Instructor’s Manual

to accompany

Health Economics
Sixth Edition

Charles E. Phelps
University of Rochester
 AUTHOR’S PROLOGUE

The Sixth Edition of Health Economics contains a fair amount of new material, as
outlined in the Preface of the textbook itself. The most important new material
centers around the Patient Protection and Affordable Care Act (PPACA). In the
Fifth edition, rather than just sticking a chapter onto the end of Health Economics,
I chose to interweave relevant material throughout the textbook, beginning in the
overview in Chapter 1, with discussion of aspects of the PPACA in many of the
chapters, most notably Chapter 16 (“universal insurance”), which was entirely
restructured to focus on the PPACA rather than some hypothetical “universal
insurance plan” as in previous editions. You will also find significant new material
in sections chapters involving regulation, insurance (both demand and supply),
managed care and pay-for-quality, and many other places.

The Sixth Edition brings another uncertainty: with the White House and both the
US House of Representatives and the US Senate in the hands of the Republican
Party, numerous political effort arose (none successful as of this writing in late
2017) to “repeal and replace” the Affordable Care Act (ACA). Even late into the
very end of 2017, discussion arise about including a repeal of the Individual
Mandate of the ACA as part of an unrelated budget resolution (not requiring a
super-majority in the Senate, as much legislation does). One of your key tasks in
teaching this material is to make it timely, which means (in this case) paying
particular attention to legislation that might alter the ACA – and to think about its
consequences with your students when you are at the appropriate point in the class.

Thus, to make this material maximally relevant to students, I suggest that you (as
the instructor) keep on top of the political ups and downs – even keeping a log of
failed attempts to change the ACA – and be ready to discuss the consequences with
your students. This may require that you delve into details of the ACA that may be
unfamiliar to you, but the teaching moments that emerge are endless.

In a totally shameless plug for another book I have coauthored, let me suggest that
a way to delve deeper into these issues is to get a copy of The Economics of US
Health Care Policy, coauthored by Stephen Parente, PhD, from the University of
Minnesota. While Health Economics, Sixth Edition gives you the tools to diagnose
the illness, Phelps and Parente suggests some specific treatments. What follows is
a brief summary of some of the key issues that we highlight in Phelps and Parente,
all of which you can use as “teaching moments” in discussing material in various
chapters of this book. Both books are published by Routledge Press (England).
1. The Individual Mandate. This is perhaps the most visibly hated portion of the
ACA among its opponents – “forcing people to buy something that they don’t
want.” A common reference was broccoli, apparently the least-favorite vegetable
of former President George Bush Sr, who famously stated after he became
President, “now they can’t make me eat broccoli.” You could analyze the
individual mandate in terms of standard demand curve analysis (forcing people to
consume more than they want of some good), but that would miss the point.
That’s the “cost” side of the mandate. The other half comes with the way the
market for insurance functions, because of the separate ACA restriction on insurers
using preexisting conditions to price insurance policies. So we must now turn to
the “preexisting conditions” question next.

A separate but related issue is the tax penalty for non-compliance with the
Individual Mandate. My own personal view is that it is far too low to achieve its
stated objective. For many – particularly young and healthy people (like many of
the students in your class will be in a few years when on their own after college) –
the logical economic choice may be to pay the penalty rather than buying
insurance. You might do a little classroom “experiment” to get the students to look
at quotes for insurance that they might have to buy to comply with the mandate,
have themselves ask how much they value that insurance, and then penalty for
non-compliance.

2. The Ban on Using Preexisting Conditions. Just the opposite of the individual
mandate, this is highly popular across the political spectrum. Quite simply, it says
that insurers cannot use preexisting conditions either to set the price of insurance or
even in the offer to sell insurance. The economic logic for this is simple – it
creates a market that otherwise cannot exist, the coverage of genetically linked
diseases over which people have no control. But if you ban the use of preexisting
conditions, many rational people would simply not buy insurance until they
became sick, and then they would buy full coverage insurance. Thus (ala the
market failures discussed by Rothschild and Stiglitz), would emerge and the
market could well collapse. So you can’t really talk about the Individual Mandate
without thinking about the way the market functions, and this is a great way to talk
about separating equilibrium in markets (Chapter 11) and market instability.

3. The Precious Metals insurance standard. Part of the ACA defines the quality of
insurance only on a single dimension – the average percent of all medical care
costs covered by the plan. Platinum (better than gold!) requires at least 90%
coverage. Gold requires 80%. Silver requires 70% and Bronze requires 60%. The
individual mandate demands that people have at least a Bronze plan.
This ”Bronze” requirement actually precludes people from buying some plans that
have had increasing popularity in the market – High Deductible Health Plans
(HDHPs). They actually closely follow the analysis by Nobel Laureate Kenneth J.
Arrow, who proved (Arrow, 1963) that the optimal insurance plan provides full
coverage above a deductible. Particularly for young and healthy people, HDHP
plans can readily have an actuarial structure that covers less than 60% of expected
costs, and hence these do not qualify as Bronze plans. This provides a good
discussion opportunity for ways to evaluate insurance plans. What else might be
pertinent to consumers? Things like risk protection (which a “stop-loss feature
such as HDHPs have), access to physicians (which many “preferred provider”
plans inhibit), and complexity of administration (which, for example, is quite
complicated in Medicare Part D insurance for prescription drugs).

Subsidies to the Health Insurance Exchanges (HIX). Earlier attempts to derail the
ACA focused on a US Supreme Court challenge to the individual mandate (which
failed). Then the Trump administration (in 2017) sought other ways to reduce the
subsidies to lower-income people. The subsidies – related to family income
relative to the federal poverty line -- use each family’s regional “Silver” plan as a
benchmark. The subsidy is a fixed amount per year, and can be spent only on
health insurance, but may be spent on a Bronze plan (but not an HDHP plan that
does not meet Bronze rules).

You can talk about this process in several ways. On what “margins” does the
subsidy work? Clearly, it subsidies health insurance relative to other goods and
services, but within the package of “health insurance” it is a flat amount, leaving
open the issue of how much coverage to get for the enrollee. This thus acts just
like a ‘flat copayment” health insurance plan that would pay $X per visit. While
this was once a common type of health insurance plan, especially for hospital care,
it is now obsolete, almost wholly replaced by the converse type of plan – the
consumer pays a flat amount per doctor visit (e.g., $25 per visit. You might have
a useful discussion about why the “flat amount” works one way with the purchase
of insurance (the way the ACA works) vs. the purchase of doctor visits (the way
much health insurance works now).

You can also, of course, bring this in as evidence about the price elasticity of
demand for insurance, although the observed data (the increased enrollment in the
ACA) mixes up the price subsidy effects and the mandate’s effects.

4. The Employer Mandate. Perhaps almost as equally disliked by ACA opponents

as the Individual Mandate is the Employer Mandate, but there is a long, long
history of attempts to create universal insurance coverage in the US, going back at
least to Republican President Richard Nixon, and including proposals from many
other Democrat and Republican proposals.

Even setting aside these political issues, there are important economic lessons to be
learned from the Employer Mandate issue. Using employer-based insurance to
extend coverage continues issues of job-lock and entrepreneurship-lock (fear of
lost coverage if you leave a big firm for a start-up), for example.

The mandate also hits small firms harder than larger firms, because of the size-
related loading fees. The ACA attempts to blunt this by offering subsidies to
smaller employers through the Small Business Health Options Program (SHOP),
but it cuts off at 100 employees, and all of the available evidence shows that there
is still an important loading-fee differential between firms above this cutoff and
that available to really large employers. This all affects the types of coverage that
firms of different sizes might offer, and puts workers at risk of losing their jobs if
the mandate affects them AND they are near the legal minimum wage. This is one
of the many ways in which health economics and labor economics intersect.

A parallel issue is the tax subsidy to employer-paid premiums, discussed in detail

in Chapter 10. A number of other issues arise around this subsidy, most notably
the wholly regressive nature of it. You can’t get it if you don’t work, higher-paid
workers have higher employer paid premiums, and the tax subsidy rises with
income (since it is defined by the marginal tax rate). The CBO has estimated that
almost three quarters of the subsidy goes to the upper half of the income
distribution.

The Phelps and Parente book argues for maintaining the Individual Mandate but
abolishing the Employer Mandate – once the tax subsidy for employer-based
insurance is removed. This makes for a great discussion near the end of the course.

One of the hottest political topics of 2017 (and likely going forward) is the price of
prescription drugs. At the core of the debate is the tension between affordability
(both to individuals and to society) on one hand and the incentives for innovation
on the other hand. Thus, when you get to Chapter 15 (Regulation), try to bring in
as much as possible of the current debate on drug price regulation.
A General Note About Handbook of Health Economics links.

The Fifth Edition of Health Economics was completed in 2011, and in 2012, North
Holland Press published Volume 2 of the Handbook of Health Economics, edited
by Mark V. Pauly, Thomas G. McGuire, and Pedro P. Barros. Their Volume 2
contains 16 new essays by top-notch health economists from around the world, and
at the end of each chapter, I have added suggestions about readings from Volume 2
where I thought them appropriate to those listed in the Fifth Edition from Volume
1 of the Handbook of Health Economics. Reading them in advance of where the
material comes in the lectures will give you the most-recent review of key issues in
the related areas.

Thank you for adopting my textbook. It’s a labor of love, and I hope it helps you
and your teaching. I try to make it better with each edition so if you have
suggestions, I welcome them. I hope you have a lot of fun teaching this course!
 CHAPTER 1

WHY HEALTH ECONOMICS

Chapter Outline

Why Health Economics

Important (if not Unique) Aspects of Health Economics

Government Intervention
Uncertainty
Asymmetric Information
Externalities
How Markets Interrelate in Medical Care and Health Insurance
Fixed Technology (Figure 1.1)
Dynamic Issues (changes through Time
Income
Aging
Technological Change
Price changes and their meaning (quality and consumer surplus)
Spending Patterns over time
Summary
Related Chapters in Handbook of Health Economics
Problems

Teaching Tips from the Author

As a general rule for the entire Health Economics Course, I tend to emphasize the
roles of uncertainty and information throughout the course. Indeed, I proposed a
title for the book to my first editor of “Uncertainty and the Economics of Health
Care,” He thought it sounded too specialized, and I now agree with him, but you
will find an emphasis on uncertainty throughout the book. I always look for ways
to bring uncertainty into the discussion, and point out whenever possible how
uncertainty rests as the basis for much of what we know in health economics as
different and interesting compared with (say) the market for sugar or salt. While I
will not point out these opportunities everywhere, I urge you to take this task on
(for me, if you will): help your students recognize the role of uncertainty,
information asymmetry, and the social rules and institutions that have emerged as a
result of the uncertainty that’s omnipresent in health care.

When I taught using this book (ECO 236, an upper division undergraduate course
at the University of Rochester), I went through Chapter 1 fairly rapidly, without
detailed reference to specific items in the text. I used a lot of questions to the class
here: it’s a great way to get the class used to answering questions I pose when the
questions have no “right” answer.

The “important (if not unique) aspects of health care section in Chapter 1 sets the
stage for what comes later, so it’s worth a bit of time to emphasize these issues.
None of them are wholly distinctive to health care markets, but the extent to which
they matter exceeds that found in almost any other sector of the economy.

Government intervention of course is not novel to health care. The government

regulates airline safety (in more ways than when I wrote the first edition), food
safety, traffic safety, and (an issue more prominent in recent times!) safety of
money you have entrusted to the banking system. Of course the government takes
primary responsibility for safety from foreign invasion (the military), as well as air
and water quality. Note the common thread of much of earlier government
intervention – safety. The logic for government intervention there is similar to the
logic in health care – possible harm or death to humans that is irreparable if it
occurs, coupled with either major informational problems for individual consumers
in achieving the same level of safety or obvious economics of scale in producing
the safety (public goods).

This leads naturally into two other areas that demand constant attention in health
care – uncertainty and informational asymmetry (which are obviously first cousins
intellectually). In few other areas of the economy can you find such an overriding
emphasis on uncertainty. The institutions we have in modern societies to deal with
uncertainty span a huge array of topics – health insurance and managed care,
malpractice insurance, FDA regulation of drugs, enormous regional variation in
medical treatment rates for specific therapies (diagnostic and therapeutic
uncertainty by providers), and now, “clinical outcomes assessment” activities.
These all are manifestations of uncertainty.

Asymmetric information also turns up commonly, most often in doctor-patient

interaction, but also in the knotty problem of possible market failure in health
insurance. The emphasis on professional ethics (e.g., as in Arrow 1963)
emphasizes the importance of informational asymmetry, and the societal norms
and rules we have (“professionalization,” licensure) are societal responses to these
issues.

I like to use these discussions to involve the students in answering questions I pose.
This works particularly well in the “important aspects of health care economics,”
where the students’ own experiences in life help to illuminate some of the key
ideas. How many have been to a doctor in the last year? What was the first thing
they were asked by the receptionist? (It’s usually something about insurance
coverage!) Some questions can be deliberately whimsical: How many of you have
worked on your own car, or do repairs around the home? OK, now how many of
you have done surgery on yourselves or removed a tooth? (This emphasizes the
information gap between providers and consumers of health care.)

The material on technological change and prices is essential for later discussions
about the future health care system. I view it as vitally important that students of
health economics learn to parse out the implications of increasing prices. First, the
measures of prices are poor, often omitting quality improvements. Second, and
more important, it is quite possible (as the chapter shows) to have a welfare-
enhancing price increase if comes with sufficient quality improvement. This
comes back full-force in Chapter 16 when the discussion turns to universal health
insurance (and the PPACA) and questions of “how much can we afford to spend”
in health care? Box 1.1 provides the key discussion.

There’s a simple way to work through the tables at the end of Chapter 1 showing
spending patterns through time. The tables “nest” in a very specific way that you
can illuminate by focusing on the numbers at the lower left corner of the table.
Look at Table 1.5 to start out. The annual total spending in 1960 was $23 billion,
and grew to $3206 billion in 2015. The ratio of those two years is 138, the number
in the lower left hand corner of the table. The same ratios are calculated for
components of health care spending as well as the “total” figure.

Now look at Table 1.6. This corrects for general inflation, so the ratio of 138
becomes 20.9. In other words, almost 7/8ths of the increase in spending (about 85
percent) is just simple inflation over the past half century. The “real” increase of
21-fold is still important, however.

Table 1.7 takes the next step, correcting for changes in population levels. The
“real per capita” increase is now down to “merely” a factor of 11.8. Table 1.8
takes a more specious step by correcting not only for general inflation but for
relative price increases (using the “medical CPI” component). This step is fraught
with difficulty because of the imprecision in adjusting for quality in the Medical
CPI, but this brings the ratio down to 4.67

Another subtle issue comes up in these tables – what percent of health care is
“prescription drugs?” Standard CMS data say “10 percent,” but it turns out that
this is incomplete. This Sixth edition recognizes the flaw: these CMS tables only
recognize the sales through retail drug markets, hence omitting all of the drugs
purchased by hospitals, doctors, infusion centers (mostly for cancer drugs), nursing
homes, etc. This turns out to miss a significant proportion of prescription drugs,
and hence understates the importance of drug prices on the overall spending
picture. A 2016 report from the Assistant Secretary for Planning and Evaluation
(ASPE) in DHHS clarified this issue.

A second issue in defining the “percent” is the choice of the denominator. The
ASPE/DHHS report uses “personal medical spending” as the denominator. Using
this as the denominator leads to saying that prescription drugs are 16 percent of
medical spending. A larger possible denominator is “total” medical spending,
which includes “personal” plus R&D costs, construction and equipment, and the
costs of operating health insurance plans (including Medicare and Medicaid). This
larger denominator gives the value of 14 percent as the share created by
prescription drugs.

Tables 1.9 and 1.10 set the stage for another important calculation – adjusting for
the average age in the population. This, of course, sets up a discussion about the
possible effects of the retiring “baby boomer” (post-WW II birth) cohort and its
effects on health care spending. The correction for an aging population between
1960 and 2010 yields a surprisingly small adjustment in the total-spending ratio.
The ratio of 4.7 in Table 1.8 would decline merely to 4.3 after the age adjustment
process. In other words, a relatively small fraction of the real per capita spending
increase over the last half century in the US has come from the aging of the
population. As the baby boomers (present author included) move into retirement,
of course, the effect will accelerate, but the calculation is important because it
shows how much of the “real” increase is related to something else.

What is that “something else?” I (and many of my colleagues in health economics)

think the answer is mostly “technological change,” which raises the importance of
regulation of and payment for new technologies (chapter 15) and their use in a
universal health insurance system (chapter 16). Many people see a “single payer”
system as a way to increase efficiency by reducing administrative costs, but I see a
more important issue: it creates a single decision point for introduction of new
technologies, which leads to the possibility of too much or too little new
technologies (or both – in different areas) coming into the health care system. One
benefit from our pluralistic health financing system is the introduction of multiple
decision points for technology introduction, hence the ability to learn from others’
mistakes.

Classroom Projects:

In this instructor’s resource, you’ll find class or individual “projects” that relate to
specific chapters in the textbook. These can mostly be done outside of class, but
may involve in-class reporting time. These projects may well be done in groups,
per the previous discussion of group learning and “workshops” as an adjunct
teaching method to augment classroom lectures.

I welcome your comments ([email protected]) about how these projects

work out and any suggestions of projects you’ve used that add to the class. (I will
add these to the project list through time with attribution to the individual first
proposing them to me.) This way, all the users of Health Economics can help each
other become more effective teachers.

Chapter 1 projects:

A. Technological change over time – the costs of treating a specific illness:

Divide the class into appropriate numbers of groups (about half a dozen per team
as a suggestion) and have them investigate the technology for treating a specific
illness 25 years ago vs. current technology. If possible, have them find out the
typical costs of treatment, but that will be harder. The key idea is to understand the
extent of technological change in treating illnesses and efficacy of those
treatments. In some cases, simple average length of stay (ALOS) in the hospital for
treatment will be very illuminating. Some examples of interesting illnesses to
consider: heart attack (involvement of intensive care units both now and quarter
century ago, but new drugs to eliminate blood clots, new diagnostic methods such
as testing for enzymes that signal damage to the heart muscles, etc.); normal
delivery (ALOS was over a week in 1975; it’s now about a day or day and a half);
allergies (antihistamine drugs in both cases, but now far fewer side effects; also
nasal sprays available such as sodium cromalyn, spray cortisone drugs that were
not available then). Mental illness (immense change, shifting from long term
hospitalization to treatment with drugs). Good reading on this issue, even if old, is
Anne Scitovsky’s 1967 article in AER. A series of new studies on cardiac
treatment (David Cutler and others) and mental illness (Richard Frank and others)
have illuminated these issues in more recent times.

B. Sectoral Shifts in Treatments:

This project looks at the different growth rates through time in the various
“sectors” of health care – inpatient hospital care, physician care, prescription
drugs, “other” and nursing homes. This project probably should avoid the step to
Table 1.8 and concentrate on 1.7 and earlier tables (because of the problems with
the medical CPI components). Look at Table 1.7. Which parts of the health sector
have increased most rapidly, and why? Most notably, of course, is “nursing
homes” but also prescription drugs. Have the students list why they think these
patterns emerged, and ask what they expect in the coming 20 years or so. I suggest
that you have them re-calculate the ratios using 1995 as the base year rather than
1960. The picture changes considerably. For example, the 2010/1995 increase for
drugs is 3.3. The nursing home ratio is only 1.3. The hospital ratio using 1995 as a
base is 1.6. The big bursts in the early years came in hospital, physician and
nursing home expenses. (Hmmmm, when did Medicare and Medicaid come in and
what did they insure?)

And separately, when did prescription drug expense start to grow faster than other
components? Three major legal changes shifted the world of biopharmaceutical
product development. First, the 1980 Bayh-Dole Act shifted ownership of
inventions from NIH-sponsored research to the organization (typically a medical
school or other university department) receiving the grant funds. Then universities
started “tech transfer” operations to promote their inventions and licensed the
patents to industry. Second, in 1984 came the Orphan Drug Act that eased the
rules for developing drugs for “rare” diseases. And in 1985, the Hatch-Waxman
Act expanded the period of market exclusivity for patented drugs from 20 to 25
years, with a nod towards the time lost to FDA approval. All of these boosted drug
research. At about the same time, Medicare started a major change in payments to
hospitals that remarkably reduced length of stay and hospital spending (see
Chapter 12 on Medicare and Medicaid). So while drug spending started to rise,
hospital spending slowed down.

Sample Exam Questions with Suggested (Cryptic) Answers

1. (15 points total) Major uncertainty looms as a key issue in many aspects of the
health care system. Identify the key social or economic institutional
arrangements that seem to be established primarily to deal with the uncertainty
described in each area (1 point each):

a) Financial risks arising from random illness.

b) Variable competence by doctors and other medical providers.
c) Demand inducement for major surgery.
d) Incomplete information about the effectiveness of medical treatments

e) Risks of dangerous or ineffective drugs.

Now, for 10 points, pick one of these items and discuss how the risks arise, and
how the social institution deals with the risks. (If you think the institution is
ineffective in doing so, discuss.)

Answers: health insurance; licensure; second opinions; prior

authorization and similar demand-management strategies of managed
care insurance; FDA drug regulations.

2. [This is a variant on the first question.] (20 points) Uncertainty enters into the
maintenance and production of health in numerous ways, including (a) Whether
and when individuals will get sick; (b) Whether a given treatment will work for a
given patient; (c) Whether on average a treatment will work for a whole class of
patients; (d) whether a specific doctor is competent to deliver treatment. Pick any
two of these and describe the major mechanisms (institutional arrangements) that
our society has created to deal with the uncertainty.

Answer: (a) insurance and related social programs; (b) expert panels such as
NIH, and managed care processes for approving treatments in advance and
retrospectively; (c) insurance plans listing procedures in advance that they will
(not) approve; (d) licensure.

3. (10 points) “Health care is so different that we can’t possibly use standard
economic theory to think about it.” Discuss intelligently.

Disagree; although the basic tools of microeconomics require revision (e.g.,

treating health, not medical care as the normal good, and understanding that the
not for profit hospital is different than for-profit organizations), we can still use
economic approaches fruitfully to study health care.
Answers to End-of-Chapter Questions
*Starred answers appear on the Companion Website.

1. a) Demand for insurance

b) Provider decisions about treatment (e.g., medical practice variations)
c) Quality of care inference by patients
d) FDA regulation of drugs and medical devices
e) In insurance markets, insurers sometimes know less about the health
conditions of their potential customers than those potential customers
know.

*2. While medical care is “different,” standard economic analysis still offers much
useful insight and analytic structuring. On the demand side, for example, the
“human capital” approach offers a way to move from the demand for “health” to
the demand for medical care. On the supply side, we must account for the
incentives and motives of not for profit suppliers in some cases. And in the
demand for insurance, we must account for the effect of insurance on medical care
use in a specific way. But these are all modifications of standard theory, not
rejections.

*3. Illness event; insurance coverage; income; price of medical care; education;
sex; life style choices and age all affect individual demand for care. The single
most important of these for an individual is the individual’s illness event (if it can
be observed). In groups, if you cannot observe individual illness events, age is
the strongest predictor of medical care use.

*4. Income, education, expected illnesses, aversion to risk, price of insurance

(loading fee) all affect demand for insurance. In particular, expected medical
events that drive demand for medical care, not actual events, since the insurance is
chosen before the actual event is realized.

5. Example: For total care, the table 1.7 number is 8.2 and the “relative price
adjusted” number in table 1.8 is 2.8. This implies a relative price increase of 2.93
(the ratio of 8.2 to 2.8). For hospital care, the same type of calculation yields 2.96,
almost identical to the total. Drugs had a ratio of 2.90. For physicians, it was 2.87.
The big surprise is that there isn’t much difference across these sectors over this
half century.
*6. The quality of care in a medical sense has surely improved a lot over this time.
The safety and treatment standards, for example, include more reliable standby
services (for medical emergencies) and the like. The typical two-person hospital
room now also includes much better communication services (tv, phones), and
probably (although some would dispute this) better meals. Since the CPI does not
reflect the increase in value, it overstates the price increase. In effect, it compares
a 1960 “apple” with a 2005 “orange” rather than comparing two identical apples.
 CHAPTER 2

Utility and Health

Chapter Outline

How to Think about Health Care

Health as a Durable Good
The Production of Health
Health Through the Lifecycle
A Model of Consumption and Health
A More Detailed Look at Consumption and Health
Summary
Related Chapter in Handbook of Health Economics
Problems

Teaching Tips from the Author

The first half of this chapter provides an excellent basis for reinforcing basic
economic concepts that the students should have acquired in Intermediate
Microeconomics (which I required as a prerequisite class when I taught this
material). The guiding light here, of course, is Michael Grossman, who created the
way of thinking about health as a durable good, which in turn stems from his own
PhD thesis advisor, Nobel Laureate Gary Becker, who (along with Kelvin
Lancaster) created the “household production” way of thinking about economic
behavior.

I view it as essential that students come away from this discussion understanding
how indifference curves “work” when the utility function has “Health” as one
argument (i.e., U = U(X,H). One could quibble, of course, with this definition,
since there’s a stock/flow issue lurking here, but I don’t think it impeded the train
of thought. (Health–H–is really a stock, and the flow of “services” from that stock
is what really creates the utility, of course. So if you want to be picky, add some
multiplier (k) to the stock to show the flow rate from the stock.) But this work
using production functions and indifference curves (mostly the latter) is central to
what comes in Chapters 3 and 4.
The second half of Chapter 2 is mostly new material, beginning with the Fourth
edition and into later editions. This work provides a much more detailed
discussion of how lifestyle affects health than in previous editions. While the age-
specific causes of death provide good “hints” about the effects of lifestyle on
health outcomes, Table 2.6 gives dramatic and direct evidence on the issue: the
“real causes of death” are lifestyle choices in a large part. We all owe a debt to
McGinnis and Foege for their pathbreaking work in their original study on this
topic. With a little good luck, somebody will repeat their idea with 2010 or even
2015 data soon, and you can update the tables from Mokdad et al. Keep your eyes
open for this! I think it is quite likely that any new studies will show obesity has
overtaken tobacco as the #1 real cause of death.

I find it important to remind my students of the goals of this analysis. I do not

intend these discussions as a moralistic “thou shalt not...” rant. After all, the
presumptive goal is utility maximization, not maximization of life expectancy.
Some of the things that harm one’s health are very enjoyable!1

I particularly found it fascinating to learn about the nuanced effects of education on

lifestyle choices. In every facet of lifestyle I delved into–smoking, obesity, alcohol
use, and even breast feeding–I found solid evidence that education affected
people’s life style choices in a positive way. More education leads to less “self-
destructive” behavior. The effect stands out most clearly in the consumption of
tobacco. The education gradient with smoking is very steep. The effects of
education on alcohol use are particularly complex. More education leads to more
“participation” in alcohol consumption (see Figure 2.7) but less binge drinking and
less heavy drinking (the latter of which are more health-harmful). Even the choice
of alcoholic beverage has a unique educational twist: more highly educated people
drink more wine and less distilled spirits, the latter of which are more health-
harmful. (Sorry, but I could not find any data on education and red wine drinking
choices.)

This all leads to the wonderfully complex question of how education has all of
these effects. Does the higher education give people more reason to “protect their
investment” in human capital? Does the education itself allow people to acquire
more information about the effects of lifestyle choices (and hence modify their
behavior)? Or (as Victor Fuchs originally suggested), does some unmeasured
difference in time preference across individuals affect both educational attainment
1
A person once said to me that humans are happiest when something is entering or leaving their
body. Think about it in the context of the lifestyle choice discussion!
and health habit choices? The puzzle, of course, is that people with low discount
rates (long time horizons) would have propensities both to invest in more
education and to avoid health-harming behaviors. But since we really don’t
understand how discount rates are formed, we must leave this set of issues to
discussion and conjecture.

Classroom Projects

A. Life style and longevity.

Have every class member write down (if they can) the age of their grandparents,
and (if deceased) years of age at death and if known, the cause of death. Have them
record what they know about the smoking habits of each grandparent. Compile the
information into simple 2x2 tables such as (alive / dead) vs. (smoker / non-smoker/
former smoker). This may be something where a call home the weekend before
will help.

B. Obesity and exercise.

Have the students write down their best-estimate of how much they weighed when
they entered college as freshmen, and their current weight. (The average freshman
student gains about 10-15 pounds.) Have them calculate their Body Mass Index
(BMI) using the entering freshman weight and the current weight. College deans
report a “freshman 15” weight effect – students gain a lot of weight in their first
year of college and access to relatively unlimited food supplies in collegiate
cafeterias. Google the phrase “freshman 15” and you’ll be amazed at the number
of web sites devoted to this topic. An a capella singing group (at Northwestern
University) has even been named after the phenomenon.

Aggregate these data (anonymously) and report the results back to the class. You
might then take another run through the mortality risk data in Figure 2.7 and have
them calculate how much their risk of death has increased with their own freshman
weight gain.

C. Use Stewart, Cutler and Rosen’s table (see bibliography of textbook) for a
discussion of quality adjustment, relating specifically to tobacco use and obesity
levels.
Answers to End-of-Chapter Questions
*Starred answers appear on the Companion Website.

*1. The leading causes of death are all “violence” related – homicide, suicide,
vehicle crashes, other accidents. The health care system can only “patch up” the
consequences of these events, not prevent them.

*2. The leading causes of death for those over age 65 are heart disease, cancer,
stroke, chronic obstructive pulmonary disease (emphysema). These are all related
to lifestyle choices, including most prominently tobacco use, obesity and lack of
exercise.

3. The leading causes of death before WW II were mostly infections, and with the
advent of antibiotics and vaccines to prevent infectious diseases, life expectancy
rose dramatically. The adverse effects of lifestyle choices (tobacco, obesity,
alcohol, etc.) have probably reduced our life expectancy compared to what it
would be if we were more prudent in these dimensions of choice. The Japanese,
for example, have a markedly higher life expectancy than do US citizens, in part
related to common dietary practices.

4. If you wanted to pick one life style choice to remove, it would probably be
smoking currently, but smoking is on the decline now and obesity is on the rise. A
forward-looking answer would probably focus on obesity.

What can change obesity? Obviously, education about the issues. But many states
and cities are considering rules that limit “fast food” chains near schools,
requirements for restaurants to put caloric content of meals on menus, and the like.
Some legislators have considered a “fat tax.”

Would this improve overall happiness? Perhaps, perhaps not. Why do people eat?
They enjoy it! Limiting access to certain foods may reduce weight, but also reduce
utility overall. A carefully considered health policy should include not only the
effects on obesity but also on overall happiness.

*5. General education (e.g. a college degree instead of stopping with a high school
degree) improves people’s ability to assimilate information. It also provides more
income in general, which in turn can lead to healthier life style choices (e.g.,
cleaner air, more tennis courts in the neighborhood). Education may also alter
people’s time horizon. If you have a lot of human capital in your head (more
education), it makes more sense to preserve the value of that by remaining healthy.
Higher education seems to lead to healthier life style choice in many areas,
including tobacco use, alcohol use, and obesity (as measured by BMI).

*6. Probably worse off, but it’s not a clear-cut case. Higher gasoline prices lead to
more use of public transportation, which in turn both increases the amount of
walking people do and also reduces smog (probably in only very small ways, but it
will reduce smog). These make people better off. But the direct costs of higher
gasoline have their own direct consequences in terms of lost consumer surplus and
transfers (through payment to oil producers) that reduce consumer well being
directly. No careful study has balanced all of these off yet.

a) Many diseases have direct genetic links, including “traditional” inherited

diseases such as Tay-Sachs, muscular dystrophy, some types of breast cancer, and
many neurological diseases (just to name a few of the many of such diseases).
 CHAPTER 3

THE TRANSFORMATION OF
MEDICAL CARE TO HEALTH

Chapter Outline

The Productivity of Medical Care

Confusion about the Production Function: A Policy Dilemma
Physician-specific Variations (Medical Practice Styles)
Extensive and Intensive Margin Differences: Are They Similar?
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 3: Marginal, Average, and Total Productivity

Teaching Tips from the Author

This chapter completes the reshuffling of ideas from the first three chapters in
earlier editions. The entire body of material on production of health now comes in
Chapter 3.

This edition also discusses some relatively new estimates of the productivity of
health care that appeared relatively recently—some very interesting work by Hall
and Jones (2007). Figures 3.1 and 3.2 present their results in different ways.
Figure 3.1 shows the results in terms of the elasticity of health with respect to
medical inputs. Figure 3.2 shows marginal costs of saving a life or adding a life
year. Figures 3.1 and 3.2 have several important facets to them: (a) the marginal
cost of adding life years has grown through time – at about a 5 - 7% real annual
rate, in fact. As our society has “picked the low hanging fruit” in adding life years,
it becomes increasingly more expensive to continue to add more. (This is, of
course, just good old “decreasing marginal productivity” in another guise.) (b) The
marginal costs of adding a life year vary a lot by age (Figure 3.2): you get more
life years for your money by investing in young children. (c) Teenagers are
particularly an expensive investment, mostly because not much goes wrong with
young bodies except violence-related events (see discussion in Chapter 2), so this
is mostly patching up people who have encountered violence of some form or
another (autos, guns, etc.). And finally, Figure 3.2 also shows how very expensive
it is to add life years to the very elderly, an issue of rising importance, and one that
demands inputs from ethics, politics, economics, religion, and other perspectives to
have any sort of a useful discussion.

This material again creates a good occasion to review some basic principles of
economics and statistics. I try to use my own upper-division undergraduate classes
to emphasize the basic tools of economics through the specific applications to
health care. Production functions are a good example of an area that probably
benefits from a review, and then application to the specifics of medical care. I use
Table 3.1 to demonstrate the ideas of intensive and extensive margins. The key
here is to note that the incremental cost-effectiveness ratios in Table 3.1 are really
the same information as that contained in a societal-level production function: the
CE ratio is the inverse of a marginal productivity measure, in some sense. (CE of
course uses dollars, not physical inputs, so it’s more akin to the inverse of the
marginal revenue product, of course.)

Basic concepts of statistics can be reinforced with the discussions on medical

practice variations, especially the ideas of population means, standard deviations,
and their ratio (the coefficient of variation—COV). The latter turns out to be
useful in an economic analysis of the consequences of variations that appears at
the end of Chapter 4, where the welfare loss from variations turns out to be a
function of the square of the COV, so this is a good time to get that concept
implanted with the students.

I find it very useful and important to invoke students’ own perceptions about
“what is health” in the issue of measuring productivity. I point out that most
“productivity” measures look at things very easy to measure (life expectancy,
mortality rates) when much medical care has nothing to do with that (headaches,
depression, allergies, knee surgery). This is a good place to get students thinking
about the problems associated with measuring the productivity of medical care
that arise from the multi-dimensional aspect of “health” itself. If you wish to
introduce a complexity to the class, you might at this point use a paper I published
in early 2017 discussing multi-criteria decision analysis (MCDA) as elaboration
on cost-effectiveness analysis: Phelps and Madhavan, “Using multi-criteria
approaches to assess the value of health care.” Value Health. 2017;20(2):251-5.
MCDA models are decision-support tools (just like Cost Effectiveness Analysis,
but with more dimension of value). The multiple dimensions of value require that
decision-makers specify the tradeoffs between the dimensions. The various
software approaches to this elicit these values rather than the usual economists’
approach of estimating them from observed behavior.

Classroom Projects

A. Intensive and Extensive Margins:

The use of any diagnostic technology is ripe for a discussion, particularly cancer
screening of various types. The class can be broken into working groups (about
half a dozen per group as a suggestion), pick a specific type of screening available
(e.g., mammography for breast cancer, fecal occult-blood screening for colon
cancer, colonoscopy or sigmoidoscopy for colon cancer, etc.). For one of the early
landmark studies on this, see Neuhauser, D., Lewicki, AM, "What do we gain from
the sixth stool guaiac?" New England Journal of Medicine, 1975: 293:226-228. (I
went to grad school with Duncan Neuhauser at the University of Chicago.) Have
them search the www and elsewhere for recommendations on use of these tests
from professional societies such as the American Cancer Society, US Preventive
Services Task Force recommendations, and those of professional physician groups
such as the American College of Physicians.

B. Measuring “Health”—Mortality and Beyond

Much of our ability to measure the marginal productivity of health care –

particularly in work such as that of Hall and Jones (2007) – uses death rates or life
expectancy as the only measure of health. Thus a useful discussion for a class
project – again perhaps in groups – would ask them to think about how these
measures might change if one took into account both life expectancy and quality of
life. For example, would newborns “saved” through application of newborn
intensive care have “full” quality of life if afflicted by conditions commonly
persisting among premature babies such as cerebral palsy, blindness (retinopathy
of prematurity), developmental disabilities, etc.? Would the cost per “quality-
adjusted” life year for the elderly go up or down (compared simply to “life years”)
if one took into account relief from pain, improvements in mobility and self-
reliance, etc.?

Sample Exam Questions with Suggested Answers

1. (25 points total) The following question explores the issue of marginal
productivity of medical care.
a) In terms your grandparents would understand, explain the concept of
diminishing marginal productivity (5 points).

In the simplest analogy, ‘too many cooks spoil the soup.’ More generally, it occurs
when you add people (or supplies, or whatever) to help produce something and it
ends up getting in the way. Too much fertilizer spoils the lawn. Too many cars
congest the highway. Too many professors create a Tower of Babel confusion.

b) Provide an example of a situation where marginal productivity would decline

because of an increase on the extensive margin (5 points). Please try to use a
real-life example for a specific medical treatment.

Extending breast cancer screening to younger and younger women would cause
this, since the underlying risk falls as you shift to younger women.

c) Provide an example of a situation where marginal productivity would

decline because of an increase on the intensive margin (5 points). Again,
please try to use a real life example.

Increasing the frequency of breast cancer screening from 2 years to 1 year to 6

months.... would demonstrate this; eventually as you did the screening weekly or
daily, you’d create more cancers with the radiation of the tests than you’d detect
in each new application.

d) Thinking broadly about the US health care system, would you expect typically
that medical care would be used at a point where the marginal productivity was
“too high” or “too low”, these terms being defined by thinking about whether the
incremental health received by patients from the marginal use is worth the societal
resources necessary to produce it? Explain your answer. (10 points)

Probably to the point where the marginal product is near zero; with full
coverage insurance, the rational decision is to drive the use to the point of zero
marginal productivity.

2. (10 points) Agree or disagree (discuss): “We know that when people who have
no health care are given access to reasonable levels of health care, their health
status improves. Therefore, we can presume that if we expanded the level of health
care available to all citizens, even those who have “adequate” care now, their
health would also improve.”
3. (15 points) Cross-regional studies of medical practice variations and
comparisons of individual physicians' "styles" show considerable differences
between the rates of use of medical care across providers. Presuming (for the
moment) that the populations being treated in these comparisons are equivalent (or
sufficiently so to ignore the differences), would you say that these studies show (a)
that some doctors treat too much; (b) some doctors don't treat enough; (c) both; (d)
neither; or (e) we can't tell from data like that. Explain your conclusion.

You can’t tell; variations arise from disagreement about proper treatment use,
but that could be correct on average or too much or too little.

4. (10 points) The attached table shows the costs per quality-adjusted life year
($/QALY) for treating various people with various interventions. In economic
terms, these estimates measure the marginal cost of producing a QALY. Identify
which of these represent increases in health care use on the extensive margin and
which represent increases in health care use on the intensive margin. [Note: attach
table 3.1 here for the student to use, or portions thereof as you wish.]

Most of these are extensive margin (wider populations). The intensive margin uses
are adding mammography to annual breast examinations and doing Pap smears
on a biannual rather than a triannual basis.

5. [Note: This needs to wait until the chapter on demand, and the concept of
welfare loss, are done. But it fits in Chapter 3 otherwise. Don’t use it as a
midterm, but rather as a final exam question.] (20 points) "Area variations"
describe differences in medical care use across geographic regions. Under certain
assumptions, one can characterize the welfare loss arising from these variations
(using a model of imperfect dissemination of information). Provide a brief
discussion (using a graph) of what how one might measure the welfare loss from
area variations. Using that graph, discuss and evaluate the underlying assumptions
necessary to make this line of logic "work" (i.e., what are the key underlying
assumptions regarding this approach to measuring welfare loss). (The key issue
here is the discussion of the assumptions, not necessarily a careful derivation of
the graph. You do NOT need to provide a formula measuring the welfare loss
itself; just use a graphical picture.)

This revolves around the discussion in Chapter 5 of “extensions of demand

theory.” See Figures 5.2 and 5.3 and surrounding discussion.
Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

*1. This doesn't seem likely. First, some studies have held the insurance coverage
constant across regions, and large variations still appear. More directly, studies of
medical practice variations within countries such as Great Britain (with a
nationalized health care system) and Canada (with a nationalized health insurance
system) still exhibit large variations in medical care use across regions. These
types of studies virtually eliminate insurance coverage as an important determinant
of the large variations that we observe, for example, in the rate of use of
tonsillectomy or surgery for lower-back injuries. Medical disagreement remains a
more potent explanation.

2. For breast cancer screening, increasing use on the intensive margin would
increase the frequency of testing for those in the population already receiving the
test. Increasing on the extensive margin would add more people (e.g., younger
women) to the list of people receiving the test.

3. That would be a risky conclusion. The comparison only shows differences in

mortality for the two populations. Other benefits of the health care system that do
not appear as differences in mortality would be masked. Thus we cannot conclude
definitively that health care has a zero marginal product, but it does suggest that we
should turn our attention to less severe measures of health outcomes to determine
what value we do receive from the additional treatment.

*4. Degree of training doesn't appear to have a lot to do with observed regional
variations. First, most regions have similar patterns of generalists and specialists
per 1000 population (except for rural areas), yet they have important differences in
use of services. Second, the Boston versus New Haven comparisons show that
even in cities where medical care is produced predominantly through university-
affiliated hospitals by medical school-affiliated doctors, large differences in
behavior can emerge. Third, one can see considerable variation even within
procedures, such as coronary artery bypass grafts, that are carried out only by
highly trained specialists.

5. We would of course under-estimate the marginal product of health care if we

ignored some dimensions of output of the health care system. It would be the same
as looking at automobiles and looking only at improvements over time in fatality
rates from crashes, while ignoring other aspects of the cars such as driving
comfort, air conditioning, sound systems, mileage, etc.

6. We can’t tell from these data whether the variations show too much, too little,
or “just right” on average. The variations only tell differences around the overall
average. Those variations could exist when the system had systematically too
much treatment, too little treatment, or “just right” on average.

*7. The best evidence would come from two strands of the literature in health
economics. One strand looks at the cost of treating specific diseases over time.
When studying (for example) treatment of people with heart disease and mental
illness, researchers have found that the either that successful treatment comes at a
relatively low and falling cost per QALY or in some cases (e.g., mental illness)
that the costs of successful treatment have actually fallen. The other strand comes
from the work on the marginal productivity of medical spending in reducing
mortality (the work of Hall and Jones, discussed in the text in the discussion
surrounding Figure 3.1 and Table 3.1).

8. The QALY approach can be derived directly from a utility-maximizing

approach (Garber and Phelps, 1997), whereas DALYs do not have this
conceptual basis. QALYs look at the gains in quality-adjusted life years
associated with a particular intervention. The quality adjustments typically come
from population survey data, using either standard-gamble approaches (the most
conceptually rigorous) or alternatives such as time tradeoff questions or
“thermometer” visual analogs.

By contrast, DALYs begin with an idealized life expectancy (currently, that of

Japanese women, the longest-lived population in the world) and look at the
difference between that idealized goal and actual life expectancy of a given
population. Second, the disability adjustment in DALYs is derived from expert
opinion, not population survey data. The other key difference is the use of an
age-profile adjustment in DALYs that gives extra weight to years normally
considered economically productive (e.g., adults before retirement age).

9. The answers to this exercise are idiosyncratic to the students’ choices. You
(the instructor) may wish to consider this as an exercise in group settings with
your students instead of a homework exercise.
 CHAPTER 4

THE DEMAND FOR MEDICAL CARE:

CONCEPTUAL FRAMEWORK

Chapter Outline

Indifference Curves for Health and Other Goods

From Indifference Curves to Demand Curves
How Demand Curves Depend on Illness Events
Demand Curves for Many Medical Services
The Demand Curve for a Society: Adding Up Individual Demands
Use of the Demand Curve to Measure Value of Care
How Insurance Affects a Demand Curve for Medical Care
Time Costs and Travel Costs
The Role of Quality in the Demand for Care
Revisited: the Price Index for Health Care
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 4: Demand Curves and Demand Elasticities

Teaching Tips from the Author

This is complicated material. While teaching it, I find myself working through the
figures in the textbook very carefully as most of my lecture time in this chapter. I
get the students involved a lot, probing and pushing them for answers about the
effects of a specific type of health insurance on the demand curve. Sometimes if
they are not offering answers in class, I’ll ask them to take a “pop quiz” where I
ask the effect of some insurance feature on demand, give them a minute or two to
write out an answer (usually a figure suffices) and then tell them the answer, letting
them “grade” themselves. It offers good practice for the types of exam questions I
give on this material.

It is very important for me to keep the same notation in the figures I use in class
and those in the textbook. If you prefer alternative notation, I strongly suggest that
you hand out a translation table to allow the students to match up with the
notation in the textbook figures.

Some of the figures in this chapter have a conceptual issue that I need to be careful
about: they show indifference curves for the same person with two different illness
states, but these indifference curves cross, which is of course illegitimate in
standard micro-economics textbooks. They’re allowable here since each set is
state-dependent. Colored marking pens on a whiteboard help a lot in class (or
colored chalk in older classrooms). I’d use color in the textbook, except that it
adds a lot to the cost of the students buying the book, and this is the one area where
I think color would be really helpful as an expository device. I use bold and
regular lines, but you can use color on a chalkboard. You could also use overhead
projectors and different colored pens. Please remember that about one out of every
twelve males is red/green colorblind (as well as a much smaller rate in women).

In the section on “how demand curves depend on illness events” I get very
melodramatic. I start a demand curve picture on the chalkboard on the far left side
of the chalkboard, and draw a demand curve for a “mild” illness near the origin.
Then I show a “moderate” illness a bit further out. Then as I start to talk about
“really serious” illnesses, I start walking across the room, dragging the chalk to
extend the horizontal axis of the picture I’m drawing. When I get to the far side of
the room, I’ll draw a “serious” illness demand curve, far, far away from the
others. I do this to emphasize how different the demand for medical care are for
different illnesses. Sometimes, if the room’s shape and blackboard configuration
permit, I take the figure around the corner and down the next wall. It’s all
melodrama, of course, but when you talk about the type of event that causes
people to spend $50- to $100,000, it’s nice to portray the difference graphically. If
a “mild” illness causes the spending of $100, then the demand curve for an event
that leads to $100,000 in spending is logically 100 times further out on the x-axis.
If your first demand curve (mild) is “merely” 6 inches away from the axis (at the
price line, say) then the “really serious event” has to be 50 feet away.

There’s an inherent confusion (some might say “trap”) associated with the
“behavioral” demand curves that the book uses to show how demand curves
change with different insurance policies. These curves cannot be used for social
welfare calculations, at least in a societal sense. Thus when I talk about using
demand curves as measures of incremental value, I try to maintain the distinction
that these are the demand curves that ignore the effects of insurance. Why, one
might ask, do I even bother with demand curves like Figure 4.9? Because they
show not only the effect on consumption, but also the effect on the price
responsiveness (and they allow a discussion about the incentives to search with
different plans), but they can’t be used for welfare analysis.

Another issue arises at the cusp of Chapters 4 and 5: the distinction between theory
and evidence. Having trained at Chicago at the knee of Milton Friedman, George
Stigler and the like, I have a fond appreciation for Friedman’s “positive
economics” views and the importance of evidence to support (or refute!) theories.
Many students do not seem to grasp this distinction, and when I ask for evidence
on exam questions, I get back “models” or “theories”. So I emphasize the
distinction, and even highlight in bold and italic the word “evidence” in the
question.

Finally, now that the material on demand curves has come and gone, you can
fruitfully revisit the material in Chapter 1 on price indexes (See Box 1.1.). This is
a good place to reinforce the overall notion of consumer surplus as well.

Classroom Projects
The following projects can be used along with Chapter 4 or Chapter 5.

A. As a class project to understand how quality adjustment works in Quality

Adjusted Life Years (QALY) methods, use one or several of the methods described
in standard QALY work (George Torrance’s 1986 and 1987 articles are excellent
resources here.) Divide them into teams and have them estimate (using themselves
or other subjects) the quality adjustment arising from a permanent disability such
as blindness, deafness, confinement to a wheel chair, loss of the use of a hand, etc.
Try to do the standard gamble approach (it’s complicated!) and then compare
against a simple “thermometer” approach (which, in my experience, often gives
very similar outcomes).

B. Find out what health insurance coverage their parents have at home, and learn
the provisions of those insurance plans for treatment of routine office care,
prescription drugs, psychiatric illness, and dental care. Summarize the results in
class in simple tables (on the fly on the blackboard if the class is small enough),
and use discussion to understand the nuances of coverage differences. It will help
if the students can actually get a copy of the insurance policy or brochures
describing coverage parameters in advance, so this is something to announce in
advance. This information can be used again in Chapter 10.

You could also lead students in a discussion – perhaps more relevant to them –
about the insurance coverage they have (often required by your college or
university) and what’s offered as basic health care in the student health service.
Ask them what seems to be “missing” and how much it would be worth to them to
have it included.

Sample Exam Questions with Suggested Answers

1. (15 points) Comment: "The welfare loss in medical care markets from a standard
“coinsurance” type of insurance policy gets bigger, the smaller the price elasticity
of demand for the service that's insured." (You may wish to draw a picture to
support your answer.)

The key idea is that welfare loss triangles get bigger, the greater the subsidy
to medical care. Indeed, the welfare loss rises with the square of the
proportion of the market price that is subsidized.

2. 10 points. Carefully draw the demand curve for an individual with no insurance
and the same demand curve for a person with insurance paying for 75% of the
medical expenses of that individual. Be sure to describe how you construct the
“75% insured” demand curve using the uninsured demand curve as a reference.

See Figure 4.9. The key discussion point is to show how one picks the quantity
consumed for each price, given a new coinsurance rate of 25%. Example: if the
price is $100, find the quantity consumed on the original demand curve for $25
price, and associate that quantity with $100 on the insured demand curve.

3. (20 points) On the following diagram, describe the optimal quantity consumed
by consumer 1 (demand curve D1) and consumer 3 (demand curve D3), and
explain your answer for each.
[Professor: Insert here diagram like Figure 4.11. You may relabel these as
consumers 1 and 2. Use of the middle demand curve is trickier, and comes in a
later problem.]

The proper amount to consume occurs where the demand curve intersects the
relevant price line.

4. (10 points) Agree, disagree, or uncertain: Defend your answer clearly,

using available evidence as you are able, with the following quote:

"Health care is so special that price will never matter. People will either buy
what medical care they need, or else they will just go without treatment if they
can't afford the fees."

On the basis of a rational consumer model, ignoring price would lead to

improper resource allocation and hence less than feasible utility. Also, in a
Chapter 5 sense, this is empirically refutable; any competent demand study
would show that quantity varies with price.

5. (15 points) Show using two separate graphs the effect of a health insurance
policy that (a) pays for 80% of the medical care purchased by an individual, and
(b) pays $20 per visit for each "unit" of medical care purchased (e.g., an MD
office visit). Describe briefly why demand curves change in the way you portray
in each case.

One is like Figure 4.9. The other is like Figure 4.10.

6. (15 points) [This question needs a figure like that of Figure 4.11] For the
demand curve shown here, what is the optimal medical spending for the
consumer shown with D2? Why?

This is a complicated question of welfare analysis. You have to compare the

sizes of the welfare triangles associated with consuming at the point where the
demand curve intersects the full price and the point where the demand curve
intersects the insured line. If triangle A is larger than triangle B, then the
optimal spending is at ma since you give up too much welfare loss in A to
offset the gain in B. If the area of B exceeds that of A, then the optimal
spending is mc. Buying m* is irrational; you’d give up area A but never gain
B.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. See Figure 4.9 for the correct approach. To make it accurate, set C = .5.

*2. It will create a direct upward and parallel shift in the demand curve by an
amount (measured on the price axis) equal to the payment per visit (e.g., $25 per
visit).

*3. Yes, they could cross, if the price sensitivity differed considerably for illnesses
of similar but not identical severity. This could happen in theory, but it seems
implausible intuitively.
4. See Figure 4.7. Yes, in this case these three demand curves can readily cross,
since they show different persons’ preferences, and they assuredly need not be
parallel. Figure 4.7 shows the horizontal aggregation. As you add more and more
people into the society, the demand curves continue to curve out flatter and flatter,
even if all of the underlying demand curves are straight lines.

*5. We have to think about demand curves for the same services as markets
supply medical services. Thus, we need to think about the demand for things such
as office visits, penicillin, and so on. The demand for penicillin represents the
aggregated demand for all people (see Problem 4) across all illnesses for which the
doctor might recommend penicillin (e.g., sore throats, sexually transmitted
diseases, and wound infections).

6. Quality increases will shift up the willingness to pay (demand curves) as in

Figure 4.12, unless the quality is invisible to the consumer (in which case the
demand curves will not shift up).

*7. Consumer surplus is the extra amount you would be willing to pay to consume
something (e.g., a hamburger) above and beyond the amount that it costs you to
acquire it (e.g., the price charged by McDonald's, Wendy's, or Burger King). If you
would willingly pay $5 for a hamburger and McDonald's will sell you one for $1,
then you have $4 consumer surplus from consuming that one hamburger.

*8. Five visits is totally irrational. To get to five visits, the consumer must forgo
the consumer surplus triangle traced out by the upper price line ($20), the demand
curve, and the vertical portion of the price line. (Call this triangle A.) It would be
better to stop at two visits and not lose this consumer surplus. Once having gotten
to five visits, the consumer could increase the overall consumer surplus by moving
to seven visits. The amount of extra consumer surplus is the triangle traced out by
the lower price line ($20 X 0.2 = $4), the demand curve, and the vertical portion of
the price line. (Call this triangle B.) Thus, either two or seven visits must be better
than five visits. To know whether two or seven visits is better, one must compare
the loss (A) with the gain (B). If B<A in area, then seven visits is better; if B>A in
area, then two visits is better. To consume exactly five visits would be stupid and
irrational.

9. Assuming that health care is a “normal” good, demand curves would shift
outward as a first approximation. What happens in addition depends on whether
the consumption of “good” and “bad” types of goods increases as well as income
increases (in other words, we must also know the income elasticities of demand for
health-affecting goods and services). For example, if demand for cigarettes was
negatively related to income (“inferior good”), then the added income could lead to
reduced smoking, which would improve health and hence lead to lower medical
care use. The net effect of income on demand for health care is the sum of all such
effects from health-affecting goods and services, plus the intrinsic effects arising
from the pure demand for health.

10. = 100 - 10p, then the slope dq/dp = -10, and the elasticity is (dq/dp)x(p/q). At
p=5, consumption is q = 100 = 10p = 100-50 = 50, so the elasticity is -10(5/50) = -
1.
 CHAPTER 5

EMPIRICAL STUDIES OF MEDICAL CARE

DEMAND AND APPLICATIONS

Chapter Outline

Studies of Demand Curves

Effects of Age and Gender on Demand
The Effects of Illness on Demand
Lifestyle and its Effects on Demand
The Demand for “Illness”
The Demand for Quality: Choice of Provider Specialization
Other Studies of Demand for Medical Care
Applications and Extensions of Demand Theory
Decision Theory: Deriving the “Right” Demand Curve for Medical Care
Cost-Effectiveness Ratios and Demand Curves
Why Variations in Medical Practice?
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 5: An Example of Medical Decision Theory

New material:
 Demand for illness (addiction, joint production of pleasure, incomplete
information)
 Cross elasticity with drugs cost offset

Teaching Tips from the Author

This is a really boring chapter if the lectures just recite the “facts” from the
empirical literature. I spend relatively little time in lecture talking about these
“results” from various studies, and instead, I engage the students in fairly intense
studies about the methodological issues associated with measuring demand for
medical care. How do you measure price? Or quantity, for that matter? If you have
observational data, how were such data generated? Is the price endogenous or
exogenous? If some of the students in the class have had econometrics, you can
involve them (at least) in questions of identification. Mostly, I try to get the
students to understand that there are a lot of ways to use data purportedly showing
price-quantity choices that are not on the same demand curve. Controlling for
severity of illness is a really important issue (and hence, how to measure it).

Since I had the luxury of having helped design and operate the RAND Health
Insurance Experiment, I can use the experimental design questions to help in this
discussion: How does the RAND HIS get around the endogeneity of price?
(Random assignment!) What about refusal to participate or attrition? (That didn’t
have much effect, because of the Participation Incentive.) But wait, doesn’t the
Participation Incentive alter the insurance plan itself? (No, this is the difference
between price and income effects.) Does the length of the study matter? (Yes,
assuredly, and a design feature helped understand that.) Is the omission of people
over age 65 important? (You betcha! But the Feds wouldn’t let RAND enroll them,
because it would have required giving up the “right” to Part A Medicare.) Is the
study still valid these decades later? (I think so; it’s behavioral information, which
I presume to be pretty stable, but of course, this study has nothing much about
managed care except the one pure HMO design point.) Etc.

The Sixth Edition adds a new discussion about the Oregon Medicaid experiment
(where they randomized newly eligible people either to receive full coverage
Medicaid or “no coverage”, and used the “no coverage” people as controls. This
newer Oregon Medicaid study has some important data as well, but also does not
include persons over age 65.

The Sixth Edition includes a significantly greater amount of information on

prescription drug insurance, most notably Medicare Part D, about which much
more is now known than was available even at the publication of the Fifth Edition.

The other new material brings in an important new cross-elasticity of demand

involving prescription drugs. This 2006 study calculated the effects on “standard”
medical care use when the costs (copay) for prescription drugs was changed in a
large insurance plan. The drug plan saved money by reduced drug use, but found
a partially offsetting increase in doctor office visits (the cross-elasticity effect).
Moral to the story: when looking at changes in insurance benefit design, one must
consider the total cost of care, not just the particular product or service at hand.
One new study shows this effect for a single drug that treats angina pectoris.

You might return to this issue when you discuss Medicare (Chapter 12), since the
Medicare Part D drug coverage plans would have no incentive to consider cost
offsets, whereas Medicare Part C (Medicare Advantage) HMO plans would have
incentives to consider the total cost of care.

Classroom Projects

The following projects can be used along with Chapter 4 or Chapter 5.

A. As a class project to understand how quality adjustment works in Quality

Adjusted Life Years (QALY) methods, use one or several of the methods described
in standard QALY work (George Torrance’s 1986 and 1987 articles are excellent
resources here.) Divide them into teams and have them estimate (using themselves
or other subjects) the quality adjustment arising from a permanent disability such
as blindness, deafness, confinement to a wheel chair, loss of the use of a hand, etc.
Try to do the standard gamble approach (it’s complicated!) and then compare
against a simple “thermometer” approach (which, in my experience, often gives
very similar outcomes).

B. Find out what health insurance coverage their parents have at home, and learn
the provisions of those insurance plans for treatment of routine office care,
prescription drugs, psychiatric illness, and dental care. Summarize the results in
class in simple tables (on the fly on the blackboard if the class is small enough),
and use discussion to understand the nuances of coverage differences. It will help
if the students can actually get a copy of the insurance policy or brochures
describing coverage parameters in advance, so this is something to announce in
advance. This information can be used again in Chapter 10.

Sample Exam Questions with Suggested Answers

1. (10 points) "Observational studies" of demand for medical care (i.e.. those
simply observing individuals' behavior) have often found demand elasticities for
individuals' use of care that are quite large, in the neighborhood of -1 or larger.
Quasi-randomized trials and randomized controlled trials have found much
smaller elasticities (closer to zero). Discuss the hazards and strengths of using
observational studies vs. randomized trials to learn the true effect of health
insurance on demand for medical care.

Observational studies have the risk that the price and quantity are jointly
determined, e.g., if sickly people buy better health insurance, hence have a lower
price, and end up consuming more medical care because of their illnesses.
Randomized trials have much stronger methodological characteristics, but you
have a risk when you extrapolate beyond the populations studied in the RCT. An
example would be over-65 persons in the RAND HIS.

2. (10 points) What would you expect to happen to student utilization of the
Health Service if the main facility were relocated from the current location to
[insert here some more distant or closer location, as appropriate to your campus]?
What would be the primary cause of any change you predict? What evidence can
you use to support your belief that your predicted change would occur?

Demand should fall. The Simon and Smith study is directly pertinent. Acton’s study
of NYC Medicaid enrollees on travel time effects, and Scitovsky and Snyder’s study
of Stanford employees at the Palo Alto Clinic are also pertinent.

3. (20 points). Discuss the major strengths and weaknesses of the RAND Health
Insurance Study. On net, would you rather trust results from this study or non-
experimental studies in the literature?

The strengths come from the randomized trial design, avoiding simultaneous
equation bias present in observational studies. A primary risk is that the finite
horizon of the study contaminated the results. There are also risks from
generalizing away from the actual cities in which the study was done, etc.

4. (15 points) Describe the relationship between a demand curve (as commonly
thought of by economists) and a cost-effectiveness analysis as carried out by
medical technology evaluation specialists.

CE studies are “normative” demand curves, specifying what a rational consumer

would do with full information (or at least as much information as you can build
into the CE study).

5. (10 points) Suppose a health insurance policy paid for 80% of all treatment
received by the patient. What do you think would happen to patient choices about
quality of care, compared to the same person but with no health insurance?

Quality would probably rise, but the effect is not guaranteed theoretically. It
would depend on the price response for various qualities and the interaction of
quality and quantity. Restaurant meals provide a good intuition here; if I had
somebody offering to pay 80% of my restaurant meals, I’d almost certainly go to
classier restaurants than when I quantity (X*) is reached. Thus that triangle
shows the loss of value for the range when marginal value exceeds marginal cost.
Triangle B is just the reverse; consumption is too large (relative to X*), and shows
the effects of using medical care when the cost exceeds the value.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

*1. The study gave everybody a fixed sum of money at the beginning of the year
covering maximum possible expenditure, but that amount came independently of
any medical care use. Thus, this "participation incentive" adds income, and we can
expect that it would be used just as any other income is. The experimental plans
varied the price of care and should provide a pure price effect. The distinction
between income effects and price effects makes the experimental evidence valid.
Also, if the assertion in the second sentence was correct, then there would be no
differences in spending between the plans, and there was.

2. The welfare loss is larger, the larger the demand elasticity (since the welfare
loss triangles get larger as the price responsiveness increases). Thus the results
from the RAND study imply lower welfare losses from the increased utilization
created by the insurance than the previous studies.

*3. The arc-elasticity of demand provides the basis for one answer (see Table 5.3).
The definition of arc-elasticity is E = [(q2-q1)/((q2+q1)]/[(p2-p1)/(p2+p1)]. Here
we can set p1=1 and p2=0.2. Table 5.3 shows (using dental care as an example)
that E = -0.39 in the price range that's relevant here. Thus [(q2-q1)/((q2+q1)]= E ×
[(p2-p1)/(p2+p1)]. We can calculate (p2-p1) = -0.8 and (p2+p1) = 1.2. Thus [(q2-
q1)/((q2+q1)] = -0.39 × (-2/3) = 0.26. Then we can solve for q2 = 1.7q1. This
means that dental care use would increase by 70 percent with that increase in
insurance coverage. Similar calculations show (for example) that the increase in
acute outpatient care use would be 54 percent, and for hospital use would be 21
percent. Comparison of the raw inpatient costs (Table 5.4) across plans gives a
very similar answer, showing the comparison between a 95 percent coinsurance
plan ($315 per year) and a 25 percent coinsurance plan ($373 per year), or an 18
percent difference. The change in coinsurance between these two plans is slightly
smaller than the ones used in this problem, so the change in spending should also
be a little smaller.

4. It says that they are complements. Both plans had the same price for hospital
care (zero) but one had a higher price for ambulatory (non-hospital) care. As the
price of the ambulatory care went up, the use of hospital care went down (in the
same direction as ambulatory care use went).
In common use language, when people “get into the system” with better
ambulatory coverage, they are more likely to get hospitalized. Thus the two
services move hand in hand, which makes them complements in the economic
sense.

5. Triangle A is a welfare loss because the consumption ended before the optimal
quantity (X*) is reached. Thus that triangle shows the loss of value for the range
when marginal value exceeds marginal cost. Triangle B is just the reverse;
consumption is too large (relative to X*), and shows the effects of using medical
care when the cost exceeds the value.

6. The study of student use of the college health service is one such study. The
studies by Acton and Scitovsky and Snyder give two more studies. These are
reported on pages
159-162.
 CHAPTER 6

THE PHYSICIAN AND THE PHYSICIAN FIRM

Chapter Outline

The “Firm”—Inputs, Output, and Cost

The Physician as Entrepreneur
The Physician-Firm and its Production Function
Nonphysician Primary-Care Providers
The Size of the Firm—Group Practice of Medicine
The Physician as Labor
The Aggregate Supply Curve: Entry and Exit
The Open Economy: U.S.- and Internationally Trained Physicians
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 6: Cost Passthrough

Teaching Tips from the Author

The following Teaching Tips apply both to Chapters 6 and 7.

I find myself wandering back and forth between the material in these two
chapters, since the boundary line is somewhat vague. There are common
threads I try to emphasize, and I try to get the students to read both chapters in
advance of the first lecture on this material.

First, doctors’ behavior has systematic elements of standard economic behavior.

The returns to an MD degree make it attractive, and ditto the returns to
specialization. The spatial location problem presents another good point to talk
about economic incentives. I always begin this discussion with Hotelling’s ice
cream vendors on a linear beach, since it’s such an easy way to represent spatial
competition. Oh, yeah, specialty choice is like spatial choice in many ways.

I try to emphasize the distinction between the physician firm and the physician as
an individual, since confusion on this point can compound itself a lot. (My old
mentor in graduate school, Reuben Kessel, confused these issues badly by talking
about entry restriction into the profession as the source of monopoly power in
individual practices. Of course, this can’t be correct. If there is monopoly power in
the final product market, it relates to search, not the entry restriction on individual
doctors. There’s a quasi-rent associated with the entry restriction, but that’s in a
factor market, not the final product market.)

This issue, of course, brings with it the distinction between the demand curve at
the market and the demand curve for the individual firm. We have precious few
estimates of the latter in the market for physician services (and indeed, again for
hospitals in later chapters). This is another reason why I like the work of Wilde
and Schwartz (and Sadanand and Wilde): it shows how search affects the demand
elasticity facing a single firm. I describe it this way to students: the entrepreneur
faces a market where some patient search for a lower price, and others don’t. For
those who don’t search, the entrepreneur can treat the demand curve as a scaled
down (random sample?) of the market demand curve (with the same elasticity),
and can price accordingly as a monopolist. After all, if somebody picks a single
firm to buy from and never searches, and the firm knows this, we have a
monopoly situation. But...if the patient does search (and finds a lower price), then
the entrepreneur loses all that business. So the pricing strategy of the firm is a
blend of the monopolistic tendencies for the non-shoppers and the urge to attract
the shoppers. Obviously, the more shopping, the more price sensitivity, and the
lower the equilibrium price. I can also work in the role of insurance, of course, on
those elasticities and incentives to search.

I also like to emphasize the potential of alternative ways of producing similar

services. If you have nurses (especially nurse-practitioners) in your class, they’ll be
happy to have you discuss nurse-practitioners as an alternative to physicians in
primary care, midwives in obstetrics, etc. I like the old RAND study from the Air
Force clinic on this issue, because it separates out the technical production
possibilities from the legal prohibitions. (The US gum’mint doesn’t have to pay
attention to state licensing laws!)

As you will see in Chapter 7, I’m a big fan of the monopolistic competition model
to help think about physician markets. It makes a lot of sense to me, and it really
emphasizes (particularly the work of Louis Wilde and various colleagues) the
importance of both the demand elasticity and search on the market equilibrium.
This ties back very nicely to Chapter 4 and the effect of insurance on both the
demand elasticity (insurance drives the elasticity towards zero, which increases the
equilibrium price in a monopolistic competition environment) and incentives to
search (most insurance plans obliterate incentives to search for price, but not
quality.) Sometimes I try to get the class involved in a discussion about how you
find a doctor, but most of the undergraduates I teach have not had much experience
in that realm. That discussion would bear more fruit in a graduate class of MPH,
MD, or other such students.

Chapter 6 introduced some new material in the Fifth edition that you may wish to
review if you’ve used previous editions. In some sense, both begin to peer within
the “physician office visit” to ask that’s being produced. One of these sections
looks at the physician as a diagnostician (the complexity of the diagnostic problem
in important ways leads to the regional variations that Chapter 3 discusses). The
other looks at the “intermediate products” that a physician visit generates,
including compilation of a medical history, creation of diagnoses, production of
empathy, conveyance of information to patients, etc. This section also takes an
excursion into the world of sociology with a discussion of physician “prototype”
behavior as discussed by Emanuel and Emanuel (1992).

Chapter 6 includes a new discussion of practice ownership patterns. In 2008, the

US crossed a great divide when over half of all physician practices became owned
by hospitals instead of physicians. This phenomenon has not slowed. It has two
implications for later analysis. First, it seems increasingly less likely that we’ll see
new information on physician practice costs. Larger organizations tend to be
VERY sensitive about releasing data that might create a “competitive advantage.”
Second, this trend has assuredly accelerated the shift to electronic health records,
which will improve quality over time at least to some degree.

Chapter 6 also updates data on applications to medical school over time showing
how the demand for medical education has changed over time and also how the
rates of inflows of internationally trained medical students have changed over time
in response to various changes in immigration rules and requirements placed on
foreign-trained students.

The Fifth Edition introduced in Chapter 7 some new material on specialization and
the possibility of barriers to entry. The new Figure 7.1 shows a strong relationship
between mean income in various specialties vs. the percent of medical residency
(training) positions are filled by US medical school graduates. Those fields with
the strongest earning potential are filled at very high rates by US graduates, while
lower-paying specialties (mostly primary care and non-procedural specialties) have
much higher rates of foreign-student participation.
Chapter 7 also broaches the topic of demand inducement. This topic is a black
hole into which entire lifetimes have vanished. My own “bottom line” comes
from the elegant randomized controlled trial reported at the end of the chapter
(Hickson, Altmeir and Parrin, HAP hereafter). “It” happens, but the importance
has to be weighed in every setting. This is a good time to talk about
methodological problems of measurement as well; for that, I love the Dranove
and Wehner (1994) article on demand inducement of child births, which
“demonstrates” using a fairly standard aggregate-data technique the OB doctors
“induce” more childbirths, an idea that leaves most people giggling at the very
thought. I use it as a way to get students thinking about the identification
problem in trying to estimate the extent of demand inducement.

On this issue (and I play this song for the RAND HIS, the Air Force para-medic
study, and elsewhere – indeed, anywhere when the occasion arises) I emphasize
the incredibly important value of a randomized controlled trial (RCT) as a device
for getting rid of these identification problems. That’s why I like the HAP article
on the pediatrics clinic: it uses randomization to get rid of the analytic problems.
My students know by the end of the semester that I have a deep love and affection
for RCTs as the most valid form of evidence.

Classroom Projects

The following projects can be used along with Chapter 6 or Chapter 7.

A. Search and prices:

You have several ways to organize this project, but the goal is to get students
searching for information in the market about prices and quality. I’ll use the idea of
teams again, but this can be done on an individual level in a sufficiently large city.
It can be done with reference do prescription drugs, physician office visits, dental
treatment, etc. An example “search” project: have the student(s) call half a dozen
different drug stores in town to find out the price of some commonly used
prescription drugs. Use drugs that have some possible meaning to the students,
such as allergy medications (Allegre, Benedryl, Claritin or Clarinex), pain relief
anti-inflammatory drugs (here, focus on branded drugs such as Motrin, Advil, and
the like and their generic counterparts, often sold by chain drug stores such as
CVS). Have them compare prices for the same drug and for generic counterparts of
branded drugs. Think also about www searches for drugs, and bring in issues such
as drug prices vs. shipping costs.
For physician or dental services, it will be illuminating for them to try to find out
the “price” of a doctor visit or dental visit. They may have to call up and pretend
that they are searching for a new doctor (or dentist), and ask what the price is for a
regular checkup (or something similar). They will be asked what insurance they
have, and should be prepared to try out several answers, including “no insurance”
or something generic like “Blue Cross”. Have them discuss what they found in
class. Many may find that the receptionist will not discuss prices over the phone,
which is part of the exercise. Note: this may be more complicated in a small town,
where there are only a few primary care physicians; think about the effect on the
doctors’ office of having dozens of students calling on the same day! This may not
be such a good project in a small college town.

B. Spend a Night Out with your students.

Get a (legal!) DVD copy of the 1981 movie “Who’s Life Is It Anyway”, starring
Richard Dreyfuss as an artist who becomes a quadriplegic in an auto crash and
wishes to die, and his doctor (played by John Cassavetes) who wishes to prevent
this outcome. Although the legal issues have changed some since the film (or the
Broadway play upon which it was based), it’s still a powerful study of legal and
ethical issues relating to intensity of treatment and physician style. The trick in the
discussion afterwards is to get somebody to defend the behavior of the doctor
played by Cassavetes, although the “style” he portrays is very common, according
to Emanuel and Emanuel and subsequent research.

Sample Exam Questions with Suggested Answers

1. (15 points) In a specific year in the mid-1990s, out of approximately 250 first
year residency positions in anesthesiology, only 47 of them "filled" (i.e., had
doctors agreeing to train at hospitals who wanted those doctors). This was far
lower than past years, where not only did every anesthesiology residency "fill," but
there were many more applicants than positions. To what economic forces would
you attribute this change? (Be clear on the link between the forces you identify and
the outcomes described.) Do you expect this change to persist for many years, or to
go away within a year or two? Why?

The perceived return on this particular form of specialty training probably had
fallen. At least three forces lead to this: First, managed care has reduced the
amount of surgery; second, new drugs have replaced surgery in some important
cases. Both of these lead to reduced demand for the “input” factor of
anesthesiology services. Finally, more and more routine surgery is done with
nurse-anesthetists rather than MD-specialist anesthesiologists, so this substitution
possibility also reduces demand for these services. All would lead to rational
reductions in doctors’ interests in training to become anesthesiologists.

2. [Note: This relates in part to Chapter 15 on Regulation, and you may wish to
defer it until that chapter has been covered.] (15 points) Agree or disagree, defend.
"Licensure of doctors and restriction on entry of foreign-trained physicians
unambiguously serves the interest of consumers by keeping bad doctors out of
practicing medicine.

The quality restriction may be a good idea if patients can’t judge the quality of
doctors readily, but if there were full information, then the quality restriction could
and probably would reduce utility of some patients. Think of requiring that ever
car sold have “Rolls Royce” or “Cadillac” quality; the cost would be prohibitive
to many consumers, who would have to forego the utility associated with owning
(say) a lower cost car.

3. [This really covers a lot of territory, including managed care, regulation,

technological change, but it fits in this chapter as a final exam question.] (25
points) Discuss what you expect to happen to the average annual earnings of
specialized surgeons (heart surgeons, orthopedic surgeons, neurosurgeons, etc)
over the coming decade or so, taking into account all the changes in government
policy, market forces, trends in health insurance markets, and innovation in
medical treatments that you know about.

See the answer in part to question 1 in this section. In addition, Medicare’s

revisions of payment to doctors rewards “cognitive” services more and
procedures less, which cuts against surgeons (pardon the expression).

4. (15 points) Deans of medical schools commonly worry considerably about the
debt level of physicians entering practice after their medical training. Do you
share their concern? If so, why? If not, why not?

It’s a big debt, but the economic returns on the education far exceed the costs of
borrowing, so the students are almost certainly better off financially by going to
medical school, even with the debt burden.

5. (10 points) What do you think would happen to the price of physician services
in the US if barriers to foreign-trained physicians were eliminated? Why?
Unless you believe in almost uncontrolled demand inducement, any standard
market model, including monopolistic competition, would lead to the conclusion
that adding more suppliers to a market will reduce the price. In the case of
physician services, we have to think about the physician-qua-labor and hence
judge its effect on final product prices. It’s possible, for example, that a backward
bending supply curve in the labor market would alter the “standard” analysis.

6. (20 points) Why do we not see more large physician group practices? [Hint:
think about aspects of the medical practice itself and also consequences for
consumers.] What evidence do you know that speaks to the relevant issues?

Travel costs of patients would make “centralization” of medical offices an

undesirable market outcome. Second, the costs of monitoring behavior and
eliminating shirking in large organizations would rise. A number of studies
speak to this latter issue, e.g., Gertler and Gaynor’s study of physician
productivity in various practice arrangements.

7. (5 points) Describe the origins of the name of the Mayo Clinic.

Pure whimsy; it’s not food poisoning, as the footnote in Chapter 6 asserts. I had
one professor email me to say that he included this as a bonus question to find
out who was actually reading the textbook. I also (to my mild embarrassment)
had a student in my Health Economics class at the University of Rochester ask
me about this footnote, saying (somewhat incredulously) “that’s not really true,
is it?” I had to politely tell him that it was all a joke, designed to keep students
from falling asleep while reading the textbook.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. AC = TC/Q. Minimum AC occurs where the derivative with respect to Q equals

zero. Thus we must find d(AC)/Q = {Q[d(TC)/dQ] - TC}/Q2 = {Q*MC -TC}/Q2
since MC = d(TC)/dQ. Now set this to zero, and we have Q*MC = TC, or MC =
TC/Q. But TC/Q = AC, so MC = AC. Thus when AC is at a minimum, AC = MC.
(Note: This is really a basic microeconomics concept, and is included here just for
reinforcement of the basic concept.)
2. a) AVC = a1 + a2Q + a3Q2
b) MC = a1 + 2a2Q + 3a3Q2.
c) As in problem 1, take the derivative of AVC and set it to zero. This occurs
when
a2 + 2a3Q = 0, so Q* = -a2/2a3.
d) Call the result Q**. Then Q** (similarly) equals -2a2/3a3.
e) Q** is 2/3rds of the size of Q* for this function.

*3. The firm competes in the final product market with other firms that may have
different numbers of doctors, staff, and so on. The firm assembles a variety of
inputs (doctors, nurses, office space, etc.) to produce the product (physician visits).
The physician laborer may individually have a backward-bending supply curve,
but in part because of the opportunities for substitution, there is no particular
reason to believe that the supply curve of the physician-firm is backward bending.
In sum, physicians are a factor input, combined with other factors to create the
final product of the physician-firm.

*4. Going to medical school and specializing in surgery is profitable, but the high
annual incomes are not appropriate evidence on this point. Rather, one must look at
the internal rate of return for the income flow from specializing in surgery versus
that from other alternatives. Surgical training requires many years (4 to 8,
depending on the specialty) after medical school, in which earnings are lower than
the market alternative, not to mention 4 years of medical school, entailing outlays
of up to $100,000 total, and the forgone earnings of the alternative occupation.
Any free market will show that those who do enter surgery will have higher
earnings than the average person during the years that they practice, to make the
investment of time worthwhile. Thus, the high annual earnings of doctors in
practice tell us nothing in particular about the profitability of the decision.

*5. The cost share of medical malpractice insurance is far too small to explain the
large increases in physician prices over recent decades. (See Box 6.2 on physician
cost structures.) Overall, malpractice insurance costs about 10 percent of all
nonphysician costs in a typical physician-firm and is easily less than 5 percent of
total costs, including physician-time payments. Thus, one could (for example)
quadruple the malpractice insurance costs and not much would happen to the total
costs of a physician-firm. For example, suppose that malpractice insurance costs
$5000; total non-M.D. costs are $50,000 including the insurance; and the physician
payments are $80,000, making total costs of the firm $130,000 and insurance costs
3.8 percent of total costs. Now let malpractice insurance costs go up by a factor of
4 to $20,000 and all other costs remain the same. Then total costs rise to $145,000,
which is an 11.1 percent increase in total costs, and in most market settings, a
smaller proportional increase in final product price.

6. Patient travel time does this. Think about a model of production where the
doctors paid for the transportation costs of patients (e.g., by sending taxicabs to
their door at home). Then fewer physician firms would obviously have (on
average) larger transportation costs associated with their business, eventually
leading to diseconomies of scale and a limit to the size of the firms.

*7. At least four sources might arise, three in the short run and one in a long run.
First, doctors can always work longer hours than they did before. We have good
evidence from HMO versus fee-for-service performance that hours of labor are
sensitive to marginal rewards to physician time. Second, some doctors might
decide not to retire when they might otherwise have chosen to do so. Third, we can
import doctors from foreign countries, as happened dramatically in the 1960s and
1970s. Finally, medical school and residency training positions can increase,
creating a long-run increase in the stock of physicians.

8. One reason: they might value their patients’ health, and believe that they can do
a better job than nurses and aides in creating “cures.” The value they place on
patients’ health would lead to this result. Second, the state’s regulatory processes
and licensure laws might inhibit use of nurses and aides. Third, the doctor might
perceive the shadow price of his/her own time differently than society.

9. In groups with shared resources, the “common good” may not be effectively
“charged out” to each doctor using it, and thus (like any “common good” problem)
the good in question will get over-used. (Think about the use of a copy machine in
the office in a world where there was a per-copy charge vs. one where the machine
was unmetered.) Another way to put this is that if a single doctor saves $X by
being careful, the savings are split among N doctors in the group in many
partnerships of doctors.

There is also a problem of observing the true effort of doctors in a large

group. A doctor can “goof off” and remain undetected, thus receiving the same
salary as a hard worker but doing less work (or, in another way of thinking about
it, taking a bit of vacation on the job while others work hard).

10. Probably the most likely explanation is that small groups have not reached a
scale sufficient to provide support services and fringe benefits to staff in the most
economically efficient way. The “buying power” of the hospital for these sorts of
things will be relatively more important to smaller physician firms. On the flip
side, smaller firm physicians (by dint of being in a small firm) have expressed
some preferences for autonomy, and giving up some of that autonomy may be
more costly to them. On balance, it seems that the economies of scale “win out,”
since smaller physician firms have been more likely to sell their practices to
hospitals than larger firms.

Another possible issue (the data are not available one way or the other to test this)
is that specialty mix is a confounder. It may be that hospitals are mostly acquiring
“primary care” practices to act as feeders for their specialty groups. If relatively
small primary care practices were most likely to be acquired, it would appear as if
it was size-related when in fact it might be specialty-related.
 CHAPTER 7

PHYSICIANS IN THE MARKETPLACE

Chapter Outline

Physician Location Decisions

Consumer Search and Market Equilibrium
Actual Search by Patients
Advertising and the Costs of Information
The Role of Licensure
Estimates of the Demand Curve Facing Physician-Firms
Induced Demand
The Role of Payment Schemes
Summary
Related Chapters in Handbook of Health Economics
Problems

Teaching Tips from the Author

The following Teaching Tips apply to both Chapter 6 and Chapter 7 and repeats
what begins the Chapter 6 discussion.

I find myself wandering back and forth between the material in these two
chapters, since the boundary line is somewhat vague. There are common
threads I try to emphasize.

First, doctors’ behavior has systematic elements of standard economic behavior.

The returns to an MD degree make it attractive, and ditto the returns to
specialization. The spatial location problem presents another good point to talk
about economic incentives. I always begin this discussion with Harold
Hotelling’s ice cream vendors on a linear beach, since it’s such an easy way to
represent spatial competition. Oh, yeah, specialty choice is like spatial choice in
many ways.

I try to emphasize the distinction between the physician firm and the physician as
an individual, since confusion on this point can compound itself a lot. (My old
mentor in graduate school, Reuben Kessel, confused these issues badly by talking
about entry restriction into the profession as the source of monopoly power in
individual practices. Of course, this can’t be correct. If there is monopoly power in
the final product market, it relates to search, not the entry restriction on individual
doctors. There’s a quasi-rent associated with the entry restriction, but that’s in a
factor market, not the final product market.)

This issue, of course, brings with it the distinction between the demand curve at
the market and the demand curve for the individual firm. We have precious few
estimates of the latter in the market for physician services (and indeed, again for
hospitals in later chapters). This is another reason why I like the work of Wilde
and Schwartz (and Sadanand and Wilde): it shows how search affects the demand
elasticity facing a single firm. I describe it this way to students: the entrepreneur
faces a market where some patient search for a lower price, and others don’t. For
those who don’t search, the entrepreneur can treat the demand curve as a scaled
down (random sample?) of the market demand curve (with the same elasticity),
and can price accordingly as a monopolist. After all, if somebody picks a single
firm to buy from and never searches, and the firm knows this, we have a
monopoly situation. But... if the patient does search (and finds a lower price), then
the entrepreneur loses all that business. So the pricing strategy of the firm is a
blend of the monopolistic tendencies for the non-shoppers and the urge to attract
the shoppers. Obviously, the more shopping, the more price sensitivity, and the
lower the equilibrium price. I can also work in the role of insurance, of course, on
those elasticities and incentives to search.

I also like to emphasize the potential of alternative ways of producing similar

services. If you have nurses (especially nurse-practitioners) in your class, they’ll be
happy to have you discuss nurse-practitioners as an alternative to physicians in
primary care, midwives in obstetrics, etc. I like the old RAND study from the Air
Force clinic on this issue, because it separates out the technical production
possibilities from the legal prohibitions. (The US gum’mint doesn’t have to pay
attention to state licensing laws!)

As you will see in Chapter 7, I’m a big fan of the monopolistic competition model
to help think about physician markets. It makes a lot of sense to me, and it really
emphasizes (particularly the work of Louis Wilde and various colleagues) the
importance of both the demand elasticity and search on the market equilibrium.
This ties back very nicely to Chapter 4 and the effect of insurance on both the
demand elasticity (insurance drives the elasticity towards zero, which increases the
equilibrium price in a monopolistic competition environment) and incentives to
search (most insurance plans obliterate incentives to search for price, but not
quality.) Sometimes I try to get the class involved in a discussion about how you
find a doctor, but most of the undergraduates I teach have not had much experience
in that realm. That discussion would bear more fruit in a graduate class of MPH,
MD, or other such students.

Chapter 7 also broaches the topic of demand inducement. This topic is a black
hole into which entire lifetimes have vanished. My own “bottom line” comes from
the elegant randomized controlled trial reported at the end of the chapter
(Hickson, Altmeir and Parrin, HAP hereafter). “It” happens, but the importance
has to be weighed in every setting. This is a good time to talk about
methodological problems of measurement as well; for that, I love the Dranove and
Wehner (1994) article on demand inducement of child births, which
“demonstrates” using a fairly standard aggregate-data technique the OB doctors
“induce” more childbirths, an idea that leaves most people giggling at the very
thought. I use it as a way to get students thinking about the identification problem
in trying to estimate the extent of demand inducement.

On this issue (and I play this song for the RAND HIS, the Air Force para-medic
study, and elsewhere–indeed, anywhere when the occasion arises) I emphasize the
incredibly important value of a randomized controlled trial (RCT) as a device for
getting rid of these identification problems. That’s why I like the HAP article on
the pediatrics clinic: it uses randomization to get rid of the analytic problems. My
students know by the end of the semester that I have a deep love and affection for
RCTs as the most valid form of evidence.

Classroom Projects
The following projects can be used along with Chapter 6 or Chapter 7.

A. Search and prices:

You have several ways to organize this project, but the goal is to get students
searching for information in the market about prices and quality. I’ll use the idea of
teams again, but this can be done on an individual level in a sufficiently large city.
It can be done with reference do prescription drugs, physician office visits, dental
treatment, etc. An example “search” project: have the student(s) call half a dozen
different drug stores in town to find out the price of some commonly used
prescription drugs. Use drugs that have some possible meaning to the students,
such as allergy medications (Allegre, Benedryl, Claritin or Clarinex), pain relief
anti-inflammatory drugs (here, focus on branded drugs such as Motrin, Advil, and
the like and their generic counterparts, often sold by chain drug stores such as
CVS). Have them compare prices for the same drug and for generic counterparts of
branded drugs. Think also about www searches for drugs, and bring in issues such
as drug prices vs. shipping costs.

B. For physician or dental services, it will be illuminating for them to try to find
out the “price” of a doctor visit or dental visit. They may have to call up and
pretend that they are searching for a new doctor (or dentist), and ask what the price
is for a regular checkup (or something similar). They will be asked what insurance
they have, and should be prepared to try out several answers, including “no
insurance” or something generic like “Blue Cross”. Have them discuss what they
found in class. Many may find that the receptionist will not discuss prices over the
phone, which is part of the exercise. Note: this may be more complicated in a small
town, where there are only a few primary care physicians; think about the effect on
the doctors’ office of having dozens of students calling on the same day! This may
not be such a good project in a small college town.

Sample Exam Questions with Suggested Answers

1. [Note: this cannot be given until Chapter 16 has been read, but it relates to
demand inducement and physician incentives.) (10 points) Agree or disagree. "The
Japanese have higher use of prescription drugs than any other industrialized
country in the world because they are the only country to include complete
prescription drug coverage as a part of the "normal" health insurance that
everybody receives.

In Japan, doctors not only prescribe but sell drugs directly in their offices and
clinics. Thus they have a strong financial benefit from prescribing, and the
minimal transactions costs for their patients to fill the prescription “on the spot”
give them a price advantage. Thus it’s not the inclusion of drug coverage (which
is commonplace now in modern health insurance) but the selling arrangement
that matters.

2. (10 points) "In markets for physician services, prices differ across providers
in a single community mostly because of quality differences."

Incomplete search will create price differences even when quality is identical.
3. [Note: This relates also to Chapter 3.] (10 points) Two concepts that have some
similarity in the literature on health economics are "variations in medical care" and
"demand inducement." Indeed, some people have suggested that demand
inducement "explains" variations in medical care. (a) In one or two sentences, what
key feature about "demand inducement" distinguishes it from "variations in
treatment rates.”? (b) What evidence can you use to support the idea that demand
inducement DOES NOT account for or "explain" variations in medical practice
use.

Demand inducement arises from some financial (or related) advantage to the
provider. Variations arise (presumably) because of incomplete information or
disagreement on the part of providers about the benefits of various treatments.
The strongest evidence that variations arise for something other than profit
motives comes from places like Great Britain, where doctors are on salary
(rather than fee for service) and hence have no monetary gain from inducing
demand; variations are common there and similar to those found in (say) the
U.S.

4. (15 points) "The way doctors get paid has nothing to do with the amount and
type of care they deliver. They mostly are concerned with healing patients."

Utter nonsense. Compensation to doctors definitely affects their treatment

recommendations and work intensity. The best RCT example is the study of
pediatric clinic doctors by Hickson, Altmeier and Perrin, sited in the references
and discussed in the chapter. Gertler and Gaynor also show how compensation
affects work effort. Comparison of HMO and regular insurance plans also shows
much less use of invasive procedures such as surgery in the salary-model HMO.

5. [This uses the concept of externalities that Chapter 14 discusses, but it fits here.]
(15 points) An externality is defined as an individual’s activity either imposes costs
or confers benefits on others, but individuals do not take account of such costs or
benefits in their own decision making.
Under that logic, would you define “comparison shopping” as an externality?
Discuss the logic of your answer in terms of market pricing models.

Comparison shopping creates a positive externality for non-shoppers, since the

shoppers “police the market” for the non-shoppers.

6. [This requires a discussion of drug use in Japan, Chapter 16, but fits here for a
final exam question.] (15 points) Japanese citizens have the highest rate of use of
prescribed drugs in the world. Similarly, citizens of Florida who see doctors who
own radiology centers get referred more for radiologic diagnosis than patients of
doctors who do not own such facilities. Discuss the common origins of such events
from an economic viewpoint.

In both cases, the doctor “owns” an enterprise that sells a service that the
doctor prescribes (drugs or therapy or diagnostic tests). Thus these practices
have the common theme that the doctor is enriched by prescribing the specific
treatment. This fits in the general heading of demand inducement well.

7. (20 points) Concepts discussed in this class about physician choices include (a)
what specialty they choose, (b) where they locate, and (c) the nature of interaction
between physician firms in a single market. Describe the essential idea(s)
associated with our understanding about how doctors make these choices. Do you
see any common themes in these ideas?

In (a) and (b), we have what amounts to a Hotelling model of spacial

competition; doctors will tend to migrate to regions (specialties) with higher
economic returns to their investment. The process of competition will tend to
equalize the returns across region (specialty) through natural forces of entry and
exit. This process, of course, is at the heart of the monopolistic competition
model of physician markets to which (c) refers, since entry is a paramount
feature of that model.

8. (10 points) Assume that you are a policy analyst for Blue Cross/Blue Shield of
Elsewhere, Nevada (population 3,500). Your boss is trying to get more specialized
surgeons to move to your area, and proposes to do so by having a lot of fancy
dinners to which prospective doctors are invited from around the country.
Comment on the likely success of his proposal.

Not likely to work; the dinners will have little effect on the lifetime earnings of
the doctor, which will probably have a more prominent effect on the choice.
Towns of 3500 population are not sufficiently large to support most specialists.

9. The market for physician services is discussed as a monopolistically

competitive market (rather than purely competitive or purely monopolistic). (5
points: a) How does the price in such a market compare with the two “extreme”
cases of pure competition or pure monopoly, both in terms of level and
variability? (10 points) b) What key forces drive the market to this type of
monopolistic competition instead of having a competitive market?
The price will fall somewhere between these two extremes (part a), reaching the
competitive market when search becomes sufficiently complete. Thus incomplete
search is the driving force keeping the market from becoming competitive.

10. [Yet another demand inducement question.] (15 points) There has been
considerable debate about the existence and extent of demand inducement by
doctors. Discuss the most compelling evidence that you know of regarding the
existence of demand inducement. How much does this tell you about the overall
extent of inducement?

The most compelling evidence to me is the RCT in the pediatric clinic (Hickson,
Altmeier and Perrin), because the research design eliminates confounding
variables. The differences in use of radiologists’ services and therapy services for
those doctors in (say) Florida who own (and do not own) the service also
provides strong evidence for the existence of demand inducement.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. The work on physician location choices after the big increase in medical school
output speaks against this quote. If they could have generated enough demand to
remain in luxurious settings, we would not have seen the relative shift of doctors to
rural areas that these data show.

2. In a monopolistically competitive market, more search increases the demand

elasticity confronting each provider, thus causing the equilibrium price to fall
closer to the minimum average cost in the market. Advertising, especially if
“informative” in nature, reduces the cost of search. Thus advertising could work
adversely to the interests of doctors in the market, and it could be in their self-
interest to induce the government to suppress advertising.

*3. With 15 doctors, the average is about 10,000 patients per doctor, and City C
(with only 6000 people) just can't compete for a doctor. The 15 doctors will divide
themselves with 10 in City A and 5 in City B, and every doctor will have nearly
the same proportional demand. Nobody has any incentive to move at this point, so
this allocation is stable.

With 30 doctors, the average is about 5000 patients per doctor, so City C is
now an attractive location with 6000 patients, and 1 doctor will locate there. The
tricky puzzle is then whether City A or City B will come up a little short to supply
the doctor to City C. Consider two alternative patterns of location:
Pattern I: City A = 19 doctors, City B = 10 doctors, City C = 1 doctor.
Pattern II: City A = 20 doctors, City B = 9 doctors, City C = 1 doctor.

Consider Pattern I first. A doctor in City B might think that City A looked
appealing, because there are 5210 patients per doctor (99,000/19) compared with
5100 (51,000/10) in City B. However, if a doctor moves from City B to City A, the
number of patients per doctor in City A will fall to 4950, less than the number
available now in City B. Thus Pattern I is stable. By the same logic, Pattern II is
not stable, since a doctor in City A would find it attractive to move to City B. Thus
Pattern I will emerge. The doctor in City C will earn unusually large returns, but
nobody else can afford to move there, because it would split a small market into
two parts that would both be smaller than the market remaining in either City A or
City B in Pattern I.

4. In an experimental context (the most solid form of data), the study of the
pediatrics clinic doctors’ behavior showed the presence of demand inducement, but
it was not unlimited in size. Similarly, the comparisons of patterns of prescribing
(for example) x-ray studies for those doctors who own (and do not) x-ray facilities
provide a non-experimental evaluation of demand inducement’s existence. Again,
while present, demand inducement is not extremely large, based on this
comparison.

The pediatrics clinic data also reveal shirking, since the salary-paid doctors
“undershot” the well-care targets of the American Academy of Pediatrics (a well
known standard of care to pediatricians).

Since shirking (or under-treatment) and demand inducement (overtreatment) have

both been shown to occur with their related payment mechanisms (salary and fee
for service), we cannot say anything a priori about which is better for patients.

*5. Doctors are central to discussions of demand inducement because there is a

large informational asymmetry between doctors and patients and also because the
law has put doctors in the special position of being the only ones who can do
certain things (such as prescribe drugs and perform surgical operations).

*6. The doctor-patient relationship refers to the building of specific knowledge by

the doctor about the patient (and vice versa). This knowledge may be contained in
records kept by the doctor or even in subtle visual clues such as changes in skin
color, anxiety of the patient, and the like. This sort of specific knowledge should,
on average, allow doctors to make diagnoses either at lower cost or with higher
quality than if they did not have such specific knowledge. The higher quality of
care that should (on average) arise from this arrangement should make the patient
more willing to continue treatment with the same doctor, creating, if you will, an
economic barrier to changing providers. This barrier in turn allows the doctor to
exploit the patient's unwillingness to change providers, e.g., by ordering more
diagnostic tests than necessary and providing them through facilities owned by the
doctor. Not all doctors will do this, of course, but the economic opportunity
certainly exists.

7. The difference comes from between-firm search by patients. The elasticity

facing a single firm contains both the market demand elasticity (patient
preferences) and the results of search by patients (those finding lower prices
moving to the lower priced providers).

8. It is difficult to conclude anything with strong conviction. Doctors are

presumably more knowledgeable about the value of treatment, and hence harder to
deceive into using “too much” care. If all else were equal, we’d say that the data
show no inducement. But doctors and their families often receive “professional
courtesy” price reductions that should increase utilization. It is possible in concept
both to have demand inducement (increasing the use of the non-doctors) and
doctors getting more treatment (because of the lower price).
 CHAPTER 8

THE HOSPITAL AS A SUPPLIER OF MEDICAL CARE

Chapter Outline

The Hospital Organization

Who Is the Residual Claimant?
Where Does the Utility Function Come From? A Political Theory Model
Hospital Costs
Long-Run versus Short-Run Costs
The Hospital’s “Cost Curve”
Another Complication: Outpatient Surgery
The Demand Curve Facing a Single Hospital
The Utility-Maximizing Hospital Manager Revisited
Summary
Related Chapters in Handbook of Health Economics
Problems

Teaching Tips from the Author

The following Teaching Tips apply both to both Chapters 8 and 9.

Here again my fondness for the monopolistic competition model emerges. Chapter
8 works a lot at helping the student understand the unique, if not bizarre nature of
the not for profit (NFP) hospital in the US. What does NFP mean? How does
behavior change compared with the for profit hospital? What evidence do we have
about that?

Of course, the other key feature (perhaps even more important than the NFP status)
is the unique role of the doctor as “captain of the ship” even if not an employee of
the hospital. (Sidebar: the only other setting remotely like this that I know of is the
Harbor Pilot who guides ships through a harbor and into port. The Harbor Pilot’s
familiarity with the dangers, geography, etc. of the harbor give him legal primacy
even over the ship captain. That’s the only thing I know of remotely like the
doctor/hospital situation.)
Sometimes I try to get students thinking about how universities and colleges would
operate if professors were independent free agents, came onto campus to give
lectures, and could “demand” that the school provide all sorts of specialized
resources for our teaching. Wistful as I might sometimes get about such an idea
(before, but not after I became Provost of the University of Rochester!), I think the
students see immediately how bizarre that world would look.

I spend a fair amount of time on the model of the hospital with a leader’s utility
function. Numerous “capture” theories have emerged (doctor’s workshop,
administrators—a variant on the principle-agent problem, nurses (or staff more
generally, with higher wages), but nobody seems to have proposed the co-op
model (i.e., that the organization serves the patients)! The idea of a utility function
for the hospital goes back to an early work by my good friend and colleague, Joe
Newhouse. I used to complain to him that the model was vacuous because we
don’t know from where the utility function comes. But since then, I’ve done some
thinking on this issue (triggered mostly by my associations with the Political
Science Department of the University of Rochester, the birthplace of Positive
Political Theory), and I’ve come to conclude that one can create such utility
functions out of the normal board of trustees arrangements of NFP hospitals.
Chapter 8 provides a brief sketch of those ideas, and Chapter 9 boldly assumes a
stable utility function for the hospital and proceeds to analyze market behavior in
that world, using (no surprise by now, I presume) a monopolistic competition
framework.

Chapter 9 offers a wide range of exercises in market equilibrium. Laying out the
basic idea is tricky and I go through it very slowly. It’s not a single demand curve
facing a hospital, but rather a family of them, one for each possible quality level.
It’s not a single cost curve (which of course means I need a concept even of “a
single” cost curve–I like the ray-average cost idea intuitively), but a family of
them. Now interpose the idea of the NFP rule and we have a market-determined
equilibrium choice set (the EE curve). The next step translates that into utility
space (and I have a utility function developed in Chapter 8), so I can move the EE
curve into a production possibilities frontier which I label the FF curve. I carefully
construct a single point on the EE curve and then shift that same point to the FF
curve, then again, and then chalk in the FF curve overall. Now I have Newhouse’s
model intact, and I can talk about market equilibrium.

This presents a wonderful opportunity to do “what if” models of hospital

interaction. What if insurance coverage increases? (All the EE curves and hence
FF curves move out.) What if a new hospital opens up in town? (The EE and FF
curves of extant competitors shift in.) What if the costs of operating a hospital
increase? (The cost curves shift up, but willingness to pay doesn’t, so the EE and
FF curves shift in.) Etc. It’s a great and glorious mutatis mutandis game, but the
fun thing is, it really seems to help understand such things as intertemporal
increases in hospital spending (income and insurance effects in part), the medical
arms race phenomenon, and (later in the chapter on regulation), why it is possible
for a regulator to improve on the competitive equilibrium (but not likely to
happen!).

Classroom Projects
The following project can be used along with Chapter 8 or Chapter 9.

A. If possible, arrange for a visit to a local hospital in the emergency department,

and to a normal “floor” for regular treatment. If several hospitals in town have very
different characteristics, send teams to each and have them report back on what
they saw. The differences between a major metropolitan “county” hospital and a
suburban 150 bed general-care hospital will provide a stark contrast. Have them
prepare a list of things to look for, such as how busy the place was, how many
people seem to be waiting for care, the age, gender and racial profile of patients,
etc.

Sample Exam Questions with Suggested Answers

1. (10 points) What single organizational feature about not-for-profit hospitals in

the US differs most from the organization of other corporations and organizations.
(Note: I'm not talking about the not for profit status itself here—many institutions,
including schools, churches, charities, social service organizations, etc. all share
the common feature of not for profit ownership.) What major consequence arises
from this regarding the functioning of the hospital as an economic entity?

The key is that most of the economic activity in the hospital is dictated by
somebody who’s not an employee (the private-practice doctor). This leads to a
disconnection between the demands for resource use and the costs of providing
them. (See discussions of work by Jeffrey Harris, MD, PhD for an early and
illuminating insight into these arrangements.)

2. (10 points) "Most US hospitals are organized as not for profit corporations.
Because they don't have the force of the pure profit motive guiding their behavior,
we can be confident of the following: Not only do they produce the wrong
amounts and qualities of care (from a social point of view), but they invariably
will do so inefficiently." Discuss intelligently.

Cost minimization should still persist, since the goal-driven NFP hospital can
accomplish more if it produces efficiently. There may be allocative inefficiencies
arising from the motives (and tax differentials compared with FP competitors) but
we have no reason to expect X-inefficiency.

3. (15 points) Discuss intelligently the relationship between doctors and

hospitals in the "normal" US not for profit hospital. Be specific in terms of the
way doctors are compensated, and the role doctors have in the demand curves
for hospitals.

Many (but not all) doctors are independent contractors who have “admitting
privileges” in hospitals. This relationship in some ways makes the doctor the
customer, rather than the employee of the hospital, but it’s a mixed relationship.
More formally, the doctor is an agent for the patient (the true customer). In some
cases now, the hospital actually owns the practice of the doctor (the doctors are
employees), but the separate arrangement is still by far the most common.
In these cases, the doctor is paid on a fee for service basis, usually with the
intervention of some third party insurer (private or government). Demand curves
for hospitals, because of this arrangement, depend on the doctors they have “on
staff,” since patients cannot admit themselves into the hospital (by law).

4. (15 points) What is the biggest difficulty in measuring the “average cost” of
hospitals?

Case mix differences greatly confound this issue.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. The profits made in the NFP hospital must be “spent” internally on some
combination of increased quality and output. Thus the NFP hospital should have
higher quality and/or higher output than an otherwise-identical FP hospital.

*2. Doctors do, and typically they are not employees of the hospital. This feature,
in many ways, is the most unusual of all aspects of the not-for-profit hospital in the
United States. The administrative directors have little way to control the use of
resources within the hospital, once they are "in place" and a doctor orders their use
for a patient. Ultimately, only a doctor's persistent practice of very low-quality
medical care makes it easy to remove a doctor's hospital privileges. No other
organization operates in this way, including not-for-profit organizations such as
colleges, churches, and the like.

*3. Nothing, to be honest. The larger hospital will almost certainly have more
complex patients with more comorbidities, and so on, and will undertake more
complicated procedures. Both of these phenomena will raise the cost per patient in
a way that has nothing to do with "economies of scale" in the usual sense. To
understand hospital efficiency, one must carefully control for the case mix of the
patients in the hospital, and even after doing that, one must be certain that other
characteristics of the patients don't affect hospital costs. Only then would an
"efficiency" interpretation have any validity, and this almost never happens in real
hospital data.

*4. (b) They could end up with worse mortality statistics because of patient
selection. Those patients with the worst complications (other diseases, advanced
state of disease, etc.) might well get referred to the specialized treatment center
because it offered the best hope of cure. If the mix of referred patients was
sufficiently "far gone," then the specialized hospital might actually end up looking
worse, not better, in outcomes. This emphasizes the importance of understanding
patient severity mixes in hospitals when comparing their outcomes.

5. Among other things, they would be less likely to induce demand, and have
more opportunity and incentive to reduce their work load. Both of these lead to the
conclusion that per-person medical care use will be lower in Britain than in the US,
subject to additional differences arising from income, physician/patient ratios, etc.

*6. Larger hospitals typically have more specialized treatment capabilities. This is
a general reflection of Adam Smith's observation that "division of labor is limited
by the extent of the market." Large hospitals need large patient bases from which
to draw customers. In those large population centers, specialization will occur not
only within doctors' practices (see Chapter 7 and discussion surrounding Table 7.2)
but also in hospitals. Put another way, a small rural hospital will never have
enough cardiac surgery patients in the area to justify a specialized cardiac surgery
capability, and if one is built, it would probably never be able to attract a
competent cardiac surgeon to the region (because that surgeon would prefer to
practice in a larger city where there were enough patients to sustain the practice).
7. The Figure would be like Figure 8.5 except that the vertical line at N* would
shift out to the right so much that the utility of Trustee B would be lower than U1
at that output. For example, if N* were set at the intersection of BB and U0, this
would be the case.
 CHAPTER 9

HOSPITALS IN THE MARKETPLACE

Chapter Outline

Hospitals and the Market for Medical Staff

Hospitals and Patients
A Model of Equilibrium Quality and Price
Insurance and Competition in the Hospital’s Decision
How Doctors and Hospitals Interact: “Goodies” for the Doctor
How Doctors and Hospitals Interact: Patients for the Hospital
Competition—“Old Style” Versus “New Style”
Entry and Exit: The Pivotal Role of For-Profit Hospitals
The Hospital in Labor Markets
Nursing “Shortages”
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 9: The Hospital’s Quality and Quantity Decision

Teaching Tips from the Author

The following Teaching Tips apply to both Chapter 8 and Chapter 9, and repeats
material provide about Chapter 8.

Here again my fondness for the monopolistic competition model emerges. Chapter
8 works a lot at helping the student understand the unique, if not bizarre nature of
the not for profit hospital in the US. What does NFP mean? How does behavior
change compared with the for profit hospital? What evidence do we have about
that?

Of course, the other key feature (perhaps even more important than the NFP status)
is the unique role of the doctor as “captain of the ship” even if not an employee of
the hospital. (Sidebar: the only other setting remotely like this that I know of is the
Harbor Pilot who guides ships through a harbor and into port. The Harbor Pilot’s
familiarity with the dangers, geography, etc. of the harbor give him legal primacy
even over the ship captain. That’s the only thing I know of remotely like the
doctor/hospital situation.) Sometimes I try to get students thinking about how
universities and colleges would operate if professors were independent free agents,
came onto campus to give lectures, and could “demand” that the school provide all
sorts of specialized resources for our teaching. Wistful as I might sometimes get
about such an idea (before, but not after I became Provost of the University of
Rochester!), I think the students see immediately how bizarre that world would
look.

I spend a fair amount of time on the model of the hospital with a leader’s utility
function. Numerous “capture” theories have emerged (doctor’s workshop,
administrators—a variant on the principle-agent problem, nurses (or staff more
generally, with higher wages), but nobody seems to have proposed the co-op
model (i.e., that the organization serves the patients)! The idea of a utility function
for the hospital goes back to an early work by my good friend and colleague, Joe
Newhouse. I used to complain to him that the model was vacuous because we
don’t know from where the utility function comes. But since then, I’ve done some
thinking on this issue (triggered mostly by my associations with the Political
Science Department of the University of Rochester, the Birthplace of Positive
Political Theory), and I’ve come to conclude that one can create such utility
functions out of the normal board of trustees arrangements of NFP hospitals.
Chapter 8 provides a brief sketch of those ideas, and Chapter 9 boldly assumes a
stable utility function for the hospital and proceeds to analyze market behavior in
that world, using (no surprise by now, I presume) a monopolistic competition
framework.

Chapter 9 offers a wide range of exercises in market equilibrium. Laying out the
basic idea is tricky and I go through it very slowly. It’s not a single demand curve
facing a hospital, but rather a family of them, one for each possible quality level.
It’s not a single cost curve (which of course means I need a concept even of “a
single” cost curve–I like the ray-average cost idea intuitively), but a family of
them. Now interpose the idea of the NFP rule and we have a market-determined
equilibrium choice set (the EE curve). The next step translates that into utility
space (and I have a utility function developed in Chapter 8), so I can move the EE
curve into a production possibilities frontier that I label the FF curve. I carefully
construct a single point on the EE curve and then shift that same point to the FF
curve, then again, and then chalk in the FF curve overall. Now I have Newhouse’s
model intact, and I can talk about market equilibrium.
This presents a wonderful opportunity to do “what if” models of hospital
interaction. What if insurance coverage increases? (All the EE curves and hence
FF curves move out.) What if a new hospital opens up in town? (The EE and FF
curves of extant competitors shift in.) What if the costs of operating a hospital
increase? (The cost curves shift up, but willingness to pay doesn’t, so the EE and
FF curves shift in.) Etc. It’s a great and glorious mutatis mutandis game, but the
fun thing is, it really seems to help understand such things as intertemporal
increases in hospital spending (income and insurance effects in part), the medical
arms race phenomenon, and (later in the chapter on regulation), why it is possible
for a regulator to improve on the competitive equilibrium (but not likely to
happen!).

Classroom Projects

The following project can be used along with Chapter 8 or Chapter 9.

A. Have your students ask their parents about their history of surgical procedures
and normal births. In particular, have them ask how long their mother was in the
hospital when they were born. Compare the average length of stay with current
estimates for common surgical procedures and normal delivery.

Sample Exam Questions with Suggested Answers

1. (20 points) Discuss how a hospital's demand curve differs from the market
demand curve for hospital services, discussing specifically the role of other
hospitals' price and quality and the role of each hospital's medical staff. Under
what circumstances would the elasticity of the hospital-level demand curve match
that of the market level demand curve? When these circumstances do not hold,
would the hospital-level demand curve have a larger or smaller elasticity (in
absolute value)? Why?

The hospital’s demand curve is more elastic than the market demand curve. It
contains not only the intrinsic price-elasticity of the market, but also elements of
provider-switching on the basis of price. The hospital and market curves look the
same (except for scale) in the case where NO consumer search takes place and
patients get randomly assigned to hospitals. When at least some shopping (for
price) takes place, the hospitals’ demand curve is more elastic than the market
curve.

2. (15 points) Agree or disagree, defend your choice: "The market should force
every hospital to produce at the same quality, reflecting the preferences of the
typical consumer in the region where the hospital is located." (You may wish to
use a brief diagram to illustrate your point.)

Disagree. Standard monopolistic competition models lead to the possibility of

market segmentation, which can actually increase consumer welfare, even if
price is above minimum average cost. (It depends on whether the costs of
producing quality variations are lower with multiple firms than a single firm,
among other things.) Adding to that the proclivities of hospital leaders to choose
different price/quality/quantity mixes makes it unlikely that all hospitals will
settle on the same quality.

3. The following diagram [Note to professor: insert diagram here like Figure 9.2,
but without labels.] shows a not for profit hospital's possible ways of operating.

a) (5 points) Label the axes and each of the lines shown in the figure. (You need
not use the same labels as in the textbook, but you should define what you mean in
each case using careful economic definitions.) There is one important line missing
from this figure: include and label it.

See the figure for correct labels.

b) (10 points) Show what happens in this diagram if the extent and generosity of
health insurance in the community in which this hospital operates increases (i.e.,
insurance coverage improves in the market). Carefully describe why you draw
the additions to this figure that you include. PLEASE think before you draw! If
you cross out lines, it greatly complicates the problem of grading.

Demand curves for all hospitals should increase for all qualities, so the EE and
FF curves shift outward. The equilibrium utility reached by the hospital should
rise as the possibilities frontier shifts outward.

4. [This relates to Chapters 5 and 7 also.] (10 points) Demand for physician
services as measured (e.g.) in the RAND Health Insurance Study has an elasticity
of about -0.2. Studies of demand curves facing individual physician firms show
much larger elasticities (in the range of -2 to -3 or so). Similarly, demand for
hospital care shows an elasticity of about -0.15, yet demand curves facing
individual hospitals are estimated to have an elasticity of about -4 or so. Explain
how these differences can emerge. One study found that the elasticity was -4 in
general, but -2.3 for Blue Cross (the best-insured) patients. Discuss why you think
this might happen?

The difference arises because of price-sensitive choices of the provider, above and
beyond the general market-level demand elasticity. It hinges on comparison
shopping; if people seek the lowest-price provider (for a given quality) then search
can drive the price down to the competitive level, which happens as the elasticity of
demand facing each provider increases to (negative) infinity. Elasticities like those
reported in the literature suggest some search, but not enough to produce a
competitive market.

5. (20 points). Consider a typical US not-for-profit hospital operating in a

community with several other hospitals. What would happen to General Hospital
(the one you are to tell me about) a grateful patient from another hospital (let’s call
it St. Elsewhere Hospital) left a large gift to the hospital in her will? Be specific in
terms of what you think the effects would be on General Hospital, and why you
think these effects would occur. Your answer will greatly benefit by the use of
several key figures that show how market equilibrium works with not for profit
organizations dominating the market.

The answer here uses figures like Figure 9.2 and 9.3, but draws them for two
hospitals. When St. Elsewhere get a gift, it becomes more attractive after the
investment, and hence some patients migrate to there from General hospital. The
student should show the demand curves for General Hospital shifting inward,
hence the EE and FF curves. The new equilibrium most likely has lower quality
and output.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. Look at Figure 9.3, and the tangency of FF to U2. This occurs at (S*, N*). Then
going back to Figure 9.2, we can find the cost of care by finding N* on the
horizontal axis and running up vertically to the highest possible intersection on the
EE curve. That will occur at the intersection of AC and D for the quality S*.
2. The demand curves in Figure 9.2 all shift inward as some of the hospital’s
business migrates to the new entrant. The cost curves are unaffected, so the EE
curve shifts inward as well. The corresponding movement in Figure 9.3 is an
inward shift in the FF curve. Thus the equilibrium output (N*, S*) will occur
“inward” from the original point, almost certainly leading to both lower quality and
quantity (unless the FF curve and/or the indifference curves have a very unusual
shape).

*3. Quality will almost certainly increase. The "frontier" of choices available to
the hospital will expand as insurance coverage expands, and by far the most likely
outcome is that hospitals will increase both quality and quantity as they move to
the new equilibrium point of production.

*4. If the hospital has any monopsony power, it will more likely emerge in the
more specialized fields. Thus, intensive care nurses are more specialized than
"general" nurses, who in turn are much more specialized than janitors. Even a
single-hospital town probably has no meaningful monopsony for janitors, since
such general-skilled labor can easily ebb and flow from and to other industries as
hospitals demand changes. General nurses are more specialized, but many of them
work in settings other than the hospital and are probably more mobile across
regions as well. Specialized nurses will exist only in the hospital setting, however,
increasing the chance of monopsony power. In the standard monopsony setting,
"unfilled positions" will always exist, and, thus, we would expect to see ads by the
hospital for those professions.

*5. Because hospital admissions rose from 1993 forward but patient days
remained stable, it must be true that hospital length of stay declined. (Chapter 12
discusses this for Medicare patients.) Because a considerable amount of surgery
shifted to outpatient surgery (wherein the patient does not stay overnight, and
hence does not count as an inpatient admission), the mix of medical and surgical
inpatients shifted towards "medical" (i.e., non surgical). The average severity of
illness almost certainly increased (using the data in Box 9.1 and logical thinking)
since the ambulatory surgery assuredly removed the patients with the least-serious
illnesses from the inpatient admissions list, leaving only the more severely ill
patients. The ability of hospitals to reduce length of stay in the face of increasing
patient illness severity indicates that technological improvements (and changes in
medical practice patterns) facilitated decreases in length of stay. Had not the less-
ill surgical patients shifted into the ambulatory surgery category (leaving only the
more seriously ill patients as inpatients) the average length of stay probably would
have declined even more than it did.
*6. Not-for-profit organizations intrinsically "hang around" longer than for-profit
counterparts, both for legal and organizational reasons. First, the legal question: If
a hospital is to close (or merge, selling its assets to another organization) in a for-
profit setting, the returns from that sale (profits, if the returns are greater than debt
and other liabilities of the hospital) go immediately to shareholders, thus providing
the proper economic incentives to sell the hospital if that's the best economic
choice. By contrast, in the not-for-profit hospital, there are no shareholders and,
hence, nobody to whom "profits" can be dispersed. Thus, any merger or sale would
really have to take place with another not-for-profit organization, otherwise the
legal charter of the original not-for-profit organization would be violated.
Sometimes a new charity is created to receive the profits of a sale of a not-for-
profit hospital to a for-profit hospital.

A second consideration is the purpose and personal incentives of the

ultimate decision makers, the board of trustees. In many cases, these people have
dedicated important parts of their lives to the continued operation of the hospital,
and it provides for them not only considerable personal satisfaction but also
important standing in their community. Thus, closing down the hospital may be
psychologically difficult.

Finally, when a hospital (and indeed, any not-for-profit enterprise) has

accumulated an endowment of any significant size, there is legally little way to
"hand over" that endowment to other purposes, so, in some sense, the only option
open to a board of trustees facing a declining economic picture is to try to "make
do" while slowly dribbling the endowment away. This is much more likely the fate
of a not-for-profit hospital than a for-profit hospital, where the powerful incentives
of the market work to shift resources to their most highly valued use.

7. Affiliation of hospitals within a region should lead to reduced duplication of

facilities, since the newly-affiliated hospitals could lower total costs of the system
by removing duplicative resources. This would allow them to take the newly-
released resources to provide other services that would add utility to the hospital
leaders.
In real-world settings, the extent of “affiliation” that is actually achieved
may fall well below the feasible level, since the “politics” of the two organizations
may interfere with their ability to consolidate services.
 CHAPTER 10

THE DEMAND FOR HEALTH INSURANCE

Chapter Outline

The Demand for Health Insurance

Reasons People Want Insurance
Choice of the Insurance Policy
The Supply of Insurance
Insurance Market Stability: The Question of Self-Selection
Income Tax Subsidization of Health Insurance
Empirical Estimates of Demand for Insurance
The Overall Effect of the Tax Subsidy on the Health Sector
“Optimal” Insurance
Other Models of Demand for Insurance
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendices to Chapter 10:
Appendix A: A Detailed Calculation of Welfare Loss
Appendix B: The Calculus of the Risk/Moral Hazard Tradeoff
Appendix C: The Statistics of an Insurance Pool

Teaching Tips from the Author

This chapter is the heaviest sledding of any in the book, at least in terms of the
mathematical demands on the student. I have yet to find many students who
intuitively grasp the concept of risk aversion, and my only hope is that careful and
repeated demonstration of the idea will get it to a point of functional usefulness for
students. The extensive exercise at the end of the chapter with a specific utility
function is an attempt to do this, but in lectures, I essentially repeat the figures
early in Chapter 10, a step at a time, with a lot of repetition, and a lot of attention
to using exactly the same notation as the book uses so students don’t have to
reconcile notes and the book.
As an attempt to ease the students into the issue, I now have a section using a
simple two-state indifference curve model, the approach used in the early literature
on demand for insurance. You’ll find this in the discussion surrounding Figure
10.1. I hope this helps.

Demand for Prevention. The section surrounding Figure 10.6 is entirely new in
the Sixth edition, bringing together a new way of looking at demand for
prevention. As we all know as economists, life is full of tradeoffs. Prevention is
another such event. The reasons why preventive care are desirable all come
together in the new Figure 10.6. This is wholly unlike demand for acute medical
care, where risk aversion and welfare loss from excess consumption dominate
the tradeoff. Here, prevention protects against financial risk by lowering the
probability of a “bad” event. The value of prevention depends hugely on the
welfare loss from standard health insurance, which in turn hinges both on the
coinsurance level (C) and the demand elasticity η. Optimal prevention is greatest
either at low levels of coverage (C near 1) to prevent the health and spending
losses, and at near full coverage (C near 0) to prevent excess welfare losses.

In the elaboration of the standard risk aversion model to health care, I have to
resurrect the welfare loss models from Chapters 4 and 5 (and the demand
elasticity information too). Here (again), it’s important to use the right demand
curve–the “uninsured” curve with the price effect drawn in. This is a careful
balancing of risk reduction against moral hazard, a problem first laid out by
Zeckhauser in 1970. He had the utility function all wrong – he used far too large
a measure of risk aversion, in my view (see Garber and Phelps, 1997), but he had
the basic idea in hand.

Two other key issues emerge in Chapter 10. One is the possibility of market
failure. The “version” of this story set out here is mine, but I think it’s faithful to
the original concepts of market failure in insurance markets laid out by
Rothschild and Stiglitz. The essential ingredient remains: insurers when faced
with informational asymmetry will try to produce insurance plans that lead to
self-revelation by high-risk consumers, and the process of doing this harms the
low risk consumers. This is also tricky stuff – the entire chapter is full of
complicated and tricky stuff, so it’s a good time to go slowly with your students.

As I noted in the Preface to the Sixth edition, the Health Insurance Exchanges
(with all of the issues associated with tax penalties for non-compliance with the
Individual Mandate and the ongoing ACA ban on use of preexisting conditions)
are fertile ground for thinking about separating equilibria.
The other issue is one of my favorite tubs to beat on: the role of taxes in the
demand for health insurance, hence in the demand for medical care, and hence to
the equilibrium size of the health sector. The importance of this, of course,
depends on the price elasticity of demand for health insurance, and (alas) we have
nothing comparable to the RAND experiment that measured the elasticity of
demand for health care. (The RAND HIS did have some hypothetical questions to
enrollees about insurance choices, but it wasn’t real observed behavior.) I’ve been
quoted as saying (and I believe it) that the most important health policy statement
in the US is a line in the tax code of the IRS, the one exempting health insurance
premiums paid by employers from the tax base. Phelps and Parente (2017)
devote the first third of their book to this issue.

Classroom Projects
A. Have the students calculate (with their parents’ help) the marginal tax rate faced
by the parents in their last tax year. Did they itemize or not? What was the home
state? Did they remember FICA, and how did they handle that? Were the parents
self employed or not?

B. Find out the health insurance premiums charged to students over the past 5 to
10 years in your school; discuss the time trends and try to understand why the
premiums have changed the way they have. Compare if possible to national trends.

C. Get the insurance options for employees of one or several large employers in
the area. (Your college or university would be a good source for at least one
employer.) Break the students into groups with highly-detailed descriptions of the
insurance options and have them choose (as if they were a single family) what
coverage to adopt, including an understanding of how much they must pay out of
pocket. Try to find at least once choice set that includes a very expensive but
“unmanaged” plan such as a “traditional” indemnity plan that has no restrictions on
utilization, no provider panels, etc. This comparison can be used after the project in
Chapter 11.

Sample Exam Questions with Suggested Answers

1. (25 points) Health insurance has been characterized (correctly) as a tradeoff

between risk spreading and moral hazard (here, "moral hazard" means the increase
in use of medical care due to the way the insurance policy works). Show how this
tradeoff occurs using a diagram of demand for medical care, on which you
specifically show the welfare loss from the "moral hazard" effect, and specifically
show the out of pocket spending for a consumer with and without insurance.
Discuss your diagram in sufficient detail to assure me that you really know how
this tradeoff works. (Hint: Begin with a diagram showing demand curves for at
least two types of illness, one "mild" and one "serious." Show the quantities
consumed with and without an insurance policy. This should get you on your way
correctly....)

The answer to this question relies on the student properly reconstructing Figure
10.3 and then showing how triangles A and B (weighted by their probabilities of
occurrence) offset the welfare gains from reducing financial risk.

2. (25 points total) The largest problem confronting a health insurance market is
that "sickly" people may mask themselves as "healthy" and thus attempt to buy
insurance that is much cheaper, from the buyer's point of view. The problem from
the seller's point of view, of course, is that if this goes on undetected by the
insurers, they will lose money, since they have priced the insurance as if people are
healthy.

(10 points) In words (plain English) what is the key idea from economists'
understanding of this problem as to the general approach that insurance
companies will take to resolve this problem?

This is an example of market failure due to incomplete information in the market.

The approach insurance plans will use is to structure plans where the most sickly
people “self-reveal” their health status through their insurance choices.

(15 points) Using a diagram with budget lines and indifference curves for sickly
and healthy people, show how this phenomenon works. Talk intelligently about
your diagram and label all the relevant lines. Explain why the lines are drawn the
way you drew them.

The correct answer here relies on the reconstruction of Figure 10.7 with
appropriate discussion.

3. (15 points total)

a) (5 points) As one looks at historical data in the US from 1950 forward, describe
the medical services that have been most commonly covered by health insurance
(for the most people, and soonest in history), and those that have been covered less
commonly (for fewer people, and later in history).
Most commonly, hospital care and physician services, most importantly for those
taking place in the hospital (such as inpatient surgery), less commonly
prescription drugs and dental care, and often excluded, preventive treatments.

b) (10 points) Discuss two economic aspects of the demand for these various
services that will help understand the historic trends you described in part (a).

Two key issues drive these patterns: One is the intrinsic variability in financial
risk associated with each type of service. Hospital services have the greatest
variance; preventive services the least. The other is the demand elasticity of the
covered service. The greater the demand elasticity, the less desirable is the
insurance coverage.

4. (5 points) Suppose a health insurance policy costs $1100, and the expected
benefits (in a statistical sense) are $1000. “The price of the insurance policy is
$1100.” Comment on the statement in quotes.

It is utter nonsense. The price is $100 (the loading fee); the remainder is a
“passthrough” of expenses for medical care.

5. [This is a variant on several previous questions.] (10 points) Would you expect
more people to purchase insurance against losses associated with dental care or
with hospital care? Why?

More likely to insure against hospital care than dental care risks. Spending
variance is higher and the demand elasticity lower for hospital than dental care.

6. (15 points) Most health insurance is provided in the United States through
employer work groups. (a) Describe what feature of government policy makes
this particularly desirable for workers and firms in the United States, and
indicate briefly how this particular government policy affects decisions about
the amount of health insurance people acquire through this method. (b) Setting
aside the government policy in the first part of this question, what other feature
about employment groups makes them valuable as a way to make the insurance
market function well?

(a) The US Tax Code exempts such payments from the taxable income definition.
This changes the relative prices of health insurance compared with other goods
and services in the US that must be purchased with after-tax income. Health
insurance thus receives a subsidy equivalent to the marginal tax rate of the
individual. (b) Employment groups bring together a group unrelated to health
risks that makes lower-cost group insurance feasible. The grouping works
because there is little risk of self-selection of sickly people into high-coverage
plans (although some of that risk remains when the employer offers different
insurance packages to its employees.

7. [This is a variant on 6.] (20 points) Some observers of the US health system
have said that the most important federal policy relating to health care is the IRS
federal tax code. What part of the IRS code is so important? Why is it so
important? What behavioral parameters (e.g., elasticities of demand or supply)
have to be "large" in order to make this a valid issue?

The exemption of the health insurance premium payments (made by the employer)
from the tax base is the important issue. It provides a subsidy to health insurance
relative to other goods and services that people buy with after-tax income. It
matters most of all when the elasticity of demand for health insurance itself is
large, since the subsidy then creates a larger shift in the equilibrium amount
insurance in the market. That promulgates into a world of increasing use of health
care because the health insurance subsidizes health care. These problems are all
made worse (in a sense of total spending) when supply elasticities in the insurance
and medical care markets are small, since the demand shifts drive up prices.

8. (10 points) “The average annual expense for hospital care in the US in 1995
was over four times the average annual expense for prescription drugs and over
eight times the average annual expense for dental care. This explains completely
why many more people purchase insurance against hospital costs than purchase
insurance against drug and dental care costs.” Discuss

Balderdash! It’s not the average expense that matters, it’s the variability (risk).

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. The demand for insurance is driven by the variance, not the mean risk. If the
average expenditure was very large but the variance was zero (think of “grocery
insurance”), then the demand for insurance would be zero as well (unless a tax
subsidy created some added demand).

*2. The tax subsidy to work-group insurance has probably increased the scope of
generosity of private insurance considerably. This in turn has increased the demand
for medical care through the usual mechanisms (see Chapters 4 and 5), which may
well in turn drive up medical costs if the supply of medical care is upward sloping
in the long run. All of these factors cause the size of the health care system to
expand considerably as the tax subsidy increases. Thus, the tax-code provision
making health insurance premium payments (by employers) exempt from all forms
of income taxation has probably affected the health care system more than most
other policy choices made by the government.

3. The triangle is WL = _ _p _q in general. _p = (1-C)p with insurance, _m =

dm/dp _p. Thus WL = _ (1-C)p x (1-C)p(dm/dp). Now multiply by p and divide by
q, and combine terms, giving WL = _ (1-C)2 _ pm, since _ is defined as (dm/dq)
(p/m).

4. This exercise will differ by state; see Box 10.2 for an extended example of how
to carry it out.

*5. Both contain an element of asymmetric knowledge. Adverse selection occurs

when the patient knows more than the insurer about potential health risks of the
patient (or spending plans, e.g., with pending pregnancy). Demand inducement
occurs when the doctor convinces the patient that some treatment is needed when it
really isn't. The doctor's superior knowledge and the patient's trust in the doctor
allow demand inducement.

6. The market would be less likely to emerge in the society where contagious
diseases were prevalent. The reason is that the risks of individuals would be
positively correlated, thus reducing the gains from insurance pools. At the extreme,
if the risks are perfectly correlated, the gains from insurance evaporate.

7. The premiums might differ, but the “owners” of the NFP insurance firm could
well use the advantage to change the quality or quantity of insurance sold.
Increasing the quantity would, of course, best happen by lowering the price (i.e.,
passing the tax savings through to the consumer), but it might also be the case that
the NFP organization would deliberately increase the quality of coverage (e.g., by
eliminating deductibles in the coverage) as a “philosophy” for insurance. At least
historically, Blue Cross firms’ managers have said in print that they espouse “full
coverage” insurance and use their market advantage to help spread that type of
insurance to wider coverage.

8. Solved in text.
 CHAPTER 11

HEALTH INSURANCE SUPPLY AND MANAGED CARE

Chapter Outline

The Supply of Insurance

Managed Care: A Response to the Incentives of Traditional Insurance
Why Managed Care?
Types of Interventions
Consumer Side
Which Interventions Work Best of Managed Care?
Long-Run Issues
Summary
Related Chapters in Handbook of Health Economics
Problems

Teaching Tips from the Author

This material continually evolves, and if there’s any part of the book where
combing of the papers and weekly news magazines for stories that link to the
work, it’s this chapter. Managed care has evolved hugely since the first edition of
Health Economics came to press, and will continue to evolve. This chapter
required by far the greatest revision of any parts of the book even from Editions 4
and 5. So my first “hint” is to keep a “shoe box” (or equivalent digital file of
PDFs) full of pertinent clippings from the papers on this topic.

The largest addition in Chapter 11 in the Sixth edition is new material on the way
the health insurance exchanges work and the income-related subsidies that flow
through them. Of course, this may all change if the Republican-controlled
Congress modifies this part of the ACA, to you’ll have to be on top of this
regularly.

The next observation is that in lectures, it’s hard to find the correct “line” through
the material. Managed care organizations involve multiple dimensions of
management, some involving consumer copayments, some involving provider
compensation, some involving direct interference with patient/doctor treatment and
diagnosis choices. There is no “best” way to pass through this material linearly. I
suggest that each professor think carefully about what will work best in your local
situation. One path through would provide some quick definitions of terms (such as
“capitation,” “prior approval,” etc.) and then march through the standard alphabet
soup organizations (IPA, PPO, HMO, POS...) and talk about how each
organization uses each type of control device.

An alternative approach would be to lay out a three dimensional picture with axes
showing type of compensation of doctor (unregulated FFS at one extreme, salary
at the other) on one axis, degree of interference in individual clinical decisions
(low to high) on another, and patient copayment responsibilities on a third axis
(although this one is easier to omit if 3D drawing is difficult for you). Then you
can “plot” each type of organization in this 3D space and use this as a vehicle to
talk about the choices.

This is also a good place to get student involvement with “anecdotes” about their
experiences with managed care. I had one, for example, where I fell and tore the
tendon off the top of my kneecap. A call to 911 led to an ambulance ride to the
nearest hospital (our own university’s hospital, mind you) for an ED visit, x-rays
and the like. Then my insurance plan denied the claim because I’d not had prior
authorization from my primary care provider for the “referral” to the ED! I finally
got it paid, but this type of experience helps students to understand what these
management controls do, and why they frustrate so many people.

Balancing a discussion of the physician’s view of the Managed Care incursion into
the market, the Fifth edition added some perspectives on the consumer side – what
types of insurance coverage have more or less favor among the people buying the
insurance? The data in Table 11.2 show some remarkable trends. Conventional
(non-managed) care has almost vanished! It has fallen from 73 percent of the
market in 1988 to less than 1 percent by 2011. Talk about the market for buggy
whips! Classic HMO plans grew in favor in the late 1990s and then ebbed,
reverting back to about the same market share in 2011 as they held in 1988. Point
of Service (POS) plans flourished briefly as an “open HMO, and then declined
again in parallel with the HMO model plans. What has filled the void? The PPO
type plan (with fewer constraints on patients than HMO and related plans) and the
new High Deductible Health Plans (formerly known as Consumer Directed Health
Plans), which have rapidly surged from essentially no market share in 2005 to 17
percent in 2011 and now account for almost a third of the under-65 market. It
should make for an interesting discussion with students to talk about these trends.
My own “take” on these data says that consumers are opting for less “managed
care” and more freedom to select their own treatment paths, even if it costs more
than more completely managed care such as HMOs.

There is an important issue related to HDHPs that is playing out in Washington,

DC in arcane ways – what is the “minimum required” plan to fulfill the individual
mandate? The ACA says “a Bronze Plan,” which must cover 60 percent of the
actuarially expected costs of enrollees. But for a young person with relatively low
expected spending, many HDHP plans could fail that test. Thus you may wish to
discuss with your students how to assess an HDHP within the context of a
“Bronze” mandate. (Phelps and Parente, 2017 also discusses this issue.)

I cannot leave the discussion of HDHPs without mentioning the passing of

Kenneth Arrow, perhaps “the” founding father of the field of health economics,
with his landmark 1963 article about the welfare economics of medical care. His
emphasis on uncertainty hugely influenced my own career. The appendix to his
1963 article proved that the optimal insurance plan would have full coverage above
a deductible, precisely what the HDHP plans provide. I had the joy of spending an
afternoon with him in 2016 (talking about voting mechanisms). There is a short
tribute to his work and life that I wrote in Value in Health, July–August, 2017,
Volume 20, Issue 7, Page 999. He was an incredibly classy person as well as an
amazing and prolific genius in economics, operations research, voting theory, and
many other fields of intellectual endeavor.

The Fifth edition added a brief section introducing the Accountable Care
Organization (as created in the PPACA), which alters the incentives of health care
providers to focus on measurable quality levels and improvements. As these
organizations evolve, you can probably bring new material into the classroom
about the effectiveness of this new approach, but as this edition went to press,
ACOs were an idea and nothing more; they did not begin to operate until 2012, and
as this Sixth edition went to press, still relatively little information is available on
their effects on utilization, costs and health outcomes.

Classroom Projects

A. Have each student either use their own experience or that of a parent or friend,
and bring to class anecdotal information about the process of getting a medical
treatment approved (or not!) by a managed care plan. This should trigger a lively
class discussion about management techniques and their effects on patient
satisfaction. Guide the discussion to point out the necessity that these rules (when
they actually affect patient behavior) must annoy some people, since otherwise the
denial of treatment can’t save any money for the insurance plan (and hence the
premium paid by the insured individual). If possible, link this discussion back to
exercise #3 in Chapter 10 and ask if they wish to reconsider their insurance
choices.

Sample Exam Questions with Suggested Answers

1. (15 points) Health Maintenance Organizations (HMOs) have substantially lower

hospitalization rates per person than do Fee for Service (FFS) insurance plans.
Would you say that this is mostly due to (a) the use of preventive medical care by
HMOs, (b) financial incentives to patients, (c) financial incentives to doctors, (d)
because the HMOs manage to enroll mostly healthy people anyway, or (e) none of
the above? What evidence do you know to support that the thing you selected has a
significant effect on the total amount of medical care patients receive?

Mostly due to financial incentives (doctors are on salary). Some of it comes from
self selection (as many comparison studies now show). Prevention probably has
little to do with it, and in fact, it’s difficult to demonstrate that HMOs use more
preventive services than FFS medicine produces. Interestingly, in the Hickson
Altmeier Perrin study, the doctors on salary produced fewer routine child-care
exams, a quintessential form of preventive medicine.

2. Discuss the major differences between a standard health insurance package and
an HMO, in terms of (a) financial risk to consumers, (b) health risks to consumers,
(c) total medical care use, (d) convenience to consumers and (e) any other key
facets of the comparison that you think relevant.

HMO plans often have little or no coinsurance, so the financial risks are low.
Health risks are higher, however, since the HMO has financial incentive to
withhold care. Thus, if care “works,” there’s a health risk. (Iatrogenic medicine
works in the opposite direction.) The HMO will use less medical care. The HMO
will be less convenient, both for “hassle” associated with the process of
managing care and if the HMO operates a central facility that all patients use
(travel costs will be larger than in a spread-out FFS system.
3. [This is a variant on 1.] (15 points). Describe the key differences between a
pure "staff model" HMO, an Independent Practice Association (IPA), and a
traditional fee for service (FFS) insurance plan in terms of (a) the main incentives
confronting providers (doctors), (b) the degree of direct control the insurance plan
can exert on provider behavior (independent of incentives), and (c) the likely
effects on medical care spending.

The HMO in pure form hires the doctors on salary, where the IPA has doctors in
their own offices, treating patients on a FFS basis using fees negotiated between
the doctor and the insurance plan. Point of Service (POS) plans use doctors
scattered around, but on capitation (fixed payments per year for caring for each
patient). Because they directly hire the doctors, the HMO should be able to
control medical care use more (monitoring, record keeping, etc. and hence
should have lower total medical care use.

4. (25 points total) Modern "managed care" insurance plans alter both the
incentives facing providers and those facing consumers of care, compared with
incentives for uninsured patients. (Note: these may be completely different
incentives; they need only affect one or the other of providers and consumers,
not necessarily both of them.)

a) (15 points) Make a list of the most important incentives that affect patients
and the incentives facing providers, and for each one, indicate the likely
consequences you foresee in terms of patient and provider behavior.

Patients:
Copayments (reduce use vs. no copayments)
Prior authorization needed (reduced use)
Some treatments not approved for payment generally are not covered
(reduces use)

Doctors: Prior authorization needed for some treatments

Payment denied for some treatments – doctor is stuck with bill or has to
explain to pts.
Doctors’ compensation shifts from fee for service to salary, reducing
incentives for use.
b) (10 points) Pick one of the items you listed in part (a) and discuss in detail the
evidence you know about that describes how much these incentives actually alter
behavior.

Any of the evidence-related articles described in Chapter 11 would count.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

*1. Staff-model HMOs have doctors on salary and typically have one major
location for providing services to patients. IPA model HMOs pay doctors on a fee-
for-service basis, and those doctors provide services from their own offices,
typically scattered across the community. The staff model HMO should have
greater abilities to reduce utilization for several reasons. First, the financial
incentives of salary (versus fee-for-service) should reduce utilization. Second, the
dispersal of offices throughout the community in the IPA model HMO reduces
travel cost, which could have some effects on utilization (see Chapter 5).

2. One could advise the insurance plan to try to find doctors with a low propensity
to use care for their patients, since this could reduce the total benefits paid to the
covered patients, and hence the premium costs that the plan would have to charge
covered individuals. However, the insurance plan would also need to take into
account the extent to which they could negotiate discount prices with those
providers (they might know they were otherwise low-cost providers and use that to
bargain for higher prices), and also to understand how the patients would react to
those doctors’ styles.

*3. Increasing technology to treat illness will add to the expenditure pressure felt
through time in health insurance premiums, and with each new treatment comes
additional "problems" of using too much care when insurance covers much or all
of the medical intervention. Thus the additional "moral hazard" problems could
well lead to increased desire to find some way to offset those problems (i.e.,
managed care). Increasing income through time should also increase the desire to
have "state of the art" medical care, perhaps at an ever-increasing rate (see Chapter
16 regarding the effects of income on international comparisons of spending rates).
However, higher income probably brings with it a desire to avoid "hassle" from the
medical care system, which leads to the conclusion that among the managed care
plans of various "flavors," those with the least amount of hassle may succeed more
than those with high degrees of intrusion into patients' care-seeking preferences.
4. Doctors in the same region will more likely share the same opinion with the
first doctor than would doctors from another region. (This must be true because of
the cross-regional variability in utilization patterns.) To the extent that patients are
unwilling to travel to find a second opinion, the program will lead to relatively
fewer reversals of treatment recommendation.

5. Gatekeeper programs have two issues affecting how they work. On the
incentives side, the hope is that when patients are required to see a gatekeeper first,
the potential financial rewards to the primary care doctor (within a fee for service
system) will encourage that doctor to treat the patient directly, rather than referring
to a (presumably more expensive) specialist. The “technical” issue that ultimately
determines the value of the gatekeeper concept is how often that happens. In the
worst case, the gatekeeper always refers the patient to a specialist, in which case
the MCO pays both the gatekeeper fee and the specialist fees and procedure costs.
If that happens, gatekeeper programs are a bad choice.

6. The primary way in which MCOs actually save money compared to traditional
indemnity and similar insurance is by negotiating lower fees for each service
provided by doctors. Thus as market penetration of MCOs rises, physician
incomes will fall, unless they can and due create sufficient demand inducement
among patients of other insurance plans (e.g., Medicare and Medicaid) to offset the
loss of income from lower fees.

*7. Because managed care organizations use a variety of methods to control costs,
some of them are bound to find ways to reduce the use of specialists and high-
intensity treatment, particularly when lower cost choices have similar or equal
results. Thus an increased spread of MCOs could well lead to a reduced demand
for the services of specialists in the long run.

8. The key difference appears in the incentive structure for the physicians
treating patients. In a “classic” HMO, the physician group shares in any savings
associated with reduced costs of treating patients. This is said to enhance
incentives for using preventive care, but it also strongly reduces the incentives to
hospitalize patients, and also eliminates incentives for demand inducement (since
the doctors are paid on salary, not fee for service). In an ACO, the same incentive
structures may well exist (although an ACO need not be an HMO), but in addition,
the payment structure directly depends upon the providers (hospitals, doctors, etc.)
meeting a set of measurable quality standards, some of which are “process” (such
as vaccination rates) and others are actual outcome measures (such as hospital re-
admissions). Thus the ACO links reimbursement to quality directly in a way that
was done only indirectly (if at all) in the classic HMO.
 CHAPTER 12

GOVERNMENT PROVISION OF HEALTH INSURANCE

Chapter Outline

The Medicare Program

Medicare HMOs (Medicare Advantage)
Operational Changes in Medicare
The Medicaid Program
Summary
Related Chapters in Handbook of Health Economics
Problems

Teaching Tips from the Author:

This chapter always strikes me as somewhat different than the others in the book
because there is (a) a lot of institutional, political and legislative detail involved
(without which the discussions make less sense) and there is not a lot of “theory”
involved, at least in terms of new theory that’s not come earlier in the book. (The
chapter begins to draw extensively on demand theory, and the chapters on
physicians and hospitals and insurance, but about the only new wrinkle is the
shift in the basis of payment from per-activity to case-based payment in PPS.) So
each instructor needs to figure out the proper balance of institutional detail vs.
conceptual work.

This chapter has been revised perhaps more than any other in the book as the
Sixth edition emerges, and the revisions continue apace. Earlier additions
involved an expanded discussion of SCHIP programs and the reauthorization of
that program that took place in 2008. Then came a new discussion about
Medicare Part D prescription drug program, and an expanded discussion of Part C
(“Medicare Advantage”) programs that takes a careful look at “risk adjustment”
models. This “risk adjustment” issue is really important to make the market work
well, so it’s worth some time in class on this issue. There is also a new section on
Medicaid, albeit much shorter than an earlier chapter that appears on the book’s
web site from the publisher. This Sixth edition expands the discussion of the
economically appropriate size of the Part A and Part B deductibles (Part A is far
too large, and Part B is far too small at present), and emphasizes the growing role
of Medicare Advantage as a voluntary voucher program to allow Medicare
enrollees to purchase private insurance with their Medicare entitlement. Almost
one third of all Medicare enrollees now use Medicare Advantage, and this
fraction should grow, since newly entering enrollees significantly seem to prefer
the “voucher” to “Traditional Medicare.”

Another feature of Medicare is that the innovation in payment concepts has

greatly differed between their treatment of the supply side of the market and the
demand side. On the supply side, major innovations have come through Part A
(where prospective payment was introduced) and Part B (where the RBRVS
system changed physician compensation patterns considerably). But on the
demand side, Medicare has been very staid, relying on 1950s style of insurance,
and adopting little of the modern cost-management philosophy of the private
insurance sector (prior approval, etc.) Also, although the Part A inpatient
deductible has moved up with time (and the average cost of hospitalization), the
Part B deductible has changed little over the nearly four decades of care.

One could get bogged down in the details of Medicare; this chapter barely
touches the range of events affecting or determined by Congress and HCFA. The
proper balance will probably depend on the purposes of your class and the types
of students you have. If you wish to go “real world” then this chapter involved
numerous opportunities for supplementing the textbook with real world clippings
(as also Chapter 11).

Phelps and Parente (2017) has an entire section discussing ways to make
Medicare and Medicaid better for enrollees, so if you want some further
discussion about the problems remaining in Medicare and Medicaid, I refer you
to Part II of Phelps and Parente.

Classroom Projects

A. For a grandparent or grandparents (or some other person they know over age
65), have each student learn what type of supplemental insurance those persons
have, how much it costs, and what it covers. Have them ask their families about
SCHIP insurance for younger children. And ask about grandparents in a nursing
home under Medicaid. And ask about grandparents’ participation in Part D.

B. Have the students acquire some available insurance programs for Part D from
common sources such as the AARP web site and compare them. Also have them
find out what Medicare Advantage (Part C) programs are available in your region
and/or their home towns and compare the offerings and premiums for various
plans.

Sample Exam Questions with Suggested Answers

1. (20 points) Discuss the essential difference between the Prospective Payment
System (DRG-based reimbursement) that Medicare now uses vs. the “original”
way Medicare paid hospitals under Part A at the beginning of the program in
1965. What evidence do you know about the major consequences of this change
in terms of (a) resource use (costs) for treating patients, and (b) health outcomes
of patients?

(a) Under PPS, the payment was per case, not per item used. This puts a large
incentive on hospitals to reduce length of stay, since they get paid essentially the
same whether the patient is in the hospital for a long or short time. This caused
LOS to decline post-PPS faster than the decline in under-65 patients. (b) Studies
directed by ProPac suggest that at least on crude measures of health outcomes,
patients’ health is not harmed. Major studies on this involve readmission to the
hospital for complications and rates of post-hospital mortality. In selected
illnesses, patients’ health has been shown to have degraded under PPS.

2. (20 points) “Medicare stinks as insurance.” (Famous health economist quoted

in discussion of Medicare.) Do you agree or disagree? Why?

Medicare follows the tradition of early Blue Cross plans – first dollar coverage for
hospitalization but limited coverage for every extended duration stays in the
hospital. Thus the risk-reducing characteristics of Medicare are poor compared
with better private insurance plans as they have evolved (e.g., catastrophic caps on
spending by families).

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. a) The other group would have an ALOS of 18 days. Thus the overall
ALOS would be 10.
b) If we eliminate the cataract patients, the remaining group would have had
an ALSO of 18 days, and it fell to 8.5 days. This represents a relative
decline of 53 percent. This contrasts with a decline of 15 percent (10 vs.
8.5 days) if one just compares the raw numbers.
 c) If only half of the cataract patients had surgery out of the hospital, the 8.5
days would represent a blend of two long stay patients and one short stay
patient. The relative decline in hospitalization would lie between the 53
percent figure and the 15 percent figure.

2. The hospital would take strong steps to reduce length of stay, since added days
of stay bring no marginal revenue but they do carry marginal costs to treat the
patients. However, the hospital would not be expected to reduce LOS to the point
where complications began to increase (see the next chapter on medical liability)
nor would it be worth spending major resources to reduce ALOS. (For example,
ALOS could be reduced by adding nursing staffing, but at some point, it would be
cheaper to stop trying to reduce ALOS and just let the ALOS remain.)

3. Part A leaves major risk for rare but expensive long-stay hospitalizations.

*4. SCHIP enrollment increased more than the number of uninsured children
decreased because of "crowding out." This occurs when people either choose not to
insure children privately once the SCHIP program is available (and enroll in
SCHIP), or they actually drop private coverage in favor of SCHIP enrollment.

5. Balance billing occurs when doctors bill patients for the balance of their regular
bill that Medicare does not pay. It occurred often in the early days of Medicare
when many doctors continued to seek their usual and customary fee from Medicare
patients. This practice obviously increased the financial risk of the patients, since
they were “uninsured” for the portions of the doctor’s price above the allowed
Medicare fee. Later changes in the Medicare program have led many more doctors
to accept the Medicare fee as “payment in full”, by agreeing to become
“participating physicians.” This eliminates balance billing for those doctors and
their patients.

*6. (a) Payment to procedure-intensive specialties will fall, and hence the
economic returns to specializing in those areas of medicine will fall. Conversely,
the economic returns to geriatric medicine should rise, because it mostly involves
diagnosis and counseling with patients. Everything we know about physician
specialty choice says that physicians-in-training respond to economic incentives.
Thus, demand for residency training in orthopedics should fall, and it should rise
for geriatric medicine. Demand for pediatric residencies should be unaffected,
since Medicare doesn't cover children. If a great number of private insurers
adopted the same payment system, then pediatric training might become more
desirable, since pediatrics, in general, involves "thinking," not doing, procedures.
(b) Procedurally based specialties tend to use hospitals more than cognitively based
specialists. The real issue hinges on how much (if at all) procedurally-based
specialists can induce demand to offset falling reimbursement rates. Hospital use
could rise or fall, depending on the extent of successful demand inducement, but in
general, demand inducement could tip the balance towards increased hospital use,
other things held equal.

7. In general, either quality of care must fall or the hospital must cut back on those
programs that are not “self-sustaining,” in a general sense. Because of the
preference patterns of hospital leaders, we can expect that they will opt to use some
of the hospital’s profits to support otherwise-unprofitable services. (The particular
choices will vary from hospital to hospital, of course.) A financial squeeze from
the Medicare system will cause them to re-evaluate those choices, as well as a
likely general reduction in quality as a way to conserve costs (e.g., lower staffing
levels).

8. The growth in hospitalizations (measured by discharge rates per 1000 enrollees)

was probably a combination of things, including population aging and new
technologies available to treat the diseases of the elderly, as well as expanding
coverage of “Medigap” plans. The precipitous drop that paralleled introduction of
the DRG system almost certainly came from a shift of “low-tech” patients to
treatment settings either outside the hospital or same-day care within the hospital
setting (“outpatient surgery”). The actual economic incentives were complex, and
one would need to carefully study the actual DRG reimbursements for such cases
as inpatient vs. ambulatory care patient vs. the incremental costs of treating
patients in those settings. But at the same time, a whole array of minimally
invasive surgical procedures were entering practice. Thus the new technologies
available (arthroscopic surgery on joints, laparoscopic surgery for other body
areas) probably had as much to do with the shift as economic incentives per se.
 CHAPTER 13

MEDICAL MALPRACTICE

Chapter Outline

Background of the Legal System in the United States

The Economic Logic of Negligence Law
Judicial Error, Defensive Medicine, and “Tough Guys”
Medical Malpractice Insurance
Evidence on Actual Deterrence
Malpractice Awards: “Lightning” or “Broom Sweeping Clean”?
Tort Reform
Tort Reform Writ Large
Vaccine Injuries: No Fault vs. Tort

Summary
Related Chapters in Handbook of Health Economics
Problems

The Sixth Edition adds a more extended discussion of the effect of tort reform on
health care costs (Avraham, Dafney and Schanzenbach, 2009), adding to the body
of evidence suggesting that there is not a lot of potential gain in lowering costs
from further tort reform. This augments a new section in the Fifth Edition on
quality of care and its relationship to negligence lawsuit risk. That section
discusses a new study that links together tort claims against nursing homes with a
large array of data about quality problems within individual nursing homes. Of the
many quality measures reported by nursing homes, only two – weight loss and
pressure sores – showed a relationship between actual rates of quality problems
and tort claims. Other quality measures (fractures, falls, dehydration, use of
restraints) were unrelated. What did matter? The litigation environment had much
more to do with claims against nursing homes than their reported quality. The
highest quality nursing homes in high-litigation states had a greater chance of
being sued than the lowest-quality homes in low-litigation states. This study
provides important new data on how well (or not!) the tort system improves
quality. My “take” on this study is that it puts more termites into the foundation of
the tort-based medical malpractice system. The final segment in Chapter 13
discusses the Vaccine Injury Compensation Program as a “no fault” alternative
now in place in the US.

There is one issue that did not make it into the Sixth Edition that I wish to raise
here – the risk of lawsuits to HMOs who do not cover treatments and then get sued
by their enrollees. Two US Supreme Court rulings in 2004 covered this issue, but
only with partial effect on the market. Two things intersect here. First, some
states have laws that are much more consumer-friendly than others regarding their
ability to recover damages arising from failure to cover treatments. Federal laws
are more favorable to insurance plans. The US Supreme Court rulings stated that
employer-sponsored insurance plans were covered by ERISA – the federal law
governing employer-sponsored insurance (ERISA = Employee Retirement Income
Security Act, although it also applies to health care plans.) So non-employer-based
plans can still be sued under the more consumer-friendly state law. So the issue of
potential lawsuits for withholding coverage has not gone away, and remains part of
the overall regulatory and legal background for understanding insurers’ decisions
about coverage and scope of benefits.

I also have a way to link this material in Chapter 13 on Malpractice Law with
discussions of the Coase Theorem on externalities that appears in Chapter 14, so
you may want to circle back to Tort Reform after discussing the Coase Theorem in
Chapter 14. Further discussion on this comes shortly!

Teaching Tips from the Author

Like other late-in-the-book chapters, this one soaks up a lot of preparatory

material from previous chapters, and I always try to point out these look-back
occasions. The physician control over resource use in the health system is
central, of course. The patient-provider relationships (including demand
inducement) are lurking here. (Interestingly, little malpractice work involves
doing too much treatment, unless something goes awry. But nobody ever has
won a malpractice case simply because too much treatment was given!) Even
the existence of malpractice insurance hinges on the notion of risk aversion
(chapter 10).

I like to point out how much the “pure” theory of negligence relies on a cost-
benefit analysis at its heart, perhaps the most obvious case where economic logic
gets built into the law. The Learned Hand Rule really is just a mandate to have
careful cost-benefit analysis done!

I find a lot of intrinsic resistance from students to the belief that there are too few
lawsuits in our system, since we tend to think of our society as being lawsuit-
happy. But the evidence from various studies is pretty striking to me that a lot of
people who are negligently injured don’t sue their providers, and hence that a lot of
the potential deterrence is missing. Those professors who rebel against this idea
(along with my students) might find it interesting to ponder why the evidence is so
hard to swallow here. Is something missing in the analysis?

Ultimately, I try to bring the class to the ultimate question posed at the end of the
chapter: is the whole liability system worth it? There’s no clear answer, of course,
but a discussion on this issue serves to help the students come to understand the
fundamental issues clearly.

Classroom Projects

A. What’s malpractice? Have the class submit a scenario they know about from
personal experience or friends that sounds like it may or may not be malpractice.
With a little pre-sorting, you can find some interesting events that highlight the
question “was this malpractice?” Organize discussion groups around each one of
these you have picked, and have them report back to the class as to their
conclusion. Each will be based on scanty facts: they’ll have to judge based on thin
information, but their discussion of what might make it malpractice or not will be
useful. You may wish to save up a newspaper clipping file of juicy “events” from
the local paper that will serve the same role as the class contributions.

Sample Exam Questions with Suggested Answers

1. (25 points total)

a) Describe clearly the two essential purposes of the medical legal liability
(malpractice) system (5 points)

Compensate those who are harmed and create proper incentives to minimize
avoidable adverse events.

b) Describe the essential features of the most common method of pricing

malpractice insurance to physicians (5 points)
Insurance premiums vary by specialty, but not much based on prior history (or
other evidence of systematically bad medical practice).

c) Discuss how the malpractice insurance system’s features (part b) help or

hinder the desired functioning of the malpractice legal system (5 points)

It blunts the incentives for doctors to be careful; they don’t face as large a
financial consequence for their mistakes in the system where the premium is the
same for every specialist of a given type.

d) Discuss what evidence you know as to how effective the malpractice system is
in achieving its two primary goals (as discussed in part a of this question) (10
points)

Several studies, most notably in California and NY, show that relatively few
malpractice events lead to lawsuits. Thus the incentives for injury prevention are
blurred. As for compensation, the current system doesn’t compensate many people
who are injured negligently, and the administrative costs of managing the system
are high compared with (say) strict liability or no-fault systems.

2. (15 points) “The malpractice system is purely random, striking good doctors and
bad doctors with equal propensity.” (A complete answer will contain reference to
some hard evidence.)

Studies such as Rolph’s show that the predictive power of prior malpractice
claims history is large. A small handful of doctors in his Los Angeles data base,
for example, accounted for 10 percent of the claims and 30 percent of the
payments. These were doctors with a prior history of 4+ previous malpractice
suits lost.

3. (15 points) There are two key goals of the medical malpractice system from a
social and economic viewpoint. Identify them (Just name them; don’t elaborate.)
Describe the role of medical malpractice insurance as it affects these two goals
(i.e., does it enhance or detract from the ability of the malpractice system itself to
accomplish these goals, including any pertinent aspects of how malpractice
insurance is sold and priced.

The two goals are to compensate injured persons and to provide incentives to
avoid preventable negligent injuries. Malpractice insurance as commonly
administered has mixed effects on these goals. By eliminating the risks of
bankruptcy by individual providers (e.g., physicians) it probably enhances
compensation. However, the pricing mechanism (charging by specialty but not by
prior history of malpractice losses) creates a non-cooperative game theory
problem where doctors to not face the full costs of their negligence.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. To the extent that lawsuits are necessary to create deterrence, this is correct.
The evidence from several studies (California, NY, etc.) shows that a relatively
small fraction of patients who were negligently injured actually bring a lawsuit, so
the deterrence level is lower than if the frequency of suing were higher.

2. Malpractice insurance provides relief to the doctors from the financial risk they
face. The malpractice legal system (not “insurance”) is designed to provide relief
to injured patients, and also to deter damage. The insurance provides relief for
injured persons if and only if they win a suit and the doctor or hospital is otherwise
unable to pay the legal damages.

3. The Learned Hand Rule says that negligence has occurred if it would have been
cheaper to prevent an injury than the expected value costs when it occurs (cost x
probability of occurrence).
In effect, this law puts a simple cost-benefit test on every activity that might
prevent an injury. If the benefits exceed the costs, then the law says that the
activity should be used commonly. If taken to a full economic analysis, the
Learned Hand Rule would look at the extensive and intensive margins of use of an
injury-preventing activity and ask (a) for which patients should it be used and (b)
how often, in both cases using an implicit cost-benefit test for the expanded
utilization.

*4. It has no major role. Malpractice law is wholly the territory of state
governments. This is advantageous for economists wishing to understand how
different malpractice legal arrangements affect doctor and patient behavior, since
the various states have different laws and change them over time, thus creating
natural experiments for the economist to study.

5. Malpractice insurance probably increases the rate of injury, particularly to the

extent that the insurance is sold in ways that do not capture the intrinsic risk of the
doctor in creating injuries. With insurance sold at the same price to all doctors (or
all within a given specialty), the relatively bad doctors receive a subsidy from the
good ones. If the bad doctors paid the full brunt of their expected costs, either they
would be driven out of practice or their costs of practice would go up. Either would
cause the amount of treatment by the bad doctors to decline, and thus the overall
injury rate would decline.

*6. A doctor's history of malpractice strongly predicts the chances that a future
malpractice event will occur. In one study of doctors in Los Angeles, 46 (out of
8000) doctors accounted for 10 percent of all malpractice verdicts and 30 percent
of the damages awarded. Each of these doctors had lost four or more lawsuits in
the period of study. Another study of insurance pricing showed that doctors' history
of malpractice could predict malpractice awards in the future just as well as a
doctor's specialty. Most insurance companies use doctors' specialty to set
premiums, but few use past malpractice claims history.

7. The primary claim of value for the negligence system is that is in concept deters
negligent injuries (malpractice). It does so by imposing a costly structure
involving litigation. No-fault systems in concept weaken the incentives for
providers to be careful (relative to the negligence system), but no-fault systems
also reduce the litigation and other administrative costs. On balance – as Ronald
Coase has reminded us in several of his writings – the optimal system must take
into account the transactions costs of achieving the desired goals.

The 2011 ruling from SCOTUS opens the door for states (who control negligence
law) to switch to a no-fault system, and in fact, the majority opinion seems to
embrace the idea in ways that do not seem to rely specifically on the market for
vaccines.

In addition, the new analysis on nursing home quality from 2011 adds further to
the ammunition of those who believe that the negligence system has little effect on
quality of care. In that analysis by Studdert et al. (2011), quality of care was
generally unrelated to the rate at which negligence claims were brought against
nursing homes. This adds to the message from previous studies in California and
New York showing (in general) low rates of negligence suit filings when apparent
negligence (as rated by experts) had occurred.
 CHAPTER 14

EXTERNALITIES IN HEALTH AND MEDICAL CARE

Chapter Outline

Externalities, Property Rights, and the Control of Externalities

Externalities of Contagion
Solutions to the Externality Problems
International Issues—Expanding the Scope of the Externality
Externalities from Tobacco
Information as an Externality
Research as an Externality
Reason for So Little Research on Medical Effectiveness
Transfusion-Induced AIDS and Hepatitis
Summary
Related Chapters in Handbook of Health Economics
Problems
Appendix to Chapter 14: Value of Life

In terms of new material, the final section on transfusion-induced AIDS and

hepatitis has shifted to include an international comparison, rather than focusing
only on the US situation. As the new Table 14.1 shows, the risk of transfusion-
related disease is strongly related to the degree of economic development of
various countries. As one might expect, blood safety seems to be strongly related
to per capita income, and perhaps would qualify as a luxury good.

Teaching Tips from the Author

This chapter is in some ways a collection of pretty varied material. This in part
happens because there are so many different types of externalities in the world
around us, many of them with effects on our health, and they can all be interesting
problems.

Nevertheless, beginning with the Fourth edition and extending on to this one, I
have weeded out some material from earlier editions in order to expand material on
vaccine programs. You will find a wholly new section on the externalities
involved in vaccine supply, for example. I have also added a new section on the
international transmission of contagious diseases and mechanisms to control such
happenings.

I usually spend some time at the beginning of this chapter talking about Ronald
Coase and the Coase Theorem. Many people misunderstand Coase’s Nobel Prize
winning work, thinking that he says that assignment of property rights doesn’t
matter. He actually says just the opposite of that! The Coase Theorem says that if
there are no transactions costs, you get the same economic outcome no matter
how you assign property rights. The conclusion that I draw from this work is to
pay attention to property rights assignments and their implications for transactions
costs. On to this chapter–the point I make with the Coase Theorem is that
externalities arise when property rights are vaguely defined and/or transactions
costs are large.

Once the idea of externalities has been put in place, I find that the most interesting
way to walk through the examples of various externalities situations is to outline
the economics of why it’s an externality and then talk about ways of controlling
them, and the various social implications of those controls. Should we use markets
(difficult in true externality situations), regulation and laws (a common approach),
or social pressure and moral suasion? This is an important question to keep in mind
whenever we confront an externality. It’s not my idea: it belongs to Robert Clark,
former Dean of the Law School at Harvard. Sometimes markets will work better
than we think. Provision of military protection has features of externalities, but
once the government decides to mount a military force, I think that most people
share my belief that we are better off with an all-volunteer military than with a
compulsory draft. People once thought that pollution control was best done with
direct regulation, but we now recognize that many situations (e.g., air pollution
from stationary sources) lead to better pollution control at less cost to society by
using markets (in this case, marketable permits). And sometimes social pressure
works best. Think about the best ways to control the spread of colds in your
school–probably social norms among the students are a lot more effective than
rules.

In economics, we often turn to market mechanisms to control behavior, and the

control of externalities is no exception. Tax-subsidy schemes, marketable permits,
etc. are all manifestations of this approach to social control. However, I prefer to
emphasize in these discussions not only the efficiency aspects of various control
mechanisms, but also to bring into the discussion a positive-economics view of
how politicians will behave in making their choices. Off-budget approaches (such
as mandating firms to do things to control pollution or to provide health insurance
for workers) have a favorable element to politicians – nobody can accuse them of
making a larger government! And various mechanisms have different distributional
characteristics, which are interesting and important even if you don’t care about
distribution of wealth, because they affect politicians’ behavior. Thus, I guide the
discussion of various externalities towards these sorts of questions. Think of drunk
driving, for one example: one can tax alcohol to reduce drunk driving, but one
could logically also tax gasoline, since both are required for drunk driving. Which
is better to tax? It turns out that the answer is alcohol, because the welfare loss
from taxing gasoline to reduce drunk driving would greatly exceed that of taxing
alcohol to achieve the same goal. But what you really want to do is tax the dual
activity of drinking and driving, and there’s no efficient way to do that without
considering enforcement costs. Etc.

Here’s a fun way to loop back to Chapter 13 and malpractice law: think about the
Learned Hand Rule and the Coase rule together. Coase says that IF transactions
costs were trivially small, you would get to a socially optimal solution no matter
which way you assign property rights, but of course we know that transactions
costs are not small in medical-legal issues. The Learned Hand Rule (Chapter 13)
is really a prescription for optimal investment in prevention of injury.

Now think about the data on the frequency at which injured people seem to file
lawsuits (from Chapter 13). For smaller injuries – not often. For really serious
injuries, more so. But still we see a low frequency of filing lawsuits, which leads
to “not enough” deterrence. This is precisely what we would expect with
transactions costs entering the decision to sue.

One partial solution to transactions costs is the creation of a legal “class” in a class
action suit, which greatly reduces the transactions costs for injured people. So
plaintiff’s lawyers love class action, and defense lawyers try to restrict the use of
this idea.

Finally, you might discuss with your class how the Coase idea of assigning
property rights in a way that minimizes transactions costs affects thinking about
the optimal medical-legal system. Would a complete no-fault system be better? It
would surely reduce transactions costs. Given the low level of deterrence that
seems implicit in the low rates of filing for damages, perhaps we’d not give up
much deterrence if we went to a no fault system. All food for a good discussion in
your classroom!
Classroom Projects

A. Find out about the vaccination policy of your college or university and discuss
it. Do you have mandatory vaccinations? If so, why? If not, why not?

B. Have the students write record (and then summarize) the rules in their home
town about smoking in public places (restaurants, bars, private offices, public
spaces indoors, public spaces outdoors, etc.). Compare with both the legal rules in
your community and those enforced on your campus.

C. Generate a discussion about some health-related behaviors such as sneezing,

nose-wiping, going to class or work when you are sick (compare and contrast a
cold, nausea, vomiting, etc.) and the social norms to regulate such behavior.

Sample Exam Questions with Suggested Answers

1. (10 points) Discuss intelligently the following quote: "Compulsory

vaccination (such as required by (Your College Here) for its students (but not
its faculty!) has no rational basis for support. Intelligent people should be left to
their own decision on this matter."

Contagious diseases in closed communities like campus dorms make

compulsory vaccination more desirable. However, any compulsory system
imposes differential costs on some who strongly object to the vaccination (e.g.,
for religious beliefs or fear of needles). It makes more sense for students living
in dorms than for faculty or staff who do not live in close proximity to each
other because of different contagion risks.

2. [This is a variant on (1), emphasizing alternative solutions to the problem.] (15

points) Using a simple diagram, discuss intelligently the fundamental issue arising
from externalities such as people's incentives to get vaccinated against contagious
diseases. Be sure to mention at least two methods by which society could "solve"
the problems caused by this externality.

This should have the student reproducing something akin to Figure 14.1.

3. (15 points) Many people have criticized the US blood supply system as being
"too commercial, and prefer a more altruistic method of collecting and
distributing blood for transfusions. The key issue is whether the blood contains
viruses that might cause disease for the recipient. Could a more commercial
blood supply system improve quality? If so (or not), why?

A system of commercial blood supply would create higher quality if (and

probably only if) the liability system functioned properly (and perhaps if bonding
were required to assure assets sufficient to pay damages, since commercial
corporations can declare bankruptcy in the US. But given a functional liability
system, commercial incentives should lead to high quality blood being available,
probably at higher prices. Commercial suppliers would vigorously test donors,
and then advertise the quality of their blood.

4. (10 points) The production and dissemination of information is often described

as "public good" that is insufficiently performed by private initiative. Provide a
simple model that would allow you to evaluate the value of providing
information to consumers in a way that affected their demand for some product
(e.g., tobacco, informing people of its dangers).

Something like Figure 14.3 and an intelligent discussion suffices for a good answer
here.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

*1. Externalities occur when Person A's action imposes costs on (or creates
benefits for) other persons, but in such a way that Person A does not take account
of these when making economic choices. A common feature of such problems is
that property rights (and hence liability) are poorly or incompletely defined.
Another common feature is that even when property rights are well defined,
transactions costs are large relative to the amount of damage. Thus, sneezing in a
crowded room creates external costs (spreading of colds).

2. Commercialization of the blood supply would increase blood supply quality if

and only if two things happened. First, the liability system would have to function
very well, so that the damages created by tainted blood would lead to lawsuits and
added costs for the providers. Any profit-maximizing firm would optimize on
finding ways to screen out tainted blood, and the risks would fall. (They would
also not carry with them the preferences of many NFP blood suppliers who, for
reasons of social norms, etc., do not actively screen against donors from high-risk
populations. See the box discussion blood from Ethiopian Jews as an example of
this issue.)
The second issue is the legal option of bankruptcy for the firm. Declaring
bankruptcy would eliminate the legal burden of a malpractice suit arising from
tainted blood. Thus to gain the full effects of a more commercialized system, a
backup system of bonding (insurance against bankruptcy, if you will) would
probably be necessary. This could be done either through regulation or with market
pressures (if a sufficiently knowledgeable public demanded such bonding as a
condition of using the FP provider.)

*3. Herd immunity is the situation when such a large proportion of host species
(e.g., humans) are immune to an infectious disease (e.g., small pox) that the
organism causing the disease can no longer find new hosts to infect, and eventually
dies out. Thus the "herd" becomes immune even if every member is not.
Vaccination in humans can create herd immunity before 100% vaccination rates
occur, thus creating a benefit to those who have not received a vaccination. To get
enough people vaccinated, society could (a) reduce the price of vaccinations
(possibly to the point of paying people to become vaccinated), (b) require
vaccinations of everybody, (c) advertise or otherwise try to induce people to
become vaccinated, or (d) find some way to spread the vaccine through food or
water supplies, or something similar. (We do this in the US with folic acid in bread
to reduce neural tube defects in unborn children.) The first solution has more "on
book" program costs, but has the advantage that the people who choose to receive
the vaccinations are those to whom it is least costly. Mandatory vaccinations
impose larger administrative costs (finding out who is and who is not vaccinated)
and (through compulsion) can create large costs to some individuals (those who
fear the treatment or who have strong religious preferences against using
medicines). (c) may be very ineffective, or may work well, depending on the
situation. (d) is not likely useful for vaccinations, since few if any vaccinations are
effective when ingested.

4. Taxing alcohol also taxes the use of a considerable amount of consumption that
is not affiliated with driving, and is hence inefficient. However, a number of
studies show that the enforcement system is rather lax in finding and identifying
drunk drivers (although this has probably increased through time). Several studies,
for example, report that police checked off “alcohol not involved” in many cases
when hospital blood tests of injured drivers show BAC levels exceeding legal
limits. This suggests strongly that police do not fully enforce the available drunk
driving laws.

*5. (a) is very private, "spilling over" to relatives and friends in terms of emotional
distress at seeing their relative/friend ill or dead. Hence that's about a 0.1. (b) Here
the illness risk to other people rises, but mostly to those to whom the smoker is
personally close, so much of the cost is still internalized, hence about a .25 here.
(c) Here the smoke bothers more people, many who might choose to avoid the
smoke if they could, so this is very public, hence about .9 or even 1.0. (d) is like (c)
except that the people in the car or airplane are fewer in number and presumably
more closely known to the smoker, hence about a 0.5.

6. The predicted consumption of 6000 and an actual consumption of 3000

cigarettes per year is a decline of 3000 per person per year (dq = -3000). At a
consumption of 6000, a price of $.10 and an elasticity of -.5, the slope of the
demand curve is -15,000 (this comes from manipulating the formula for an
elasticity; see Chapter 4). Thus the change in marginal value (dp) associated with
the reduction in consumption of 3000 is dp = -3,000/30,000 = -.1 Thus the welfare
triangle we need is of size _ dp x dq = _ x .2 x 3000 = 150. The triangle measures
dollars of value, so the information is worth $150 per year per citizen. Of course,
the value is spread across those who smoke now (more value) and those who do
not (only secondary-smoke related value, and a value not captured in the
calculation).
 CHAPTER 15

MANAGING THE MARKET:

REGULATION AND TECHNICAL CHANGE IN HEALTH CARE

Chapter Outline

A Taxonomy of Regulation
Licensure
“Certificate of Need” (CON) Laws
Drugs and Devices: The New Wave of Medical Care
Summary
Related Chapters in Handbook of Health Economics
Problems

Chapter 15 is perhaps the most extensively-revised part of the Sixth Edition, with a
dozen significant new sections added – mostly about payment mechanisms and
about the prescription drug market and its regulation. I give this “heads up” to
those who have used previous editions – this chapter has a lot of new material that
you will want to know before your students read it.

Teaching Tips from the Author

Regulation flows directly from externalities, so the flow of these two chapters is
quite natural. As with externalities, there are more interesting examples of
regulation in health care than I can possibly deal with in the lecture time available,
so I pick and choose. Through the years, I have restructured the sections on
Certificate of Need (CON) programs and price controls, and the same holds in this
edition. Price controls may seem archaic as well, but it’s easy to think about the
Medicare Prospective Payment System as a price control, which makes me want to
spend some time with that topic.

I like to spend time on Medicare PPS because it helps to reinforce the model of
hospital behavior established in Chapter 9. That’s a general goal in my teaching of
these last few chapters – loop back to the earlier “fundamentals” and bring them
into the discussion. Another example is the discussion surrounding Figures 15.3
and 15.4. While hospital bed capacity limits are a thing of the past, an
understanding of how they work (using these two figures) reinforces some standard
production theory analysis. Depending on your goals, this is worth more or less
time in the classroom.
This edition further expands the changes begun in the Third edition relating to the
focus on technical change and our society’s management of it (mostly through the
FDA). I see this as an area ripe for “current events” work in the classroom; there is
almost never a week going by where I don’t see something in the newspaper that
provides a good example of technical change in health care.

This edition has a new section on the role of insurance coverage decisions and their
effects on technological change, and also a new section on Electronic Medical
Records as a new technology in health care.

Classroom Projects

A. With the students in teams, have them identify as many areas as possible where
regulation has affected their own lives in matters involving health and safety over
the last (say) 5 years. Give a prize to the team that comes up with the longest list
(you are the judge and jury about whether items are redundant, whether they
“count,” etc.) The lists should be quite long: professional licensing, FDA
prescription drug rules, FDA rules about drug advertising, speed limits, public
health laws, drunk driving laws, possibly mental health rules, workers
compensation, OSHA, etc.

Sample Exam Questions with Suggested Answers

1. (10 points) Drug regulation seeks to protect individuals from having harmful
drugs marketed in the U.S. Does this system unambiguously increase consumers’
well-being in the US, or does it possibly work in the other direction? If you think
there are possible harms to consumer welfare, what is the mechanism by which this
happens?

2. [This is a more directed question than 1.] (10 points) Agree or disagree. "FDA
drug regulation definitely makes US citizens worse off by preventing them from
being able to buy drugs that are available in other countries, including our near
neighbors Canada and Mexico.

3. [This relates back to Chapter 8.] (20 points) What would you expect to happen
to the quantity and quality of care produced in a typical US not for profit hospital if
the government imposed a price control on the hospital? Specifically, would you
expect shortages to occur (as they would be expected to occur in a typical
competitive market with for profit providers)? Use relevant diagram(s) to assist in
your answer.

The answer here hinges on reproducing something like Figure 15.5.

4. [A variant on 3.] (20 points) Provide an intelligent discussion about how you
would expect a binding price control to affect a typical not-for-profit hospital in
the United States. (A simple diagram of the hospital’s equilibrium behavior will
probably help.) Be sure to include in your discussion the issues of whether you
would expect shortages to occur (as you would expect in the “classical” free
market with a price control) and what you would expect to happen to the quantity
and quality of care provided by the hospital after the price control became
effective (compared with before the control was in place).

Figure 15.5 again.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

1. The 5 percent annuity is worth $11.888 million, and the 10 percent annuity is
worth $8.824 million. The first 8 years of the 5 percent annuity are worth $6.786
million, and for the 10 percent annuity, $5.868 million. Thus the last nine years
capture only 43 percent of the possible 17 year annuity at 5 percent, and only 33
percent of the annuity at a 10 percent rate.

2. At 5 percent, the present value is $20 million, and at 10 percent, the value is
$10 million. A perpetuity minus the first 8 years is worth $20 - $6.789 million =
$13.2 million at 5 percent. At 10 percent, it is $10 - $5.868 million = $4.132
million. At 5 percent, a 17 year pure annuity is worth $11.888 million, so the better
deal for the drug company would be a perpetuity without the first 8 years. At 10
percent, the 17 year pure annuity is worth $8.824 million, so that’s also a better
deal than the perpetuity without the first 8 years.

3. The rule eliminates dangerous drugs (safety) and those that don’t work
(efficacy), but the market would also screen out both of these errors through time.
However, errors in the system allow some bad drugs to slip through (which both
the market and the regulators will eventually catch) but keeps some good drugs out
of consumers’ hands, thereby reducing consumer well being. The net effect cannot
be determined in advance, but it is plausible that consumers are worse-off with the
regulation than without it. The same issue arises here as in the discussion of
commercialization of human blood supply in Chapter 14’s questions. The ability of
a drug company to declare bankruptcy alters the story some.

*4. DTC advertising makes patients more aware of both the benefits and risks of
prescription drugs than they would be in the absence of advertising. The key issue,
of course, is whether the information is unbiased; the FDA regulates the content of
the ads, but the visual imagery may convey information beyond the actual script
spoken by the actors. In concept, DTC advertising could be beneficial form a
societal perspective if it increased the use of a drug that was previously under-
used, and it could be detrimental if it increased inappropriate use. The results of
such a comparison could well differ from drug to drug.

5. Hospitals with a price below that offered by the DRG already will have no
direct effects on their behavior (hospitals that might be characterized as low-cost,
low-quality, such as simple rural hospitals). Hospitals with unusually high prices to
begin with will feel more of a binding constraint from a newly imposed DRG, and
will have to alter both quality and output accordingly. Their choices have some
indirect effects on the market facing the otherwise-unaffected “low cost” hospital.

6. In concept, this can be correct; the monopolistically competitive equilibrium

model shows that with completely free entry, the ultimate market tangency has an
EE curve consisting only of points of tangency of demand curves with one point on
the left side of AC curves for each possible quality. A savvy regulator could block
entry at a point – if done just right! – to prevent demand curves from shifting
inward all the way, and thus in concept would be able to pin costs down to the
minimum AC point for every hospital in the market. But this is a tall order for the
regulator!

7. Licensure does restrict entry, but it also provides a floor on quality that the
consumer can depend upon. If quality is very difficult for the consumer to monitor,
this quality-floor can have economic value. Whether that value overcomes the
costs created by the entry restriction cannot be determined from theory; empirical
work would be needed.

*8. Because regulations vary by state, one can look at the rate of hospital cost
increases in states with and without various forms of regulation and infer the
effects of those regulations. Studies by Sloan and Steinwald, Salkever and Bice,
and others, for example, show that CON laws have little effect on hospital costs,
despite the potential for "saving" unnecessary bed construction costs. Several
studies, including those by Sloan and Steinwald and also Dranove and Cone, show
that price controls do have some effect on cost growth. The most effective price
control was the "freeze" imposed by the Nixon administration during the mid-
1970s.

9. The monopoly pricing rule to maximize profits says to expand output until MR
= MC, or put slightly differently, to operate where P/MC = η/(η+ 1). This is the so-
called Lerner markup rule. Solving this for η with a markup of 18 percent gives η =
-1.18/.18 = -6.55. However, this “rule” uses MC, not AVC, and AVC are not MC
in general. (Generally, AVC is lower than MC.) The published “markups”
represent revenue above AVC. Suppose that half of the manufacturer’s costs are
variable (related to production) and half are fixed (research, independent of
output). Then the markup is 36 percent relative to variable costs. In that case, the
same formula gives η = -1.36/.36 = -3.777... If 80 percent of the costs are fixed,
then the markup to variable costs is 5 times 18 percent = 90 percent, and the
formula gives η = -1.9/.9 = -2.11. Etc. Thus we cannot infer much about the
elasticity of demand from these markup data without knowing the cost structure of
the pharmaceutical manufacturers.
 CHAPTER 16
UNIVERSAL INSURANCE ISSUES AND
INTERNATIONAL COMPARISONS OF HEALTH CARE SYSTEMS

Chapter Outline

General Considerations for a National Health Policy

Review of Health Policy and Systems in Selected Countries
Snapshots of Four Countries
Aggregate International Comparisons
Growth in Costs and Health Outcomes
A Final Conundrum
Summary
Related Chapters in Handbook of Health Economics
Problems

Chapter 16 has undergone yet another substantial revision, mostly in the material
about the PPACA. When the Fifth Edition appeared, the PPACA had been enacted
but was not really “underway” meaningfully. The transition is still incomplete –
parts of the law only came into effect in 2016 and 2017, and some remain deferred
into the future. So the material describing the PPACA has been updated to bring it
into “present time” as of the book’s publication (late 2017), but much ambiguity
remains, particularly what the Congress and President Trump might do to unravel
parts or all of the Act. I can only urge that you keep alert to the comings and
goings of efforts in the Congress to alter or remove the PPACA (and possibly
replace it with something else).

Teaching Tips from the Author

This chapter brings together almost everything that has preceded it in the book, and
offers an opportunity to reinforce those ideas and bring them into focus for public
policy discussions very nicely. I use my lecture time at the end of the semester to
focus on “big picture” questions in health care, and a “debate” about universal
insurance provides a natural way to do this. Depending on how good your students
are (or how brave they are in class), you can stimulate a lot of interesting
interaction with students in these chapters. I get the class going on the main
questions posed about universal insurance, and I warn them that I will “quarrel”
with almost anything that any of them put forth. It’s not that I am naturally
combative in class, but rather I want to show them that there are no “absolutely
correct” viewpoints in this debate. Any “solution” has problems. Doing nothing
has problems too! I also try (and usually succeed) to make these discussions work
in a way that the students have no idea of my personal viewpoints about universal
insurance in the US by the end of class.

The key idea is to focus on the things that any universal health plan seeks to
accomplish. Chapter 16 (as written) discusses these key issues and then talks
about how the PPACA deals with them. If the Congress repeals the PPACA, then
the discussion simply morphs into analyses of proposals to replace it, and if we see
partial repeals (e.g., the individual or employer mandates), then you can talk with
your students about the likely consequences in the markets for insurance.

Ideas worth recapturing in the discussions arising here in Chapter 16 go back

through virtually every chapter in the book. How do we deal with health-affecting
human behavior (such as tobacco, alcohol, diet)? What should be covered in the
insurance, and what will the effects be on consumption and hence total system
costs? How can we structure copayments (if any!) to help control costs? How do
we provide incentives for doctors and hospitals to produce the proper quantity and
quality of various types of care? How will they respond to various incentives?
What should the structure of a universal insurance plan look like? Should it be
uniform or just have a minimum standard? (This goes back to Chapter 10, why do
people demand risk reduction?) What role for managed care? How does the
system interact with Medicare for the elderly? How does it interact with the
medical legal system? What externalities does the system solve? Does it create
more (including the costs of public finance)? How do new technologies get
introduced? This is the material of Chapters 1–15! And it’s all pertinent to a
discussion of universal insurance.

Many people ask for my own preferences regarding universal insurance. I have
tried hard to avoid letting such preferences leak into the textbook material itself.
These choices hinge on many other things beyond economic analysis and logic.
But if somebody put a gun to my head, I’d probably say “Bernie Sanders’ plan,
with a one-word change. He wants “Medicare for All.” I’d propose “Medicare
Advantage for All.” I would include income-related vouchers and a mandate that
all individuals carry coverage, but I would also revise the “precious metal”
standard to make HDHP plans allowable.
Classroom Projects

A. If you have students who have lived abroad (e.g., junior year) or come from
overseas, have them discuss briefly the health care system in the country of
visitation or origin, highlighting one or two themes (cost, quality of care,
perceptions about equity, etc.)

B. Organize a debate between two teams, where they spend some class time
preparing both sides of this debate topic: How best to do universal coverage? Of
course, the ACA attempted to get universal coverage, combining a Medicaid
expansion with an individual mandate and an employer mandate, plus subsidies
through newly established Health Insurance Exchanges (HIX). So maybe the best
way to do it is to pit “repair” the ACA with a more extensive “replace” and ask
how that might best be done.

Again, there is fertile ground in Washington, D.C. Senator Bernie Sanders (VT)
has introduced a “Medicare for All” plan that would have a single payer (unless it
also included something like Medicare Advantage, which is really a voluntary
voucher conversion of the entitlement to Medicare). Or perhaps (my own personal
favorite) would disagree with Sanders only with one word: The basis of my
favorite plan would be “Medicare Advantage For All.”

The important thing in leading such a discussion is to remain neutral on the key
issue (should we have universal insurance) and ask probing questions to make both
proponents and opponents of the plan think, perhaps even squirm.

Sample Exam Questions with Suggested Answers

1. (10 points) "The US stands out as having far too much medical spending. We
know this because international comparisons show that the US spends far more
per capita than any other country in the world on medical care." Discuss
intelligently.

The most obvious point is that the spending patterns of countries are closely tied
to per capita income, so the discussion should begin with something like Figures
16.1 and 16.2, noting that we are “on” or “above” the regression line depending
on functional form. A more refined discussion adds the tax incentives for health
insurance in the US, noting that this will overstimulate demand for insurance
(and hence for medical care).
2. (25 points) What mechanisms are available to finance a universal health
insurance system in any modern industrialized country? (You should name at
least three for complete credit.) What are the economic costs of using each type
of financing that you mention?

These mechanisms are discussed in detail on pages 553-555.

3. (20 points) Throughout the 1970s to the 1990s, various US presidents

indicated a desire to achieve "universal insurance" in the US. Discuss at least 3
ways you know of that might achieve this goal, including the approaches used (a)
by the Nixon and Clinton administrations (10 points), (b) the Canadian system (5
points), (c) the British system (5 points).

(a) These two (and other) administrations used a tri-legged “stool” relying on
mandated health insurance for workers and their families, plus Medicare for the
elderly, and some sort of federalized version of Medicaid as the fallback for those
not covered by either of the first two options. (b) The Canadian system uses taxes
to finance a universal “Medicare” program. Taxes are mostly value added, a low-
distortion tax system. Most of the providers (doctors, hospitals, and the like) are
private, as in the US. (c) The British system uses taxes (mostly income) to finance a
state-run health care system. While there are some private doctors, most of the
provision of care is because people are eligible to receive care within the British
National Health System.

4. (20 points) There are several reasons whereby a society might think about
mandating "universal insurance". Provide at least two commonly offered of
such reasons, and discuss intelligently the strength of logic underlying those
ideas (i.e., do these "reasons" stand up to scrutiny?).

One of the reasons is basically the “merit good” argument (which some would
describe as paternalistic). Another approach says that we have a system now of
free-riders with no insurance who get care for free when they get sick (because
laws mandate at least emergency care and require hospitals to provide some
charity care for their tax-exempt status). Requiring mandatory insurance, in this
view, eliminates these free riders. Those concerned about the income distribution
effects of this approach note that as a separate issue, the government can subsidize
the insurance purchases of those with low income.

A third reason is the possibility of market failure (due to asymmetric information

in the insurance market between buyers and sellers). With some types of market
failure, social welfare can be improved, at least on average, by mandating
universal insurance at a “community” rate. In some cases (perhaps rare), the
improvement can even be Pareto-improving, where everybody is made better off
(but this takes some special circumstances that may well not exist in the real
world.

5. [This relates back to Chapter 2.] (10 points) In international comparisons of

medical care spending and health outcomes, the US seems abnormally poor in
health outcomes, given the medical spending patterns, while Japan seems to have
abnormally good health outcomes. What do you think explains these two
disparate outcomes?

Japan in part benefits from a diet very low in fat and hence little disease
associated with fat (including heart disease, and the like). They also have an
aggressive and excellent pre-natal care system. In the US, most observers note the
extreme heterogeneity of the population as one source of poorer outcomes, and the
large disparity between the quality and amounts of care available to those at the
upper and lower ends of the income distribution.

6. (15 points). Would you define health care as a necessity or a luxury good,
using normal economic definitions of those terms (Income elasticity below or
above 1.0)? What evidence do you have from aggregate (e.g., international)
and/or microeconomic (individual-level) data to support your statement? If these
types of data are inconsistent, how do you reconcile them?

On the basis of international comparisons, health care is clearly a luxury good,

with an income elasticity far above 1. In cross sectional studies, the elasticity is
below 1, and hence a “necessity” by usual definitions. One reason to account for
the difference: national income levels determine the large-scale quality of the
system, but health insurance (within that system) removes financial barriers to use,
and hence drives down the income effect on demand within any single country.

7. (15 points) Describe a profile of the types of persons in the United States who
are most likely to be without health insurance coverage. Here, "without health
insurance" means that the individual neither has private health insurance nor is
the person eligible for any public program (such as Medicare, Medicaid, etc.)
Include in your discussion the economic forces that lead to the decisions of these
individuals not to purchase insurance.
Young, low income, working in a low-wage job for a firm that does not offer health
insurance as a fringe benefit.

8. [A variant on 4.] (25 points) Proponents use a number of arguments to justify

the US adopting a policy of universal insurance in the United States. Discuss at
least two of these arguments that depend not on the belief that the government can
make better decisions than individuals, but rather that something exists in health
insurance markets that prevents the market from working well without
government intervention. (Hints: Think about markets for long term care
insurance. Think about bad debt and charity care.) In your discussion, include
some commentary on appropriate governmental “remedies” to solve the problems
that you identify in your earlier discussion.

See the answer for problem 4 in this chapter. The “hints” are designed to aim
people in the right direction. Long term care insurance, for example, is our best
working model of market failure. Bad debt and charity care raise the free-loader
issue. A third argument comes not from the idea of universal insurance, but rather
of common administrative practices: transactions costs could be reduced, says
this argument, by having a single standard payment mechanism (like the
Canadian system). Other ideas will likely emerge in classroom discussion of these
issues.

Answers to End-of-Chapter Questions

*Starred answers appear on the Companion Website.

*1. The most common uninsured person is young and (to many people's surprise)
employed. Typically, the worker has low education and hence low market wages
and commonly works for a small firm (e.g. a dishwasher in a restaurant). Such
small firms face much higher loading fees for health insurance and, hence, are less
likely to offer insurance as a fringe benefit than are larger firms. The low earnings
of these individuals also reduce their demand for insurance, since health insurance
is a normal good (i.e., the income elasticity exceeds zero and, in fact, is moderately
large). Finally, their earnings are large enough to make them ineligible for public
insurance plans such as Medicaid. Thus, they fall in between the two major sources
of insurance in the United States — public insurance for low-income people and
work-group insurance for high-income people. Some people call this group
"tweeners" because of their in-between position.
2. The outcome would depend on the nature of the mandatory insurance and its
pricing. If the law required insurers to sell to anybody who wanted at a fixed price
(one form of “mandatory” insurance), then it would effectively become a
community rated insurance (the insurers would charge a rate that would recover
their average outlays). If the mandate only required that every consumer acquire
some insurance, then the market would not differ from the separating equilibrium
market described in Chapter 10.

3. The demand elasticity is probably too large to make it valuable to include dental
coverage in any universal package, and the variance of the risks of dental injury
and illness is not sufficiently large to overcome that problem. In other words, the
political system is responding in a way similar to what private demand for
insurance would suggest, namely that without the present tax subsidy, few people
would buy dental insurance.

4. When a firm is required to provide insurance to its workers, it has to recover

those costs somehow, by raising its prices. In the worst case, the firm is in fierce
competition with foreign companies that do not face the mandate, and it simply
cannot sustain the increase and will leave the industry (hence costing the US
workers their jobs). In a milder version, the competition is only within the US, but
the higher price leads to demand shifts away from that industry, and some jobs are
lost. Ultimately, in either case, the other possibility is that other wages paid to the
workers shrink to compensate for the higher health insurance costs.

One group where this cannot happen is those up against the minimum wage
law limits. For that population, wages cannot fall to offset the higher costs, and job
loss is assured.

*5. These data show a generally positive link between density and earnings. Here
are the data:
Primar
y Primary Ortho Ortho Relative Relative
Densit
Density Earnings y Earnings Density Earnings
31.1111
AUS 14 93 0.45 188 1 0.494681
CAN 10 125 0.32 209 31.25 0.598086
FR 17 96 0.34 154 50 0.623377
GER 10 132 0.44 203 22.7272 0.650246
 7
UK 7 160 0.28 324 25 0.493827
15.1515
US 10 187 0.66 442 2 0.423077

If you graph relative earnings vs. relative density, you will see an upward sloping
line with Germany as an outlier to the upper left. With Germany included, the
regression line has t = 1.27. Excluding Germany, you get t = 2.99. Germany is
clearly an “influential observation” in this small data set. Germany is unusual
because of the high relative earnings of primary care doctors and the low relative
density. Something in Germany gives primary care doctors more earnings clout
than their relative density would explain.

6. This quote ignores the very strong relationship between per capita income in a
country and its medical spending patterns. The US would have the highest medical
spending among all countries just because of its income, so the high spending
alone proves nothing about “waste.”

7. Japanese citizens have almost no dietary fat, and their hospitalization for
diseases related to fat consumption (heart surgery, gall bladder disease) are
extremely low compared to that of other countries.

*8. Almost all of it. If one plots spending per person versus average per capita
income, the United States lies almost exactly on the line, perhaps slightly above it,
depending in part on whether one fits a straight line or a logarithimic curve (see
figures 16.1 and 16.2).

9. Within any single country (such as the US), health insurance will tend to
obliterate effects of income differences. At the extreme, a fully insured poor person
and fully insured rich person should have quite similar consumption of health care
(subject, of course, to the effects of their distinctive life style choices). But across
nations, no matter whether the health care system is competitive, regulated, or
state-owned, political and market forces will lead to investments in quality and
availability so that, at least on average, consumers’ preferences are met. (Neither
the market nor an intelligent political system would lead to other outcomes.) Thus,
the income differences across countries could well lead to large differences in
health care spending across those countries, while within each country, the income
effects could be quite small.