Click for next page ( 225


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 224
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 SL 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 224
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 224
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 PD'&TR1CS 20 t I 10 1 ol o ANST - 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 224
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 224
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 224
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