the enemy of the good. We have got 
some insights into this question by interviewing the benefits 
managers of large firms, actually some years ago, asking them 
about medical savings accounts.
    It was kind of interesting. They sort of fell into two 
camps: The slightly smaller camp were a set of people who said, 
``We are terribly worried about adverse selection and risk 
adjustment and we don't think we'll offer it, or we'll go 
slow.''
    Another group said, we tried to offer any benefit plan that 
a sizable fraction of our employees like. We said, ``Well, 
aren't you worried about risk segmentation,'' and they said, 
``Well, no. For one thing, we can pretty well control that the 
total amount of difference in well being is pretty small, 
relative to our worker wages, and although the young workers 
might gain from the spending account in their health insurance, 
they're losing on the way we do pensions, without vesting the 
pension.'' In some ways, the sort of theoretical idea that 
you'd like to make things turn out perfectly, was trumped by 
the view that our primary role here in offering benefits is to 
offer a variety of things that our workers like, and not worry 
about small gains and losses, especially when they are more 
potential than actual.
    If they turn out to be larger than small, the world--well, 
life is long and you can re-adjust things later.
    Senator Reed. Thank you very much.
    Dr. Pauly. In fact, a lot of firms do.
    Senator Reed. Thank you, Doctor. Thank you, Mr. Chairman.
    Chairman Bennett. Thank you very much. Having been an 
employer and wrestled with these challenges before I came to 
the Senate, I'm resonating with what you're saying, Dr. Pauly. 
We had a cafeteria plan in which we said to each employee, 
``You have $300 a month in what we called `flex box.' You can 
spend them any way you want.''
    It was a fascinating kind of experience to discover the 
different situations with different employees. One employee 
said, ``Are you nuts? I've got six kids--and Dr. Cardon, you 
may have only two, but I have six--I've got to have every one 
of those dollars in a health plan, in order to cover what might 
happen if a kid falls out of a tree, rides a motorcycle, 
whatever all else. Yeah, I'm healthy, but all that money has to 
go to a health plan.''
    The next employee comes in and says, ``My husband works at 
Hill Air Force Base. He's covered by the Federal employee 
program. A health insurance program in this company would be 
redundant to the coverage that I already have through my 
husband, so can I use that money for daycare?'' We said, 
``Sure, you know, give us the name of your daycare provider, 
and we will send that money every month to your daycare 
provider.''
    Next employee comes in and says, ``My husband works at 
Kennicott, same thing, I don't need health care coverage, it 
would be redundant for me. My kids are all grown. Can you put 
that money in my 401(k)?'' Yeah.
    And so on. Interestingly enough, the benefit administrator 
who ran our plan said: ``Don't offer life insurance. Life 
insurance is a bad investment. It's a bad mistake; don't offer 
it.''
    I said, ``We're going to offer it.'' Well, we just told you 
it's a bad mistake. I said, ``That's your determination. There 
may be employees who make a different determination, who may 
feel, for whatever reason, they want their benefit dollars in 
life insurance.''
    He said, ``Well, they're making a mistake.'' I said, ``I'm 
not going to make that decision for them; I'm going to allow 
them to make the decision.'' There was a small percentage of 
our employees who said, ``If I have this much--these many 
benefit dollars to spend as I see fit, for my piece of mind, I 
want some life insurance.''
    Their determination of what amounted to peace of mind was 
different than the administrator's determination as to what 
amounted to peace of mind. I trusted the individual employee to 
make that decision, even though I might not have counseled them 
to buy life insurance.
    It produced a much happier employee and a much higher level 
of employee morale, which is, after all, what I wanted. You 
know, I offer benefits in order to get people to come to work 
for me, and in order to make them feel like they want to stay 
working for me, instead of going off to work for my competitor.
    I offered them a package of salary, and I offered them a 
package of retirement, and I offered them a package of 
benefits. That's what you do as employers, you compete in the 
pool for the best employees you can get, and then you act in 
ways that will hang onto them. You don't want to punish them or 
drive them away.
    It was a very interesting experience to go through the 
cafeteria plan and discover how different people had different 
ideas, and, frankly, none of them struck me as irrational. 
Everyone had reasons for wanting what they wanted, and everyone 
came from a different situation and different circumstances.
    The vast majority, of course, spent most, if not all of 
their benefit dollars for health care, but there were these 
other examples of people who said, ``In our situation, it makes 
more sense.'' That experience convinced me that the old canard 
that the average person is incapable of making an intelligent 
decision with respect to health care, is just that; it's a 
canard; it's not the truth. I think we're smarter than many 
policymakers give us credit.
    This has been a very helpful panel, in helping us 
understand the benefits, the opportunities, the challenges, and 
the pitfalls of expanding a degree of consumer choice with 
respect to health care. I thank you all for coming, and I thank 
you all for your contributions. Your full presentations will be 
in the record, and based on what we've seen at some of these 
other hearings, you'll be read by some very interesting people 
that you might not have anticipated when you made your 
submission to the Committee.
    Again, thank you all. The Committee is adjourned.
    [Whereupon, at 11:20 a.m., the hearing was adjourned.]
                       Submissions for the Record


Prepared Statement of Senator Robert F. Bennett, Chairman, U.S. Senator 
                               from Utah

    Good morning and welcome to our hearing on consumer choice in 
health insurance. Many consumers would like to have greater choice and 
control of their health care and health insurance coverage. They know 
from their experience with many other types of goods and services that 
choice and competition helps match their different tastes and 
preferences to those options providing the best value.
    Recent efforts to increase consumer choice include:
     providing multiple health plan options with employer-
sponsored coverage,
     offering new consumer-driven health care arrangements such 
as Health Savings Accounts,
     reforming Medicare financing to strengthen private plan 
alternatives in what was originally called the Medicare+Choice program 
back in 1997, and was changed to Medicare Advantage in last year's 
Medicare Modernization Act, and
     trying to level the playing field for those purchasers who 
select individual insurance market products rather than employer group 
insurance coverage.
    To be fair, some of those initiatives have advanced further than 
other ones. In part, that's because various plans to increase consumer 
choice in health insurance often face criticism that they will trigger 
a host of purported dangers, usually starting with what's called 
adverse selection. That term tends to be loosely defined and widely 
used in health policy debates. Its most accurate definition is when 
consumers know significant private information about their expected 
expenses that their insurers do not know. In that case, it's more 
likely that insurance buyers who know that their risks are greater than 
average will want to purchase more insurance when it's priced based on 
average risk. Buyers who expect their risks to be lower than average 
will prefer less insurance coverage.
    This simple description of adverse selection then projects that 
insurance premiums for the original coverage offered will increase more 
than otherwise, because low risks either switch to other types of 
insurance or, in the extreme, drop coverage entirely. The end result is 
presumed to trigger a ``death spiral'' of rising claims costs and fewer 
paying customers to finance them under the initial insurance policy. In 
the worst-case scenario, the death spiral extends to the overall health 
insurance market, which can break down completely.
    Those who believe that more consumer choice in health insurance is 
just another risky scheme are likely to handicap it, if not prohibit 
it, through such policy measures as community rating, standardized 
benefits, coverage mandates, and preferential subsidies.
    Today's hearing will examine how employers and insurers can offer 
better choices to consumers in practice without producing the sort of 
adverse consequences sometimes predicted in theory. We want to 
determine what really happens in insurance markets in the pooling and 
pricing of risks and sort out real problems from imagined ones. It 
appears that there are natural limits on the scope and scale of 
potential adverse selection problems. Employers and insurers seem to 
manage remaining ones rather effectively in most cases. Nevertheless, 
there may be policy opportunities to improve access to the care and 
coverage that consumers value most.
    Health insurance coverage, of course, is just a means to an end. 
The real objective is better health and better, outcomes from medical 
treatment. But improving the value of insurance that's available to a 
diverse population of consumers is an important part of that process. 
Increasing their choices, rather than reducing them, seems to be a 
fundamental starting point for upgrading the status quo.
    Our panel of witnesses today includes Dr. Mark Pauly, one of the 
nation's leading health economists from the Wharton School at the 
University of Pennsylvania. He has written extensively about the 
operations of health insurance markets and how public policy may shape 
them.
    Dr. James Cardon of Brigham Young University has examined whether 
consumers have private information advantages over insurers that could 
trigger adverse selection and distort health insurance markets. His 
most recent research focuses on the effects of new consumer driven 
health care choices like Health Savings Accounts.
    Jeffrey Closs is president of BENU a company that provides benefits 
choice center services to mid-sized employers. BENU uses risk 
adjustment tools to encourage insurers to compete more vigorously for 
portions of an employer's business and provide more meaningful choices 
for covered employees.
    Linda Blumberg of the Urban Institute has studied issues of risk 
selection and risk segmentation in voluntary insurance markets, 
particularly those involving small employers and individual consumers.

                               __________
Prepared Statement of Senator Jack Reed, U.S. Senator from Rhode Island

    Thank you, Chairman Bennett. I want to thank the Chairman for 
holding today's hearing on the issue of ``adverse selection'' in health 
insurance markets. This is obviously an important issue given the 
amount of attention that has recently been given to high-deductible 
health plans, such as Health Savings Accounts (HSAs).
    This hearing gets to the heart of the debate over HSAs, which is 
whether or not they will encourage healthy and wealthy people to opt 
for higher-deductible plans, while less healthy people are left in 
increasingly expensive traditional insurance plans. Unfortunately, 
illness affects everyone, regardless of age, race or economic status.
    Since HSAs appeal to a healthier population with fewer health care 
costs, they could actually have negative consequences for less-healthy 
people seeking insurance. The clear danger is that HSAs will divide the 
insurance market between healthy and less healthy people, making the 
health care system even more inequitable than it is today as insurers 
adjust pricing to reflect the risk pools in each type of insurance.
    If HSAs attract the healthiest people, those Americans with 
traditional insurance will face higher premiums and increased cost-
sharing. Higher premiums will put tremendous pressure on companies to 
stop offering comprehensive, traditional insurance. Companies will 
either pass on the higher costs to employees, make them switch to an 
HSA or simply drop coverage. While proponents of HSAs argue that they 
offer consumers more choice, those may not be terribly attractive 
choices to many people.
    President Bush has proposed spending $41 billion on HSAs and high-
deductible plans, which will at best extend health insurance to a tiny 
fraction of the 44 million who don't have coverage today. Clearly, this 
policy is not directed toward insuring the uninsured. It looks more 
like HSAs are another tax shelter for the wealthy--who have no trouble 
affording insurance or quality health care--rather than an innovative 
approach to expanding health care coverage.
    I'm skeptical of the benefits of HSAs, which probably won't reduce 
costs or increase health coverage. Nevertheless, I hope today's hearing 
will shed some light on whether or not HSAs will make it easier for 
patients to get the care they need.

                               __________
    Prepared Statement of Mark V. Pauly, Ph.D., Bendheim Professor, 
 Professor of Health Care Systems, Insurance and Risk Management, and 
  Business and Public Policy in the Wharton School, and Professor of 
     Economics in the College of Arts and Sciences, University of 
                              Pennsylvania

    Thank you, Mr. Chairman, and members of the committee, for an 
opportunity to testify on adverse selection in health insurance and 
related issues.
    Private health insurance would be far less controversial if we 
lived in a world where everyone was similar in terms of risk. Then 
insurers would charge similar premiums to everyone who put similar 
effort into shopping for a given policy, and would be equally eager to 
sell insurance to anyone. After the fact, those lucky enough to have 
low actual health expenses would have paid in more than they got back 
from insurance, but this redistribution from those who did not become 
sick to those who did would be something that everyone would agree 
before the fact was both fair and attractive, and all would be eager to 
buy insurance as long as the premium was not too much higher than 
expected benefits.
    The world in which we do live, it is obvious, is different. It is 
one in which ``risk'' varies before the fact, in the sense that 
different consumers reasonably expect to collect different amounts in 
benefits from a given policy because they expect to get sick with 
different frequencies and severities. Insurers can identify and measure 
some characteristics that they know predict above or below average 
benefits, characteristics such as age, location, and the presence of 
chronic conditions. Insurance markets can still function in such a 
world, but now either premiums or purchases will be different for 
different people.
    What will happen depends crucially on whether insurers have and can 
use the same information that predicts benefits as consumers can use. 
If everyone has the same information, and the information does predict 
different risk levels, then insurance theory (Rothschild and Stiglitz, 
1976) tells us that insurers will choose to charge below average 
premium to the lower risks and above average premiums to the higher 
risks. Someone who has four times the expected benefits from a given 
policy compared to someone else will be charged about four times the 
premium. At those premiums, insurers will be equally eager to sell to 
low and high risks. In insurance theory, this situation of proportional 
risk rating will be stable and probably will be one in which low risks 
are no less likely to buy insurance than high risks. (Some very high 
risks with low incomes may find that premiums are so high and expenses 
so near certain that they are just as well off not buying insurance 
they cannot afford and paying those expenses directly when and if they 
can.)
    Insurance markets, the same theory also tells us, will be very 
different if insurers do not have equal information to what buyers 
have, or if insurers are not allowed to use the information they do 
have in setting premiums and bidding for business. In the extreme case 
in which insurers either cannot distinguish among risks or are not 
permitted to do so, they will be forced to charge the same premium to 
everyone who buys insurance. But if the insurance purchasers know their 
risk levels, their willingness to buy insurance at this premium will 
vary. Higher risks will be very enthusiastic about buying, since they 
can on average collect in benefits more than they pay in premiums. But 
low risks may, in the limit, decide not to buy insurance at all because 
it looks like a bad deal to them, or may at least seek to buy less 
generous coverage than the high risks desire. This situation of 
community rating will be one in which the low risks are less likely to 
buy insurance than under risk rating. In the limiting case in which the 
low risks bail out altogether, the so-called death spiral, the premium 
insurers end up charging to the high risks will be the same as they 
would have charged under risk rating; the effect of community rating 
will only be to drive out all of the low risks (which is definitely not 
the same as no risk) from the insurance market, with resulting adverse 
effects on access to care and financial stability. It is in this sense 
that community rating can be inefficient compared to risk rating, since 
it can make the low risks worse off and not make the high risks better 
off (Pauly, 1970). In the less extreme case in which some low risks 
might continue to buy, the high risks could be better off but the low 
risks will still be worse off than they would have been under risk 
rating. There will still be inefficiency compared to the ideal because 
the low risks will choose less coverage than they would have chosen if 
they had faced premiums reflective of the true cost of their coverage.
    Whether there will be cream skimming, in which insurers are more 
eager to sell to low risks than to high, depends on whether the adverse 
selection-community rating is essential (caused by insurer inability to 
tell risks apart) or inessential (caused by regulations or policies 
which forbid insurers from using information they have to set lower 
premiums for lower risk and higher premiums for higher risks). In the 
case of essential adverse selection, as in the case of risk rating, 
there should be no cream skimming because all potential purchasers look 
equally profitable to insurers. Insurers might want to cover only the 
low risks, but they cannot tell who is who. In the less extreme case of 
regulation-required community rating, insurers will try to avoid 
selling to high risks they can identify, on whom (as a group) they are 
sure to lose money; there will be cream skimming.
    For these kinds of reasons, some insurance analysts think risk 
rating is better than community rating. But many policymakers, and some 
other analysts, do not look at it that way. They do note the downside 
of community rating in terms of squeezing out the low risks, even to 
the extent of a death spiral in which at some point only the highest 
risks end up buying. (This should really be called a near-death spiral 
because at that point it will be profitable for some insurer to enter 
and offer a less generous plan at a much lower premium that can pull 
some of the lower risks back into the market; the market will rise, 
phoenix-like, only to go into another spiral.) But policymakers also 
find much not to like in risk rating, precisely because the higher 
premiums for higher risks may bite into their ability to consume other 
necessities for life if they have low income, and sometimes because 
observing higher income high risks paying more than higher income low 
risks still looks unfair, especially compared to a policymakers' dream 
world in which everyone pays a low premium. That this is impossible in 
a world of competitive but unsubsidized insurance markets only 
marginally dampens their ardor.
    The most obvious way to deal with these problems is to use 
regulation. Require insurers to charge similar premiums (or limit 
premiums for high risks), but forbid low risks from buying less 
generous policies. Then require insurers to sell policies to high risks 
they know will be causing losses, and, when there is enough political 
nerve, forbid insurers and the low risks from dropping out by mandating 
insurance purchasing. Measures short of this draconian one can still 
lead to bad adverse-selection type outcomes, especially when community-
rating rules force insurers to ignore information they have and thus 
lead to inessential adverse selection. Then, when insurers respond to 
community rating regulations with cream skimming, one needs to write 
yet more regulations to require open enrollment and guaranteed issue. 
To avoid the death spiral, we move to a regulatory spiral. As with 
other kinds of health care regulation, how bad (or good) the regulatory 
outcome will be seems in practice to vary across states, depending on 
the characteristics of their potential insureds and the form and 
administration of the rules. In some states such rules seriously 
curtail the size of the insurance markets, while in others the main 
effect is only discontent among the low risks and the insurers who 
would like to sell to them.
    The main novel point I want to make here is that recent research 
suggests that, in both theory and practice, there are ways alternative 
to regulation to get closer to what policyrnakers want (or should want) 
when risk rating and adverse selection are possible. Compared to 
perfect regulation administered with perfect regulation, or even to the 
wise and prudent regulation that occasionally happens, these 
alternatives may still leave something to be desired. But compared to 
the kind of regulation we have had or can generally expect to have, 
they at least deserve equal billing and equal consideration. These 
alternatives may work better if some other government actions are 
curtailed and some modest regulation applied to encouraging the 
alternatives.
    To be specific: one might suppose that, as is often the case, 
policymakers must choose between two undesirable outcomes--unfair risk 
rating or inefficient community rating--in order to deal reasonably 
well with risk variation. New developments in research (Pauly, 
Kunreuther, and Hirth, 1995; Cochrane, 1995) suggests that, in theory 
and in fact, in many circumstances realistic competitive insurance 
markets can avoid much of both bad situations, and that a relatively 
modest amount of public intervention can deal with the cases that fall 
through the remaining cracks. The fundamental reason for this market 
behavior is that potential insurance consumers also dislike the more 
negative aspects of either kind of behavior, and competitive insurers 
have developed methods to avoid them. The fundamental reason for the 
political behavior is that some policymakers have already developed 
some well-tailored solutions that leave the market intact but rein in 
the worst cases.
    The three kinds of ``solutions'' to which I want to draw your 
attention are (1) guaranteed renewability at uniform premiums, (2) 
group insurance, and (3) high risk pools. Because the first is much 
less well understood than the other two, I will discuss it in more 
detail, but I will also comment on the other two devices.
    The great majority of people who are high risk today were not 
sicker than average at all times in their lives. Data shows what common 
sense tells us: even people who are in excellent health have higher 
medical expenses on average as they age, and some pick up chronic 
conditions. The age-related part of increasing risk is perfectly 
predictable; what is not predictable is the random onset of a chronic 
condition that makes a person high risk not only initially but for some 
time to come, possibly for life.
    Most medical expenses for people under 65 are not related to 
chronic conditions; they come from the ``bolt-from-the-blue'' event of 
an accident, a stroke, or a complication of pregnancy that we know will 
happen on average but whose victim we cannot (and they cannot) predict 
well in advance. This is precisely the kind of low probability, high 
cost event for which insurance works extremely well as a device for 
substituting a smaller certain payment for an unexpected rare but large 
payment. Sometimes, however, what strikes unexpectedly is a condition 
from which the person is unlikely to recover rapidly; such random but 
then chronic conditions make future medical expenses higher for people 
who have them. If insurance premiums were proportionately risk rated to 
the risk prevailing for the next year (the usual time period for health 
insurance), people who are well today and have no chronic conditions at 
the moment would face the chance of contracting such a condition with 
two bad financial outcomes. Not only is diagnosis usually associated 
with high immediate medical expenses, it would also be associated with 
a sudden and serious jump in premiums.
    Risk averse people should want to have protection not only against 
high current period expenses but against the unexpected onset of a 
condition that might entail high lifetime premiums; they would seek 
protection against ``the risk of becoming a high risk.'' In some real 
world health insurance markets such protection exists and was quite 
common even in the absence of regulatory rules. Specifically, most 
health insurance policies bought on an individual basis contained a 
provision also common in individual term life or disability insurance: 
guaranteed renewability at class average premiums. With this provision, 
the insurer promises not to single out insureds whose risk has 
increased more than average for high premiums when they renew their 
coverage. Instead, they are to be charged the same premiums as are 
charged to everyone else who was in the same initial (usually low) risk 
class as they and bought the same type of coverage. Administering such 
a guarantee is easy for an insurer: it promises to base its future 
premiums only on whatever information it collected about risk when it 
initially sold the coverage; it promises not to revisit the question of 
risk based on new data that might be obtained from the person or even 
based on the claims history data that the insurer has; it promises not 
to ``re-underwrite.'' This provision does not guarantee constant 
premiums; premiums can rise if expected medical expenses rise for 
everyone in the risk class (say, because of higher medical prices), and 
premiums may rise according to a schedule specified in advance as a 
function of perfectly predictable things, like growing older. But the 
person with coverage with this feature is protected against the bad 
luck of becoming riskier than average, and therefore will not pay a 
higher premium on becoming a high risk. This feature is not free, of 
course; policies that contain it must have higher initial premiums 
(``frontloading'') than would premiums for a policy for which the 
insurer retained the right to increase premiums for people who 
contracted a chronic illness. But it is easy to see why rational, 
foresighted people would prefer the slightly more expensive but surer 
policy to the cheaper but riskier one.
    Federal law now requires states to ensure guaranteed renewability 
for individual (but not group) insurance policies. But even before the 
spread of such State laws, industry observers estimated that about 80 
percent of policies voluntarily (on the parts of both buyers and 
sellers) contained such provisions. (Pauly, Percy, and Herring, 1999) 
There is, however, considerable debate about how they work in practice, 
debate which is assisted by the absence of nationwide comprehensive 
data on practices in insurance markets, especially in the individual 
market, so that evidence tends to consist of anecdotes and problematic 
surmises. There certainly have been cases in which insurers were caught 
engaging in re-underwriting even when they were forbidden to do so, and 
a number of State insurance departments have said that they would 
prohibit risk rating at renewal even in the absence of specific State 
law under their general authority to limit arbitrary and excessively 
discriminatory premiums (Patel and Pauly, 2003). Some insurers are said 
to have gotten around the requirement to continue to cover high risks 
by raising premiums for all insureds so that all drop out of the risk 
class, and then selectively re-enrolling only those low risks who have 
not been put off by this behavior. Insurance brokers and agents insist 
that they pay attention to this kind of behavior and steer customers 
who come to them for advice away from insurers who engage in semi-shady 
practices. We know that this feature does not work perfectly everywhere 
for everyone, but how well does it work on average?
    Research has provided some data that is highly consistent with 
guaranteed renewability generally operating as the theory and the 
intent of the contractual provision suggests (Pauly and Herring, 1999; 
Pauly and Herring, 2001). This finding is striking enough that it 
deserves to be emphasized even beyond the issue of guaranteed 
renewability. To be specific, there is very strong empirical evidence 
that the premiums higher risk insureds pay are much lower than would be 
consistent with proportional risk rating. Stated slightly differently, 
while high risks do pay higher premiums than low risks, the increase in 
premium with risk is much less than proportional to the increase in 
risk.
    This result has been obtained in a large number of studies using 
large nationwide data sets from different time periods. Depending on 
the measure used of risk, the ``elasticity of premiums with respect to 
risk'' in multivariate analysis of data ranges from about 20 percent to 
less than 50 percent; never higher. That is a person whose risk is 
twice as high as average will pay a premium only 20 percent higher. 
Table 1 shows more intuitive evidence for this proposition. It uses 
data from the late 1980's before there was widespread premium 
regulation in the individual insurance market or requirements of 
guaranteed renewability, but when that feature was common nevertheless. 
The risk level for a person in the data set is characterized by the 
person's age, gender, location (to measure differences in medical 
cost), and pre-existing chronic conditions. Statistical models were 
used to relate the actual medical expenses, and the actual insurance 
benefits received for each person, to that person's values for these 
variables; the estimate of risk for that person is then the ``predicted 
value'' of their medical expenses (that is, the average medical expense 
for a large number of people with the same values for these 
characteristics as they). Those risk estimates were then used to select 
a sample of people with individual health insurance expected to have 
medical expenses in the top 10 percent of possible values of risk, and 
another sample of people in the bottom half of those values. As the 
first line of the table shows, the expected expenses, the actual 
average expenses, and the actual average insurance benefits were much 
higher for the high risks than the below-average risks. The average 
benefits for the high risks were 11 times greater (at $2054 per person) 
than for the lower risks (at $187). The premiums were higher for the 
higher risks too, but the key point is that the premium for these very 
high risks (at $1150) were only 1.4 times greater than that of the low 
risks ($825); there was a substantial amount of averaging of risk in 
the premium structure.
    While there are doubtless many causes for this phenomenon, one of 
them probably is guaranteed renewability. People with such provisions 
would not be paying premiums that were higher than average because they 
became higher risks. Of course, some people in the data were new 
purchasers of insurance whose premiums would be risk rated, but 
apparently by no means all. There is even stronger evidence. We looked 
at how premiums and risk varied with age for similar policies. Insurers 
certainly can determine a buyer's age, and they certainly can determine 
that, other things held constant, expected expenses and benefit 
payments will rise with age (especially for men). What we found, 
however, was that the premium paid by the average older man was only 
about 40 percent higher than that for the average younger man when the 
expected expenses differed by a factor of two to one. But this pattern 
of overpayment relative to expected expense for the younger people who 
would generally be the new buyers of insurance is exactly the 
frontloading that would be predicted to arise under guaranteed 
renewability (but that would be unstable in competitive insurance 
markets under proportional risk rating). We have further examined the 
path of premiums and benefits with age in this market and find that it 
corresponds rather well with the path that would be consistent with 
guaranteed renewability. In doing this analysis, we adjusted for the 
fact that people often do not keep their individual coverage from a 
given firm but drop it because they have taken a job that carries 
coverage or because they switch insurers. Because the low risks have 
already prepaid their contribution to the high risks, their dropping 
out does not cause any problems for the ability of insurers to continue 
to maintain protection for higher risks. Some high risks do drop out as 
well but, as expected, at a much lower rate.
    In our analysis of individual insurance data we found that only the 
locational and demographic variables were consistently related to 
higher premiums. The person's health status when they bought insurance 
(measured by the presence of a pre-existing chronic condition) was not 
statistically related to premiums, but the scarcity of observations on 
people with such conditions means that our estimates are themselves 
necessarily imprecise (Jack Hadley and James Reschovsky, 2003) using a 
different risk measure (contemporaneous health status) and a more 
sophisticated but somewhat delicate statistical technique, did find 
that people in poorer health paid higher premiums, but even there the 
increase in premiums was much less than the increase in risk. I 
therefore conclude that individual insurance markets (even when they 
were unregulated) provided a substantial amount of protection against 
the adverse effects of risk rating to people who did what we want them 
to do--bought insurance before they became high risks, and stuck with 
their insurance rather than becoming uninsured.
    Risk rating can only occur if insurers can determine risk levels; 
under perfect risk rating, there can be no adverse selection. However, 
in a world in which buyers of insurance may sometimes know more than 
sellers, it is interesting to note that guaranteed renewability 
provides potentially important protection against adverse selection. If 
people buy this coverage early in life (as they should to take 
advantage of the provision), they are likely to be much more similar in 
risk levels than they will become later on. And since it is rational 
for the people who remain healthy to stay in their original policy 
where they have already made transfers to those in their cohort who 
became higher risks, it is less likely that they will drop out and 
start a death spiral. Finally, if those who remain lower risk do drop 
out or are lured away, because they have already prepaid their transfer 
to the high risks, the insurer does not need to raise premiums to the 
high risks.
    We have investigated some of the other reasons why higher risks pay 
premiums that are less than proportional to their relative risk levels. 
There is evidence that higher risks search more intensively to find a 
premium that is low relative to the expected benefits; it makes more 
sense to checkout many insurers (or use a broker to do so) when one is 
paying $400 a month for insurance, than when one is 25 and paying less 
than $100 a month for insurance (Pauly, Herring, and Song, 2003). And 
it probably is true that some risk factors, like the decision on the 
timing of the next child or the repair of an old football injury, is 
better known to the insured than to the insurer. But this phenomenon 
may be partially offset by the fact that insurers actually have more 
accurate data on risks than typical insurance consumers do.
    Another feature of insurance that can protect against uncertain 
jumps in premiums and adverse selection is group insurance. The great 
bulk of Americans obtain their health insurance as group insurance 
related to their employment. Probably the main reason they do so does 
not have to do with any risk variation factors, but rather to the 
substantial tax subsidy to workers (not to employers) present in the 
exclusion of compensation received as health benefits from income and 
payroll taxation. But group insurance probably does have some features 
that deter the kind of behavior theory was earlier said to predict.
    Most simply (but not most obviously), group insurance offers a much 
better deal for your money for a given policy than does individual 
insurance. The difference between the premium one pays and the benefit 
one should expect on average to get in group insurance is lower than 
for individual insurance both because of economies of scale associated 
with group purchasing (especially lower selling and billing costs) and 
because of the tax subsidy. These features in effect may make insurance 
such a good deal for the wealthiest low risks (who get the biggest tax 
subsidies) that they will not be motivated to drop coverage and start a 
death spiral even if their premium is not properly tailored to their 
risk. As long as a low risk's net premium is low enough after the tax 
benefits are taken into account, the fact that there is some cross 
subsidy to higher risks may not matter.
    A more complicated issue is whether or not employment-based group 
insurance in some sense ``pools risk'' more than other arrangements. 
For large groups, there is no explicit individual underwriting, but the 
cost of that function is only a tiny fraction of any insurer's 
administrative cost. There can be variation in premiums with risk 
across small groups; a firm of three 25-year-olds in good health will 
pay much less than a firm of three 60-year olds who are out of shape. 
Moreover, the requirement that one be able to work to qualify for one's 
own employment based insurance serves to automatically screen out the 
highest risks and those unable to take a job because they are caring 
for a dependent with high risk. But the key determinant of access to 
insurance and net payments for insurance is the policies employers 
follow with regard to this benefit:
    One thing that employers are motivated to do is to try to keep as 
many of their employees in the insurance plan as they can, because the 
premium, or even the availability of group insurance, depends on the 
participation level of workers in the firm. Let too many of them drop 
out, and the group insurance may not be offered by an outside insurer. 
Even self-insured employers (who cover the majority of workers 
nowadays) want to achieve economies of scale. Thus employers should 
want to avoid death spirals and widespread non-participation.
    Probably most importantly, workers in group insurance almost never 
pay an explicit total premium that is related to their precise risk 
levels; they almost always all pay the same employee premium if they 
choose the same policy for the same-sized household unit. (There is 
explicit risk rating for the higher risk associated with having more 
people covered under a family policy relative to an individual policy). 
However, economists believe that workers pay for the bulk of their 
group insurance not through explicit premiums but through lower wages, 
and generally money wages are not explicitly adjusted based on an 
individual employee's risk level.
    The evidence does, however, strongly suggest that worker wages are 
adjusted to some extent to reflect the different cost of insurance as a 
function of risk (Pauly and Herring, 1999; Sheiner, 1994). Wages vary 
by seniority, and more senior workers are usually older. What we found 
was that, other things equal, wages increased significantly less 
rapidly with seniority for workers who obtained job-based insurance 
than for those who did not; we interpret this as the effect of higher 
insurance costs taking away some of what would have been the usual 
raise associated with more experience and seniority. Moreover, common 
sense tells us that an employer cannot take the typical $6000 
``employer contribution'' out of the wages of younger workers and still 
expect to compete to hire those workers with other firms that offer 
higher cash wages and no coverage.
    There is no evidence that wages vary with health status given age 
and gender (though the lower wages of women could in part reflect their 
higher medical costs). But remember that with guaranteed renewability, 
premiums in individual insurance also need not vary with health status. 
Thus I would conclude that the amount of risk pooling in group 
insurance is at best only very modestly greater than in individual 
insurance on average. The difference would be greatest between a high 
risk person able to get a job at a firm that offers benefits and what 
that person would be charged as a new applicant for individual 
insurance. But the job with insurance is by no means assured to a high 
risk, and the typical buyer of individual insurance is renewing, not 
buying new, so this difference tends to average out to a small number 
if it is present at all.
    The main virtue of group coverage in terms of risk variation is not 
risk pooling per se but rather that it discourages adverse selection. 
It does so in several ways. Most obviously, the range of insurance 
choices a person has within a firm is usually much smaller than the 
range of choices in individual insurance, and any opportunity to choose 
less generous coverage (whether it is a high deductible plan or a cost 
constraining HMO) offers a chance for low risks to separate themselves 
out. The downside of this advantage is less choice, but firms and their 
workforces are free to make this choice not to have many choices.
    Equally, if not more important, is the fact that the worker who 
chooses to decline group insurance while remaining in the firm almost 
never recaptures the full premium for that coverage. Instead the worker 
will get back any employee premium and (in some firms) a small bonus 
for refusing coverage, but that reward is almost always much less than 
the value of the insurance even to a low risk. We do have a problem 
with more workers offered employment-based coverage rejecting it, 
especially as the average explicit employee premium has risen, but 
there are almost no cases where rejecting coverage to save the employee 
premium would be rational behavior if the person thought that without 
coverage they would have to pay for all of their medical care out of 
pocket. (They might drop and expect to rely on family assistance or 
charity care, and the still tiny fraction of people offered coverage 
who reject it may just be the minority of any population who are 
irrational or unthinking.)
    So there is very little total dropping out by lower risks, but do 
they inefficiently drop back to less generous coverage? Not 
necessarily, because employers can if they wish control adverse 
selection and risk rating. The simplest way to do this is to offer only 
one plan. But even when employers offer several plans, the key to 
controlling selection is to properly set the difference in employee 
premiums (or in the contribution to spending accounts) across plans 
(Cutler and Reber, 1998; Pauly and Herring, 2000). If employers 
foolishly make the premium much lower for the less generous plans, then 
all but the highest risk will join them, leaving the few remaining high 
risks in a more generous plan. But research shows employers how to 
calibrate the premium difference to reflect the premium cost reduction 
associated with the low risks (not the average and certainly not the 
difference in expected benefits when the low risks have already sorted 
into the less generous plan). So employers who want to control adverse 
selection can do so to a considerable extent (though not perfectly), 
especially if they self insure all of the plans they offer. Things are 
somewhat more complex if multiple outside insurers are used and those 
insurers are not given the data they need to estimate the risk levels 
of the people who will choose their plans. Risk adjustment of the total 
premium the insurer gets combined with appropriate setting of premium 
differentials will prevent adverse selection if that is an employer 
goal.
    Research (Pauly, Percy, Rosenbloom and Shih, 2000) suggests that 
some employers try to limit the choice of options and set the premiums 
to control adverse selection, while others take the view that any 
redistribution away from older workers in their health plan offering is 
probably offset from redistribution toward such workers in their 
pension plan or in other benefits, and that the total amount of 
redistribution (and inefficiency) is small. As long as the least 
generous plan offered is still a decent plan even for higher risks, 
there probably need be little policymaker concern about adverse 
selection in group insurance. Personally I would only be concerned 
about offering a health savings account type plan to very low income 
workers, or offering a very restrictive HMO to workers who would react 
strongly to limits on access, but I would not be much concerned in 
general.
    How does the rate of takeup of insurance vary with risk level in 
group and individual insurance? Are higher risks more likely to have 
coverage than lower risks (which would be consistent with adverse 
selection), are they less likely (which would be consistent with very 
strong risk rating), or is coverage nearly universal and independent of 
risk (which would be ideal)?
    Research on this subject is far from definitive. Studies that have 
looked at people in households where someone is a full time employee 
(and therefore potentially eligible for group insurance if the person 
chooses or is able to get a job at a firm offering coverage), the 
strongest and most consistent finding is that the size of firm in the 
industry or occupation of the worker is by far the most important 
predictor of having coverage (along with the size of the tax subsidy 
and therefore income) (Pauly and Herring, 2000). People who work in 
industries dominated by larger firms are much more likely to end up 
with coverage than those who work in small firm industries. The 
relationship of coverage to risk, given firm size, is less well 
understood. What we observe seems to depend on what measure of health 
risk we use. If we use chronic conditions as the measure, employed 
higher risks are more likely to have coverage than employed lower 
risks. If we use self reported health status, coverage may be less 
likely for high risks. Analysis of the late 1980's data showed that 
high risks were significantly less likely to have group coverage only 
if they were low-income people working in small firms, but not 
otherwise (Pauly and Herring, 1999). There is little evidence that 
employers in general have difficulty in continuing to offer coverage to 
people who become high risks, and no evidence at all that they have 
problems with people who have unexpected high expenditures.
    It is much harder to determine how risk levels affect the 
likelihood of having coverage in the individual market because anyone 
can participate in that market, but most people do not do so and 
instead obtain group insurance. We have looked at people in households 
where no one is a full time employee--the household's income comes from 
self employment, part time work, or non-work sources. The relationship 
here depends even more on the measure of risk. Len Nichols and I (2002) 
found that if we measure risk by age, controlling for income, older 
people in ``non-group'' households were much more likely to have 
individual coverage, despite higher premiums, than younger people. We 
also found that people with chronic conditions were more likely to have 
coverage, although the relationship was not as strong. On the other 
hand, when risk is measured by self-reported health status, people who 
label their health as fair or poor are less likely to have individual 
coverage controlling for income; this is the opposite of adverse 
selection. One puzzle in the data is that many of those with insurance, 
who say that no one in their household works full time, still list 
themselves as having obtained group insurance coverage; there is no 
clean division of the population between those with access to group 
insurance and those who must use the individual market.
    Precisely for this reason one should be very cautious in trying to 
draw conclusions about the comparative performance of individual and 
group insurance markets. If I was forced to do so, I would conclude 
that there may be differences in the likelihood of obtaining individual 
insurance coverage by people who are very high risks when they seek 
coverage, but that if the group market does better, the differences are 
small, and are limited by the fact that many very high risks do not 
have access to employment-based insurance. It would be nearly 
impossible to provide those currently without a group option access to 
that option on the same terms as the current users. I think the 
differences in the extent to which net premiums do (or could) vary with 
risk are small, and any stronger relationship in the individual market 
is attributable to its small size and marginal or add-on character. For 
example, a person who had group insurance, who contracts a high risk 
condition, loses their job and insurance, and uses up their COBRA 
coverage, will be recorded as a high risk trying to buy new coverage in 
the individual market. But one could argue that placing the person in 
that situation is as much the fault of the link between tax subsides 
and group insurance which does not provide guaranteed renwability 
protection to individual workers as it is the fault of individual 
insurance.
    Fortunately, there is a device available to pick up the pieces 
without requiring the imputation of blame: high risk pools. I do not 
intend to discuss the actual working of these pools in detail. Instead 
I want to point out that the concept of having a subsidized, decent 
though limited coverage policy available to high risks unable to obtain 
or retain individual or group coverage makes great sense as a safety 
net. Since the number of high risks is by definition low, it avoids 
having to distort insurance markets for the great majority who are not 
high risks in order to make transfers to a few unlucky people. Some of 
the more anecdotal research shows that almost any risk can obtain 
individual coverage if they persist at searching long enough, but those 
who have already been rejected or quoted very high premiums perhaps 
ought to have another option than spending their time with insurance 
brokers. In idealized concept, a high risk pool ought to offer coverage 
at premiums somewhat higher than those charged for good risks but still 
at reasonable levels to people who have tried and failed to obtain 
coverage on their own. The financing of these pools should be generous 
enough to accommodate those who need to use them, and that financing 
should be raised by general revenue taxation, not by requiring insurers 
to contribute and thus raising premiums which drive more people out of 
regular insurance. The terms of coverage (premiums, type of coverage) 
should be only moderately attractive, because we want to preserve 
incentives to people to obtain voluntary coverage before they become 
high risk, rather than wait to pick up attractive subsidized high risk 
coverage when and if that happens. I am hopeful that it is possible to 
design a plan that walks this fine line and still preserves an 
opportunity for people to obtain coverage that will give them financial 
protection and access to care. Coordinating high risk pools with 
guaranteed renewability provisions would seem to be desirable.
    To sum up: the important problems with private health insurance in 
the United States are not associated with the risk variation-risk 
segmentation issues that are so prominent in insurance theory and many 
policy discussions. Our problem is not that the insurance is expensive 
and unattractive for high risks; it is that in some cases it is 
expensive and unattractive for all risks. It is true that the largest 
single segment of the uninsured population is low risk healthy twenty-
somethings, and some adverse selection in group and individual 
insurance may modestly contribute to this. But I believe that a much 
larger contributor is the absence of generous subsidies and the absence 
of marketing efforts targeted at this group; there may actually be too 
little effort at cream skimming those low risks who remain uninsured.
    This is especially the case for people who are discriminated 
against by being ineligible for generous tax subsidies when they buy 
insurance (the non-self-employed in the individual market) and those 
who could have access to products with lower across-the-board 
administrative costs but do not currently have such access. Finally, 
the key background issue of what if anything we want to do when 
premiums are rising not because of insurance market behavior but 
because medical care is becoming both more costly and yet much better 
should really be front and center in the policy debate.

  Table 1.--Expenses in Nongroup Individual Coverage, by Risk (Expected
                                Expense)
------------------------------------------------------------------------
                                                         Bottom
                                                          50%    Top 10%
------------------------------------------------------------------------
Actual benefits.......................................     $187    $2054
Premiums..............................................      825     1150
Actual expenses (total)...............................      555    3504
------------------------------------------------------------------------
Source: Pauly and Herring (1999), based on 1987 NMES data.

                               References
    J. Cochrane, ``Time-consistent Health Insurance,'' Journal of 
Political Economy, 1995, 445-473.
    D. Cutler and S. Reber, ``Paying for Health Insurance: The Tradeoff 
between Competition and Adverse Selection,'' Quarterly Journal of 
Economics, 1998, 433-466.
    J. Hadley and J. Reschovsky, unpublished paper, 2002.
    L. Nichols and M. Pauly, ``The Non-group Health Insurance Market: 
Short on Facts, Long on Opinions and Policy Disputes,'' Health Affairs 
Web Exclusive, October 23, 2002.
    V. Patel and M. Pauly, ``Guaranteed Renewability and the Problem of 
Risk Variation in Individual Health Insurance Markets,.'' Health 
Affairs Web Exclusive, September 8, 2002.
    M. Pauly, ``The Welfare Economics of Community Rating,'' Journal of 
Risk and Insurance, 1970, 407-418.
    M. Pauly, H. Kunreuther, and R. Hirth , ``Guaranteed Renewability 
in Insurance,'' Journal of Risk and Uncertainty, 1995, 143-156.
    M. Pauly, Health Benefits at Work, University of Michigan Press, 
1997.
    M. Pauly and B. Herring, Pooling Health Insurance Risks, AEI Press, 
1999.
    M. Pauly, A. Percy, and B. Herring, ``Individual Versus Job-based 
Insurance: Weighing the Pros and Cons,'' Health Affairs, 1999, 28-44.
    M. Pauly, A. Percy, J. Rosenbloom, and D. Shih, ``What Benefits 
Specialists Think about Medical Savings Account Options for Large 
Firms,'' Benefits Quarterly, 2000, 39-46.
    M. Pauly and B. Herring, ``An Efficient Employer Strategy for 
Dealing with Adverse Selection in Multiple-plan Offerings: An MSA 
Example,'' Journal of Health Economics, 2000, 513-528.
    M. Pauly and B. Herring, ``Expanding Health Insurance Through Tax 
Credits: Tradeoffs and Options,'' Health Affairs, 2001, 1-18.
    M. Pauly and B. Herring, ``Premium Variation in the Individual 
Health Insurance Market,'' International Journal of Health Finance and 
Economics, 2001, 43-58.
    M. Pauly and B. Herring, ``Incentive-compatible Guaranteed 
Renewable Health Insurance,'' NBER Working Paper 9888, July 2003.
    M. Pauly, B. Herring and D. Song, ``Health Insurance on the 
Internet and the Economics of Search,'' unpublished paper, 2003.
    M. Rothschild and J. Stiglitz, ``Equilibrium in Competitive 
Insurance Markets,'' Quarterly Journal of Economics, 1976, 130-149.
    L. Sheiner, ``Health Care Costs, Wages, and Aging: Assessing the 
Impact of Community Rating,'' Federal Reserve Board, December 1994.

                               __________
  Prepared Statement of James H. Cardon, Ph.D. Associate Professor of 
      Economics Department of Economics, Brigham Young University

    Mr. Chairman and Members of the Committee:
    I have been asked to comment on the problem of adverse selection, 
both generally and as it might apply to consumer-directed health plans, 
such as Health Reimbursement Arrangements (HRA), and the new Health 
Savings Accounts (HSA).
    Adverse Selection is a term borrowed by economists from the 
insurance industry to describe a possible problem in the functioning of 
insurance markets. Insurance is valuable to people because it allows 
them to make a fixed premium payment in exchange for reducing risk. 
Adverse Selection is caused not by imperfect information about future 
expenditures but by asymmetric information: buyers or sellers of 
insurance may have private information about risk. There is potential 
for adverse selection any time either buyers or sellers have 
significant informational advantages.
    George Akerlof (1970) first illustrated the problems of private 
information advantages in the used car market, the market for 
``Lemons.'' Cars are either good or bad, and only the owners--having 
driven them for some time--can tell the difference. Buyers cannot tell 
the difference, and will therefore be unwilling to pay a ``Cream Puff 
'' price for a car that might be a Lemon. Cream Puff owners are 
unwilling to sell at less than Cream Puff prices, but Lemon owners are. 
Then only Lemons are sold, and the used car market unravels in what is 
sometimes called a `death spiral'. This is great economic theory 
because it is simple, intuitive, and seems to be supported by casual 
experience. Best of all, it would seem to apply to a wide range of 
markets. It is tempting to start seeing Adverse Selection everywhere. 
On the other hand, this simple, stylized model ignores important 
details of real markets.
    Michael Rothschild and Joseph Stiglitz (1976) extended the argument 
to the case of insurance. In this case, consumers have private 
information about risk status that they withhold from insurers. High 
risk consumers cannot be distinguished from Low risk consumers. The 
authors identify a simple market solution to this information problem 
that effectively identifies and separates High and Low risks. The 
insurer offers 2 plans to all customers. One is a high cost, high 
coverage plan and the other is a low cost, low coverage plan. Premiums 
and coverage levels are carefully chosen so that High risks choose the 
High coverage plan and Low risks the Low coverage plan. Risk types are 
fully revealed, and the only deviation from a world of symmetric 
information is that the Low risk types are forced to accept less 
coverage.
    A possible alternative outcome involves Pooling of risk types into 
a single plan. Both risk types receive the same coverage and pay the 
same premium, which reflects the average risk in the pool. Risk types 
are not revealed in this outcome. Low risk types subsidize High risk 
types and there is potential for an outside firm to engage in ``cream-
skimming'' by offering a plan that only Low risks will prefer. In this 
case, the pooling outcome does not occur. Pooling outcomes can and do 
exist in the group market, where the possibility of cream-skimming by 
outside firms is limited by employer subsidies, tax subsidies, and the 
fact that, on average, group insurance is cheaper than non-group 
insurance per dollar of coverage. Factors that can limit worker 
mobility between firms increase the potential for a pooling outcome, 
and Crocker and Moran (2003) show that more generous and comprehensive 
coverage is feasible with decreased mobility.
    The separating outcome, in which each risk type is correctly 
identified and rated, troubles some analysts because superficially it 
seems to defeat the risk-pooling function of insurance. This is a 
mistake, since health care expenditures are wildly unpredictable even 
given detailed information about demographics and medical condition. 
Because all risk types face substantial uncertainty about actual 
expenditures, insurance with premiums that accurately reflect those 
risks will always be desirable. The separating outcome is a possible 
market solution to severe informational asymmetry.

                     SOME LIMITATIONS OF THE MODEL

    The model above assumes that consumers have the informational 
advantage. This might not be true. After all, insurers have data on 
perhaps millions of consumers as well as a reserve of medical expertise 
not available to the average consumer. New customers might have an 
advantage over insurers, but for the cost of a physical the insurer can 
obtain a great deal of information to reduce its disadvantage. It seems 
likely that both sides of the market have private information of some 
type.
    Also, the private information consumers have might be of little 
practical use. To be useful the information must be specific about 
near-term expenditures. I believe that part of the reason that adverse 
selection seems obviously true is that we often mistake vague worries 
about family history for reliable information. We probably have less 
useful information than we think.
    The model assumes that there is a single year of coverage and no 
chance for learning over time. Yet many consumers stay with the same 
insurance company for years, and claims data are a gold mine of 
information on current usage and diagnoses of acute and chronic 
conditions that should help insurers identify a consumer's risk type.

       EVIDENCE OF ADVERSE SELECTION IN VARIOUS INSURANCE MARKETS

    As used and as useful as this model is, there is something of a 
divergence between the theory and its application to real markets, and 
this has led to widespread misinterpretation of statistical evidence. 
There is a crucial difference between selection based on private 
information (unobservable information) and selection based on public 
information (observable information, including demographics and 
income). Theoretical models that lead to adverse selection are 
concerned with private informational advantages.
    In a paper published in 2001, Igal Hendel and I built a statistical 
model to test for the presence and importance of asymmetric information 
in health care markets. The question is whether there is evidence of 
private information that can produce adverse selection. The test we 
used is based on the link between insurance choices and subsequent 
consumption of health care. We distinguish between mutually observable 
information, such as demographics and income, and information which is 
private to the consumer. The unobserved information links insurance 
choices and health care expenditures, as those consumers more likely to 
need health care purchase more generous insurance coverage. 
Intuitively, the test is based on whether the link between insurance 
choice and health care consumption can be explained by the observed 
information. If observables account for the link, then we can rule out 
the importance of private information in the joint insurance/health 
care decision.
    Much to our surprise, we found that the link between health 
insurance choices and health care consumption is mostly explained by 
income and other demographics\1\. As is normally the case, expenditures 
do vary predictably with income and demographics, but most of the 
variation in expenditures is purely random and unpredictable. Our 
research shows no evidence of private information leading to adverse 
selection in the health insurance market.
---------------------------------------------------------------------------
    \1\ One simple numerical measure of the private information is what 
we might call a ``signal to noise ratio'', or the ratio of the 
estimated variance of private information to the estimated variance of 
the purely random component. The higher the ratio, the more important 
private information is. The ratio is .27 if we artificially exclude all 
demographic variables and .004 if we include those variables. By this 
measure, the amount of true private information is trivial.
---------------------------------------------------------------------------
    Evidence from related insurance markets can be used to assess the 
importance of private information. Two recent studies examine adverse 
selection in the auto insurance market. Chiappori and Salanie find no 
evidence of adverse selection among new drivers in the French market 
(2000). Dionne, Gourieroux, and Vanasse (1998) and find that there is 
no adverse selection in the Quebec market once observable demographics 
are controlled for.
    The life insurance market is similar in many respects to the health 
insurance market. There is much at stake for consumers, the underlying 
risk is partly health-related, and there exist both group and 
individual submarkets. Cawley and Philipson (1999) use data on actual 
premiums and quantities as well as consumer perceptions about risk. 
They find that, contrary to predictions of the basic model, there is a 
negative relationship between risk and the amount of insurance 
purchased (people who believe they are at risk purchase less 
insurance). They also find evidence of bulk discounting: the cost per 
dollar of coverage becomes cheaper for higher coverage. Both of these 
findings are inconsistent with private information on the consumer 
side. The authors suggest that, in this case, the insurers have the 
information advantage:
    Some studies claim to find adverse selection. My own paper cited 
above is sometimes cited incorrectly as having found evidence of 
adverse selection, when in fact the opposite is true. This 
classification is consistent with common but incorrect usage. The 
confusion in this case and in many others is the distinction between 
true adverse selection as it is used in theory (selection based on 
private information) and adverse selection as it is loosely used by 
policymakers in practice.
    For example, an excellent recent paper by Cohen (2003) claims to 
find evidence of adverse selection in the Israeli auto insurance 
market. Cohen finds a positive relationship between insurance coverage 
choices and the frequency of subsequent accidents. A peculiar feature 
of that market is that insurers do not use driving histories to set 
premiums for new customers. The so-called private information in this 
market is only private because insurers ignore available information 
that is commonly used in other countries. Even so, the insurance market 
still functions reasonably well.
    The papers cited here should cast some doubt on the severity of the 
problem. A failure to find evidence of informational advantages leading 
to adverse selection in a given market does not mean, of course, that 
it cannot or does not occur; rather, it means that the problems that do 
exist are swamped by other factors or that the problem has been managed 
by consumers and insurers in some other way.
    To return to the original example of adverse selection, the used 
car market is supposed to break down due to severe adverse selection, 
and yet it is clear there is a robust market for such cars. Obviously 
when buying a used car a consumer must consider the Lemons problem. But 
buyers and sellers have arranged institutions to control the problem. 
Warranties, inspections, seller reputation and the prospect of repeat 
dealing are examples of how markets deal effectively with a potentially 
serious problem. People are clever, and they adjust in order to make 
things work. So the market that inspired concerns about adverse 
selection is in fact a fairly good example of market success. Ebay is 
another example of a market that should suffer from informational 
problems, and yet it continues to grow. Buyer and seller reputation 
play an important role here.
    I maintain that `death spiral' concerns are exaggerated, and that 
informational advantages are often either small or two-sided, with both 
buyers and sellers having private information. Many cases of so-called 
adverse selection are due to deliberate neglect of available 
information. In health insurance markets, several factors mitigate the 
problem of residual private information. Benefits managers adjust 
premiums and benefits to maintain stable enrollment. There are also 
non-price remedies available. For example, my own benefits plan 
includes a low cost, higher cost-sharing option. Enrollment in this 
plan is for a minimum of 2 years, and this provision prevents employees 
from frequent switching from high to low coverage.

       POTENTIAL FOR ADVERSE SELECTION IN CONSUMER-DIRECTED PLANS

    Archer MSAs have been available to small businesses for several 
years in a very restrictive way. Health Savings Accounts (HSAs) were 
introduced as part of the Medicare Prescription Drug, Improvement and 
Modernization Act of 2003. With HSAs, consumers and their employers are 
able to contribute pre-tax dollars into these accounts to use for out-
of-pocket medical expenses. To qualify, consumers must be covered by a 
health plan with a relatively high deductible of between $1,000 and 
$5,000 for an individual and between $2,000 and $10,000 for a family. 
Preventive care is excluded from this restriction and can receive 
first-dollar coverage:
    These plans offer consumers and employers greater flexibility in 
plan options, and there is potential to improve the delivery of health 
care and increase insurance enrollment by lowering costs. Part of the 
reason for rising health costs is that insured patients will over-
consume health care because they often pay only a small portion of 
health expenditures. HSAs seek to reduce this inefficiency by combining 
higher cost-sharing with a tax-preferred saving account. Catastrophic 
coverage is the most important component of any insurance plan because 
it protects us from financial ruin. Coverage for small, predictable 
expenditures is largely a result of a tax code that encourages us to 
pay for such expenses through an insurer instead of out-of-pocket.
    There is some confusion about what HSA balances represent. 
Accumulated balances are wealth that reasonable people will use wisely. 
As such, there would seem to be little concern that individuals with 
large balances will overspend. In general, the perceived cost of using 
$1 from the account will reflect the cost of replacing that $1 the 
following year, which depends in part on the individual's tax rate. For 
example, if the tax rate is 30 percent, then the cost of replacing the 
dollar is $.70. In effect, these plans are low cost, less-comprehensive 
plans with deductibles to limit risk. Unused balances can eventually be 
withdrawn as retirement income. Because of this provision, even very 
large balances will not be spent carelessly.
    Concerns have been raised that these plans benefit the wealthy and 
offer another tax shelter. This is true, but all rules that allow 
income to be sheltered from taxes benefit the wealthy, since they face 
higher marginal tax rates. An employee's share of employer provided 
insurance is already paid using pre-tax dollars. Retirement savings 
receive the same tax treatment, but putting money in an HSA is 
preferable to putting it in an IRA because HSA offers the option of 
using balances for health care.
    One commonly-made argument against HSAs has been that they will 
lead to a segmentation of health insurance markets that will exacerbate 
the standard adverse selection problem, leading either to increased 
risk segmentation in a separating outcome or to the premium `death 
spiral' in which exit of the healthy from comprehensive plans raises 
premiums to the point that the market for such insurance collapses.
    At an intuitive, common sense level, I believe concerns that HSAs 
will distort markets are greatly exaggerated. So far as risk 
segmentation is concerned, HSAs are similar to existing high-deductible 
or other plans with high levels of cost-sharing, and benefits managers 
know how to manage enrollment among a variety of plans by adjusting 
premiums and plan benefits.
    There are two possibilities that we should consider. First, adding 
an HSA option to menu of plan offerings is like adding a less-
comprehensive plan to the menu. This may be in addition to or in place 
of an existing low-coverage option. Again, this is nothing new, and 
should be manageable. I believe it is more likely that introducing the 
HSA might drive out the alternative low-coverage plan, leaving a choice 
between more comprehensive options and the new HSA.
    Second, a firm that offers a single plan option might be replacing 
a traditional fee for-service plan or an HMO with an HSA. That is, the 
comprehensive plan in the pooling outcome is replaced with an HSA. This 
case might cause greater concern because this would leave employees 
with no alternative. However, employers can vary the generosity of the 
HSA by changing' premiums and the employer contribution to the account.
    A move to an HSA might reflect a trend toward offering lower levels 
of coverage in the face of rising health care costs. Worker 
compensation consists of a combination of cash wages and benefits, and 
will be determined by worker productivity. Tax policy, regulations, and 
employee preferences determine the precise mix between wages and 
benefits. Cutting benefits makes firms less competitive in attracting 
and retaining workers, so firms must have a good reason for cutting 
benefits. The availability of new style of plan does not seem to be 
such a reason unless the firm believed the new plan was more efficient.
    Health economics is a very challenging field, and the models and 
language involved tend to induce headaches. After all the analysis, 
markets will provide the final test: If HSAs work, then they will 
become popular. If they do not work, then they will disappear. After 
all, traditional plans will continue to be available, and decisions are 
usually biased against change. If firms find that HSAs are not a good 
match for their employees, they will drop HSAs. HSAs will likely become 
a useful alternative to less-comprehensive insurance or managed care, 
and they are worth a try.

                               REFERENCES

    Akerlof, G. A. (1970). ``The Market for `Lemons': Quality 
Uncertainty and the Market Mechanism,'' Quarterly Journal of Economics 
84, 488-500.
    Cardon J. and I. Hendel (2001). ``Asymmetric Information in Health 
Insurance: Evidence from the National Medical Expenditure Service,'' 
RAND Journal of Economics 32, No. 3, pp. 408-427.
    Cawley J. and T. Philipson (1999). ``An Empirical Examination of 
Information Barriers to Trade in Insurance,'' American Economic Review 
89, pp. 827-846.
    Cohen, A. (2003). ``Asymmetric Information and Learning: Evidence 
from the Automobile Insurance Market,'' Harvard Law and Economics 
Discussion Paper No. 371.
    Crocker, K. and J. Moran (2003). ``Contracting with Limited 
Commitment: Evidence From Employment-Based Insurance Contracts,'' RAND 
Journal of Economics 34, No. 4, pp. 694-718.
    Rothschild M. and J. Stiglitz (1976). ``Equilibrium in Competitive 
Markets,'' Quarterly Journal of Economics, 90, 629-49.

                               __________
 Prepared Statement of Jeffrey M. Closs, President and CEO, BENU, Inc.

                            1. INTRODUCTION

    Good morning Mr. Chairman and members of the Committee. I am Jeff 
Closs, President and Chief Executive Officer of BENU, Inc. I am pleased 
to be here today to participate in the hearing on ``Expanding Consumer 
Choice and Addressing `Adverse Selection' Concerns in Health 
Insurance''. This topic is exactly what my company, BENU, addresses for 
small and mid-size companies today. We have a relationship with CIGNA 
Health Care and Kaiser Foundation Health Plan in Oregon, and CIGNA and 
Group Health Cooperative in the State of Washington, to offer choice of 
health plan delivery systems for employers to offer to their employees, 
yet reallocate premium to insurers to correct for the adverse selection 
that inevitably occurs.
    Health insurers compete aggressively for the business of the 
employer. What they cannot do is compete aggressively for the consumer. 
Let me give you an example. The marketing executive for Group Health 
Cooperative told us of the wonderful way they treat diabetics. He spent 
considerable time describing their prescription system which flags a 
new insulin prescription, which triggers a nurse to call the diabetic 
person for education on the best methods of monitoring and controlling 
their blood sugar, to make an appointment with a dietician to review 
their nutrition and to schedule follow-up appointments to screen for 
additional diseases. I marveled at the comprehensiveness and 
effectiveness of their care. But when I asked ``Why not encourage all 
diabetics to join Group Health'', he said ``Of course we would love to 
care for all the diabetic people, however, our current payment of the 
average' premium will not cover the cost of treating the diabetic 
person no matter how efficient the care!
    Our health insurance system is broken. The problem is that we 
expect our health insurance carriers to be more than plain old 
insurance. I define insurance as a financial vehicle that spreads the 
risk of financial calamity from rare, unpredictable events--not 
predictable events--among a large group of people. If I tried to apply 
for home insurance while my house was on fire, and I was turned down, 
would you be surprised? Of course not. But when a woman with leukemia 
can't get health insurance, we find that unacceptable. We expect our 
health insurers to be part social program. Do we expect insurers to be 
paid the same rate for bad drivers as they receive for good drivers? Of 
course not. But to engage an insurer to compete for the diabetic as 
well as the healthy we need to compensate them appropriately. The truth 
is we expect our health insurance carriers to be part service plan, 
taking good care of the healthy and chronically ill alike, and part 
social program, spreading the cost of health care evenly among all 
participants. Unsurprisingly they are having a hard time being either.
    Why is consumer choice so important? It is so we can create an 
efficient, competitive consumer market whereby insurers have the 
incentive to provide the service plan component. If insurers are paid 
appropriately, they will have the incentive to enroll the chronically 
ill as well as the healthy since they have the potential to make a 
profit. If they fail to provide high quality care, consumers can `vote 
with his or her feet' and change to another insurer that will care for 
them appropriately. In this model, aligning insurer payment to enrolled 
risk creates an incentive for insurers to provide efficient, high 
quality health care.
    What we need is an ability for consumers to make choices among 
competing delivery systems, to make value judgments between cost and 
quality when assessing their choices. If one system provides better 
care at an appropriate price, they should have the ability to choose 
that delivery system. If the diabetic feels Group Health offers 
superior care for their needs, they should be able to enroll with Group 
Health, without Group Health fearing they are going to create 
unsustainable losses.
    But what if I told you that there was a way to fix this system, 
whereby we could keep the social program aspects of our system, give 
consumers choices they need, while at the same time engage insurers to 
compete for all consumers and control costs for employers? In fact, 
BENU does this today by reallocating premium using risk assessment 
tools available today.
    What is wrong with the current system is not how we FUND health 
care, but how we PAY insurers. We FUND health care by charging everyone 
the same premium for the same plan, no matter how sick they are, what I 
call the AVERAGE COST MODEL. That's how we retain the social program 
part. But instead of paying the INSURER this average cost model 
premium, we should adjust payments to insurers based on the chronic 
illness of those who they enrolled, what I call the RISK-
ADJUSTED MODEL. In other words, employers can still offer employees a 
premium-subsidy based on the AVERAGE COST MODEL, but insurers should be 
paid using a RISK-ADJUSTED MODEL.

                2. EVIDENCE: LACK OF INSURER COMPETITION

    Very few employers offer a choice of health plan, let alone choice 
of insurers. In 2004, 84 percent of all United States employers offered 
only one health plan to their employees\1\. The percentage of employers 
that offer more than one plan increased with employer size; however, in 
most cases, the additional options were simply different plans offered 
by the same insurer. For example, an employer might offer an insurer's 
point-of-service (POS) plan as well as their preferred provider 
organization (PPO) plan. Typically such plans are served by the same 
provider networks, so consumers are not offered competition among 
different delivery systems but rather different financing mechanisms 
for the same delivery system. Very little data exists regarding how 
many employers offer more than one insurer, but it is certainly less 
than 16 percent, which is the percentage of employers that offer more 
than one plan.
---------------------------------------------------------------------------
    \1\ The Kaiser Family Foundation--Health Research and Educational 
Trust, Employer Health Benefits Annual Survey (Chicago and San 
Francisco 2004): 59.
[GRAPHIC] [TIFF OMITTED] T7228.001

                      3. CAUSE: AVERAGE COST MODEL

    Insurers charge a premium that closely matches the average member's 
expected cost to the insurer for the upcoming year. But individual 
members' expected costs vary dramatically. Someone with chronic heart 
failure is expected to cost much more than a healthy twenty-year-old. 
From a financial standpoint, the insurer prefers to enroll the healthy 
and not the chronically ill. Of course, this runs counter to the 
commonly assumed mission of insurers to cover the cost of those who 
need medical care. Every chronically ill member enrolled costs the plan 
more than his or her premium. Therefore, there is a disincentive to 
recruit the chronically ill, and it is this average cost model that 
creates the misaligned incentive\2\.
---------------------------------------------------------------------------
    \2\ R.E. Herzlinger, Consumer-Driven Health Care: Implications for 
Providers, Payers, and Policymakers (San Francisco: Jossey-Bass 
Publishers, 2004); 77-83.
---------------------------------------------------------------------------
                   4. ETIOLOGY: HOW DID WE GET HERE?

    How did the health insurance industry arrive at an average cost 
model? Because health insurance is more than just plain old insurance. 
It is also a service plan and a social program, which make the current 
average cost payment model inefficient and costly.
    First, health insurance is insurance: a financial vehicle that 
spreads the risk of financial calamity from rare, unpredictable events 
among a large pool of members. Health insurance originated in the 
1930's primarily as a means of protecting individuals from unexpected 
hospital costs.\3\ These costs were due primarily to acute conditions, 
and thus unpredictable. A casualty insurance model for health care 
financing was therefore appropriate at the time.
---------------------------------------------------------------------------
    \3\ P. Starr, The Social Transformation of American Medicine (New 
York: Basic Books, Inc. Publishers, 1982): 295-306.
---------------------------------------------------------------------------
    While the casualty model made sense in the 1930's, advances in 
medicine have created a new group of individuals with chronic illnesses 
who live much longer, requiring expensive ongoing care. For the 
insurer, this has meant that loss expectations include not only claims 
due to unpredictable events, but also some due to predictable events as 
well. A patient with kidney disease, who did not have a life expectancy 
of more than a few months in the 30's, now may live many years thanks 
to costly dialysis treatments. This service plan component of modern 
health insurance, in which one pre-pays for anticipated services in the 
coming year, does not exist in other lines of insurance. A purchaser of 
life insurance does not expect to die next year when he buy's term life 
insurance, nor does a homeowner expect that her house will burn down 
when she buys homeowner's insurance. (If they did, it would be fraud!) 
But with health insurance, the insured expects to consume services and 
file claims in the contract year. A component of costs has become 
predictable.
    Modern health insurance is also unique because of the expectation 
that the known healthy will subsidize the cost of care for the sick. A 
recent newspaper article described the case of an uninsured woman who 
was diagnosed with leukemia. The article lamented that she could not 
buy insurance to cover the costs of chemotherapy treatments. This 
sounds reasonable to us. But, by the same token, it would not seem 
reasonable to us for a person whose house is on fire to buy fire 
insurance. Why do we think differently about health insurance? Because 
as a society we view health insurance as part social program.
    The social program aspect of health insurance has created the 
average cost model for insurer payment. It is the social program aspect 
of health insurance that prevents us from charging the person with 
cystic fibrosis his or her full expected cost in the upcoming year. 
Instead, the cost is spread amongst the rest of us who are fortunate 
enough not to have been born with the illness.

        5. RESULT: SINGLE INSURER, FULL-REPLACEMENT HEALTH PLANS

    The average cost model has perpetuated employers' use of a single 
insurer, full-replacement approach in the health insurance they offer 
to employees. Insurers market aggressively to employers, competing for 
a company's entire membership. But if an employer wishes to offer an 
additional insurer's health plan to their employees (called `slice 
business') the original insurer resists, not just because the original 
insurer wants to retain the business, but because they fear enrolling 
the costlier portion of the group, a phenomenon called adverse 
selection:
    Adverse selection makes it difficult, if not impossible, for 
insurers to compete effectively at the consumer level. Historically, 
insurers have pursued slice business as a means of writing more 
business. But this extra business is unprofitable if the new members 
are sicker than the group by which the average cost premium was set. As 
a result, most insurers will not share enrollment of the same employer 
group with a competing insurer.
    Another way of looking at it is that adverse selection occurs when 
consumers are offered a choice of insurers and health plans and are 
exposed to significant cost differences between those plans. A consumer 
who does not expect to need much health care in the coming year will 
not see value in choosing the costlier plan. The chronically ill 
member, who does need a lot of care in the coming year, will likely 
consider that costlier plan.
    When insurers allow slice business, they implement strategies to 
create an equal sprinkling of the healthy and the chronically ill among 
all of the insurers offered. They do this to create an enrollment with 
each insurer with an average cost potential equal to the average cost 
of the group. One way to achieve this is to standardize benefit designs 
across insurers to lessen the cost variance between insurers. Another 
way is to require the employer to subsidize a major portion of the cost 
difference between insurers.
    Unfortunately for employers and employees, the mechanisms insurers 
use to mitigate adverse selection eliminate the reasons why employers 
want to offer choice in the first place: a meaningful choice of 
insurers and plans with meaningful price differences that allow 
consumers to make value assessments between cost and quality. Add to 
this the administrative complexity for employers of offering more than 
one insurer to employees, and one can see why the average cost model 
leads to a single insurer, full-replacement model of health insurance 
coverage.

              6. IMPLICATION: INCREASED HEALTH CARE COSTS

    In a single insurer, full-replacement model, the employer is the 
one choosing the insurer, not the employee. But employers are not as 
effective as employees in making value assessments because individual 
needs and preferences differ. In the late 1980's and early 1990's many 
employers controlled double-digit health care inflation by forcibly 
moving their employees into managed care. With restricted networks and 
tight utilization controls, managed care slowed health care inflation 
dramatically. While many employees did not mind this style of care, 
others disliked the restriction of services that used to be abundantly 
available. The managed care backlash led employers to negotiate with 
their insurers to lessen the utilization controls and to be more 
inclusive in their networks. Employee satisfaction increased, but costs 
again skyrocketed.\4\
---------------------------------------------------------------------------
    \4\ A.C. Enthoven and S.J. Singer, ``The Managed Care Backlash and 
the Task Force in California,'' Health Affairs 17, no. 4 (1998): 95-
110.
---------------------------------------------------------------------------
    The single insurer, full-replacement model of health insurance 
coverage does not control costs. It leads instead to a demand for all-
inclusive networks, forcing the insurer to include the efficient and 
the inefficient, and the good and the poor quality provider. These wide 
networks are not the cohesive provider organizations needed to 
efficiently take care of the chronically ill.\5\
---------------------------------------------------------------------------
    \5\ A.C. Enthoven, ``Employment-Based Health Insurance Is Failing: 
Now What?'' Health Affairs (Web Exclusive May 2003): 237-249.
---------------------------------------------------------------------------
7. SOLUTION: CONSUMER CHOICE OF COMPETING INSURERS AND PAYING INSURERS 
                           FOR RISK ENROLLED

    The best way for employees to become engaged in value assessments 
is to have employers offer them a meaningful choice of health plans 
from competing insurers. Competition among insurers creates incentives 
to provide value to consumers and maximizes consumer satisfaction. If 
consumers are exposed to the true cost differences between insurers, 
they will have a reason to choose less expensive delivery systems or 
costlier options if they see value in doing so. This is called a 
defined-contribution approach because employers offer all employees a 
fixed-dollar subsidy to their health plan choice. This approach is 
necessary for consumers to make value assessments. It yields savings 
for the employer by allowing them to fund only the lowest cost plan, 
employees then buy-up to the options they desire.
      
    [GRAPHIC] [TIFF OMITTED] T7228.002
    

    Figure 2 demonstrates how an employer who currently offers only one 
moderately priced, one-size-fits-all PPO can save significantly by 
introducing a lower cost, comprehensive HMO plan from a competing 
insurer. In the single insurer situation, the PPO plan premiums are 
$250 and the employer pays 90 percent of that, or $225. In the package 
with choice, the HMO costs $200 per month and the PPO is still $250. If 
the employer subsidizes $200, then it yields a $25 savings per covered 
employee. The employees now have a no-cost option, but they can keep 
the PPO if they are willing to spend $50 per month, the cost 
differential between the plans.
    Insurers, however, need to be kept whole in this process. While 
average cost payments from employers can be maintained (social 
program), an intermediary, such as BENU, must reallocate payments to 
insurers proportional to chronic illness burden, or `risk', that 
enrolls (service plan). In the example in Figure 2, healthier employees 
will be attracted to the low cost HMO option, raising the average per-
employee-cost of those remaining in the PPO. Risk assessment tools that 
predict future costs based on clinical diagnoses can reallocate the 
average cost rates funded by employers into risk-based rates paid to 
insurers.
    Paying insurers risk-adjusted rates allows employers to offer a 
choice of insurers while pursuing a defined contribution strategy that 
was not sustainable when the employer paid the insurer the average 
cost. Employers protect themselves, but employees are empowered to make 
the value assessments critical for efficient competitive markets.

      8. ADDITIONAL BENEFIT: MORE ATTENTION TO THE CHRONICALLY ILL

    When employers offer choice to employees without risk adjusting 
payments to insurers, powerful incentives are created for insurers to 
figure out how to enroll low cost, healthy members and not to enroll 
high cost, chronically ill members. One cannot blame insurers for this 
strategy. When employer's offer a choice of insurers in an average cost 
model, it creates financial calamity for insurers that actively recruit 
the chronically ill. Consider an HMO that may have an excellent 
diabetes care pathway, including an early detection system that 
identifies new enrollees with insulin prescriptions, an education 
program taught by nurses, a nutrition program in which a dietician 
contacts patients with nutritional advice, and a followup care program 
with specialists who help with co-morbid disease prevention. The HMO 
then markets this excellent program to an employer that will offer it 
to employees. But when it comes time to enroll members, there is no 
incentive for the insurer to enroll the diabetics. Why? Because the 
average premium is not sufficient to cover the costs of the diabetic, 
no matter how good the care is.
    If employers pay insurers premiums commensurate with the chronic 
illness burden of enrollees, it will actively encourage these plans to 
compete for all members, effectively removing the underwriting profit 
incentive. Insurers will have the incentive to provide high quality 
care to the chronically ill because they represent greater revenue. If 
they fail to do so, the chronically ill member can vote with his or her 
feet and change to another insurer that will care for them 
appropriately. In this model, aligning insurer payment with enrolled 
risk creates efficient, high quality, cost effective health care.

    9. SOLUTION FOR EMPLOYERS: BENU'S RISK-ADIUSTED PREMIUM PAYMENTS

    BENU is currently the only independent 3rd party market-maker that 
allows employers to maintain average cost premiums for their employees, 
yet pays risk-based premiums to insurers. The key to BENU's method is 
to present rates to employers that the insurer would quote if each plan 
were to receive the entire enrollment, what BENU calls the group 
neutral risk level. After enrollment, BENU calculates the insurer-
enrolled risk level and adjusts the premium paid to each insurer 
proportionately. Essentially, the rates that BENU pays the insurers are 
what the insurers would have quoted had they known in advance the 
enrollment they eventually received.
    The rates BENU charges and collects from the employer for insurers 
differ from the rates that BENU pays the insurer, but the total premium 
the employer pays BENU equals the total premium paid to insurers.
    Figure 3 shows how the average enrolled risk for insurers can 
differ from the group neutral risk. 
[GRAPHIC] [TIFF OMITTED] T7228.003

                    10. RESULTS: EXPERIENCE AT BENU

    BENU currently operates in two states, Oregon and Washington. In 
Washington we currently offer Group Health Cooperative and CIGNA Health 
Care, while in Oregon we offer Kaiser Foundation Health Plan of the 
Northwest and CIGNA Health Care.
    How does BENU assess risk? BENU uses predictive modeling tools 
developed over the last decade. Specifically, BENU uses DxCG software, 
the same company that the Medicare program currently uses in 
determining payment to insurers in the Medicare+Choice program. The 
software was developed by using claims data from a large data set of 
over two million members over a period of 2 years. By tracking 
diagnoses that are recorded for members in the first year with costs 
those members generate in the second year, a statistical model was 
created where future year costs can be predicted based on prior year 
diagnoses. To use the software, one simply enters the diagnoses for 
each member and the software will generate relative cost factors for 
each member. We call this a prospective risk factor.
    For example, a member diagnosed with diabetes in the first year may 
have a prospective risk factor of 3.2. This means that next year, we 
can expect, on average, that this member will incur 3.2 times the cost 
of the average cost per member of the two million members in the 
original reference data set.

[GRAPHIC] [TIFF OMITTED] T7228.004


    Figure 4 demonstrates the amount by which prospective risk factors 
can vary for a typical BENU employer. The graph shows prospective risk 
factors for each member in a 275 member group, ordered from highest to 
lowest. The prospective risk factor at the extreme left is 15.33, 
representing a member diagnosed with cancer. The factor at the extreme 
right is about 0.08, representing a completely healthy individual that 
never needed to see a physician. The most costly member in this group 
is expected to cost 192 times the cost of the least costly member in 
this group. This example demonstrates a 192-fold difference between 
what the costliest and least costly member is expected to cost. Yet the 
insurer is paid the average premium whether the member with the 
prospective risk factor of 15.33 or the one with 0.08 enrolls.
    How much has BENU reallocated premium among insurers? Figure 5 
answers this, showing the results for the first 13 employers to 
purchase health insurance through BENU.

[GRAPHIC] [TIFF OMITTED] T7228.005


    Notice how in several groups the adjustment altered premium more 
than 5 percent, which is significant because insurers operate on less 
than 5 percent profit margin. Adverse selection can easily turn slice 
business into an unprofitable venture. Without risk adjustment, the 
insurer that received the higher proportion of chronically ill would be 
forced to raise premium, making the cost share to the employee higher, 
further exacerbating the adverse selection, eventually making the 
affected insurer leave the offering--a situation called the death-
spiral. Risk adjusted premiums to insurers prevent the death-spiral.

            11. RISK ADJUSTMENT IN OTHER GOVERNMENT PROGRAMS

    As mentioned above, BENU uses the same predictive modeling software 
as currently used by the Medicare program in determining payments to 
insurers in the Medicare+Choice program. Several Medicaid programs 
across the country are using similar predictive modeling tools in their 
programs as well. If participants in these programs have a choice of 
insurers, it bodes well for creating efficient health care since 
insurers will actively compete for all participants, the chronically 
ill as well as the healthy, and yet create economic pressures (i.e., 
loss of patients) on the most costly alternatives to innovate to 
contain and reduce cost. The most efficient plans gain market share and 
are rewarded for being economical.

                             12. CONCLUSION

    Our current system of paying insurers perpetuates a single-insurer 
full-replacement model of health insurance coverage that leads to 
higher costs. While the current system may be an appropriate way to 
fund health care, it is not an appropriate way to pay insurers. BENU's 
risk adjustment method sensibly reallocates the funding of health care 
to pay insurers in a manner that creates a competitive consumer market 
that lowers costs for employers, satisfies employees and motivates 
insurers to provide value for the chronically ill.

                               __________
    Prepared Statement of Linda J. Blumberg, Ph.D., Senior Research 
                     Associate, The Urban Institute

    Mr. Chairman, Mr. Stark, and distinguished Members of the 
Committee: Thank you for inviting me to share my views on adverse 
selection in health insurance and its implications when expanding 
consumer choice in the private market. The views I express are mine 
alone and should not be attributed to the Urban Institute or any of its 
sponsors.
    I applaud the Committee for taking the time to carefully consider 
these issues, which are of paramount importance to individuals' access 
to health care coverage and medical services. In brief, my main points 
are:
     In order to understand health insurance markets, there is 
one overarching fact that must be understood. The distribution of 
health expenditures is highly skewed, meaning that a small fraction of 
individuals account for a large share of total health expenditures. 
Because of this fact, the gains to insurers of excluding high cost 
people swamp any possible savings from efficiently managing the care of 
enrollees. The incentives for insurers to avoid high cost/high risk 
enrollees are therefore tremendous.
     Greater risk segmentation of the market means setting 
individuals' health insurance premiums to more closely reflect each 
individuals' expected health care costs. Conversely, greater risk 
pooling implies increasing the extent to which individuals with 
different expected health care spending levels are brought together 
when determining premiums. Providing new health insurance options is 
one way, intentionally or not, that the extent of risk segmentation can 
be increased.
     Reforms that increase risk segmentation are appealing to 
some because they promise, and sometimes deliver, lower premiums for 
currently healthy persons and because the majority of people are 
healthy. However, gains from segmenting healthy groups can occur only 
if premium costs for the unhealthy are increased, or if the unhealthy 
are excluded from the market to a greater extent than is true today.
     Examples of proposed and already implemented reforms that 
will increase risk segmentation in private markets are: health savings 
accounts (HSAs); tax deductions for the premiums of high deductible 
policies associated with HSAs in the private non-group market; 
association health plans (AHPs); and tax credits for the purchase of 
non-group insurance policies.
     While risk segmentation increases the costs of coverage 
for the unhealthy, the isolated instances where states have forced 
greater risk pooling have not been successful either. Efforts at 
pooling have been limited to a small population base and have been 
foiled by individuals and groups that opt out of our voluntary private 
insurance markets.
     Addressing the problem will require subsidization of the 
costs associated with high cost individuals, with the financing source 
being independent of enrollment in health insurance--ideally, all 
taxpayers. In this way, the unhealthy could be protected from bearing 
the tremendous costs of their own care while there would be little to 
no disincentive for the healthy to give up coverage.
     Three examples of policies that would move us closer to 
such a paradigm are:
          Dramatically increasing funding for State high risk 
        pools and making the coverage both more comprehensive and 
        easier to access;
          Having the Federal Government take on a roll as 
        public reinsurer, particularly for the private non-group market 
        and for modest sized employers;
          A more comprehensive strategy would allow groups to 
        continue to purchase insurance in existing markets under 
        existing insurance rules, while each State provides structured 
        insurance purchasing pools. Through these new pools, employers 
        and individuals could enroll in private health insurance plans 
        at premiums that reflect the average cost of all insured 
        persons in the state.
     For the following reasons, introducing greater choice 
within existing insurance pools will not solve the problems I 
described. In fact, doing so will likely exacerbate them, even given 
the best available risk adjustment mechanisms:
          First, it is not sufficient to spread risks only 
        within a particular insurance pool.
          Second, benefit package design is an effective tool 
        for segmenting insurance pools by health care risk--offering 
        less than comprehensive insurance will tend to attract 
        healthier enrollees.
          Third, in private markets, where differences in 
        actuarial value of plans can be quite larger and where people 
        have the opportunity to opt in or out of the market, risk 
        adjustment becomes substantially more difficult. Risk 
        adjustment has been used in the Medicare program and is 
        universally considered to be inadequate.
          And finally, it is not even clear that employers will 
        have a strong incentive to want to risk adjust across plans. 
        Although most employers want to lookout for the well-being of 
        all their workers, they face incentives to keep health care 
        premiums down while keeping their highest paid workers 
        satisfied. HSAs may provide employers with an effective tool 
        for responding to these incentives, but place a greater share 
        of the health care financing burden directly on the sick while 
        higher paid employees can be compensated via the tax subsidy.
    Further segmentation of risk will not improve social welfare in the 
U.S. Addressing the health care needs of all Americans and protecting 
access to needed services for our most vulnerable populations--those 
with serious health problems and those with modest incomes--will 
require broad-based subsidization of both those with high medical costs 
and income-related protection for those unable to afford even an 
average priced insurance policy.

       I. THE SCOPE OF RISK-RELATED HEALTH INSURANCE PROBLEMS IN 
                           THE CURRENT MARKET

    While estimates differ, by all accounts the number of uninsured 
persons in the U.S. is large and prone to grow, both in absolute terms 
as the population increases and as a percentage of the population. The 
most recent estimate based upon the 2004 March Current Population 
Survey is 45 million uninsured persons below age 65, or almost 18 
percent of the non-elderly population. There is a substantial body of 
evidence that shows that the uninsured have reduced access to medical 
care. Many researchers have also determined that those without coverage 
have worse outcomes in the event of` an injury or illness.
    The distribution of health expenditures is highly skewed. Only a 
small fraction of individuals account for most of our nation's health 
care spending. In fact, the top 10 percent of the population, ranked by 
expenditures, accounts for about 70 percent of total expenditures in 
the country.\1\ The lowest 50 percent of spenders account for only 3 
percent of expenditures. Because of this, insurers have strong 
incentives to avoid enrolling high cost individuals and to aggressively 
pursue enrollment of low cost individuals. The potential gains to 
insurers of excluding the high cost cases swamp any possible savings 
from efficiently managing the care of enrollees. The small group and 
individual insurance markets are of greatest concern with regard to 
adverse selection, since their variability of expenditures year-to-year 
is much higher than for large groups.
---------------------------------------------------------------------------
    \1\ ML Berk and AC Monheit. 2001. ``The Concentration of Health 
Care Expenditures, Revisited.'' Health Affairs. March/April; 20(2): 9-
18.
---------------------------------------------------------------------------
    Fears of adverse selection and the natural drive to maximize 
profits, drives insurers in unregulated markets to use strategic 
behavior in the pursuit of a disproportionate share of low cost 
enrollees. These strategic behaviors can take a variety of forms, 
including: excluding preexisting medical conditions from coverage for 
defined periods; attaching riders that exclude specific conditions, 
procedures, or body parts from coverage for the life of the policy; 
engaging in medical underwriting (the process whereby insurers assess 
an applicant's relative health risk and then charge higher premiums to 
those whose risk is deemed to be higher than normal); or refusing to 
sell an applicant insurance altogether.\2\ Another technique is 
designing insurance benefit packages in such a way as to be more 
attractive to healthy persons than to unhealthy ones. Harvard health 
economist Joseph Newhouse demonstrated how insurers, in order to 
protect themselves from adverse selection, can offer less than complete 
insurance.\3\ This approach can take the form of offering coverage with 
higher deductibles, higher limits on out-of-pocket liability, tighter 
provider networks, and caps on benefits, among other things. In 
essence, insurers use lower value benefit packages to help them 
selectively appeal to the low risk.
---------------------------------------------------------------------------
    \2\ MA Hall. 2000. ``An Evaluation of New York's Reform Law.'' 
Journal of Health Politics, Policy and Law. 25(1): 71-99; K Pollitz, R 
Sorian and K Thomas. 2001. ``How Accessible is Individual Health 
Insurance for Consumers in Less-Than-Perfect Health?'' Menlo Park, CA: 
Henry J. Kaiser Family Foundation; U.S. General Accounting Office 
(GAO). 1996. Private Health Insurance: Millions Relying on Individual 
Market Face Cost and Coverage Trade-Offs. HEHS-97-8. Washington, DC: 
U.S. General Accounting Office.
    \3\ JP Newhouse. 1996. ``Reimbursing Health Plans and Health 
Providers: Efficiency in Production Versus Selection.'' Journal of 
Economic Literature. 34: 1236-1263.
---------------------------------------------------------------------------
    The result of these various strategies is to create a market that 
is segmented by health care risk. This leads to markets in which 
premiums faced by generally healthy persons are determined as a 
function of the expected costs of a similarly healthy population, and 
the premiums for the unhealthy are determined as a function of the 
expected costs of the similarly unhealthy. The markets with the 
greatest risk segmentation are those for small employers and for 
individual purchasers, the markets where the insured groups are 
smallest and the year-to-year variation in expenditures is the 
greatest. While market segmentation benefits the currently healthy by 
providing them lower premiums than they would face otherwise, it 
increases the premiums faced by the relatively unhealthy, and sometimes 
excludes them from the insurance market entirely.
    Risk segmentation has made insurance more affordable for the 
healthy and less affordable and accessible to the sick, contrary to the 
classic theory posited by Rothschild and Stiglitz\4\ This result is 
consistent with the framework posed by Newhouse.\5\
---------------------------------------------------------------------------
    \4\ M Rothschild and JE Stiglitz. 1976. ``Equilibrium in 
Competitive Insurance Markets: An Essay on the Economics of Imperfect 
Information.'' Quarterly Journal of Economics. 90(4):629-50.
    \5\ JP Newhouse. 1996. ``Reimbursing Health Plans and Health 
Providers: Efficiency in Production versus Selection.'' Journal of 
Economic Literature. 34(3):1236-63.
---------------------------------------------------------------------------
    The best example of how risk selection can lead to barriers to 
coverage for the unhealthy can be found in the private non-group, 
insurance market. With a limited number of exceptions, State laws 
permit non-group insurers to exclude individuals from coverage entirely 
based upon health status and to set premiums as a function of health 
status. They may also discontinue particular insurance products as a 
consequence of the insurance pool becoming too expensive, and only make 
alternative products available to the healthier individuals that had 
been in that pool. In many states insurers are also allowed to severely 
limit any coverage related to a pre-existing condition. For example, a 
study of the accessibility of non-group insurance for people in less 
than perfect health found examples of insurers offering one applicant a 
policy which excluded any care related to his circulatory system, and 
another excluding his entire respiratory system.\6\
---------------------------------------------------------------------------
    \6\ K Pollitz, R Sorian and K Thomas. 2001. ``How Accessible is 
Individual Health Insurance for Consumers in Less-Than-Perfect 
Health?'' Menlo Park, CA: Henry J. Kaiser Family Foundation.
---------------------------------------------------------------------------
    A recent empirical study published in the journal Inquiry found 
that the probability of buying non-group insurance goes down 
significantly as a person's health deteriorates.\7\ Using this 
information to adjust for selection bias, an important econometric 
correction that has been neglected in all other studies of premiums in 
the non-group market, the authors also found that people with 
significant health problems would face non-group premiums roughly 50 
percent higher than their healthier counterparts. Without the 
adjustment for selection bias, the data suggest that premiums do not 
vary with health status and support the misleading inference that poor 
health does not make the cost of non-group insurance unaffordable.
---------------------------------------------------------------------------
    \7\ J Hadley and JD Reschovsky. 2003. ``Health and the Cost of 
Nongroup Insurance.'' Inquiry. Fall; 40:235-253.
---------------------------------------------------------------------------
    Risk selection incentives and dynamics can also be found in 
situations where individuals are offered a choice of health insurance 
benefit packages with significantly different actuarial values. While 
with most other products, choice is considered beneficial to all 
consumers, the case of health insurance benefit packages is 
considerably more complicated. Initially, multiple options allow 
individuals to choose the package that is most consistent with their 
preferences. However, the tendency for individuals' preferences to be 
highly correlated with their health care risk means that choice in this 
market will tend to separate individuals into different packages by 
their health status. Due to the pricing differences that result, 
certain options may eventually be priced out of existence, because they 
become too expensive for people to afford. The end result may very well 
be a market that has no more choice than it had originally, but with 
the options tailored to those preferring less comprehensive coverage.
    An example of this in the group insurance market can be found in 
the recent history of the Federal Employees Health Benefits Plan 
(FEHBP). For years, Federal employees had a choice of a ``high option'' 
Blue Cross coverage and a ``standard option'' with a slightly higher 
deductible and a few other limitations. For the typical employee, high 
option was worth a little more, and, initially, premiums were slightly 
higher. Young, healthy employees risked having to pay the higher 
deductible in exchange for the small premium savings. Older, sicker 
employees preferred the high option. But the premium difference grew 
larger over time as more healthy people shunned the high option. When 
last offered in 2001, the high option family premium was $1500 more 
than the standard option. In 2002, the high option was dropped from the 
plan.\8\
---------------------------------------------------------------------------
    \8\ L Burman and LJ Blumberg. 2003 ``HSAs Won't Cure Medicare's 
Ills.'' The Urban Institute. November; http://www.urban.org/
url.cfm?ID=1000578.
---------------------------------------------------------------------------
    Over the last 10 to 15 years, well-intentioned reformers, hoping to 
provide protections in private insurance markets for high risk 
individuals and groups, have enacted legislative mechanisms for forcing 
more risk pooling than private insurance markets would have done on 
their own. In their most extreme forms, such as pure community rating, 
and particularly within the private non-group insurance market, such 
approaches appear to have increased premiums and have led to a 
reduction in the number of healthy individuals choosing to purchase 
health insurance. In some cases, the effect has been sufficiently great 
that the insurance in the community rated market may not be sustainable 
in the long run.\9\
---------------------------------------------------------------------------
    \9\ See, for example, AC Monheit, JC Cantor, M Koller and KS Fox. 
2004. ``Community Rating And Sustainable Individual Health Insurance 
Markets In New Jersey.'' Health Affairs. July/August; 23(4): 167-175.
---------------------------------------------------------------------------
     II. RATIONALE FOR CHANGING OUR HISTORICAL APPROACH TO POOLING 
                            HEALTH CARE RISK

    Equity judgments inevitably arise in any discussion of the optimal 
level of risk pooling. Many would consider lack of available coverage 
for high risk people as inequitable, while others consider it 
inequitable to force healthy persons to pay higher premiums than they 
would under stronger market segmentation conditions. I argue that 
neither our historical experience with the largely unregulated market 
outcome of risk segmentation nor with forced pooling within small group 
and non-group markets truly serve to maximize social welfare for the 
following reasons:
    First, we know that individuals with their own medical problems or 
who have family members with medical problems often have difficulty 
accessing needed care if they do not have employer-based or public 
insurance available to them. But, additionally, all individuals age and 
medical expenses tend to increase over time as a consequence, and 
currently healthy people might face high costs someday because of 
illness or injury. With segmented markets, their premiums would then 
rise, perhaps beyond their ability to pay. Broad-based pooling 
preserves access to reasonably priced health insurance over time. This 
gives even currently healthy people reason other than pure altruism to 
be concerned with effective access to care for the sick, and makes the 
pursuit of risk segmentation much less than ideal.
    Second, competition to avoid high-cost groups, and benefit designs 
structured to place heavier financial burdens on the sick can foreclose 
options that most consumers are willing to pay for if priced on a 
broad-based average.\10\ This is an efficiency loss to the society. If 
the risk pool were guaranteed to be sufficiently broad-based, consumers 
might be eager to buy coverage that was more comprehensive, for 
example, shorter pre-existing condition exclusion periods or lowering 
out-of-pocket maximums. Additionally, pharmaceutical benefits and 
rehabilitation benefits in the non-group market are often either 
severely limited or excluded altogether. Because there are many more 
healthy than sick. people, these types of options could be available 
for a small premium increase--if (and this is a big if) the size of the 
pool over which these risks were to be spread was sufficiently large.
---------------------------------------------------------------------------
    \10\ LJ Blumberg and LM Nichols. 1996. ``First, Do No Harm: 
Developing Health Insurance Market Reform Packages.'' Health Affairs. 
Fall; 15(3): 35-53.
---------------------------------------------------------------------------
    Third, sporadic efforts across various states to force pooling in 
the smallest of private health insurance markets--those for small 
groups and individual purchasers--have often not been constructive 
largely because the financial burden for covering the high cost in 
these markets can be avoided completely by the healthy by simply opting 
out of the market and not buying coverage there. The price to consumers 
of health insurance in these markets is a function of the health care 
risk of those who voluntarily decide to enter them. Because the sick, 
having greater health care needs, are more likely to enroll in 
insurance, and because these markets are quite small in total, placing 
the burden of the excess costs associated with bad health entirely on 
those voluntarily enrolling in these markets is a primary cause of 
their ineffectiveness at providing worthwhile coverage to individuals 
of all health care statuses.
    I suggest that none of our policy efforts to date have focused 
properly on the source of the risk issues in our small group and 
individual markets. Therefore, sticking with what we have, or 
exacerbating risk segmentation relative to what we see in markets today 
will not solve our problems either. It is not that broad based 
spreading of health care risk is inappropriate, as demonstrated by the 
fact that all individuals have some stake in maintaining access to 
coverage for the unhealthy and that market efficiencies result from the 
battle of insurers to avoid adverse selection. The problem is that our 
efforts at pooling thus far have been limited to too small of a 
population base and have been foiled by the ability of individuals and 
groups to opt out of sharing risk by exiting particular insurance 
markets, a dynamic that we know is related to expected health care 
risk.
    Addressing the problem, therefore, will require subsidization of 
the costs associated with high cost/high risk individuals, with the 
financing source for doing so being independent of enrollment in health 
insurance. Ideally, the source of funding would be all taxpayers. In 
this way, the unhealthy could be protected from bearing the tremendous 
costs of their own care precisely at the time that they are both 
medically and financially at greatest risk, while there would be little 
to no disincentive for the healthy to avoid or drop health insurance 
coverage due to the presence of high cost cases.

III. POLICIES WHICH WOULD ADDRESS OUR NEED FOR EFFECTIVE INSURANCE FOR 
                         ALL HEALTH CARE RISKS

    There are a number of policy options that would either begin to 
lead us toward such a paradigm or move us most of the way there, 
depending upon our current level of ambition and willingness to pay.
    First, we can dramatically increase funding for State high risk 
pools and make the coverage both more comprehensive and easier to 
access. These pools are available to individuals who have been refused 
insurance coverage in the private market, and who do not have offers of 
employer-sponsored insurance. While many states currently have high 
risk pools, due to the limited public funding through State sources 
(frequently premium taxes on private insurance policies), these pools 
may have enrollment caps and usually charge premiums that are well in 
excess of standard policies in the private market.\11\ Some high risk 
pools offer very limited benefit packages and maintain pre-existing 
condition exclusion periods. This means that, in order to enroll, some 
individuals with high cost medical conditions must be able to afford to 
pay the high risk pool premium and, simultaneously, all of their 
medical costs out-of-pocket for a year. All of these limitations hamper 
the pools' effectiveness in absorbing risk from the private market. 
However, broadening the base for financing these pools, loosening 
eligibility criteria for enrollment, making the insurance policies 
themselves more comprehensive, and offering income-related premiums 
have the potential to make these high risk pools powerful escape valves 
for the high cost in private insurance markets.\12\ Allowing employers 
in the small group market in particular to buy their high risk workers 
into well-funded high risk pools would decrease the level and 
variability in the expenditures of the remaining small group workers 
and, consequently, would lower their premiums. The cost of subsidizing 
the medical care of the high risk could be spread across the entire 
population, using a broad-based tax.
---------------------------------------------------------------------------
    \11\ ``D Chollet. 2002. ``Perspective: Expanding Individual Health 
Insurance Coverage: Are High Risk Pools the Answer?'' Health Affairs, 
Web Exclusive, October; W349-W352.
    \12\ LJ Blumberg and LM Nichols. 1996. ``First, Do No Harm: 
Developing Health Insurance Market Reform Packages.'' Health Affairs. 
Fall; 15(3): 35-53.
---------------------------------------------------------------------------
    A second strategy is to have the Federal Government take on a roll 
as public reinsurer, particularly for the private non-group market and 
for modest-sized employers. In this capacity, the government could 
agree to absorb a percentage of the costs of high cost cases, once a 
threshold level of health expenditures had been reached.\13\ 
Reinsurance of this type would not only lower private premiums 
directly, due to the broader financing of these expensive cases, but 
would reduce the variance in expenditures considerably and therefore 
should reduce risk premiums charged by private insurers.\14\ Focusing 
on small employers and the non-group market could target government 
spending where costs are highest and insurance markets most unstable.
---------------------------------------------------------------------------
    \13\ K Swartz. 2003. ``Reinsuring Risk to Increase Access to Health 
Insurance.'' AEA Papers and Proceeding. May; 93(2).
    \14\ LJ Blumberg and J Holahan. 2004. ``Government Reinsurer: 
Potential Impacts on Public and Private, Spending.'' Inquiry. 41(2): 
130-143.
---------------------------------------------------------------------------
    While private reinsurance does exist in some markets, such products 
do not address the critical issues which are the focus of a public 
reinsurance approach.\15\ Voluntary private reinsurance policies are 
subject to the same selection concerns as are the insurers that they 
are designed to cover. Those insurers who have historically attracted 
high cost individuals and high cost groups find the private reinsurance 
products either very expensive or inaccessible to them. In addition, 
the costs of the reinsurance products must be passed back to the 
individuals and groups purchasing the original insurance, again 
creating incentives for low risk individuals and groups to avoid the 
burden of risk sharing by opting out of the insurance completely.
---------------------------------------------------------------------------
    \15\ LJ Blumberg and J Holahan. 2004. ``Government Reinsurer: 
Potential Impacts on Public and Private Spending.'' Inquiry. 41(2): 
130-143.
---------------------------------------------------------------------------
    A third option is to develop purchasing pools which would combine 
the concepts of administrative economies of scale with direct 
subsidization of the high cost.\16\ This proposal allows groups wishing 
to purchase insurance in existing markets under existing insurance 
rules to continue to do so. However, it would provide structured 
insurance purchasing pools in each state, through which employers and 
individuals could enroll in private health insurance plans at premiums 
that reflect the average cost of all insured persons in the state. 
Broad-based government funding sources would compensate insurers for 
the difference between the cost of actual enrollees and the statewide 
average cost.
---------------------------------------------------------------------------
    \16\ J Holahan, L Nichols, and LJ Blumberg. 2001. ``Expanding 
Health Insurance Coverage: A New Federal/State Approach.'' In Covering 
America: Real Remedies for the Uninsured, J Meyer and E Wicks, eds., 
Economic and Social Research Institute.
---------------------------------------------------------------------------
    Comprehensively addressing the problems of the uninsured would 
require additional subsidization of the low-income population, aside 
from techniques, such as those described above, which are aimed at 
addressing the problems of risk selection.

 IV. POLICIES THAT ARE LIKELY TO INCREASE RISK SEGMENTATION IN PRIVATE 
                                MARKETS

    A number of policies, some already written into law, would tend to 
increase the segmentation of health care risk in today's insurance 
markets and/or would increase the share of medical expenses left 
uncovered by health insurance, without providing protections for the 
high risk or the low income. The implications of implementing such 
changes could be very harmful to these already vulnerable populations. 
Some could come with sizable Federal price tags, without necessarily 
increasing health care coverage on net.
    Health Savings Accounts (HSAs), passed into law along with Medicare 
legislation last year, are one such example. The legislation provides a 
generous tax incentive for certain individuals to seek out high 
deductible health insurance policies. Individuals and families buying 
these policies, either through their employers or independently, can 
make tax-deductible contributions into an HSA account. Annual 
contributions are capped at the amount of the annual deductible for the 
plan in which they enroll. Money in the account and any earnings are 
tax-free if used to cover medical costs.
    These accounts are most attractive to high income people, and those 
with low expected health expenses. The tax subsidy is greatest for 
those in the highest marginal tax bracket and is of little or no value 
at all to those who do not owe income tax. Higher income individuals 
are also better able to cover the costs of a high deductible, should 
significant medical expenses be incurred. Additionally, those who do 
not expect to have much in the way of health expenses will be attracted 
to HSAs by the ability to accrue funds tax free that they can use for a 
broad array of health related expenses that are not reimbursable by 
insurance (e.g., non-prescription medications, eyeglasses, cosmetic 
surgery). Those without substantial health care needs may also be 
attracted to HSAs because they can be effectively used as an additional 
IRA, with no penalty applied if the funds are spent for non-health 
related purposes after 65. Young, healthy individuals may even choose 
to use employer contributions to their HSAs for current non-health 
related expenses, after paying a 10 percent penalty and income taxes on 
the funds; a perk unavailable to those enrolled in traditional 
comprehensive insurance plans.
    The idea of lower premiums under high deductible policies also make 
these recent reforms attractive to some employer purchasers. However, 
the savings can only be modest for a fixed group of enrollees. Because 
the majority of spending is attributable to the small share of 
individuals with very large medical expenses, increasing deductibles 
even to $1,000 or $2,000 from currently typical levels will not 
decrease premiums dollar for dollar. The vast majority of medical 
spending still will occur above even those higher deductibles,\17\ 
therefore premium savings can only be modest. The reduction in premiums 
from moving to higher deductible plans cannot go far in encouraging 
more employers to offer insurance or more individuals to take it up.
---------------------------------------------------------------------------
    \17\ LJ Blumberg and L Burman. 2004. ``Most Households' Medical 
Expenses Exceed HSA Deductibles.'' Tax Analysts Tax Facts. Tax Policy 
Center: Urban Institute and Brookings Institution. August; http://
www.urban.org/url.cfm?ID=1000678.
---------------------------------------------------------------------------
    The real premium savings from HSAs can occur by altering the mix of 
individuals who purchase coverage. By providing incentives for healthy 
individuals and groups to purchase HSAs with high deductible policies, 
insurance risk pools can be further segmented by health status. The 
average medical costs of those purchasing the new plans will be 
substantially lower if the high risk population is left in more 
traditional comprehensive plans. The practical effect, however, is that 
the most vulnerable populations (the sick and the low income) are left 
bearing a greater direct burden of their health expenses.
    Another proposal, contained in H.R. 3901, and included in the 
President's fiscal year 2005 budget,\18\ would make the premiums 
associated with individually purchased high deductible health insurance 
plans deductible from income taxation. The deduction would be allowed 
regardless of whether other itemized deductions are taken. This new 
deduction would be available for policies purchased with HSAs.
---------------------------------------------------------------------------
    \18\ Department of the Treasury. 2004. ``General Explanations of 
the Administration's Fiscal Year 2005 Revenue Proposals.'' February; 
http://www.treas.gov/offices/tax-policy/library/bluebk04.pdf.
---------------------------------------------------------------------------
    This policy would provide a non-group insurance product whose tax 
advantage is almost as great as that available in the group market and 
which is most attractive to those with high incomes and low health care 
risk. Low cost/high-income purchasers, armed with yet another subsidy, 
would be likely to find price advantages in most states' non-group 
insurance markets. But as low cost purchasers leave the group market, 
the average cost of those staying in the group market will rise, making 
group insurance more difficult to afford for higher risk and lower 
income populations. In addition, since small employers and higher wage 
employees will be able to get tax breaks for the high-deductible health 
insurance purchased individually in the non-group market even if the 
firm does not provide coverage to their other employees, there will be 
even less incentive for them to take on the hassle, expense, and risk 
of offering insurance to their workers. The net result could be less 
insurance coverage among small businesses in particular.
    Legislation to create Association Health Plans (AHPs) and similar 
employer-based risk-pooling entities have also been introduced 
repeatedly over the years, most recently in 2003. Supporters of AHPs 
hope the legislation will encourage professional and trade associations 
to offer health insurance plans, thereby providing an alternative 
source of coverage and new mechanisms for pooling health insurance risk 
for employers. They expect such mechanisms to prove more attractive to 
small employers who currently do not offer health insurance, thereby 
increasing the number of workers with coverage. However, legislation 
promoting AHPs generally includes Federal exemptions from some State 
regulations governing existing commercial insurance products. As a 
consequence, the new plans would likely be more effective than existing 
commercial insurance products at segmenting health care risk for 
purposes of setting premiums. They will tend to attract relatively 
healthy individuals and groups, and will tend to increase premiums 
faced by those remaining in the residual commercial insurance market. 
Some (the relatively healthy) can be expected to gain from such 
policies, while others (the less healthy) will tend to lose. Estimates 
of the impact of AHPs suggest that while some employers will respond by 
offering coverage for the first time, others will stop offering the 
plans that they sponsored prior to reform. Accordingly, there would be 
virtually no net change in health insurance coverage.\19\
---------------------------------------------------------------------------
    \19\ LJ Blumberg and Y Shen. 2004. The Effects of Introducing 
Federally Licensed Association Health Plans in California. A 
Quantitative Analysis. Report prepared for the California HealthCare 
Foundation. www.chcf.org.
---------------------------------------------------------------------------
    New tax credits to subsidize the purchase of non-group insurance 
policies will also tend to increase market segmentation. As is the case 
discussed above with regard to deductibility of high deductible 
policies associated with HSAs, new incentives that draw individuals out 
of the employer-based market and into the private non-group market as 
it is structured today, tend to exacerbate segmentation. This occurs by 
virtue of the fact that there is less risk pooling in most states' non-
group markets than in employer-based markets. In addition, tax credit 
proposals do not usually vary the amount of the subsidy provided with 
the health status of the recipient; doing so is widely considered too 
administratively difficult for the IRS. But as discussed earlier, 
insurance premiums and outright access to coverage in this market do 
vary substantially with health status. Consequently, a tax credit that 
might cover a significant share of a premium for a healthy young person 
would most likely cover a much smaller share for someone with a current 
or past health problem.\20\ Risk-pool issues may be a primary factor in 
the outcome of such policy proposals, with some individuals unable to 
access the targeted market at all, and others potentially unable to 
find an affordable premium/cost-sharing combination.
---------------------------------------------------------------------------
    \20\ LJ Blumberg. 2001: ``Health Insurance Tax Credits: Potential 
for Expanding Coverage.'' Health Policy Briefs, The Urban Institute. 
August; No. 1.
---------------------------------------------------------------------------
               V. CHALLENGES TO BROAD-BASED RISK POOLING

    Some will suggest that we can prevent the selection concerns I have 
outlined by providing greater choice of health insurance plans while 
implementing a risk adjustment system that would spread the costs 
associated with the high cost/high risk insureds across a particular 
insurance pool. As already discussed, I do not believe that spreading 
such costs within any particular insurance pool is sufficient. 
Additionally, after many years of experimentation and study, the 
technology available for accurately making risk adjusted payments to 
insurers is still not as effective as we would like.\21\ Ideally, 
insurers would be compensated for the excess costs of the care of their 
unhealthy, enrollees, without compensating insurers for inefficiency in 
the delivery of services. As the Federal experience with risk 
adjustment of payments to HMOs under the Medicare program has revealed, 
such a task is a difficult one. All empirical analyses to date have 
suggested that the risk adjustment formula used to determine payments 
to Medicare HMOs have exceeded efficient payment levels given their 
healthier than average enrollees. Analysts have suggested that the best 
risk adjustment approach would be a blend of prospective and 
retrospective payments.\22\ But even in the most ideal of situations, 
the maximum variation in expenditures that can be explained` is roughly 
20 to 25 percent.
---------------------------------------------------------------------------
    \21\ JP Newhouse, MB Buntin, and JD Chapman. 1997. ``Risk 
Adjustment and Medicare: Taking a Closer Look.'' Health Affairs. 16(5): 
26-43.
    \22\ JP Newhouse, MB Buntin, and JD Chapman, 1997. ``Risk 
Adjustment and Medicare: Taking a Closer Look.'' Health Affairs. 16(5): 
26-43.
---------------------------------------------------------------------------
    The technologies currently being used in the Medicare program which 
account for slightly over 10 percent of the variation are still 
considered inadequate, as evidenced by the dissatisfied reactions of 
participating plans and their continued aggressive pursuit of healthier 
enrollees. However, even if we could agree that the most recent 
approach to risk adjustment works reasonably well in the Medicare 
context, that does not imply that it would work sufficiently well for 
adjusting plans in private markets. Key differences between Medicare 
and private insurance are that Medicare coverage is virtually 
universal--the whole population of elderly are in the risk pool, and 
that the actuarial differences between plans are very small in 
Medicare. In private markets, where actuarial values of different plans 
can be quite large, and where people have the opportunity to opt in or 
out of the market, risk adjustment becomes substantially more 
difficult. For example, where variation in benefits is allowed--more or 
less of a drug benefit, mental health benefit, etc.--selection can be 
more targeted. In addition, when the actuarial values for plans differ 
substantially, it becomes much more difficult to determine what is the 
appropriate reference for any redistribution.
    A very important issue with regard to employers and risk 
adjustment, however, is less technical in nature. That is--is there a 
strong incentive for employers to do effective risk adjustment and 
maintain plan choice over time between comprehensive and high 
deductible policies? Although most employers want to look out for the 
well-being of all their workers, in a competitive environment they face 
incentives to keep health care premiums down while keeping their 
highest paid workers satisfied. If employers can keep premiums down by 
having a healthier risk pool or leaving more of the costs of care 
directly on the sick, then they will have more dollars to put toward 
paying higher wages, thereby making them more competitive in attracting 
and keeping the workers they would like to employ. HSAs may just 
provide employers with an effective tool for responding to these 
incentives, by placing a greater share of the health care financing 
burden directly on the sick while the most valued employees can be 
compensated via the tax subsidy. This may be a real improvement over 
the past in the ability of employers to discriminate between the 
healthy and the sick, because reducing the value of employer-based 
packages in the past would have been potentially detrimental to all 
workers, and this would have hampered employers' ability to attract 
high wage workers. If this conjecture proves to be accurate, there may 
be little incentive for employers to avoid having choice of plan 
devolve to HSAs and high deductible policies being the only option. If 
no other reforms are implemented, the lower income and higher cost 
populations will then pay a larger share of their income toward medical 
care than they did previously, perhaps impeding their access to 
necessary services.
    The most important challenge facing implementation of, a broad-
based approach to risk sharing, such as those that I have outlined, is 
the financing required to implement the proposals discussed. Each of 
these 3 proposals--increasing funding to high risk pools and making 
their coverage more comprehensive; public reinsurance; and creating 
purchasing pools with public subsidies for both the high risk and the 
low income--would require new funding in a current context of enormous 
Federal budget deficits. However, as a first step, each proposal could 
be structured to limit benefits to particular groups, for example 
individual purchasers and/or small groups. This would limit the size of 
new revenues to be raised, but would also limit the benefits. In 
addition, each proposal should lead to some private savings, as 
insurance premiums go down, thereby decreasing the net costs to some 
extent.
    In conclusion, a wise person once said, when you find that you have 
dug yourself into a deep hole, the first thing you should do to save 
yourself is to stop digging. The tools that we have been using in 
private insurance markets--segmentation by health care risk, and at 
times, forced pooling within small enrollee populations--have gotten us 
into this hole. It is time to set those shovels down (in addition to 
policies which provide higher subsidies for higher income people), and 
seriously consider an approach that would separate the excess costs of 
caring for our most vulnerable neighbors from the decision to purchase 
health insurance.

                                  
