Journal of Health Economics 4 (1985) 187-193. North-HoKand
3DITBRIAL
The Theory of Physician-Induced Demand
Reflections after a Decade
Uwe E. REINHARDT
Princeton University, Princeton, NJ 08540, USA
Received March 1985
Throughout the past decade or so, American health economists have debated
the question of whether or not physicians can create demand for their
services. The debate has been lively and intense, for at least three reasons.
First, in the role of policy analyst, one would certainly wish to know the
correct answer to this question. The issue bears importantly upon public
policies concerning the training of health professionals.
If physicians could create demand for their services at will, then they could
counteract a drop in the market’s population-physician ratio simply by
increasing the utilization of their services without lowering their fees.
Increases in physician density could thus lead to increased health-care
expenditures without commensurate benefits to patients. Government-
imposed limitations on the supply of medical manpower or controls on
utilization might therefore be legitimate components of any cost-containment
strategy.
On the other hand, if the price-output decisions of physicians were
constrained by exogenously determined demand, then increased utilization
could be achieved only by lowering fees, by lowering the time price of
accessing physician services, or by changing the quality of physicisin services
in ways that are valued by patients. In a market economy, such increases in
utilization would be economically legitimate, even if they resulted. im higher
health care expenditures. It would therefore be more dificult ‘La make a case
for government intervention in the market for physician services.
Our profession’s perennial search for truth, then, could amply justifv the
decade-old debate on this issue. But one suspects that vested interest in
neoclassical economic theory has added at least some fuel tee the flames.
astery of the neoclassical framework requires a heavy personal investment
0167-6296/85/$3.30 @ 1985, Elsevier Science Publishers B.V. (North-Holland)
188 U.E. Reinhardt, Editorial
e part of the analyst. Among the payoffs to that investment is entree
into a fraternity whose power has derived in good part from the unity of
thought forged by this shared analytic paradigm. One need not be an utter
cynic to believe that, quite apart from our profession’s yearning for truth, the
defense of that unifying framework can take on a life of its own. As Yale
economist Richard Nelson has observed on this point:
‘Powerful analysis requires strong analytic structure.. . However, an
intellectual tradition of the sort required to develop a strong analytic
structure usually develops an explicit or implicit commitment to a
particular point of view.. . [in other words], intellectual traditions tend
to involve a greater commitment to particular [analytic] structu ...
1. Journal of Health Economics 4 (1985) 187-193. North-HoKand
3DITBRIAL
The Theory of Physician-Induced Demand
Reflections after a Decade
Uwe E. REINHARDT
Princeton University, Princeton, NJ 08540, USA
Received March 1985
Throughout the past decade or so, American health economists
have debated
the question of whether or not physicians can create demand for
their
services. The debate has been lively and intense, for at least
three reasons.
First, in the role of policy analyst, one would certainly wish to
know the
correct answer to this question. The issue bears importantly
upon public
policies concerning the training of health professionals.
If physicians could create demand for their services at will, then
they could
counteract a drop in the market’s population-physician ratio
simply by
increasing the utilization of their services without lowering
their fees.
Increases in physician density could thus lead to increased
2. health-care
expenditures without commensurate benefits to patients.
Government-
imposed limitations on the supply of medical manpower or
controls on
utilization might therefore be legitimate components of any
cost-containment
strategy.
On the other hand, if the price-output decisions of physicians
were
constrained by exogenously determined demand, then increased
utilization
could be achieved only by lowering fees, by lowering the time
price of
accessing physician services, or by changing the quality of
physicisin services
in ways that are valued by patients. In a market economy, such
increases in
utilization would be economically legitimate, even if they
resulted. im higher
health care expenditures. It would therefore be more dificult ‘La
make a case
for government intervention in the market for physician
services.
Our profession’s perennial search for truth, then, could amply
justifv the
decade-old debate on this issue. But one suspects that vested
interest in
neoclassical economic theory has added at least some fuel tee
the flames.
astery of the neoclassical framework requires a heavy personal
investment
3. 0167-6296/85/$3.30 @ 1985, Elsevier Science Publishers B.V.
(North-Holland)
188 U.E. Reinhardt, Editorial
e part of the analyst. Among the payoffs to that investment is
entree
into a fraternity whose power has derived in good part from the
unity of
thought forged by this shared analytic paradigm. One need not
be an utter
cynic to believe that, quite apart from our profession’s yearning
for truth, the
defense of that unifying framework can take on a life of its own.
As Yale
economist Richard Nelson has observed on this point:
‘Powerful analysis requires strong analytic structure.. .
However, an
intellectual tradition of the sort required to develop a strong
analytic
structure usually develops an explicit or implicit commitment to
a
particular point of view.. . [in other words], intellectual
traditions tend
to involve a greater commitment to particular [analytic]
structures,
which may or may not obtain, than their practitioners believe.‘l
Those who qui= I &on the existence of a stable demand
constraint on the
physician’s price-output decisions are assaulting one of the
crucial pillars of
the neoclassica. framework. It is eminently understandable that
4. the guardians
* . .
or tnat frame-work parry such assaults with vigor.
A third reason for vigor in debate on the issue may be its
linkage to
political ideology. While individual economists rarely tailor
their own value
systems to their clients’ wishes, members of the profession are
arrayed along
a wide stretch of the ideological spectrum. An economist’s
position on this
spectrum is bound to color his or her interpretation of data,
from the
selection of an econometric model to judgements on the
‘plausibility’ of
estimated coefficients. As everyone surely appreciates, the
typical report on
econometric research tends to be as much an essay in persuasion
as it is
positive science. In the present context, economists who are
relatively more
comfortable with government intervention in private markets
are more likely
to countenance econometric heresies on the neoclassical model
of physician
behavior than are economists who are ideologically opposed to
such
intervention.
This much can be conceded - and probably should be conceded -
without in the least besmirching our profession. It merely
exposes economists
as ordinary human beings. But it also points to the need for
careful audit of
5. the policy analyses produced by our profession.
The accounting profession recognizes the existence of ‘creative
accounting’,
by which those who wish to tell a preferred story with numbers
can usually
do SO. TO counterbalance this potential for bias, accountants
have developed
and adopted very strict rules of audit. An accounting audit
wouid be
inadequate were it confined merely to checking the additions
and footnotes
in a corporation’s published annual report. A good audit
requires access to
the raw data, and such access is routinely granted.
‘Nelson (1977, pp. 15 and 23).
U.E. Reinhardt, Editorial 189
Similslrly, the audit of an economic analysis would be
inadequate if it were
kept strictly to assessing an author’s published econometric
model and
estimates. Here too, access to the raw data used in research is
the sine qua
non of a quality audit. We would therefore do well to emulate
the
accounting profession’s tradition of making our raw data
available when we
publish empirical policy analyses.
All of this is by way of introduction to the subject at hand. Let
us now
6. turn to the questions that lie at the heart of the matter.
The empirical question
The debate over
distinct questions:
physician-induced demand can be decomposed into three
(1) Can physicians manipulate their patients’ demand for
physician services?
(2) If so, do physicians ever do this for the sake of pecuniary
gain?
(3) If so, will physicians at any time have fully exploited the
potential to
maximize their hourly income in this way, or is there
unexploited slack
whose extent varies with the economic pressure on the
physician?
One must wonder whether anyone outside the American
economics
profession would ever answer the first question in the negative.
Indeed, one
wonders whether even American economists would doubt that
the answer is
‘yes’.
Physicians, in particular, seem fully persuaded that they possess
consi-
derable leeway to manipulate their patients’ demand for health
services.2
Indeed, the present author has never yet met a physician who
would deny
the existence of that power (nor has he ever met one who would
7. admit to
exploiting it for personal gain). Rather persuasive systematic
evidence of the
physicians’ perception can be found in the now-famous
‘Wennberg variations’
- the vast interregional differences in per-capita utilization of
elective medical
procedures, which Wennberg and other researchers have so far
been able to
explain only with recourse to variables such as ‘physicians’
preferred practice
style’.3 To explain such variations with appeal to autonomous
consumer
preferences (‘women in area A simply crave hysterectomies
more than do
women in area B’, and so on) would stretch one’s credulity, as
would
attempts to explain the variation with sppcal to time- or money-
prices.
In their writings on physician-induced demand, economists have
typically
sidestepped the first question altogether, and the second one as
well. The
focus of their work has instead been on the third question. Thus,
instead of
inquiring into the medical appropriatensss of entire medical
treatments under
2See, for example, Menken (1983).
%e Wennberg (1984).
190 U.E. Reinhardt, Editorial .
8. varying degrees of physician density, economists have inquired
whether or
not there exists an ultimate, effective constraint on physicians’
price-output
decisions. This condition is referred to in the literature as a
‘stable demand
function for physician services’.
It has never been quite clear to this author just what is meant by
a ‘stable
demand curve’ in this context. Consider, for example, a profit-
.maximizing
firm selling its products in a market constrained by the demand
function
Q=Q(P,KA), Q&k Q&O, (1)
where Q denotes the quantity demanded per period, P signifies
the own
money price of the commodity, vector X incorporates other
relevant
arguments (including, perhaps, the time price of accessing the
commodity
and the amenities that accompany the process of accessing it),
and A denotes
some physical measure of ‘persuasion’. For an ordinary
business firm, one
might measure A by the firm’s advertising effort, in terms of,
say, its budget
for that purpose. In the context of medical practice, A might
include (of
recent) explicit advertising as well. But it might also represent a
physician’s
efforts to shift, by other means, the demand curve in P-Q space
at given
levels of P and X - e.g., by scaring patients into accepting what
9. fully
informed persons would reject as dubious medical procedures.
For an ordinary business firm, the price of a dollar’s worth of
advertising
would be $1.00. In the case of medical practice, the price of
raising A might
include, as Evans (1974) has argued, the psychic pangs of
conscience a
physician may develop when, for the sake of pecuniary gain,
(s)he persuades
patients to accept services of doubtful or negative medical
value.
What now do we mean by a stable demand curve? Is it the
penultimate
frontier at which aQ/iYA =O at all points in P-Q space, or is it
function (1) as
a whole? In practice, of course, an ordinary business firm
constrained by eq.
(1) would be unlikely to operate on the penultimate frontier, so
long as there
are positive marginal costs associated with increased
advertising effort. The
precise position of the demand curve in P-Q space remains
endogenous.
As the present writer und.erstands Evans - the quintessential
proponent
of the view that physicians1 do create demand when put under
increased
economic pressure - that author proposes that t?Q/iJA is
positive for many
physicians in utility-maximizing equilibrium. Evans further
argues that the
psychic ‘price’ per unit of A falls with rising economic
10. pressure, that ethical
conduct is a normal good. If so, the demand curve faced by the
individual
physician is likely to shift toward the penultimate frontier
(where i?Q/iYA=O)
as the population-physician ratio falls to the point of putting
downward
pressure on the physician’s income. This shift would represent
demand
creation in the pejorative sense of that term.
It appears that the neoclassical critics of Evans (and of other
members of the
U.E. Reinhardt, Editorial 191
supplier-induced demand (SD) school) assume that, at any time
we observe
the physician, i?Q/aA =O at any level of physician density. In
our debates on
this issue, members of the neoclassical school have always
raised the
question: if physicians can create demand for their services as
physician
density rises, why haven’t physicians done that even before
density rises? In
other words, it is assumed that physicians will at all times
maximize their
hourly income. Such an assumption, however, must imply one
of three
propositions:
0 a
11. m
0 C
Physicians have no conscience; therefore the psychic ‘yrke’ of
A is zero
at all times, and the demand will at all times be shifted to the
point at
which aQ/aA=O.
Physicians do have a conscience, ai d it is totally impervious
pressure. Shift variable A is engaged solely in the patient’s
never just in defending the physician’s income.
to economic
interest and
It is technically impossible for physicians to manipulate the
demand for
their services, because patients are just too smart to be seduced
in this
way. In other words, the answer to the first question raised
earlier is
simply ‘no’.
Just which of these propositions is believed by the neoclassical
school
remains something of a mystery. If one had to bet on the matter,
one would
probably put one’s money on proposition (a). As already noted,
it is doubtful
that even economists would defend proposition (c) (although
some might, in
a pinch, to sustain an argument). Proposition (b), on the other
hand, is just
not in keeping with received economic doctrine. To hold
professional ethics
12. completely impervious to economic incentives borders
dangerously on soei-
ology or other forms of ad-hocery. No self-respecting economist
would
engage in such ad-hocery. This leaves as most conformable with
neoclassical
thought proposition (a), that physicians really have no
conscience. That
seems to be our profession’s working axiom in this area. It is an
appealing
proposition, if o&y because it is so tantalizing.
Is the physician’s persmwive power udidted?
The hypothesis that physicians can manipulate the demand for
their
services has two distinct historical roots.
One root is located in the national health insurance systems of
Canada
and Western Europe. There it has been observed that, whenever
fees are
depressed, the per-capita utilization of physician services has
tended to rise in
partial or total compensation for the lowered fees. It has also
been found
that per-capita utilization of physician services in these systems
tends to rise
more or less in step with increases in the physician-population
ratio. There
192 U.E. Reinhardt, Editorial
are some neoclassical explanations for this correlation, but the
13. SiD hypo-
thesis is data-compatible as well.
Bn the United States, where fees are not constrained by fixed
schedules, the
iqpothesis arose from observed, puzzling positive correlations
between
physician fties and physician density. As none other than this
journal’s
distinguished editor observed in his article ‘A Model of
Physician Pricing’
(1970):
‘Satisficing behavior may explain the high positive partial
correlation of
the number of ptRctit;oners with price. Suppose physicians have
a
cer;&r income target. As the number of physicians in an area
increases,
visits per physician will tend to fall. To achieve any given
target income,
each physician will then have to charge higher fees.‘4
Newhouse was in good company in his early espousal of this
‘target-
income’ hypothesis. Many now well-known health economists
flirted with
that hypothesis during the late 1960s and early ‘7Os, the present
author
included?
In those days of a physician shortage (and a general paucity of
data all
around) it was sometimes pretended that there might be no limit
at all to the
power of physicians to create demand for their services and to
14. raise their fees
- that no imaginable increase in the supply of physicians would
ever push
physicians into under-doctored areas or would ever depress
their incomes in
preferred locations. Health economists can take credit for
having, in the
meantime, dispelled such wild notions through painstaking and
often so-
phisticated research. Thomas K. McCarthy’s paper in this issue
falls within
that genre.
McCarthy finds that the market for @rzary-care physician
services in large
SMSA market areas is monopolistically competitive, and that
the individual
physician’s pricing and scheduling choices are effectively
constrained by a
downward-sloping demand function exhibiting an own-price
elasticity in
excess of unity. Furthermore the author finds that, ceteris
paribus, an
increase in physician density in the relevant market area shifts
the demand
curve faced by the individual physician to the left - implying
that, ceteris
paribus, increastis in density produce downward pressure on
prices. It is an
intriguing paper, one whose underlying data base one would
love to have at
hand.
This author finds McCarthy’s paper illuminating, even though,
along with
Mark Pauly’s earlier work,” it relies on ‘patient visits’ as the
15. output measure
and on ‘fees for a routine follow-up visit’ as the relevant price
variable.
Furthermore, it is confined solely to primary ambulatory care
for which
4Newhouse (1970, p. tbi j.
‘See Reinhardt (1975, appendix B).
%ee Pauly (1980).
U.E. Reinhardt, Editorial 193
insurance coverage has traditionally been thin in this country.
Adherents of
the supplier-induced-demand school will surely claim, and
rightly so, that the
story will never have been fully told without research on the
response of
entire medical treatment for given episodes of illness to changes
in physician
density. They also are concerned with all physician services,
including
relatively well insured non-primary care. The debate on
inducement will
obviously continue.
McCarthy himself points out one important limitation of his
study,
namely, that while his findings indicate a binding demand
constraint at the
margin, there may be substantial, already fully exploited
demand creation at
that frontier and that ‘health policy analysts may not wish to
exclude
16. utilization review from their regulatory agendas based on these
results’. The
author is to be commended for registering this caveat explicitly.
Too many
defenders of the faith pretend as if that issue were not
important, that
identifying some ultimate limit is enough. It may be enough to
save the
reputation of the neoclassical framework. If one’s interest goes
beyond that,
one would worry about what happens on the -way to the
ultimate frontier.
Indeed, one wishes that he had ended his paper right at that
point.
Although competition in health care may well turn out to be this
nation’s
salvation, McCarthy’s celebration of that approach in the latter
pages of his
paper burdens somewhat the empirical pillars he has constructed
in it. Why
go that treacherous extra mile?
References
Evans, Robert G., 1974, Supplier-induced A,___,__,_ em=+
Some cmuirical evidence and implications, in:
M. Perlman, ed., The economics of health and medical care
(London).
Menken, Mathew, 8983, Consequences of an oversupply of
medical specialists: The case of
neurology, New England Journal of Medicine 308, May, 1224-
1226,
Nelson, Richard, 1977, The moon and the Ghetto (New York).
17. Newhouse, Joseph P., 1970, A model of physician pricing,
Southern Economic Journal 37, no. 2,
174-183.
Pauly, Mark, 1980, Doctors and their workshops: Economic
models of physician behavior
(Chicago, IL).
Reinhardt, Uwe E., 1975, Physician productivity and the
demand for health manpower
(Cambridge),
Wennberg, John E., 1954, Dealing with medical practice
variations: A proposal for action,
Health Affairs 3, no. 2, 632.