Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 224
Chapter 9
THE INFLUENCE OF ECONOMIC FACTORS ON
MEDICAL STUDENTS' CAREER CHOICES
Sunny G. Yoder
A medical career may be viewed as a series of private, personal
choices made by an individual. At the outset is the choice to enter
medicine, a choice that becomes manifest with the application to
medical school. If admitted' the student faces subsequent choices of
a specialty and residency program in that specialty. Upon completing
post-M.D. training, the individual chooses an initial practice
location and mode. The physician may later choose to change
specialty, location, or mode of practice. Each of these choices is
based on the individual 's unique set of values, ob jectives,
perceptions, abilities, and circumstances. Because the choices also
have important societal consequences--in the aggregate they determine
how many physicians there will be, in what specialties, and
where--considerable research has been undertaken to identify the
factors that influence these choices, particularly those factors that
might be amenable to manipulation through public policy.
The purpose of this chapter is to review the portion of that
research literature representing the contribution of economists. As
the output of a single discipline, this literature has its own
particular characteristics that should be made explicit at the
outset. In common with researchers in other social sciences,
economists look to f ind systematic relationships between certain
outcomes or behaviors and variables hypothesized to affect those
outcomes. Idiosyncratic behavior, although it may be important, is
treated as random noise; thus, some of the richness and complexity is
lost in the attempt to explain the average response of a dependent
variable in relation to several independent variables.
Economic research relies more heavily on observed behavior than on
stated intentions or motivations. Thus, although a principal objective
is to predict behavior, studies tend to draw on retrospective data.
For example, as discussed in detail below, the analysis of physicians'
specialty choice is based on data f ram physicians who have already
chosen a specialty rather than on students' future plans. Motivations
are inferred from actual choices, characteristics, and backgrounds.
This literature starts from a fundamental premise that people act
rationally in order to attain the highest level of well-being
possible, and that economic factors play a part in that well-being.
Economists do not assert that these factors dominance all others. It
is assumed that medicine is chosen as an occupation by individuals who
perceive it as compatible with their abilities and interests, and with
the kind of future personal and professional life they want. For most
people, however, more than one occupation might be a viable candidate
on this basis. Economists, as the following statements illustrate,
224
OCR for page 225
f ind it dif f icult to imagine that economic considerations have no
inf luence on the choice:
It is most improbable that the costs of education and the
income of physicians are irrelevant to the decision to
pursue medicine.]
It would be hard to believe. . .that modern medical students
in the United States and elsewhere typically chose their
careers completely in abstraction from pecuniary reward.2
Finally, economic research on these matters has considerable
relevance for public policy, since many economic variables are subject
to manipulation through the policy process. Thus, even if these
variables only matter at the margin, they may be useful policy revere.
This chapter reviews economic research on the decision to enter
medicine and on the choice of specialties and suggests areas in which
further research would contribute to a better understanding of how
economic factors affect the size and specialty composition of the
physician pool.
Medical Education as an Investment
The conceptual framework underlying research by economists on
medical students' career decisions is that medical education is an
investment.3 That is, medical education involves an expenditure of
time and money in the present that yields a stream of benefits in the
future. The time and money costs of medical education act to
discourage the potential student from making the investment; the
future stream of income and other benefits encourage the investment.
The desirability of the investment can be determined by comparing the
present discounted value of its expected costs and benefits.*
*Discounting is necessary because today's dollar can be invested to
yield a greater amount in the future, and therefore the value today of
one dollar to be received in the future is less than $1. As an
example , if the current interest rate is 8 percent per year, then 61
invested today will yield $1.08 in one year. Over two years at a rate
of 8 percent, the yield would be t1.17, which is $1 x (1 + .08) x (1 +
.08), or $1 x (1 + .08~2. In general, the total yield, T. at the
end of n years on an investment of Y dollars at a constant rate of
interest r would be
T ~ Y(1 + ryn.
The higher the rate of interest r, the greater the yield, and,
conversely,.the less the present value of a future dollar. At an
interest rate of 8 percent per year, the present value of $1 to be
obtained in a year would be $1 divided by (1 + .08), or $0.93. If the
interest rate were 12 percent, the present value would be $0.89. As a
consequence of discounting, returns and costs that occur in the near
term carry much greater weight than those farther in the future.
225
OCR for page 226
The investment framework is useful for analyzing how individual
students, or potential students, respond to economic incentives,
because it simultaneously accounts for the costs of education and the
subsequent stream of earnings. If medical school tuitions rise, we
would expect the number of students applying to medical School to
fall, all else equal. However, rising physician incomes might offset
the negative ef feet of the tuition increase. The relative desirability
of an investment in medical education, may be characterized by the net
present value (NPV) of the investment. NPV is calculated as the dif-
ference between the sum of all costs of the investment, appropriately
discounted, and the sum of all benefits, also discounted. Any
investment with a positive net present value is desirable; how
desirable depends on the comparison of its NPV to shone of alternative
investments.
A second measure of the relative desirability of an investment in
medical education is the rate of return. This is the discount rate
that equalizes the present value of the costs and benefits of the
investment.
Economic Returns on Investments in Medical Education
A number of estimates have been made of the rate of return to
medical education and its net present value. The rate of return
provides an easily understood measure of the financial attractiveness
of a medical career, since it can be compared with rates of return
available from alternative investments and with market interest
rates. On the other hand, sometimes the rate of return can lead to
erroneous conclusions about the relative value of investments
embodying different levels of costs and returns. The better of two
investments is the one yielding the larger net present value, which
may not be the investment yielding the higher rate of return.4
Using either criterion, the literature consistently has found
medical education to be a good investment. The findings and methods
are reviewed below, following a brief discussion of the methods used
to estimate costs and earnings.
Although methods for estimating the rate of return or NPV vary in
their details, the basic approach is essentially the same. First, a
comparison group is selected. The most common comparison group is
college graduates, in which case the result is the return on the
additional investment in medical education. The comparison group also
can-be chosen from other occupational alternatives such as law or
dentistry.
- Once the comparison group is selected, cost and income streams are
constructed for the two alternatives: medicine and the comparison
group. Out-of-pocket education costs generally are based on average
tuition charges, or tuition and fees. Some researchers reduce tuition
and fees by subtracting out average scholarship awards or loan amounts
226
OCR for page 227
that are used to pay these out-of-pocket costs. Other components of
costs are the earnings forgone by medical students and by residents.
Usually medical students' forgone earnings are estimated by the
average earnings of college graduates, although some researchers have
used higher figures on the assumption that medical students' have
higher than average ability. Forgone earnings for residents may be
estimated by the earnings of general practitioners, since (in most
states) a physician may enter general practice after one year of
post-M.D. training.
Future income streams are estimated from cross-sectional data on
income by age at a point in time. For physicians, such data are
available f ram periodic surveys of practitioners conducted by the
periodical Medical Economics and by the American Medical Association.
The U. S. Bureau of the Census publishes estimates of income by age
and sex for different educational attainments based on its Current
Population Surveys.5
Using these cross-sectional age-income figures to construct a
future income stream requires assumptions about (1) the length of the
working life (most researchers assume retirement at age 65), (2) the
future rate of income growth, and (3) life expectancy. In addition,
earnings streams may be adjusted for hours of work, for inflation, and
for taxes. Some studies include an estimate of earnings while in
school; others disregard such earnings, assuming they would be
negligible.
Estimates of the Rate of Return
Sloan estimated the internal rate of return to medical education
relative to college graduates for a number of years between 1941 and
1966e6 Education costs were estimated from medical school tuition
and fees, from which student stipends were subtracted. Census data on
median incomes of white male college graduates served as estimates of
forgone earnings of medical students. Physician earnings for this
period were estimated with data from Medical Economics. The income
data were deflated, but no ad ju~tments for taxes were made. Medical
students were assumed to earn, while in school, one-fourth of the
income of white male college graduates. All medical school graduates
were assumed to take three years of residency training; it does not
appear that their forgone earnings are included as a cost. They also
were assumed to spend some time in the military service, although the
number of years of service assumed is not reported. The school-work
life Was assumed to cover ages 22 through 65, while survivorship rates
were used to account f or deaths before 65.
In the most recent period, 1966, S. oan's estimate of the rate of
return was 18.2 percent . According deco his estimates, the rate had
risen gradually since 1962, when it was 16.6 percent. During this
period applications to medical school increased sharply. These rates
of return may have underestimated actual returns, since Sloan used
227
OCR for page 228
real income (ad jus ted income f or inf ration based on the Consumer Price
Index) but used discount rates more closely approximating nominal
market interest rates.. Had he used real interest rates, which tend to
be in the range of one to three percent, or nominal earnings, the
estimates would have been higher.
Using a similar methodology, Feldman and Scheffler estimated that
the rate of return to medical education, using male college graduates
as the comparison group, was 22 percent in 1970.7 Physicians'
average pretax incomes by age were taken from the AMA Seventh Periodic
Survey of Physicians. Apparently these incomes were not deflated,
although the method for constructing the earnings stream is not
described in detail. Physicians were assumed to enter practice after
one year of internship, three years of residency, and two years of
military service. The physician work-life was assumed to age 6S;
however, unlike Sloan's study, age-specific survivorship rates were
not applied to the earnings stream. Costs of education were estimated
from Association of American Medical Colleges (AAMC) data on medical
student expenses, less estimates of scholarships and interest
subsidies on student loans. By assuming that all physicians live
until retirement age, Feldman and Scheffler somewhat overestimated the
rate of return in 1970. Since Sloan's estimate may be slightly low,
the increase in the rate of return between 1966 and 1970 was probably
less than this research indicates. Presumably it did increase.
A third estimate of the rate of return to medical education (in
comparison with a college degree) was made by Fein and Weber.1
Actually, this study produced several estimates under different
assumptions about (1) expected growth in physicians' earnings, (2) the
rate of general inflation, and (3) expected growth in earnings of
white males with bachelor's degrees, and whether these earnings
reflect armed services experience. All physicians were assumed to
spend two years in military service between internship and residency
training. For 1966, in comparison with white male college graduates
with no military service, ant assuming no real growth in earnings (the
most conservative set of assumptions), the rate of return was
estimated at 15 percent. Under assumptions of moderate growth in
earnings for physicians and college graduates, the rate of return was
17 .8 percent ~ Under the same earnings assumptions, but assuming that
college graduates do not spend time in the military, the rate of
return was 21.7 percent.*
Estimates of Net Present Value
A recent paper by Lee and Carlson presents a summary of estimates
of the net present value of medical education (Table 1~.8 frost
investigators have attempted to adjust their estimates to reflect
dif f erences in hours worked by physicians and by incumbents in the
alternate occupation. One method for doing this is to adjust
*These results are summarized in Appendix C, Table C-3, of Reference
#1.
228
OCR for page 229
o
co
CJ
-c
c
cn
z
o
en
cn
In
LJ
<¢
C9
~ -
u' ·"
~ ¢
In
In
v ~ Jo
In ~
In ~ .,
au
Id ~ ¢
an
In
o 0
:~:
as ¢
~ o
lo
~ a
¢ 0
o
C)
Go ~
C. Cal
o
u
¢
~ 0
0 ~
on
r_ O
Is In
Cal Cal
to t
o o
c)
)
o o
u, cn
~c ~
~ u~
a
o o.
r_ co
1
~*
o o
~ ~o
u~ ~
0 ~
ct ct
~ c~
.
~o
o o
~o
~ ~ c~ 0
c~ ~ ~ ~
o' ~ o~ o
o ~ c~ o
e~ ~ c~ a'
0 r`+ 0
~ ~ o
~ o c~ ~
~ ~ ~ c~
~ ~ o ~
r~
cr' ~ I_
ce
c,
~o
o
3
C)
U,
0 ~ C~
CO C~
o o'
o
C~
o' ~
oo a,
CO
o ~ o
U~ o
omo ~
~ ~ ~ .
r_
C~ U~
C~
:~
u ~
~ V
U: ~
0
C.
s~
C'
so
~ o o
.
o~ ~ ~ ~ ~ o~
U~
CS' ~
:^
Ct
0 0
- ~
o
C)
Q}
tn
o
~C
o
v
·
V
a) ~
S" n,
~ o
· pt
-
~u
s~
co ~
3
~ . ,.
O ~ ~
C~ O
~ 3
N ~ V
= -
O
o0
~ ¢
S" O
CO ~ ~
3 0
·
0`
~ ~ ~ S~
0
~ ~ V
I, ~ ~
O ~ Ct
Ct ~ ~ ¢ Cl)
~ ~ ~ 0
0 3 ~ ~ ·
O ~Y ~ ~ ~ ~ ~
- 1 ~ Ll O
~ ^ GO Ct ~ CO
"= C~
V U ~ ~ 5~ ~ O
CU-~
O
r, ~ ~ ~ ~
t0 ~ 3 Ca ~ pc
0 ~ ~ ~ O
t0
C
CO
~
L' C
O ~ ~ ~
~ ~ ~ ~ 00 0
C~ ~ U) O
C~ ~ O O
~ b0 b0 U)
0 C~
000 0 ~ ~ O O
~ ~ ~ O · O
Sal >, 0 3 0 3 U' 0 ~
t~ ~ ~ ~ ~ ~ ~ ~ 0
=; ~ =:
V~ ~ - b0=
~0
^ V
~ O
cn
0
C. O
U' C~
CO
~ o
o v ~
og o o
cq
~ o ~
cn
a'
3 ~ (V
C~
S" O 0
0`
3
S" O
~ S~
¢ CO
00
·
Ct ~
Q}
U~
O ~ U~
1` ~ 1`
S"
:~:
229
^ _1
oo
so ~
~ o
s"
Ct
C.
CO
o
~C
c
C~
s~
Pt
C~
1
-
~u
·.
3
o
u,
OCR for page 230
physicians' earnings downward to compensate for their longer hours
(Lespeyre adjustment); the other is to adjust earnings in the
alternate occupation upward (Paasche adjustment). Table 1 presents
net present values unadjusted for differences in hours and adjusted by
both methods, where possible. The data year, in the second column,
may be thought of as the year in which the choice was being evaluated.
Thus, according to Sloan, from the perspective of a college graduate
in 1955, the present value of the additional investment in medical
education was just over $79,000, assuming a medical career as a
general praceitioner.6
The significance of differences in hours is illustrated, perhaps
too dramatically, by Lindsay's estimate.l° Using the same data that
Sloan used, but adjusting for hours worked, Lindsay found the net
present value to be negative, approximately -S15,700. However, the
physician work-hours used for the adjustments may have been high and
thus the net present value too 1OW.l3
The precise dollar figures in Table 1 are less important than is
the general message: a number of researchers using different data and
methods have found that, on average, medical education is an excellent
investment; only law is reasonably close to medicine in its economic
returns. This f inding holds, whether the comparison is with college
graduates or with college faculty, in which case total years of
schooling are much closer to those for medicine.
The most recent and detailed analysis of the relative returns to
education for the practice of medicine in comparison with a number of
other occupations was done by Dresch with data from the 1976 Health
Care Financing Administration survey of physician practice costs and
income and from the Census Bureau's 1977 Current Population Survey.l2
Dresch' s results, summarized in Table 2, show that in 1976 the invest-
ment in medical education had a present value of well over 6100,000
relative to eight out of 14 alternative occupations. Again, the
investment is only slightly favorable when lawyers are the comparison
group ~ the net present value of the medical education is jus t over
$11,000~. However, in comparison with natural scientists, a career in
medicine was worth $198,000 in 1976, even though the two occupations
had similar educational requirements.
Although these studies have produced varying estimates , on the
whole the findings are fairly consistent : the economic returns to
medical education have been found to be substantial. As the
investment framework allows us to see, not only have physicians'
earnings been high--a fact that surprises no one--but, even with the
substantial education costs taken into account, on average the
prospective physician could expect to receive large net returns. Do
these returns influence the decision to enter medicine? The following
section reviews research bearing on this question.
Economic Returns and the Demand for Medical School Places
Several studies have addressed the question of the relationship
between economic returns and student demand for medical education.
230
OCR for page 231
oh
-
o
g
at:
2
Cal
v
:~:
c
c
0
c
o
In
Cal
z
lo:
I Cal
-
V
o
~ o
0 o
Z
D
-
o
A:
1 _.
l
I_
0`
A:
a,
C
At: C
rat
0\
u3
0
O
O.
S
en U.
_
D ~
U
C
O
· e
~ O
~ en ~ ~ ~ ~ 0 0 ~ ~
· ~ ~ ~ ~ ~ ~ ~ · ~
O ~ ~ ~ ~ Cat ~ O ~ \0
~ ID ~ ^1 0~ ~ _1 _d _1
_ ~ ~ ~ ~ ~ ~ ~ _ ~ ^ ^ ^ ^
- 0 ~ ~ 0 0 0 ~ e~ ~ ~ ~ 0 ~ ~
O ~ ~ ~ ~ ~ o'
U~ ~ ~ ~ ~ ~ ~ ~ ~ C~
_ _~ _ _ _ _ _ _ _ _ _ _ _ _ _
C~ o' 0 ~ ~ oi ~ ~ ~ ~ ~ C~
C~ 0 0
~ ~ ~ ~ 0
C~ C~ ~ ~ C~ ~ ~ ~ C~ ~ C~ C~
O ~ ~ ~ ~ 0 U~
u~ ~ 0 ~D ~ ~ ~ ~ ~ ~ ~ a
Co ~ ~ ~ ~ ~ ~ U~
oo ~ ~ 0 o' 0 ~o ~ o' O
C~ ~ ~ ~ C~
1 1 1 1 1 1 1 1 ~ 1 1 1 1 1 1
o' 0 ~ 0 0
C~ D ~J ~ ~ C~ ~ O ~ ~ 1 - _
0 ~ 0 0\ _' O ~ 0 0 0 ~ ~J ~J ~ CJ'
0 \0 ~ ~ 0 C~ O
_~ C~J _~ _I
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _
0` ~ ~ O ~ 0 ~ ~ ~ ~ O ~ U~ o' O
O ~ ~ ~ ~ ~ 0 ~ ~ t_ ~ C~
~J 0\ ~ ~ _1 ~ ~ ~ O O
O ~ ~ ~ O
C~
0` 0 0 ~ O
C~
0 ~ cr' ~ ~ 0 ~ 0 ~ 0
~1 O ~ ~ ~ ~ —~ 1` ~ 0
C~ U~ ~ ~ ~ C~
a,
C)
·
aC 0 ·
~ V ~ ~ O ^
· so ~ 0 ~ ~ oo ~n
~ 0 0 v ~ ~ e ~ ~ ~
~ ~ ~ ~ q~ ~ c: · t:
so 0 ~ ~ ~ ~ ~o
—~ ~ :^ O V ~ —~ 6
V C 0 ·
0
(V ~ ~ X ~ ~ ~ ~ ~ 1 ~ ~
S C) ~ V ~ ~ b0 ~ ~ C
C V 0 V ~ —
~ ~ ~ 0
Ll ~ ~ ~ ~ ~ ~ C
V S ~ ~ 0 ~ O ~ 1
~ C: ~ L4 _~ ~ ~ S ~ ~ ~ X
O ~ :^ ~ 0 ~ ~ V · ~ ~ ~ V
~ t~o 3 ~ :^ ~ ~ ~ b0 ~ ~ ~ ~ — ~ ~
V C ~ te S dV O O ~ ~ ~ ~ ~ ~ O
¢ 133 ~ Z C~ ~ L) ~ ~ P~ ~ ~: 3: E_ C~
231
:^
o
U' o
0 ~
U, V
0
o
v
o . -
o ~o
C)
a'
v
C) ~
CL
~:
· sq
0
0
oo
SJ o
o V
o
Co
V
o
3
· =, _
O
tn ~
~ 0 0
0
O ~
~ =: CL
· ~ ~
C) V
00 ~ ~ V
t0 ~ O
CL
C)
_~ pc,
e 0=
~ ~ 1 0
S~ 0 C
0 ·^
P* ~ ~ ~
0
0
CL ~
~ ~ V CO
so ~
e~ o~
V
= 0
c ~ ~
0 ~ ~
t
c 0 ~
~ - ~ ~
e 0 ~
oo
0
{:
C
C
Q)
C C: ~
0
=
D O
C)
~:
~4
~5
.
C
.
X
C'
Ct
o
OCR for page 232
Sloan6 estimated the demand for medical school places for the years
1936 through 1965 as a function of several independent variables:
GRAD
PRICE
the number of male college graduates;
the price of medical education defined as tuition and
fees less student stipends;
INCOME median income of physicians three years previously;
FORGONE forgone earnings as measured by starting salaries of
male college graduates in general business or by
stipends for Ph.D. students in the biological sciences;
ALT alternative earnings as measured by median incomes of
Ph.D.s in the biological sciences;
APPt-1 the number of applicants to medical school in the
previous year, a measure of perceived probability of
success.
The dependent variable was the number of applicants in year t.
For the postwar years 1948-1965 the model took the following form:
APPt ~ ho + blGRADt + b2PRICEt + b3INCOMEt_3
+ b4FORGONEt-l + b5ALTt-l + b6APPt-1 +
+ et,
where the subscript t represents the year and et represents random
variation in APPt due to random differences among individuals and to
any systematic variation not accounted for by the variables in the
model. The unknown parameters b were estimated by ordinary least
squares regression.*
Sloan found that the number of applicants in a year declined as
the PRICE increased.6 However, the response was relatively small.
With various model specifications, the elasticity of demand with
respect to price ranged from -0.39 to -0.85, implying that a 10
percent increase in the price of medical school (measured, as noted
above, as tuition and fees less stipends) would result in a 3.9
percent to 8. 5 percent decline in the number of applicants. The
largest effect on the number of applicants was exercised by GRAD, the
number of male college graduates, a rough measure of the size of the
pool from which medical students were drawn at that time. As a single
predictor, this variable explains approximately 40 percent of the
*In the analysis of time -series data there is the possibility of
correlation among the error terms et, especially when one of the
explanatory variables is the value of the dependent variable in a
previous period. Such correlation causes the parameter estimates to
be biased and inconsistent. Sloan experimented with a method
suggested by Wallisl4 to remove the effects of the autocorrelated
error term, but the results were unsatisfactory. Therefore, only
ordinary least squares results are reported here.
232
OCR for page 233
variation in the number of applicants. However, as Sloan points out,
this still leaves 60 percent of the variation to be explained by other
factors, including costs and returns.
Physician income three years earlier had a significant positive
effect on the number of applicants. Alternative earnings (ALT),
measured by the earnings of Ph.D.s in the biological sciences, had a
significant negative coefficient; thus, it appears that the
attractiveness of alternatives has an effect on the number of
applicants.
The number of applicants in the previous year had a significant
positive effect on current applicants, suggesting that the hypothesized
"discouragement effect," or, in this case, "encouragement effect," is
a factor. However, the size of the coefficient is very small.
Sloan's analysis, by including measures of the costs of medical
education and physician earnings, implicitly employed the investment
framework discussed above.6 Feldman and Scheffler explicitly
incorporated it in their analysis of the number of applicants to
medical school for the period 1956 to 1966.7 They tested the
hypothesis that the number of applicants in a given year t was a
function of the average rate of return for that year and to the
perceived probability of acceptance with the model:
APPt ~ be + blRATEt ~ b2ENROLLt + eT
where the probability of acceptance is measured by current medical
school enrollments (ENROLL). The authors report that their best
results were obtained when they included the rate of return with a
three-year lag, and with all variables included in logarithmic for=.
This specification yielded elasticities of 0.922 and 0.882,
respectively, for RATE and ENROLL. That is, a 10 percent increase in
the rate of return would increase the number of applicants by just
over 9 percent, while a 10 percent increase in enrollments would
increase the number of applicants by just under 9 percent. Their
model predicted 36, 638 applicants for the year 1973; the actual number
of applicants that year was 40,506.7
Rather a different approach to analyzing the demand for medical
education was taken by Lee and Carlson.8 Their unit of observation
was the medical school, and the measure of demand was the change in
the number of applicants f or two periods: 197 7/78 and 1978/79.
Observations for the two periods were pooled, with dummy variables
included to control for period-specific factors and for public
schools, where the number of applicants is historically lower than for
private schools. The change in the number of applicants was
hypothesized to be a function of DTIJITION, the school ' s change in
tuition charges, and DCLASS, the change in the size of the entering
class. The authors estimated the parameters of the model
DAPP = boDTuITIoN + blDCLASS + b21978 + b31979
+ b4PUBLICl978 + bSPUBLIC1979 + e
for six income/minority status subgroups.
233
OCR for page 234
Results differed among the subgroups in an interesting way. The
number of nonminority applicants decreased with an increase in
tuition, irrespective of whether the students were in the low (under
$10,000), moderate ($10,000 to $19,999), or high (over $2O,000) income
group. The magnitude of the effect was greatest for applicants in the
high-income group and least for applicants in the low-income group.
-
tor minorities, on the other hand, the estimated coefficients on
VTUITION are close to zero and not statistically significant,
suggesting that irrespective of income group, an increase in tuition
would have no discernible effect on the number of minority
applicants. AS the authors suggest, availability of financial aid
makes this result plausible; nominal tuition charges are not actually
paid by most minority students.
To the extent it is the case that financial aid covers tuition,
DIUITIO1i would not measure the costs actually faced by these
students. The small effect of OTUITION on low-income, nonminority
applicants may be attributable to the same cause; these students also
may expect to receive need-based aid.
The variable DCLASS, the change in the size of the entering class ,
is generally positive but not statistically significant except for one
group, minority applicants in the high-income category. This result
suggests that this group is more responsive to the probability of
acceptance than are minority applicants from lower-income families.
Summary and Discussion of Findings
Two principal findings emerge from this research. First, the
average rate of return to the investment in medical education has been
very high; it was probably rising at least from the early 1960s
through the mid-1970s, when college graduates are the comparison
group. Second, the demand for medical education, as measured by the
number of medical school applicants, is positively influenced by
economic returns: as returns rise, the number of applicants rises.
Several points should be made about these findings. While the
average economic returns to medical education are very good in
comparison with those to other educational investments, there is a
great deal of individual variation in education costs and in
physicians' lifetime earnings. Several studies have computed net
present values based on several different levels of education cost,
but none has examined differences in earnings except those resulting
from specialty choice (discussed below). Other sources of earnings
variation that have been explored are urban versus rural locationl5;
seXl6,l7; and mode of practice. 17 Based on her analysis of pooled
data from the 1973, 197 4, and 1975 surveys of physicians by the I,
Langwelll5 concluded that , for most specialties, net present values
were lowest in rural counties, somewhat higher in semirural counties,
234
OCR for page 235
and highest in urban counties.* Thus, the returns to medical
education vary depending upon the doctor's choice of location.
Wolinsky and Marder's recent analysis of AMA survey data from 1980
showed that, controlling for specialty, hours of work, years of
experience, and mode of practice, female physicians earned slightly
less than did their male counterparts. 17 Langwell, replicating an
earlier study by Kehrerl8, had similar f indings, although she did
not control for mode of practice in as much detail.l6 Finally,
Wolinsky and Harder found annual earnings to be significantly lower
for physicians in solo fee-for-service practice than for those in
fee-for-service group practice. Physicians in prepaid group practices
also had lower earnings, but not at a level of statistical
significance.l7
These results suggest the need for more disaggregated analysis of
the returns to medical education in order to account for physician
characteristics and for different career paths. Such analyses could
indicate that, as a consequence of relatively high education costs and
relatively low lifetime earnings, some physicians experience low or
even negative economic returns. Such findings might serve to further
inform discussion on education financing policies.
While average returns were rising until the mid-1970s, it appears
that more recently they have been falling, as education costs have
been rising more rapidly than physicians' practice earnings. From
1977 through 1981, median tuition and fees both in public and private
medical schools more than doubled, while mean incomes of practicing
physicians rose by about 150 percent.** The research findings on the
demand for medical education suggest that these trends would be
accompanied by a decline in the number of applicants to medical
school. This has occurred (Table 3~. It would be useful to replicate
earlier demand studies with more recent data in order to improve our
understanding of the relationship between economic returns, other
factors, and the demand for medical education.
Economic Inf luences on Specialty Choice
The analysis of specialty choice also may employ the investment
framework. A medical School graduate may obtain a license to practice
after only one year of residency, or may continue residency training
in a specialty for up to five years. While specialization leach to
*The exception to this finding was the group Langwell called "other
primary care specialties, " which included internal medicine,
pediatrics, and obstetrics-gynecology. Their discounted lifetime
earnings were lower in semirural and urban count lee than in rural
counties. Note that these earnings are ad jus ted for hours of work.
**For the sake of comparison, the Consumer Price Index rose 148
percent over this period.
235
OCR for page 236
TABLE 3 Trends in Medical School Applicants, Tuition and Fees, and
Physician Incomes, 1977-1982
Median Medical
Applicants to School Tuition Mean Net Incomes
Medical School and Fees of Practicing
Year (Number) Public Private Physicians
,
1977 40~557 61~200 $4~150 ¢61~200
1978 36 ~ 636 1 ~ 473 5 ~ 994 65 ~ 500
L979 36~141 1~750 6~725 78~400
1980 36 ~ 100 2 ~ 079 7 ~ 910 80 ~ 900
1981 36'727 2 ~ 458 9,337 93 ~ 000
1982 35~730 2~916 10~650 105~000
aEstimated from mean net incomes during the first two quarters of 1982.
SOURCES: Association of American Medical Colleges, Office of Student
Finances.
Association of American Medical Colleges, Office of Public
Relations.
David L. Goldfarb, ea., Profile of Medical Practice 1981.
Chicago, AMA, 1981.
AMA, Socioeconomic Monitoring System Report, Vol. 1, No . 5,
June 1982 and No . 6, July 1982.
higher earnings after the residency, the physician must forgo
substantial earnings during the training period. Thus, the economic
value of the investment in specialty training is calculated as the
present value of the difference in lifetime earnings between the
specialist and general practitioner (or between one type of specialist
and another), leas the costs of training in the form of forgone earnings.
~ There are large differences in average practice income among
the specialties. In 1981, according to data from the AMA, mean net
income from practice ranged from about $65,000 for pediatrics to almost
$120,000 for surgery and anesthesiology (Table 4~. The rate of growth
in income also differs. From 1976 through 1981, incomes of anesthesi-
ologists almost doubled; incomes of internists and pediatricians rose by
236
OCR for page 237
or to
~ o ~ o
~ 0 ~ ~D
u~ ~ ~ ~
- ~
~ ~o mo
~S: ~ O ~ iD
~ ~ r~ O
C?
~ oo u~ ~ - ~
1 ~ _~
r~ g
~:
0
c)
e _
~
o
~ oo
-
~ C~
_ t_
C) o~
C~
~ r~
C}
U ~4
C~
0 ~
'_
D
C ~
_
C ~
_
V
0 0 e
c 0
_ ~
0 `'
0 ~ C)
~ ~ ~ C
D _ _
~ ~ O
Z ~ ~
C
~4 ~ ~ C
CL
~ C)
D O
~ U)
0 0 00
cr' ~ ~ 0
0m ~ O
0`
O 0\
~4
\0 ~D
r~
~ O
O O
O
~ O
0`
0`
~ O ~ O
C~ O _ O
~ ~ ~ ~1
O ~ L~
~ ~ 0\
0` 0 ~ O
O ~ O
O ~ ~ ~1
O ~ O
~o~ m
_ _
~o
~o 0 ~ 0 ~ 0 0 0 ~ 0
_ 0 U~ ~ ~ 0 ~ ~
O
~ ~ ~ ~ 0 ~ 0 ~ C~ ~
- ~ ~ _ _ _
~D O ~ O ~ O _ O O O
0 ~ 0 U~ 0 ~ 0
~o 0 0 ~ ~o I
U~ ~ ~ o, ~ 0
~ ~D ~ U~ ~ ~ ~ 0 C~ ~
+~ _ _ _ _
~ 0 o~ 0 ~ 0 ~ 0
r~ 0 ~ 0 ~o 0 ~ 0
~ e~ _ ~ ~ ~ r~ O
_ ~ ~ ~ ~ ~ ~ ~
_ ~ ~ ~ U~ _ ~ ~
+_ _ _ _
~D O
~ 0
U~
e~
0 U~
0
~ 0 ~ 0
~r 0
~ ~ ~ 0
U~ ~ ~ ~
_
C)
_ ~
V ~ V
V ~ ~ 0 ~ 0 ~
~o C 0 C 0 C 0
=_ ~ _
~ V C)
:~_
0 ~
C :^ 0 :^ 0
s ~ _ ~ V ~ ~
C ~ ~ C ~ C
_ - 4 _ _1
0
C ~ ~ ~ ~
~ ~ c ~ e ~ c
—^ C ~ 15 C :~= C
C
Z x: ~ Z x: ~OZ ~:
C
~ C
C, ~ U:
o U~ o
O ~ O
~D _ ~ O
0 ~ ~
_ _
~ 0 ~0
t_ O ~D O
D
0
0\ ~ ~
_ _
O
~ O
O
02
-
O
O
O
-
0 0 0
O ~ O
0 ~ `0
- 0 0 0
C~ ~ ~ ~
_ _
~ O ~ O
00 ~ O
~D ~ O ~
C~ 0` ~ `0
_
O ~ O
_ O ~ O
O ~D
O ~
_ _
~ O
`0 0
O
-
-
O
O
0`
0`
-
0 0
O
o'
U~
o'
O
\0 0
U~ _
_ ~
`0 0 0 0
0\ 0 ~ O
o,
0 ~ ~
_ _
O `0 0 ~ O ~D O ~ O
O- ~ O C~ O ~ 0 0 0
C~ ~ O C~ 0 0 ~ ~ ~
~o ~ ~ ~ C~ ~ ~ ~ ~ O
r~ ~ ~ ~ e~ ~o ~t ~ _. ~D
~ ~ ~ ~ _ C
C)
C O
0
C)
0 ~
: - O
S ~
C
-
O
~
~ C
~
C
x:
237
CL
~o
o 0 ~
C o
o ~ ~
~4
~ V
C
0 ~
so ~ o
C
C
o
0
V ~ C
aD C
~ :,
"Z
o
C)
o.
C o
_
v
0 ~
:^ o
V
CL C
o ~
C
D C
~ :~ 09
V Z ~:
:^
CJ
~V
ca
~4
C)
0 ~
:^ O
S£L y
tlO —
O
O
O ~
C
0 ~
Y ~ ~
V
0 Z Z:
C
~i
O
0
0
0
o
O
0
X
~d
ID
0 ~
40 0
C D
0
C ~
~ ~ ~0
_ 0 t0
C
CL tlO
X ~
C
0
0
O
0
C V
-4 ~
V 0
0 ~ 0
:^
aC =
O C
~d
:'
-4
C O C
C,, _
0
~ 0 ~
~ 0 ~
so o oo
~ ~ -
8
o
C
o
tn
-
-
-
.
::
S
e
o
:,
-
L.
~ 0
O _
os C:
a'
C
C~ ~
0
v
0 ~
C~ o'
c
:^
CJ ~
0 C
:^
~:
O.;
0
0
L'
C 0
~ 0
C~
C
0
·
~:
· ~
C' C
~ :^
0
_
eo
0
-
_1
C)
-
:~
~4
o
~4
o
L'
.
.
a
8
·—
U,
C~
~:
o
U,
~0
-
:^
C
~:
cn
Cd
o
CO
CI:
v
:^
U:
o
C
o
-
0
-
-
0
-
C
-
0
:^
c~
o
a,
c~
x
:
e~
c
u~
L.
D
z
-
o
:.
o
~:
U,
~:
U,
OCR for page 238
roughly 40 percent. Did the higher growth rates attract more
physicians into some specialties? Comparing the growth in the numbers
in the speciality to the growth in income does not suggest that it
did. The relationship, portrayed graphically in Figure 1, clearly is
not linear. Two of the fastest-growing specialties, internal medicine
and pediatrics, had the smallest growth in income. Obstetrics and
gynecology, with above-average income growth, had below-average growth
in numbers. In fact, the only specialty (in thin very aggregated
grouping ~ to have higher-than-average growth in incomes and in numbers
over this period was anesthesiology.
The most recent estimates of the returns to different specialties
in comparison with those to general practice are by Dresch.l2
Adjusting for hours worked and for age-specific mortality, his
estimates suggest that the discounted returns to neurosurgery and
orthopedic surgery (5 post-M.D. years), to obstetrics and gynecology
(4 post-M.D. years), and to internal medicine (3 post-M.D. years) were
in excess of $100,000. In contrast , the investment in specialty
training in pediatrics, psychiatry, or allergy/dermatology had a
negative return, with discounted lifetime earnings, on average,
insufficient to offset the costs of training.*
Sloan's estimates for 1965 showed internal medicine and obstetrics
and gynecology, ig addition to pediatrics and psychiatry, to have
negative returns.
Economic Returns and Specialty Choice
Sloan9 studied the influence of economic returns on specialty
choice, using ordinary least squares regression to estimate the
parameters of the following model:
Residentsit ~ he + blLifetime Earningsit_
+ Residenciesit + b2 FMGS + en'.
In this model, i indexes the specialty and t the year. Thus, the
number of residents in a specialty in year t were hypothesized to
depend upon the value of lifetime earnings in the specialty in the
previous year, on the number of residency positions in the current
year t, and on the number of foreign medical graduates enrolled in
U.S. residency programs. The results suggested that the number of
residency positions had the largest effect; lifetime earnings were
positive and statistically significant, but their effect was quite
small. Thus, a large increase in lifetime earnings would be required
to alter substantially the numbers of physicians entering a given
specialty.
*These findings are sensitive to the method of adjusting for hours
worked. When earnings were assumed to be linearly related to hours,
only pediatrics was unprofitable.
238
OCR for page 239
FIGURE ~ Sates of Growth In Net Income from Practice and Abets of
Pbysiclans, Selected Specialtles, 1976-1981.
z
c 50 _
40 _
30
z
at
-
c
A:
it
it
lo'
a:
1
1
1
tNT£~N L HE D'CtNE
· P£D'&TR1CS
20 t
I
10—
1
ol
o
AN£ST - ES lOLO~ Y
TALL S.ECIALTIES
PS~C - tAtRv —
R^D'OLOG~ ~
1 1 1 ' 1
1 0 20 30 40 50
I · · O8-G~N
S~RGER Y
· GENER&L PR^CT.CE
I ~ ~ ,, 1
90 1 00
60 70 80
PERCENT INCREAS£ 1N NET INCOM£ FROM PRACTICE
SOURCES: Goldfarb, D. L., ea., Profile of tledica1 Practice 1981.
Chicago, "erican Iledical Association, 1981.
Bedise C.. M. ant Danais, D. G. Physician Charecteristice and
DIstribution in the U.S., 1981 ed. Division of Survey and Data
Resources, "erican Hedical ~isociatlon. January 1983.
Excerpts froe the AMA Physiciao ~haterfile, Di~rision of Survey and
Da ta R~ sources, ~, 3anuary 19B3.
~, SMS ~rt, Vol . 1, Number 5, June 1962.
239
OCR for page 240
A very different approach was used by Medley in analyzing
specialty choice; however, his conclusions were similar.l9.20 Data
for this study came from the AAMC Longitudinal Study of Medical
Students of the Class of 1960, which included personality, value,
background, ability, and preference data collected at intervals
between their entry to medical school in 1956 and the final follows
in 196S. Data from the AMA, including current specialty and history
of postgraduate training, were added from their 1972 physician
masterfile. Hadley estimated the probability of selecting one of five
specialties (general practice, medicine, surgery, other primary
specialties, and other secondary specialties ~ as a function of
characteristics of the specialty (income, length of training,
prestige ), and characteristics of the individual, medical school, and
internship hospital. Linear probability functions (linear regressions
with a dichotomous dependent variable taking the value 1 if the given
specialty is selected and O otherwise ~ were estimated for each of the
f ive specialty groups .
Relative earnings were found to have a positive and statistically
signif leant ef feet on the probability of choosing medicine; for the
other specialties, this variable was not significant, had a negative
sign, or both. Although length of training, a measure of the cost of
training, ~ would be expected to have a negative influence on the
probability of selecting a specialty, in fact, it had a positive and
signif leant effect. This finding, which implies that the longer the
training in a specialty, the greater the probability of its selection,
is not consistent with the hypothesis that specialty choice is
inf luenced by the rate of return.
The results of these two studies have led most students of the
subject to conclude that economic costs and returns have a negligible,
e f feet on specialty choice . However, a recent study by Hay may reopen
the issue for examination.21
Hay analyzed the relationship between earnings, physician
characteristics, and specialty choice using data f ram the 1970 ALA
physician survey. The model assumed that physicians choose among
specialties on the basis on expected earnings, given their medical
school background and individual characteristics (e.g., family
socioeconomic status ~ . Correcting for selection bias, stemming from
the existence of unobserved characteristics of the physician
influencing both earnings and choice of specialty, Hay found earnings
to exert a significant positive influence on the probability of
choosing a given specialty. This analysis employed rather old data
and pioneering statistical methods that have yet to be widely used,
and thus the f indings should be viewed as preliminary. Further work
on specialty choice, employing discrete choice models, is needed to
evaluate the inf luence of economic returns.
240
OCR for page 241
REFERENCES
1. Fein, R. and Weber, G. I. Financing Medical Education. New
York: McGraw-Hill Book Co., 1971.
2. Reinhardt, U. E. Financing of medical education. Health
Communications and Informatics 5:287-303, 1979.
3. Becker, G. S. Human Capital, 2nd ed. New York: Columbia
University Press, 1975.
4. Hirschleifer, J. On the theory of optimal investment decision.
Journal of Political Economy 66:329-352, l9S8.
5. Bureau of the Census . Money income of households, families, and
persons in the United States: 1980. Current Population Reports,
Series P-60, No. 132. Washington, D.C.: U. S. Government Printing
Of f ice , 1982.
6. Sloan, F. A. Economic Models of Physician Supply. Unpublished
doctoral dissertation. Harvard University, 1968.
7. Feldman, R. and Scheffler, R. The supply of medical school
applicants and the rate of return to training. Quarterly Review
of Economics and Business 18: 91-9S, 1978.
8. Lee, R. H. and Carlson C. The Effects of Reducing Federal Aid
to Undergraduate Medical Education. Working Paper 1439-1.
Washington, D.C.: The Urban Institute, June 1981.
9. Sloan, F. A. Lifetime earnings and physicians' choice of
specialty. Industrial and Labor Relations Review 24:47-56, 1970.
10. Lindsay, C. M. Real returns to medical education: A comment.
Journal of Human Resources 8:331-348, 1973.
11. Mennemeyer, S. T. Really great returns to medical education?
Journal of Human Resources B:76-90, 1973.
12. Dreach, S. P. Marginal wage rates, hours of work, and returns to
physician training and specialization. In Issues in Physician
Reimbursement. BCFA Publ. No. 03121, Washington, D.C.: U.S.
l
Department of Health and Human Services, Health Care Financing
~ ministration, 1981.
13. Sloan, F. A. and Lindsay C. M. Real returns to medical
education: comment and reply. Journal of Human Resources
11: 118-130, 1976.
Wa~lis, K. F. Lagged dependent variables and serially correlated
errors: A reappraisal of three-pass least squarer. Review of
Economics and Statistics 49:555-567, 1967.
241
OCR for page 242
15. Langwell, K. M. Real returns to career decisions: The
physician's specialty and ~ ocation choices. Journal of Human
Resources 15: 278-286, 1980.
16. Langwell, K. Factors affecting the incomes of men and women
physicians. Journal of Human Resources 17: 261-275, 1982.
17. Wolinaky, F. D. and Harder, W. D. The organization of medical
practice and primary care physician income.
Public Health 73: 379-383, 1983.
American Journal of
18. Kehrer, B. H. Factors af fecting the incomes of men and women
physicians. Journal of Human Resources 11: 526-545, 1976.
19 . Hadley, J. An empirical model of medical Specialty choice
Inquiry 14: 384-401, 1977 .
20 . lIadley, J. Models of Physicians ' Specialty and Location
Decisions. Technical Paper No. 6.
U. S. Public Health Service,
National Center for Health Services Research, December 1975.
21. Hay, J. "Selectivity Bias in a Simultaneous Logit-OLS Model :
Physician Specialty Choice and Specialty Income, " presented at
Harvard Conference on Econometric Modeling, April 1981.
242
Representative terms from entire chapter:
college graduates
