PRICE SETTING IN THE MARKET FOR PHYSICIANS' SERVICES A REVIEW OF THE LITERATURE ¢ by David A. Juba This report was made pursuant to contract no. SSA-600-76- 0146 between the ‘Health Care Financing Administration, Department of Health, Education, and Welfare, and Pennsylvania Blue Shield. The author is affiliated with Carnegie-Mellon University, School of Urban and Public Affairs, Pittsburgh, Pennsylvania. The Project Officer for this report was Jon Gabel, Office of Research, Demonstrations, and Statistics. ; £73 LS% 4g Pui + iF : : = : a NLA A ie RA & 3 1 R328 eS ABSTRACT J £2 Pui c. ? This study reviews recent health services research concerning supply, demand and price in the physicians' services market. Data are presented to illustrate the growth in prices, expenditures and utilization of physicians’ services. Increases in the levels of such expenditures paid by third parties are documented. Basic competitive and monopolistic models of the physicians' services market are developed and critiqued. The physician's ability to function as the patient's "agent" is discussed, as are the principal behavioral objec- tives often attributed to physicians. The demand for physicians' services and the supply of such services are examined separately. Empirical evidence is presented which shows how each is related to market price. The elasticity of demand with respect to time (acting as a price) is also discussed along with the physician's purported ability to generate demand for his own services. Evidence of "backward bending'" supply curves is presented. Price equations are developed which serve to relate prices to various market conditions. Empirical evidence suggests that revenues per visit are positively related to the extent of insurance coverage. Evidence linking prices and the local supply of physicians is still inconclusive. The rela- tionship between practice organizational form and prices is not clear, although some evidence suggests that revenue-sharing physicians' groups charge somewhat lower prices. The study also discusses the relationships between physicians' prices and the quality of their services. wg fe ACKNOWLEDGMENTS The author wishes to acknowledge the guidance and tutelage provided by Professor Judith R. Lave during preparation of this report. An earlier version of this paper was reviewed by Gene Markel, Jon Gabel and Judith Lave. The author is grateful for the valuable comments and suggestions generated by these reviewers. -iii- ABSTRACT TABLE OF CONTENTS ACKNOWLEDGMENTS I. 11. 111. IV. INTRODUCTION BACKGROUND DATA A. B. Growth of Expenditures, Prices and Utilization Physicians' Services and Third Party Payments Price Controls and the Market for Physicians’ Services The CPI as an Index of Fee Inflation Variation in Physicians' Fees Growth in the Supply of Physicians Summary MODELS OF THE PHYSICIANS' SERVICES MARKET A. B. The Competitive Model Monopolistic Models Critique of Competitive and Monopolistic Models Objectives of Physicians Critique of Models of Physicians' Pricing Objectives DEMAND FOR PHYSICIANS' SERVICES Health Status and the Demand for Physicians’ Services -—jv- iii 10 13 13 17 21 23 23 26 27 33 40 43 44 TABLE OF CONTENTS (CONTINUED) B. Price and the Demand for Physicians' Services C. The Influence of Physician Supply Upon the Demand for Services D. Income and the Demand for Physicians' Services V. SUPPLY IN THE PHYSICIAN SERVICES INDUSTRY A. Determinants of the Aggregate Supply of Physicians B. Supply Behavior of Individual Physicians VI. ECONOMETRIC STUDIES OF PRICES IN THE PHYSICIAN SERVICES MARKET A. Price Equations Derived from Competitive Assumptions B. Price Equations Derived from Monopoly Conditions C. Identification and the Price Equation D. Empirical Findings E. General Criticisms of Econometric Pricing Models VII. CONCLUSIONS APPENDIX REFERENCES -- Page 46 48 51 54 54 57 62 62 63 65 66 83 85 87 89 I. INTRODUCTION Expenditures and price levels have increased rapidly in the medical care sector of the economy over the past decade. Partly in response to these increases, political pressures have been mounting in support of some form of comprehensive National Health Insurance. Health policy makers and planners realize, however, that countervailing forces exist which make health planning a difficult process. In particular, policies which are designed to address the problem of expenditure and price inflation may not be compatible with goals of assured access to medical care for various population groups. It is apparent that rational health planning requires an understanding of the operation of the various components of the medical care sector. One such component is the physicians' services sector. The importance of the physician in contemporary medical care is often emphasized in health services research. In addition to providing direct medical service, the physician also often acts as the patient's agent. In this role, the physi- cian translates requests and/or symptoms presented by the patient into a "demand" for a particular set of medical services which includes hospital and physicians’ services. This report is an exploration of the economic literature pertaining to the physicians' services market and especially to price setting behavior within that market. Specifically, this report contains: (a) background information on the growth of prices, expenditures and utilization of physicians' services, as well as informa- tion on other trends in this market; wl (b) a description and critique of several prototypical models of the physicians' services market; (c) a summary of econometric analyses of demand and supply in the market for physicians' services; (d) a review of econometric analysis of the prices of physicians’ services. The information summarized in this review should provide valuable back- ground for planners engaged in health services policy design and evaluation as well as for researchers whose task it is to establish an agenda for future research with respect to the market for physicians' services. II. BACKGROUND DATA A. GROWTH OF EXPENDITURES, PRICES AND UTILIZATION Annual increases in expenditures for physicians' services have contributed substantially to the growth in expenditures on personal health care over the past two decades. Gibson and Mueller (1977) reported that, between fiscal years 1960 and 1976, the percent of GNP attributable to personal health expenditures increased from 5.2 to 8.6 percent. Over the same period, expenditures for personal health care increased from $25.9 billion to $139.3 billion per year. Of the total expenditures for health care in 1976, 18.9 percent ($26.4 billion) was spent on physicians' services. Fiscal year percent increases in both per capita and total expenditures on physicians' services and estimated rates of utilization of such services are presented in Table I. Also presented in Table I are data on the rates of increase for selected components of the Consumer Price Index (CPI). Lave (1978) observed the following with respect to similar data on prices and expenditures for physicians' services. (a) The annual percentage change in the physician fee component of the CPI has been higher than that for prices in general (except during the Economic Stabilization Program, 1971- 1974). (b) The rates of increase in total and per capita expenditures are very similar; and thus expenditure increases are more likely the result of increases in prices and utilization rates than of increases in population. TABLE I. FISCAL YEAR PERCENT INCREASES IN PRICES, EXPENDITURES, AND UTILIZATION OF PHYSICIANS' SERVICES Expenditures on Utilization of Prices Physicians' Services Physicians' Services CPI CP1 Fiscal Medical Care Physicians Year CPI Component Fees Total Per Capita Total Per Capita 1965- 2.2% 2.9% 3.9% 5.5% 4.27 1.5% 0.3% 1966 1967 3.0 6.5 7.4 9.8 8.6 2.2 1.1 1968 3.3 6.4 6.1 10.2 9.1 3.9 2.8 1969 4.8 6.5 6.1 10.3 9.3 4.0 3.0 1970 5.9 6.4 7.2 13.5 12.4 5.9 4.8 1971 5.2 6.9 7:5 12.3 11.2 4.5 3.4 1972 3.6 4.7 5.2 9.5 8.4 4.1 3.1 1973 4.0 3.1 2.6 8.9 8.0 6.1 5.7 1974 9.0 5.7 5.0 9.7 8.9 4.5 3.7 1975 11.0 12,5 12.8 16.1 15.2 2.9 2:1 1976 7.1 10.2 11.4 14.9 14.0 3.1 2.3 All data are from (or calculated from) Marjorie S. Mueller and Robert M. Gibson, "National Health Expenditures, Fiscal Year 1976," Social Security Bulletin, April, 1977. (c) The increases in physician expenditures per capita are some- what higher than increases in the fee index, an indicator that actual utilization of physician services has increased. As the foregoing suggests, rising expenditures on physicians' services may result from fee inflation, increases in total and/or per capita utiliza- tion, or some combination of these. While reliable data on the nationwide rates of utilization of physicians’ services for the 1965-76 period are not readily available, it is possible to derive rates of change in utilization from the price and expenditure data in Table I (see Appendix for details). The derived rates of change of utilization (shown in Table I) reveal that aggregate utilization of physicians' services may have been increasing as much as three to six percent per year between fiscal years 1965 and 1976. Per capita increases over the same period were somewhat lower. Dyckman (1978, p. iv) suggested that the CPI physicians' fee index may underestimate the true rate of fee inflation. If so, the actual rates of increase in the utilization of physicians' services may not have been as great as the estimates presented in Table I. The CPI may underestimate physicians' fee increases because it does not adjust for changes in the frequency with which physicians discount the fees they charge poorer patients. As Dyckman suggests, physicians (for a variety of reasons) may no longer base their charges on the patient's income level. This decline in discount pricing means that physicians' average (or transaction) fees may have grown faster than their customary fees which form the basis of the CPI fee estimates. Dyckman believes (p. v) that "this source of bias has resulted in physicians’ fees increasing an estimated 20 percent more rapidly per year during the 1950-76 period than the increase reflected in the CPI data." If so, the utilization increases presented in Table I may overestimate the true rates of change in utilization of physicians' services. Since the product of the physician firm is a service, it may be more appropriate to compare temporal changes in the physicians' fee index to changes in the services price index than to changes in the overall CPI. Between 1960 and 1976, the annual rate of increase in physicians' fees generally exceeded increases in both the overall CPI and the services price index. Figure 1 shows that the increase in fees was more closely matched by the "increase in services prices than by the increase of the overall CPI. Hence part of the increase in physicians' fees may be the result of market pressures that affected all classes of services. Theordore and Warner (1976) showed that labor costs comprised the largest expense item for office-based practices. Wages accounted for 45 percent of total expenses in 1968 and 52 percent in 1970. Given the rising salaries of clerical workers and other ancillary staff members it is not surprising that prices in the physicians' services industry (and other service industries) should be rising in part to offset increasing labor input costs. B. PHYSICIANS' SERVICES AND THIRD PARTY PAYMENTS Since 1960, the extent of health insurance coverage in the United States has grown tremendously. The public medical insurance programs (Medicare and Medicaid) now provide coverage for approximately 40 million aged and low income individuals (Dyckman, 1978, p. 14). Furthermore, in fiscal year 1975, Medicare accounted for 20 percent and Medicaid payments accounted for 5 percent of total national physician expenditures. Of the remainder, out-of-pocket payments by patients represented 35 percent and FIGURE 1. RATIO OF AVERAGE ANNUAL PERCENTAGE INCREASE IN PHYSICIANS' FEES TO AVERAGE ANNUAL PERCENTAGE INCREASES IN: (A) ALL ITEMS (LESS MEDICAL CARE) AND (B) ALL SERVICES COMPONENTS OF THE CPI 1961-1976 Ratio at 0 “ ’ -t dH 00 PE) [I] -f ¢ . 8 ’ ' . ' o . , - hn . Po 2 < . a 4 v q \ ’ 4 5 3 ’ ’ J Lf : Moi » 4 4 - . ’ » . . » . ¢ el ’ ’ . - : 2s \ ¥ . A A & a . 7 . 7 sg ¢ + [J y ¢ bo Ee v A ’ ¢ 2 J & = bN » - \ 8 . \ . - Yoo” $ B =] . i \ a A / 1.6 V . ’ od ’ - — \ A o el | \ ’ 0 - “eo AS v boa RELEASE Trrr yori rr rr rriy Tere TT T77 1961 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 Year Physicians' Fees Relative —-=--=---- to All Items (less Medical Care) Physicians' Fees Relative to All Services Source: Adapted from similar table in Theodore and Warner (1976). Data from Social Security Bulletin; April 1977. private insurance payments covered 40 percent of expenditures for physicians’ services (Gabel et al. 1976). Within the market for private health insurance, enrollments in major medical plans (which provide comprehensive coverage of medical expenses) have grown faster than enrollments in basic surgical or medical plans. Data presented by Dyckman (1978, p. 15) show that the number of individuals covered by major medical insurance (carried by private insurance companies, Blue Cross-Blue Shield Plans or independent group practice plans) grew from 30-33 million in 1960 to 130-140 million in 1975, an increase of approximately 300 percent. The data in Table II illustrate the impact which the growth of public and private health insurance coverage has had upon the physicians' services market. Over the period 1965-1970, the percent of total expenditures for physician services paid by third parties (public and private) jumped from 36.8 percent to 55.1 percent. In this same period total expenditures for physician services increased by almost 60 percent. The rapid growth of health insurance coverage may have exacerbated increases in fees and expenditures in several ways. First, insurance coverage dramatically reduced the proportion of expenditures paid directly (out-of-pocket) by consumers. Depending upon the service in question and the deductible provisions of the particular insurance plan, the effective money price of physician services faced by the consumer was substantially below the total price of the services. This placed upward pressure on the demand for medical care which (assuming rising supply curves) resulted in price increases. Second, physicians' billed charges may be increasing because a growing number of insurers now reimburse physicians under a UCR (usual, customary, and reasonable) plan. Unlike fee schedules which provide the physician TABLE II. DISTRIBUTION OF PHYSICIAN EXPENDITURES BY SOURCE OF FUNDS, FISCAL YEARS, 1969-76 Aggregate Physician Expenditures (in Millions) Percentage Distribution of Payments | Direct Total* Direct Total® Fiscal Consumer Insurance Private Consumer Insurance Private Year Total Payments & Public Insurance Public | Total Payments & Public Insurance Public 1950 |$ 2,689 $ 2,279 $ 410 $$ 270 $ 133 | 100.0% 84.87% 15.2% 10.0% 4.9% 1960 5,580 3,685 1,895 1,524 362 | 100.0 66.0 34.0 27.3 6.5 1965 8,405 5,315 3,090 2,554 527 | 100.0 63.2 36.8 30.4 6.3 1967 9,738 5,415 4,323 2,898 1,415 | 100.0 55.6 44.4 29.8 14.5 1970 13,443 6,034 7,409 4,468 2,931 | 100.0 44.9 55.1 33.2 21.8 1972 16,527 7,113 9,414 5,754 3,649 | 100.0 43.0 57.0 34.8 22.1 1973 17,995 7,290 10,705 6,559 4,134 | 100.0 40.5 59.4 36.4 23.0 1974 19,742 7,877 11,865 7,192 4,659 | 100.0 39.9 60.1 36.4 23.6 1975 22,925 8,946 13,979 8,257 5,708 | 100.0 39.0 61.0 36.0 24.9 1976 26,350 10,198 16,152 9,502 6,632 | 100.0 38.7 61.3 36.1 25.2 *Includes a small amount from other than insurance and public sources, generally less than 0.1 percent of total expenditures. SOURCE: 1960-72: Marjorie Smith Mueller and Robert M. Gibson, "National Health Expenditures, Fiscal Year 1975," Social Security Bulletin,February 1976, Table 7, 1974-76: M. Mueller and R. Gibson, "National Health Expenditures, Fiscal Year 1976," Social Security Bulletin, April 1977, Table 3. This table appeared in Dyckman (1978), p. 17. with a fixed rate for any procedure, the UCR plans reimburse the physician the lower of (a) the physician's own usual fee for the procedure, or (b) the area's "prevailing" fee for the procedure. Under certain conditions the reasonable reimbursement may even exceed usual or prevailing rates. Since physicians have little incentive to set their usual fees at a level below what they believe to be the regional "prevailing" rates, the UCR method of reimbursement may exacerbate fee and expenditure inflation. Expanded insurance coverage also led to expenditure increases in another way. As reported earlier, physicians had previously based their fees partly upon their client's ability to pay. Specifically, poor patients were charged at rates less than the physician's customary fee. The introduc- tion of Medicaid and Medicare (coupled with expanded private health insurance coverage) obviated the need for physicians to charge low income patients a fee below the customary rate. C. PRICE CONTROLS AND THE MARKET FOR PHYSICIANS' SERVICES The data in Table I are indicative of the rapid price inflation which characterized the American economy in the late 1960's and early 1970's. To combat this inflationary trend, the Federal Government instituted a system of wage and price controls which spanned the period August 1971 through April 1974. This system of controls, known as the Economic Stabili- zation Program (ESP), essentially limited annual increases in prices and wages (in the health care sector) to 2.5 and 5.5 percent respectively. Annual growth rates of physicians' incomes, revenues, fees and expenses during the ESP are given in Table III. -10- TABLE III. GROWTH IN INCOMES, REVENUES, FEES AND EXPENSES OF OFFICE-BASED PHYSICIANS Annual Percent Change in: Expense as Mean Net Mean Total CPI Annual Mean Professional Percent of Income Revenue Fees Expenses Total Revenue 1970 — vs - - 38.0% 1971 8.3% 9.6% 6.9% 11.6% 38.6 1972 4.3 5.9 3.1 8.3 39.5 1973 2.8 3.4 3.3 4.2 39.8 1974 7.0 6.5 9.2 5.6 39.5 SOURCE: Dyckman (1978) Tables II-5 and IV-6. Table IV-6 was based on AMA, Profile of Medical Practice series. During the period 1970-74, (which was dominated by the ESP), physicians’ reported expenses grew at an average annual rate of 7.4 percent. This exceeded the growth rates of fees and total revenues for office-based physi- cians over the same period. Note, however, that expenses maintained a rela- tively stable relationship with total revenues. Between 1971 and 1974, expenses were consistently between 39 and 40 percent of total physicians’ revenues. Thus, even though physicians' expenses grew at a rate in excess of the growth rate of total revenues during the ESP, the absolute growth in physicians' expenses was not enough to cause a significant shift in the ratio of expenses to total revenues. As a result, physicians' incomes continued to grow during the ESP. The data in Table III do suggest, however, that the rate of growth of physicians' incomes was depressed during the «lle years 1972 and 1973. The data also indicate that total revenues grew notably faster than fees in 1971 and 1972. This provides some support for the hypothesis that physicians protected their incomes against loss during the ESP by changing the volume and/or mix of services which they provided, perhaps shifting to more remunerative procedures. A study of the California Medicare-Medicaid systems conducted by Holahan et al. (1978) addressed this issue. The study revealed that the charge per standard unit of physicians' services (between 1972 and 1975) increased less than the ESP target of 2.5 percent annually. During the same period, however, the number of standard units of physicians' services per claim (i.e., the intensity of services) increased at an annual rate of two to three percent. This is in keeping with the data in Table I which showed relatively large increases in utilization during the 1971-72 period and smaller estimated increases in the post-ESP era. The evidence cited above does not imply that only California (or other American) physicians may protect their incomes by varying service intensity. Reinhardt (1976) relates that, in 1975, organized medicine in West Germany agreed to limit increases in fees to an average of 2.3 percent as a contri- bution toward cost containment. Despite this, Reinhardt (p. 39) reported that: "the utilization of services increased by almost 10 percent during the year and overall expenditures for physicians' services, assuming that the percent above 1974 expenditures. The increase in services consisted almost wholly of diagnostic procedures." The lesson to be learned from these figures is that controls or limits on fees alone may not be sufficient to halt the growth in expenditures on physicians' services, assuming that the latter is a policy goal. The data also reinforce the notion that the physician is the key figure in controlling the rising costs of medical care. «12- D. THE CPI AS AN INDEX OF FEE INFLATION In an earlier section of this review, the decline in the frequency with which physicians discount fees for the poor was presented as evidence that the physicians' fee component of the CPI may underestimate fee infla- tion. The contention that physicians systematically vary the mix and volume of the services they provide in order to maintain their income levels also has serious implications for the use of various indexes, such as the CPI, as indicators of inflationary trends. If physicians regularly vary the mix and volume of the services they provide, then indexes (like the CPI) which are developed from fixed, base year service weights may give misleading estimates of the rates of price changes. This was one of the major findings reported by Holahan et al. in their 1978 study of the California Medicare- Medicaid system. The physicians' fee component of the CPI may be inaccurate for other reasons. Since the index is calculated from a sample of reported physicians’ customary fees, the index is sensitive to errors in sampling methodology. For example, the sample may not be truly representative of the urban-suburban mix of physicians. In addition, the physicians' fee index does not adjust for changes in quality over time. Hence, some portion of price increases suggested by the index may be due to changes in the quality of the physi- cians' services which are used to establish the fee index. For detailed criticisms of the CPI's medical care indexes and suggested alternatives, see Dyckman (1978) and Barzel (1969). E. VARIATION IN PHYSICIANS' FEES Physicians' fees vary within (and between) market regions as well as across specialty types. Any implications drawn from observed fee variations wl Fo must be tempered by the realization that economists do not agree on an appropriate level of fee variation in this market. Redisch et al. (1977) point out that too little variation in fee levels may indicate collusive behavior on the part of physicians, while large variations in fees are indicative of consumer ignorance and, hence, of imperfections in the market. The latter condition may indicate that physicians face downward sloping demand curves for their own services and may therefore possess monopoly powers relative to their own practices. The data in Table IV provide evidence which suggests that specialists’ fees are (on average) higher than general practitioners' fees for the (nominally) same procedures. The cross-specialty variation in fees is especially significant given the decline in the number of GP's in recent years. Between 1960 and 1975 the total number of patient care physicians has increased 27.2 percent, while the number of full time specialists increased 77.9 percent. Over the same period, the number of general practitioners has decreased 41 percent (Dyckman, 1978, p. 48). It is possible that the higher fees charged by specialists reflect a higher "quality" of service which the specialist may deliver due to his or her additional years of training and experience. An unchecked trend toward specialization may, however, effectively eliminate the consumer's option to choose a (possibly) lower quality of service at a lower price. Hence, a continuing emphasis upon specialization by physicians is likely to exacerbate the level of inflation in the physicians' services market. Differences in fees charged among G.P.'s or specialists may be at least as significant as differences in fees between these groups. Holahan et al.(1978) found much larger differences in charges (for various procedures) within than between physicians' specialty groups. A statistic -14= -C1- TABLE IV, PHYSICIAN-POPULATION RATIOS (1973), GP AND SPECIALISTS FEES (1975), BY COUNTY DEMOGRAPHIC CLASSIFICATION Follow-up Office Visit Appendectomy County Physician=- Demographic Population GP Mean Specialist GP Mean Specialist Classification Ratio Fee Mean Fee Mean Fee Non-metropolitan Counties 0 to 9,999 1/2232 $6.78 $10.08 $222.48 $282.50 10,000 to 24,999 1/2092 7.51 9.21 227.69 262.21 25,000 to 49,999 1/1412 8.07 10.00 232.71 256.91 Greater than 50,000 1/1087 8.95 10.35 235.62 276.74 SMSA's . 50,000 to 499,999 1/822 9.09 11.29 237.60 288.04 500,000 to 999,999 1/724 9.68 12.63 231.92 290.53 1,000,000 to 4,999,999 1/605 10.23 13.45 261.42 321.24 Greater than 5,000,000 1/518 11.37 16.17 347.68 426.51 SOURCE: Cantwell (1977) commonly used to measure within-group (or within-region) variation in price levels is the ratio of the price measure's standard deviation to its mean. Small values of this ratio (known as the coefficient of varia- tion or C.V.) imply that prices within that group or region are relatively concentrated. Coefficients of variation have been determined for physi- cians' fees at both city and state levels. Some examples are presented in Table V. A priori, one might expect a wider variation (larger C.V.) in fees within a state than within a city. A city may closely approximate a single market region for physicians' services, while a state represents a collec- tion of market areas. Hence, measures of fee variation across regions in a state are likely to reflect fee differentials between market areas. As Table V shows, the coefficients of variation within states and within cities primarily fall within the range .10 to .50. Additional estimates of fee variation are provided by Redisch et al, (1978) who reported C.V.'s for routine follow-up hospital visits in New York City to be 0.2 for internists and 0.9 for OB-GYN's. They believed these results (based on 1975 survey data) to be much higher than the coefficients of variation for most other products priced in local, competitive markets. Newhouse and Sloan (1972), for instance, presented examples of C.V.'s for new car and coal prices in 1959. These were in the .017 to .068 range. Using these examples as a benchmark, the physician services market is apparently characterized by marked price dispersion. Cantwell (1976b), on the other hand, suggested that C.V.'s less than 0.5 should be considered "concentrated," in which case the allegation that physicians' prices are dispersed would not be supported by the data in Table V. -16- TABLE V. COEFFICIENTS OF VARIATION FOR PHYSICIANS' FEES IN SELECTED PLACES Specialty | California Georgia Chicago New York City General Practitioner Initial Visit .31 +32 .364 +332 Follow-Up Visit «25 +25 .209 +205 Specialist First Visit 49 74 L424) 261% Appendectomy «19 .25 .148 .115 * Chicago and New York specialist data are for General Surgeon. SOURCE: New York City and Chicago - Newhouse and Sloan (1972) California and Georgia - Cantwell (1977) F. GROWTH IN THE SUPPLY OF PHYSICIANS Recent trends indicate that the supply of practicing physicians will increase significantly during the next decade. Reinhardt (1976) estimated that in 1975 there were 175.7 physicians per 100,000 population in the United States. As his projections in Table VI show, this figure could increase to 236.9 physicians per 100,000 population by 1990. The estimated growth in the number of physicians (excluding foreign medical graduates) over the next several decades is illustrated in Figure 2. ww] To -8T- TABLE VI, ALTERNATIVE PROJECTIONS OF THE SUPPLY OF PHYSICIANS (MDs AND DOs) IN THE UNITED STATES* 1980-1990 1980 1985 1990 Total No. Rate per Total No. Rate per Total No. Rate per FORECAST of 100,000 of 100,000 of 100,000 Physicians Population | Physicians Population Physicians | Population a. Low Forecast 433,600 191.1 494,100 206.5 552,000 220,2 b. Basic Forecast 446,800 196.9 519,100 216.9 593,800 236.9 c. High Forecast 459,900 202.7 544,300 227.4 637,100 254,2 Assumed Percentage of FMGs in Total Supply 1980 1985 1990 a. Low Forecast 22.8% 23.9% 24,87 b. Basic Forecast 25.1% 26.6% 27.6% c. High Forecast 27.27 29.0% 30.0% re———t eee be inononso;oo;o;o Wooo ko oe SS re eee—————————— *The main reason the forecasts differ lies in the assumptions made on the influx of FMGs. assumptions are made: SOURCE: Low estimate: Basic estimate: High estimate: The Supply of Health Manpower: The following Influx of 5,200 per year FMGs during 1970-71, and of 3,800 per year during 1972-80. Influx of 5,200 per year FMGs during 1970-90. Influx of 5,200 per year FMGs during 1970-71 and 6,600 per year during 1972-90. 25, 27 and 28, pp. 48, 53 and 54. This table appeared in Reinhardt (1976). U.S. Department of Health, Education and Welfare, Bureau of Health Resources Development, 1970 Profiles and Projections for 1990 (1974), Tables FIGURE 2. PROJECTED IMPACT OF EXISTING HEALTH MANPOWER LEGISLATION ON THE SUPPLY OF ACTIVE U.S.-TRAINED PHYSICIANS, UNITED STATES, 1970-2010 § 600 230 B g 3 i | E . 550 | ~~ =4220 . TOTAL NUMBER OF ACTIVE rad g i” PHYSICIANS -- 8 = 500 L” “4210 o ” (USE SCALE ON LEFT L” . 2 HAND SIDE) L’ zZ S 450 , 200 B wv ~ wv ! . 3 A ’, . v 7 wn = 350 72 | 180 5 5 pd PHYSICIAN-POPULATION RATIO g § 4 § an 300 (USE SCALE ON RIGHT HAND SIDE) +170 4 & & £ 250 J160 E g : g 200 1 4 A 1 i A 4 150 g 1970 1975 1980 1985 1990 1995 2000 2005 2010 YEAR Source: Reinhardt (1975), page 5. -19- Traditional economic models dictate that such an increase in the supply of physicians ought to have a depressing effect (ceteris paribus) upon price levels. The fee data presented in Table IV do not support this conjecture, however. It is evident from Table IV that general practitioners' and specialists' fees for follow-up office visits and appendectomies are generally positively correlated with regional popula- tion and with the mean physician-population ratio. Thus, in regions where there are more physicians per population, fee levels tend to be higher. In a simple market model, one would expect fees to be negatively related to the physician supply, provided that demand remains stable. The apparent positive relationship between physician density and fees may be explained in several ways. First, physician-population ratios are higher in the densely populated metropolitan areas. It may be that heavily populated areas are also characterized by relatively high wage rates and/or extensive levels of health insurance coverage. Higher wages and insurance coverage levels may act to increase the effective local demand for physi- cians' services. Even if the local supply of physicians' services is high (due to the large physician-population ratio), high levels of effective demand may keep physicians' fees at a high level. Second, although fees and physician-population ratios appear to be positively correlated, travel and waiting times associated with the consumption of physicians’ services may be lower in regions of high physician density. If so, the total costs of physicians' services (time plus money) may be more homo- geneous over regions of varying physician density than is evident from the data in Table IV. Third, the unique position of the physician as agent for his patient in the market for medical services gives the physi- cian the power to affect the demand for his own services. The depressing «Iw effect upon fees of an increase in the local supply of physicians can be offset by the physician's ability to provide each patient with more and/or higher quality services. These issues will be explored in more detail in Section VI. G. SUMMARY The foregoing data are indicative of the dimensions of the problems facing policy makers (and the public) concerned with the financing of medical care. The physicians' services market has experienced increases in price levels and expenditures which have outpaced the rate of inflation in the economy as a whole since 1965. This trend is especially apparent during the years following the termination of wage and price controls in 1974. While such controls had some success in limiting price increases, expenditures for physicians' services still grew at substantial rates during the control period. The past decades have also witnessed the expansion of health insurance coverage to the point where insurance payments now account for over 60 percent of total expenditures for physicians' services. In addition, the UCR method of reimbursement has increasingly replaced the fee schedule as the method by which physicians are reimbursed by third- party payers for their services. Expanded insurance coverage and UCR plans serve to further insulate insured individuals against rising prices and local variations in physicians' fee levels. As was noted, physicians' fees vary considerably within and across specialty types and geographic regions. Fees tend to be higher for specialists than for primary care physicians. In addition, fee levels are higher in regions characterized by high physi- cian-population levels. Recent trends toward increased specialization by physicians may exacerbate fee and expenditure inflation. Also, the 3 Ye correlation between physician supply and fees implies that increasing the regional physician-population ratio may not be an effective means by which to curb rising prices for physicians' services. «Zw III. MODELS OF THE PHYSICIANS' SERVICES MARKET In the previous section, data were presented which illustrated important fiscal trends and relationships within the physicians' services market. In order to study such phenomena systematically, economists often begin by postulating sets of assumptions about the market's structure. These assump- tions implicitly or explicitly describe: (a) the role of price as a rationing device; (b) the mechanism by which prices are sot in the market; (¢) the control over prices possessed by the various actors in the market. In Table VII, some of the contemporary studies of physicians' fees are listed along with the underlying market structure assumed therein. It is apparent that empirical research has relied heavily upon traditional economic precepts with respect to market structure. As will be indicated later, several of these studies reported statistical evidence contradicting traditional economic notions of behavior. In the following sections, brief descriptions of the traditional market forms are presented, followed by a critique of their applicability to the analysis of price setting behavior in the market for physicians' services. Following that is a discussion of various objectives attributed to physicians with respect to their pricing behavior. A. THE COMPETITIVE MODEL The market for physician services is often compared to the so-called -23- -2Z- TABLE VII. EMPIRICAL STUDIES OF PRICE SETTING Study Market Structure Data Type Time Span Specified of Data Newhouse (1970) i) simple monopoly cross-section 1961,1966 ii) competitive equilibrium (cities) Feldstein (1970) competitive equilibrium time series 1948-1966 (static and dynamic) (U.8.) Fuchs and Kramer (1972) competitive equilibrium cross-section 1966 (cities) Steinwald and Sloan (1974) none specified cross-section 1971 (physicians) Kehrer and Knowles (1974) competitive equilibrium cross-section 1971 (physicians) Cantwell (1976) none specified cross-section 1975 (SMSA's) Newhouse and Phelps (1974, 1976) none specified cross-section 1963 (individuals) Sloan (1976) monopolistic firms cross-section- 1967-1970 time series (states) competitive market. Arrow (1973) ascribes the interest in the competi- "ee tive model partly to its "presumed descriptive power" and partly to "its implication for economic efficiency." Baumol (1965) characterized a perfectly competitive industry as one in which there exists: (a) many firms, each of which controls so small a proportion of total output that its addition to or removal from the market has little or no effect on the market price; (b) homogeneity of products -- i.e., all firms must be known by buyers to produce identical products; (c) freedom of entry and exit; (d) independent decision making, with no collusion. Other economists -- e.g., Ferguson (1972) -- suggest that perfect competition also is characterized by: (e) perfect knowledge which (minimally) includes knowledge on the part of consumers, producers and resource owners with respect to prices, wages, costs and other relevant information. Figure 3(a) below illustrates the familiar competitive market equilibrium. Here, prices are determined by the interaction of producers and consumers acting in their own self interest. Once an equilibrium price and quantity are established (e.g., Py» 4.) the market should remain in this stable condition unless conditions of aggregate demand and supply shift. For example, if demand increases to D_, a new price-quantity 2 equilibrium, ®, 4, will be determined. Once market clearing prices are determined, individual firms and consumers react to them in a manner consistent with their objectives. 2 FIGURE 3, COMPETITIVE MARKET EQUILIBRIUM (a) Competitive Market (b) Competitive Firm Price Price Sy MC p Price = ain agen, nS re a inn Sor fs ee - — — Marginal D, | Revenue | | { Quantity Quantity Qa For example, the output decisions of a profit maximizing firm are illustrated in Figure 3(b). Profits will be maximized whenever the cost of producing the marginal unit of output (MC) is equal to the (marginal) revenue realized from the sale of that unit. In the competitive model, marginal revenue is, of course, equivalent to the market clearing price, p. B. MONOPOLISTIC MODELS From the foregoing, it is evident that the competitive paradigm (with its strict assumptions about market structure and price setting behavior) may not be a totally satisfactory model of the physicians' services market. Rather, models which allow physicians some degree of monopoly (i.e., price setting) power may be more realistic. In a market characterized by monopoly power, suppliers set prices in " order to achieve a particular output level. In monopolistic markets, suppliers face their own downward sloping demand curves from which they derive the power to set prices (or quantities of output sold). There are several variants of the basic monopolistic theme in the economic literature. Each corresponds to a different set of assumptions with respect to the degree to which producers are isolated from each other in price competition. In a cartel or collective monopoly model, physicians act in unison and set prices based upon group goals (see Kessel, 1958). Alternatively, physicians may set prices without considering other physicians at all. That is, they may act as simple monopolists (see Sloan, 1976). Finally, physicians may face a downward sloping demand for their own unique product but may still be affected by pricing decisions made by large numbers of similar firms in the market area. Such monopolistically competitive situa- tions have been suggested by Masson and Wu (1974) and by Hadley and Lee (1978). Cc. CRITIQUE OF COMPETITIVE AND MONOPOLISTIC MODELS 1. Special Market Characteristics The physicians' services market has several characteristics which purportedly distinguish it from the competitive paradigm. These character- istics are: (a) product heterogeneity, (b) restrictions on entry, and (c) imperfect knowledge. Although few (if any) markets are characterized by the conditions necessary for perfect competition, the deviations of the physi- cians' services market from competitive requirements often are cited by observers of this industry as especially severe. a. Product Heterogeneity As the product of the physician firm is a service, there are potentially as many types of products manufactured as there are ww physicians. ~During any patient contact, the physician provides a mix of services. Although various physicians provide nominally equivalent products (e.g., first office visits) the content and quality of the mix of services provided may vary significantly across physicians. The simultaneous nature of production and consumption with respect to physicians' services may make technically homogeneous (from the physician's viewpoint) services appear heterogeneous to the consumer. That is, the physicians' product is defined by the outcome it effects as well as by the technical aspects of the treatment. From the consumer's point of view, a set of services which provides symptomatic relief is a much different product than a (technically identical) set of services which does not provide relief. b. Restrictions on Entry In most of the industrialized world, the practice of medicine is a restricted activity. In the U.S., state licensing boards and medical organizations like the American Medical Association monitor entry into the medical industry. (See Kessel, 1958; Rayack, 1971; P. Feldstein, 1977.) Organized medicine restricts entry into the profession via its powers to affect: (1) licensure of physicians; (2) accreditation of medical schools; (3) levels of Federal aid to medical education; (4) acceptable length of training period for physicians; (5) practice restrictions on foreign medical graduates One example of restrictive behavior in this market is the well-documented conflict between the medical establishment and prepaid practice groups. Dyckman (1978, p. 13) summarized this conflict as follows: -28= "Starting with its condemnation of proposals for organized group medical practices of the Committee on the Costs of Medical Care in 1932, organized medicine has, until recently, consistently opposed the formation of prepaid group medical practice. This opposition included expel- ling HMO participating physicians or those seeking to participate from local medical societies (a move that could result in loss of staff privileges in local hospitals), directly preventing them from acquiring or retaining staff privileges in local hospitals and even attempting to deny them a license to practice medicine in the state." Additional detailed descriptions of organized medicine's resistance to prepaid group practices can be found in Kessel (1958), Rayack (1971) and P. Feldstein (1977). c. Imperfect Knowledge The difference between the patient's and physician's know- ledge with respect to various aspects of medical care may be the most important characteristic of the physicians' services market. A priori, the patient often is uncertain about the efficacy of alternative medical procedures and may have difficulty assessing the quality of the services provided by the physician. It is possible, however, that "imperfect knowledge" does not characterize all physician contacts. Using Pauly's (1977) estimates, approximately 38 percent of expenditures on physicians' services are for services purchased under conditions in which consumers might be described as reasonably well informed. Pauly indicated furthermore, that competition «2G in the market for physicians' services would not necessarily be threatened in instances where physicians and patients were mutually uncertain with respect to relevant aspects of medical care. Pauly did feel that competi- tion might be thwarted by conditions of "reducible" uncertainty, i.e., co. ditions in which physicians possessed relevant information which patients did not. The difficulty which consumers frequently experience when assessing both physicians' quality and the expected efficacy of treatment may have given impetus to the institutional arrangement whereby the physi- cian is ethically bound to act as the patient's agent. This implies that (a) the physician will make decisions which are in the best medical interests of his client, and (b) the decisions made by the physician should correspond to the decisions the patient would make if he possessed perfect information. (See Evans 1976, Arrow 1963, and Monsma 1970 for a variety of views on this subject.) The outcome of this agency relationship is that price setting is perceived by both parties as an ancillary (albeit an important) issue. As Arrow suggests, ''purely arms-length bargaining behavior would be incompa- tible, not logically, but surely psychologically, with the trust relations" (p. 965). The physician-agent frowns upon price advertising for "ethical" reasons. Patients, because of conditioning with respect to their role in the agency relationship, are somewhat loath to "shop around" for the lowest prices for physician services. Of course, this "shopping" behavior is also limited by the nature of the need for physician services. An individual with an inflamed appendix may be in neither a bargaining mood nor a bargain- ing position. Thus, even when considerable intra-market variations in physi- cians' fees does exist, patients must often rely on informal sources (e.g., «30- relatives and friends) for information regarding local prices of physi- cians' services. Newhouse (1970) postulates that the isolation of physicians from each other with respect to price competition leads to low cross—-elasticity of demand for their services. He also suggests that patients' inability to rank physicians on a quality scale may cause clients to use price as a proxy measure of quality. Therefore, even if patients are cognizant of cross-physician price differences, demand for low-priced services may be depressed because of the assumed positive correlation between quality and price. 2 Empirical Evidence with Respect to Competition and Monopoly Generally, empirical evidence supports the monopolistic rather than the competitive model of the physicians' services market. The degree of monetary price dispersion in local markets, even after adjusting for major differences in physicians' credentials, led Sloan and Feldman (1977) to conclude that physicians possess at least some degree of monopoly power. At the same time, Sloan and Feldman rejected the notion that price-fixing cartels are widespread, both because of the dispersion in local market prices and because of the difficulty such a cartel would have in policing the price decisions of its member physicians. Several economists have attempted to test monopolistic and competi- tive models econometrically. Generally, the competitive models are rejected, but so are the monopolistic models on occasion. For example, Newhouse's (1970) finding of a positive correlation between prices and physicians per capita was inconsistent with his specifications for both competitive and monopolistic models (also see Newhouse and Sloan, 1972). Feldstein (1970) reported positive coefficients on net price variables in his estimated <3] demand equations which led him to reject an annual (competitive) equilibrium model of the physicians' services market. He also found evidence of "backward bending" supply curves. Feldstein's results are open to criticism, however. Newhouse and Phelps (1974) indicated that Feldstein's demand enuabions were under-identified, hence no strong conclusions about the structure of the market could be drawn from his results. Likewise, Fuchs and Kramer suggested that Feldstein's failure to account for technological changes over time may have led to his perverse coefficient estimates. Kehrer and Knowles (1974) found consistently significant coefficients for variables associated with monopolistic competition and oligopoly in their study of physicians' fees. They indicated that some variables associated with imperfect competition might also reflect differences in the quality of physicians' services. Overall, the statistical significance of the non- competitive market variables could be viewed as further evidence of the inadequancy of the competitive paradigm to explain fully physicians' pricing decisions. Physicians themselves provide evidence that they possess a degree of monopoly power -- i.e., that physicians can set fees over a wide range. In physicians' trade journals, Haddock (1968) and Tharp (1975) described their own techniques for setting fees. Neither relied on market determined prices. In the past, physicians frequently discounted their fees for poor patients. That is, physicians charged fees based upon their patients’ income levels. This capability to charge different fees to different patients (for the same service) was often presented as evidence that physicians possess a degree of monopoly power over fees (see Kessel, 1958; Masson and Wu, 1974). Ruffin and Leigh (1973) developed a charity-competition model to explain «3 Zw price descrimination based upon incomes of patients. Masson and Wu, however, point out that the Ruffin and Leigh model makes excessively strict assumptions about physicians' "disutility" for serving the rich. They also indicate that the Ruffin and Leigh model (p. 64) "will have corner solutions negating their results unless (almost) all physicians have very similar utility functions." The foregoing indicates that, in the market for their own services physicians possess considerable freedom to set fees as they choose. The fees actually set will depend upon the objectives of physicians, which is the subject of the following subsection. D. OBJECTIVES OF PHYSICIANS The existence of monopoly power implies that suppliers (physicians) have the ability to set prices. Any price setting decision should, of course, be compatible with the physicians' goals or objectives. Researchers are divided, however, with respect to the issue of physicians' goals and objectives. At one level,disagreement exists with respect to the set of objectives which are maximized by physicians. At another level, the usefulness of maximiza- tion itself as a model of behavior is questioned. Some researchers believe that "satisficing" (or income targeting) may be a more realistic description of physician behavior (e.g., Newhouse and Sloan, 1972; Evans, 1973). Maximiz- ing models generally employ the concept of utility as the maximand. Utility is often defined to be a function of one or more variables purported to influence the physician's behavior. Univariate utility functions often equate utility maximization with profit or income maximization. Kessel (1958), for example, argues that physicians collectively set prices to maximize (group) profits. Steinwald and Sloan (1974) argue that a utility function in leisure as well as profit can be subsumed under a profit maximization model through «33 judicious transformation of physicians' leisure time into a monetary equivalent. Sloan's ( 1976) model of physician pricing behavior is founded on such profit maximizing behavior. Strict profit maximization may be an unsatisfactory description of physicians' behavioral objectives. Arrow (1963) believes that a "trust relationship" exists between physician and client. The essence of the relationship is that the physician will use his knowledge to the patient's best advantage. Arrow suggested that actual (or apparent) profit maximizing behavior on the part of the physician would be deleterious to this trust pact. Reder (1969) described the "professionally approved doctor-patient relationship" as one in which the physician is the controlling agent. Medical precepts take precedence over the patient's wishes in treatment decisions. Reder stated that this "approved" relationship is most likely to occur whenever the physician has more patients than he needs (financially) in which case the physician will be indifferent to the possible loss, as a client, of any particular patient. Reder claims that this approved behavior is inconsistent with total profit maximization. He also proposes two additional reasons why physicians may not be profit maximizers. First, the growth of medical insurance has made it increasingly difficult for doctors to charge different individuals different prices (based on patients’ incomes) for the same services. Second, due to professional considerations, certain physicians may have clear preferences for treating certain types of cases. In order to secure an adequate supply of such cases, the physician may often charge less than "what the traffic will bear." As the foregoing indicates, physicians may weigh factors other than income or profits when they set prices or make other practice-related «Bl decisions. However, multivariate utility functions often lead to complex specification of price determination, as the following discussion illustrates. 1. Formal Model of Utility Maximizing Physicians Reinhardt (1975) developed a general utility maximizing model in which leisure, income, and the size of the auxillary staff are arguments of the physician's utility function. The following model is adapted from Reinhardt's. In this model, however, an additional argument (G) is added to the utility function to address non-pecuniary aspects of physician utility. The physician's objective is to maximize utility (U), where: (1) Utility = U(X,T,G; Z) where X = physician's hours of leisure per period; T = physician's practice profits; G = other aspects of the physician's practice which affect his utility; Z = a vector of exogenous factors which may influence the physician's practice decisions. Furthermore: (2) X=H-H where H is total hours available to the physician per period; H is the physician's hours of work. (3) m= P(Q)*Q - C(Q) (4) G = G(H,L; A) and G20, G20 (5) Q = f(H,L,K; A) 35 The physician's maximizing behavior is captured by the five equations above. The physician maximizes a composite function in leisure (X), profits (wm), and a general variable (G). A priori, the marginal utility with respect to all three may be assumed to have the following signs: U = 3U/3X>0; U = aU/am>0; U_ = 3U/3Gs0 X i! g Some additional definitions follow: Q = f(+) = the rate of output (e.g., patient visits) produced by the physician's practice per period. P(Q) = demand schedule for physician services; average price per visit of output of physician services. c(Q) = cost per unit of output of physician services. A common cost curve is a quadratic (Walters, 1963, p. 45): CQ =a + aQ+ aq? 0 1 2 L = size of the physician's ancillary staff K = physician's practice capital not complementary with aides; assumed constant in this (short run) model. A = a vector of exogenous variables which affect output of physician services. The physicians' decision variables are hours of work (H) and the number of aides to employ (L). The physician's objective is to maximize utility subject to the conditions in equations (2) through (5). The physician will work enough hours and employ a sufficiently large staff so that: aU * * 6 <— =-U + |U (P" - Cf + U *G = 0 or (6a) oH x | x ) N g h -36- (6b) I @* - ty) + UG =U m g x h h 3 * * _ (7a) oT [v,c C )t) + ub, ¢, = 0 or (7b) (P*.f ) +s G = c*.f 2 U 2 2 m These results are analagous to Reinhardt's (1975). In equations 6(b) and 7(b): c* = CQ) marginal cost associated with the marginal unit of aQ physician services produced; P* = marginal revenue associated with the marginal unit of output sold; PQQ) Q + P(Q) in models where the physician sets 3Q a single price (e.g., simple monopoly); p* = P (or fixed price) in model where the physician is a price taker (e.g., competition); P(Q) = the demand curve, in models where the physician is a perfectly discriminating monopolist. Also, f, = ed and f, = AC) Equation 6(b) states that the physician will increase or decrease hours of work so that the marginal disutility from an hour of foregone leisure (Uy) is equal to the marginal utility gained from the additional units of the factor (G) and profits (m) produced by the marginal hour of work. BJ Equation 7(b) states that the physician will employ auxiliary staff up to the point where the marginal cost of additional staff cf) is equal to the marginal revenue equivalent of the extra profits and additional units of G which are induced by the marginal employment of staff. a. Variations of Maximizing Behavior Equations 6(a,b) provide a vehicle with which to examine two alternative assumptions with respect to physician behavior. First, the expected behavior of physicians who maximize a general utility function will be explored under the assumption that certain types of services (e.g., "interesting" cases) provide utility to the physician. Following this is a discussion of the simple profit maximizing physician. Let G(H,L;A) be a function which describes the physician's. utility with respect to treatment of certain patient subgroups (e.g., see Ruffin and Leigh; Masson and Wu). If marginal utility from treatment of this subgroup (Ug) is strictly positive, then equation 6(a) may be satisfied even if marginal cost exceeds marginal revenue (i.e., c*>p%). Conversely, if the marginal utility from treating the patients in this subgroup is strictly negative 0, <0) the physician may ration services to "undesirable" patients by keeping the size of practice at a level where marginal revenue is greater than marginal cost (i.e., p*>c"). If the physician is assumed to be a profit maximizer, equa- tions 6(a,b) and 7(a,b) may be simplified greatly. This is because profit maximization implies that utility is a function of only one vari- able, i.e., profit. In that case, equations 6 and 7 reduce to the familiar condition that marginal profit is zero (at optimal levels of staff size and hours of work). That is, marginal revenue (**) equals marginal cost (c*). It should also be apparent from equations 6 and 7 that a ''price «3G equation" based upon multivariate utility maximization may involve non- linear terms (e.g., Ug/Uy). Thus, utility maximization provides a useful framework for ascribing simultaneous goals to physicians. However, this leads to equations which are often intractable from an empirical standpoint. 2 Satisficing Models Several researchers have suggested that satisficing rather than maximizing behavior characterizes the physician. Newhouse and Sloan (1972) raised the question of whether empirical analysis based on the assumption that the market continually clears is useful at all. They reiterated Newhouse's earlier (1970) suggestion that physicians may be satisficers (or have an income target). Evans (1973) appears to have carried the satisficing approach farther than other economists. He developed a simultaneous model with equations for: (a) the price of medical services; (b) the per capita rate of medical services; (c) net income per physician; (d) a shortage measure (workload per physician minus physician's desired workload). In Evans' formulation, price does not clear the market. Rather, (Evans, 1973, p. 32) "market clearing takes place through a complex of informal rationing mechanisms in which the role of the supplier as agent for the consumer is a central feature. The persistence of non-equalibrating price leads the model to those of its conclusions which are novel —- i.e., that expanding the supply of physicians will drive up prices as well as costs and that increasing the technical efficiency of physicians may «39 reduce unit prices but will drive up total costs." Data limitations precluded Evans from empirically testing his model. E. CRITIQUE OF MODELS OF PHYSICIANS' PRICING OBJECTIVES As yet, no consensus exists among researchers with respect to the appropriate model of physicians' behavior with respect to price setting. Profit maximization is advocated by Steinwald and Sloan because of the relative consistency between their coefficient estimates and profit maximizing behavior. However, evidence exists which is inconsistent with profit maximizing behavior on the part of physicians. For example, Reinhardt (1972) showed that physicians consistently employ fewer than the profit maximizing number of aides in their practices. Arrow (1963) observed that a profit maximizing monopolist must operate within the elastic region of the demand curve. Otherwise the monopolist could increase profits by raising prices and reducing output. Evidence from demand studies shows the (money) price elasticity of demand for physician services to be inelastic -- i.e., consistently less than 1.0 in absolute value. This evidence does not conclusively rule out profit maximizing behavior, however. Redisch et al. (1977) gave several reasons for low (money) price elasticities of physicians' services. First, due to pervasive health insurance coverage, time cost now assumes a larger share of the total cost of physi- cians' services. "A large proportional reduction in money prices may there- fore only represent a small decrease in total costs faced by the "patient" (Redisch et al., page 18). Second, if the demand for physicians' services is viewed as derived from the demand for health (a la Grossman), then it is reasonable to expect a low price elasticity for an input, such as physicians’ -40- services, which has few close substitutes. Not all models of monopoly power require the physician to operate within the elastic region of total market demand. For example, Chamberlain's monopolistic competition model permits inelastic market demand to exist simultaneously with relatively elastic demand for the product of the individual physician-firm (see Ferguson, 1972, pp. 317-333). Some evidence does exist that income targeting (satisficing) may be more appropriate than profit maximizing as a description of physicians’ price setting behavior. When physicians describe their own procedures for setting fees they sometimes refer to an "annual net income goal" or an "acceptable growth income' as the basis for determining charges (see Haddock, Tharp). Newhouse (1970) concluded that physicians set prices to achieve income targets after he obtained insignificant values (and "wrong" signs) for coefficients of demand shift variables in his estimated price equations. He also indicated that "satisficing' behavior was adequate to explain the positive correlation he observed between prices and physician density. Evans' (1973) model of physician price setting made use of income targets for physicians. The Evans model shows that such satisficing behavior is compatible with the "agency" relationship between physician and patient which enables physicians to create demand for their own services. Evans (1974) criticized his own approach, however, as "dis- tressingly fuzzy in its failure to explain the formulation of target incomes or the distribution of target shortfalls between price increase and demand generation" (p. 168). Feldstein (1970) rationalized his findings reported earlier by post- ulating that physicians set prices (below market clearing levels) in wile order to generate excess demand for their services. In this way, physicians are able to maintain an adequate supply of "interesting" cases. Evans (1973, p. 35) indicates that it may be difficult empirically to distinguish between "excess demand" and "physician generated" demand, however. mb 2 IV. THE DEMAND FOR PHYSICIANS' SERVICES The traditional economic relationship between prices and demand was illustrated previously in Figure 3(a). In that simple model, prices served to equate supply and demand. An increase in demand (ceteris paribus) from D, to D, should lead to increases in both prices and the quantity of physi- cian services provided. In the event that prices do not serve a market clearing function, the demand schedule indicates the maximum quantity of services which consumers will purchase at any given price level. The process by which demand for physicians' services is generated is still somewhat unclear. One recent model treats the demand for physician services as derived from the demand for a higher order commodity: health. In this model (see Grossman, 1972) individuals gain "utility" from "health" and a set of other commodities. Physicians' services are one factor which an individual may employ to produce health. The individual's ability to produce health depends upon the availability of inputs (factor price infor- mation, health status, income levels and other exogenous factors) and upon the patient's technical ability to produce health from these inputs. The latter may be influenced significantly by the strength of the agency rela- tionship between the patient and the physician. Grossman's model is a useful paradigm through which one can formalize the relationships among the various market forces relevant to the demand for physician services. Most empirical studies of the demand for physician services are not as sophisticated as Grossman's work. Rather, these econo- metric studies usually estimate a linear demand relationship, sometimes in ely Be conjunction with a linear supply equation and/or other relevant equations. The quantity of physicians' services demanded, measured by visits or expenditures, is related to a set of explanatory variables such as health status, income, price and the current supply of physicians. The effects of a variety of factors upon the demand for physician services are discussed below. A. HEALTH STATUS AND THE DEMAND FOR PHYSICIANS' SERVICES Since 1970, several cross-sectional studies of demand have appeared which contained one or more measures of individuals' health status among the explanatory variables. Among these health status measures have been: (a) the individual's perceived health status (excellent, good, fair, poor); (b) the number of chronic health conditions which limit an individual's activities; and (c) the number of restricted activity days experienced by an individual over the relevant time period. Generally, self-perceived health status measures are among the most significant variables affecting physician utilization. (See Acton, 1975; Andersen and Benham, 1970; Phelps and Newhouse, 1976). Coefficients that have been obtained indicate that utilization of (and expenditure on) physi- cians' services increases as self-perceived health status declines. Like- wise, the number of chronic health conditions is positively related with physician utilization. Acton (1976, p. 188) reported: "the health status variables are among the most consistently significant predictors of demand for care. Greater numbers of chronic health conditions produce higher utilization of all forms of medical services (with t ratios ranging from 4 to over 16)". willy David and Reynolds (1976) found health status measures significant in their study of physician utilization by the poor. They created measures of "good," "average' and "poor" health status from the individual's reported number of restricted activity days and chronic conditions. Using these measures David and Reynolds found that, as health status deteriorated from "good" to "poor," physician visits by public assistance recipients increased by 74 percent and visits by other low-income persons increased by 90 percent. When direct measures of health status were not available, researchers have relied upon proxy variables (e.g., age, sex, race) as surrogates. Dunlap and Zubkoff presented summary findings from over 200 research studies which examined, among other issues, consumption and utilization of physician services. They reported that females apparently demand more physician services, or have a higher utilization rate than males. Being non-white was reported to have a negative impact on the demand for most health services. Finally, the demand for general physician services seems to decline with age until middle age. Beyond that time, demand becomes positively associated with age. One must be careful about the use of such proxy measures of health status. Leveson (1970) warns that such proxy measures will explain only part of the variation in health status among individuals. A large portion of an individual's health status may not be correlated with these surrogated variables since a substantial portion of health variation is determined by inheritance, environmental conditions and prior levels of immunization against disease. Leveson further speculates that omission of a health measure in demand studies can lead to biased estimates of the income elasticity of demand. ely Ge B. PRICE AND THE DEMAND FOR PHYSICIANS' SERVICES One of the most fundamental notions of economics is the negative relationship between prices and the quantity of goods which consumers will demand. As expected, the various empirical studies of demand for physician services include one or more price measures among the dependent variables. The effective price of physicians' services has two components: (a) the out-of-pocket monetary cost of the service, and (b) the cost in waiting and/or travel time. The following are results on the monetary price elasticity of demand for physician services from a variety of empirical studies: Study Elasticity Fuchs and Kramer (1972) -.10 to -.36 Davis and Russell (1972) -.98 to -1.03 Newhouse and Phelps (1972) -.137 Newhouse and Phelps (1976) -.07 to -.08 Most published price elasticities of demand for physicians' services are below 1.0 in absolute value and, as expected, negative. Several demand studies also have included cross-price measures as independent variables. In Davis and Russell's study of demand for outpatient services, elasticities with respect to inpatient price levels were calculated to be between .85 and 1.46. Newhouse and Phelps (1976) found much smaller cross-price elasticities in their study of demand for physicians' office visits. They found elasticities with respect to hospital visit prices between -.055 and -.14. Newhouse and Phelps interpreted these negative cross-elasticities to indicate that hospital and office services are complements. The Davis and -46- Russell results provide tentative evidence that some hospital inpatient services and hospital outpatient visits are substitutes. The second component of the price of physicians' services is the time cost. Acton (1976) suggests that time price may eventually replace money price as the chief determinant of demand. He attributes the decline in the importance of money price to: (a) the secular trend in third party coverage; (b) the rising opportunity cost of time; (c) increases in time required to receive care; (d) the prospect of national health insurance. In his 1976 study, Acton examined the effects of waiting time and travel time upon the demand for ambulatory care from public and private sources. Acton found travel time elasticities of the demand for public ambulatory care in the range -.619 to -.958. He found corresponding travel time elasticities for private ambulatory care to have values between -.252 and -.337. In the same study, waiting time elasticities were -.12 for public ambulatory care and -.05 for private care. Acton concluded: "travel and waiting time appear to be operating as normal prices, producing a nega- tive own-price elasticity and a positive cross-price elasticity of demand for medical services" (p. 190). Phelps and Newhouse (1972) explored the effect of several variables upon the number of visits made to physicians by members of a group health plan. They found that female dependents used, on the average, 2.53 more visits per year than male subscribers and 1.66 more visits per year than female subscribers. They interpreted these differences to be the result of variations in the opportunity costs of time across the different user groups. wl Fe Wise and Zook (1977) viewed the relationship between waiting time and the demand for physicians' services from a different perspective. They suggested that physicians react to variations in the demand for their services by varying waiting times within their practices. Wise and Zook estimated elasticities (with respect to the number of patient visits to a physician's office per year) of wait-to-appointment and of wait-in-office to be .65 and .61 respectively. They reported (p. 5): "The reduced form as well as the structural estimates provide direct evidence that wait to appointment, and not price, is the primary device used by physicians to ration visits." The foregoing data indicate that time costs can have as significant an impact as does monetary cost upon the demand for physicians' services. Policy makers need to be cognizant of the fact that the effective price of physician services has both monetary and time dimensions. As third-party coverage of physicians' services becomes more pervasive, the relative importance of time cost as a rationing device most likely will increase. C. THE INFLUENCE OF PHYSICIAN SUPPLY UPON THE DEMAND FOR SERVICES As previously indicated, an agency relationship frequently exists between patient and physician. As the patient's agent, the physician provides information and advice which affects the patient's perception of his medical condition and his need for care. If the local availability or supply of physicians increases, an individual physician may find his work- load (and perhaps his income) decreasing, assuming the local (inelastic) demand for physicians' services remains unchanged. Facing a smaller work~ load and possible income losses, the physician could attempt to induce additional demand for his services via his role as the patient's medical advisor. 4 8= The positive correlation between physicians per capita and various indicators of per capita consumption of physicians' services is often cited as evidence in support of the supply induced demand concept. For example, Gable and Redisch (1977) reported: "The 1974 Study of Surgical Services for the United States (SOSSUS) examined surgery rates in four metropolitan areas and found the surgery rate per population positively related to surgeon density." Gabel and Redisch offered this finding as evidence that an increase in the number of physicians locally leads to an increase in per capita utilization of physicians' services. Reinhardt (1976), however, warns of the dangers inherent in making inferences about supply induced demand from first order correlations between physicians per capita and utilization rates. Reinhardt maintains (p. 34): "What is being observed in such correlations is not demand but utilization, and utilization may be below actual demand where supply is taut. An observed correlation between health care utilization and supply may therefore reflect either demand- inducement on the part of providers or successive adaption to a hitherto unsatisfied demand that has always been there." It is apparent, then, that physician-induced demand is not the only possible rationale for the observed association between physician "avail- ability" and utilization. Both Pauly (1975) and Fuchs and Kramer (1972) developed lists of explanations for this association. Pauly's list is repeated below: (a) The availability effect arises from statistical properties of the estimation procedure. (b) The availability effect represents the response of use to changes in the time or convenience cost of care. ly Que (c) Medical services are subject to chronic excess demand, and care is rationed by physicians on the basis of the interest or severity of cases. (d) Physicians create demand by manipulating the information they provide to consumers. Researchers are divided with respect to the correct explanation of the association between demand and "availability." Feldstein (1970) for example argues the case for excess demand, while Evans (1973, 1974) is one of the chief exponents of demand creation via the physician's manipu- lation of information. Several econometric studies of the demand for physicians' services have included a measure of physicians per population as an independent variable. The elasticities of demand for physicians’ services with respect to physicians per capita from selected studies are presented below: Source Elasticity Fuchs and Kramer (1972) .335 to .507 Pauly (1975) -.285 to .767 Newhouse and Phelps (1976) -.03 Pauly, in his (1975) preliminary work, reported a positive elasticity of demand (with respect to the supply of physicians) for individuals who had less than 12 years of education. Pauly concluded (p. 22) "if physician availability affects use, it affects the use of persons in households where the head has little education." Newhouse and Phelps (1976) reported small, negative elasticities of demand with respect to combined (linear and squared) measures of physicians per population. In their demand equations, the -50- elasticities associated with M.D.'s per population were between .38 and .48. Corresponding elasticities of demand with respect to the squared term were .003 and -.22. They reported (p. 279) that the negative sign of the coef- ficient on the squared physician-population variable "supports the hypothesis that additional physicians reduce the fraction of time that demand exceeds capacity." Finally, Fuchs and Kramer believed that the demand for physicians’ services is significantly affected by the supply of physicians. They stated (p. 2): "The most striking finding of this study is that supply factors (technology and number of physicians) appear to be of decisive importance in determining the utilization of and expenditures for physi- cians' services. This conclusion stands in sharp contrast to the widely held belief that utilization and expenditures are determined by the patient, and that information about income, insurance coverage, and price is sufficient to explain and predict changes in demand." D. INCOME AND THE DEMAND FOR PHYSICIANS' SERVICES If physicians' services are a normal good (in the economic sense), an increase in individuals' incomes ought to result in an increase in the amount of these services demanded. The responsiveness of demand to income shifts is determined by two forces. A change in income affects the indivi- duals' wealth (income effect) and the relative "price" of leisure time (sub- stitution effect). Thus, as income increases, the individual's purchasing power and the opportunity cost of his time both increase. Under a variety of contractual agreements between individuals and employers, time spent by individuals in the consumption of physicians' services may not result in an actual income loss. Hence, the opportunity cost of time faced by an -51- individual depends upon factors other than wage rates. Some empirical estimates of the income elasticity of demand for physicians' services are presented in Table VIII. There exists a wide variation in the elasticity estimates reported in Table VIII. However, these data suggest several hypotheses. First, expenditures are more elastic than visits with respect to shifts in income. If price is positively related to quality in the physician services market, increases in income may enable consumers to purchase higher quality services (e.g., specialists rather than G.P.'s). Second, the negative elasticity of outpatient clinic utilization with respect to earned income may be related to the purported high waiting times associated with these facilities. As incomes increase, patients need not pay the high time price associated with outpatient services. In a study similar to the one reported in Table VIII, Acton (1976) found positive elasticities with respect to earned income for the demand for private physician services. He also found a negative elasticity for public care and concluded (p. 186) "the price effect of a wage change dominates in the demand for public care and the income effect dominates in the demand for private care." In Table VIII, elasticities are well below 1.0 in absolute value. Hence, the demand for physicians' services is apparently inelastic with respect to income. A portion of this inelasticity is undoubtedly due to the spread of public and private health insurance during the past decade. Some reported income elasticities are sufficiently far from zero to indicate that consumers treat some types of physician services as a discretionary item. 5D TABLE VIII. INCOME ELASTICITY OF DEMAND FOR PHYSICIANS' SERVICES Source and Demand Measure Income Measure Elasticity Andersen and Benham (1970) expenditures observed income .22 to .41 expenditures estimated permanent income .17 to .63 vigits estimated permanent income .01 to .31 Fuchs and Kramer (1972) visits family income .04 to .57 Pauly (1975) visits family income .011 to .116 Acton (1975) outpatient clinic visits earned income -.01 outpatient clinic visits non-earned income -.01 private visits earned income .07 private visits non-earned income .08 Phelps and Newhouse (1976) visits wage income .028 to .029 non-wage income -.00008 to -.004 «58 V. SUPPLY IN THE PHYSICIAN SERVICES INDUSTRY In traditional economic analysis, markets are made up of those agents who demand the product and those who supply it. Theoretically, these two sides of the market are responsive to different sets of exogeneous forces. The only interaction of supply and demand occurs during the determination of market clearing prices and quantities. The physician services market has at least two characteristics that make it difficult for researchers to estimate the supply responses of physicians. First, the commodity of interest in this market is a service. Hence the production (supply) and consumption (demand) of products are often simultaneous and indistinguish- able. Second, the "agency" relationship between physicians and patients places the physician on both sides of the market. Despite these difficulties, it is useful to examine what is known about the relationship between market characteristics and measures of physicians' services. Although accurate measures of physician productivity may not exist, certain proxy measures of physician output are available. These are: (1) the total number of physicians practicing in a market (the physician stock); (2) the time (weeks and hours) spent by physicians at work in their practices; and (3) the number of patient visits or the aggregate billing for services generated by physicians' practices over some period of time. A. DETERMINANTS OF THE AGGREGATE SUPPLY OF PHYSICIANS In Figure 3(a), a "typical" graph of the physician's market was Bly displayed. The most important factor affecting the location of the supply curve will be the stock of physicians practicing in that market. The regional stock of physicians will increase if in-migration exceeds out-migration, death and retirement of physicians over some time period. Projected increases in the national physician stock were presented in an earlier section of this report. The regional or local market stock of physicians will depend upon the location preferences of doctors. In Table IX, the results of much of the literature on physician location is presented in summary form. As the information in Table IX indicates, there is some weak evidence to support the contention that physicians are attracted to areas of rela- tively high incomes and/or degree of urbanization where the physician can charge higher prices. In particular, Benham et al. (1968) examined the determinants of physician supply among the states (1950) and concluded that "in large part, medics have displayed a tropism for higher incomes that has caused them to migrate with the effective demand for their services." Fuchs and Kramer's study of physicians per 100,000 population in states indicated that the significant elasticities of physician supply with respect to an average price measure were between .83 and 1.14 in most of their regressions. However, Fuchs and Kramer reported that collinearity between price and other explanatory variables caused significant fluctuations in the coefficient of price. Finally, in their study of the Canadian medical insurance system, Berry et al. (1978) concluded that the average gross payment for general practitioners in a region had a positive influence upon in-migration of G.P.'s, although the magnitude of this effect was sensitive to the particular specification of the model employed. -55- TABLE IX. SELECTED PHYSICIAN LOCATION HYPOTHESES AND SUMMARY EVALUATIONS Hypothesis Summary Evaluation 1. In U.S. cities, there are relatively few physicians Supportive in ghetto or barrio areas. 2. In U.S. cities, there has been a net outmigration Supportive of physicians from ghetto and barrio areas. 3. Percent of the area's population which is nonwhite is a Supportive significant determinant of physicians' office location. 4, In U.S. cities, there are relatively few physicians Contradictory in low-income areas. 5. In U.S. cities, there has been a net outmigration of Supportive physicians from low-income areas. 6. Income of the area's population is a significant Weakly Supportive determinant of physicians' office locations. 7. Physicians are attracted to areas of a city with a Supportive high number of hospital beds per capita. 8. Physicians are attracted to areas of a city that have Supportive a high proportion of commercial zoning. 9. Physician-population ratios rise with the degree of Supportive urbanization. 10. Physicians tend to locate in areas similar in size to Supportive those in which they were reared. 11. Physicians tend to locate in areas similar in size to Supportive those of medical school attended. 12. Training experiences and preceptorships in rural areas Weakly Supportive will influence physicians to locate in rural areas. 13. Physicians are attracted to areas which they perceive Weakly Supportive to have a "high quality of life". 14. Physicians are attracted to the areas surrounding a med- Weakly Supportive ical school as an aid to their professional development. 15. Physicians who attend "higher quality" medical schools Supportive tend not to locate in rural areas. 16. Existence of hospital facilities in a rural area tends Contradictory to draw physicians to that area. 17. Group practice opportunities attract physicians to Supportive rural areas. 18. Physicians tend to locate in urban rather than rural Contradictory areas because of a relatively greater income potential in urban areas. SOURCE: B. Eisenberg and J. Cantwell "Policies to Influence the Spatial Distribu- tion of Physicians: A Conceptual Review of Selected Programs and Empirical Evidence," Medical Care, June, 1976, Vol. XIV, No. 6. -56- B. SUPPLY BEHAVIOR OF INDIVIDUAL PHYSICIANS The aggregate supply of physician services in a local region is the sum of the services supplied by each individual physician. The supply responses of individual physicians are a function of local prices and additional factors (e.g., the local physician-population ratio and the physician's age). Although the literature with respect to micro-level supply decisions in the physician services market is small, some consistent findings exist. 1. Prices, Wages and Output per Physician Two early studies of services supplied per physician were performed by Feldstein (1970) and by Fuchs and Kramer (1972). In both, the quantity of services supplied per physician was calculated by deflating expenditures on physician services by some measure of the average price of these services. These two studies hinted that the individual physician's supply curve might be backward bending; i.e., that individual doctors supplied fewer services as the price of these services increased. In Feldstein's study, the reported supply elasticities with respect to an estimated average price index were between -1.91 and -.28. Similarly, Fuchs and Kramer found output (per physician) elasticities with respect to average price to be between -.828 and .012. They warned that only a very small degree of confidence could be attached to the initial findings of a negative price elasticity. They believed, however, that their results lent no support to the hypothesis that higher prices induce additional services from physicians already located in a state. Sloan (1974, 1975) and Vahovich (1977) examined the relationship between physicians' work hours (and weeks) and physician wages. Generally, = they provided weak support for the hypothesis that the individual physi- cian's supply curve is backward bending. Sloan (1975), for example, used linear and squared wage variables in his estimation of physicians' weeks- worked per year. He found (p. 560) "evidence of a backward-bending supply curve for weekly wages of $800 or more in 1960 (approximately one standard deviation above the mean for that year).'" Similarly, Vahovich found that physicians increase their hours (and weeks) of work as their wage rate increases. However, once a "critical" wage is reached, additional wage increases result in decreased hours and weeks of work. Vahovich concluded with the remark that his empirical results were somewhere between those obtained by Feldstein and those obtained by Sloan. He found evidence of backward bending supply (like Feldstein) but low elasticities of supply (like Sloan). 2 Physician-Population Ratios and Output per Physician In their estimates of services provided per physician, Fuchs and Kramer found (p. 38) '"the number of physicians per capita is the only vari- able tested that clearly has a significant impact on physician productivity." Their estimated elasticities were on the order of -.66. Sloan (1974, 1975) found consistently negative and significant coefficients on a physicians per population variable in equations predicting physicians' hours of practice per week. These coefficients indicate that aggregate physician services may not grow at the same rate as the aggregate stock of physicians. On the other hand, Wise and Zook reported that: (a) physicians' hours of practice per year, (b) the number of visits per year made by patients to the physi- cian's office, and (c) physician time per visit all were essentially unaffected by the local physician-population ratio. They did find, however, .-58- that the number of days a person must wait for a non-emergency appointment was negatively related to the number of G.P.'s per 1,000 residents in the county where the physician practices. The elasticity of wait-to-appointment with respect to G.P.'s per population was -.71. Wise and Zook suggested that physicians facing shifting supply conditions in the local market for their services may stablize their workloads by varying wait-to-appointment times. The anticipated growth in the supply of physicians (reported earlier in this review) also will shift the distribution of physicians toward the younger end of the age distribution. Both Vahovich and Sloan found an inverse relationship between physicians' hours (and weeks) of work and physician's age. Wise and Zook reported that older physicians worked fewer hours per year but provided more physician time per visit. They also estimated (p. 13) that "on the average, when other variables are held constant, a physician reduces the number of patients he sees per year by 94 with each additional year of practice." Thus, a decline in the mean age of physicians within a region may have a positive effect upon the aggregate supply of physicians' services. 3 Employment of Ancillary Personnel Existing research shows that auxiliary personnel (e.g., nurses, physician assistants, office aides) can significantly increase the produc- tivity of physicians' practices. Lave et al. (1973) described several studies of the effects of paramedical personnel upon physicians' (generally pediatric) practices. Those studies indicated that physicians could increase the number of patients they treat by more than 30 percent by employing para- medical personnel. -59- Reinhardt (1972, 1975) developed a production function for the physician firm. He estimated his function from cross sectional survey data on more than 2,000 physicians. Reinhardt's production function related out- put measures (total patient visits, office visits, patient billings) to a set of variables which included measures of the number of registered nurses, technicians and office aides employed by general practitioners in private practice. He found that coefficients on the three types of aides were similar in production function estimates. Reinhardt interpreted this to indicate that "whatever the tasks being delegated may be, employment of either type of aide tends to result in roughly the same saving in physician time" (1972, p. 62). Reinhardt then calculated the elasticity of output with respect to the employment of aides in general. His elasticity estimates were on the order to .34 at the sample average of 1.96 aides per physician. He found these elasticities compared favorably with those calculated by Kimbell and Lorant (1973) who report "an estimated aide-elasticity of out- put between .31 and .34 for production functions based on patient visits as an output measure, and between .38 and .39 for estimates with output measured by gross revenue" (Reinhardt, 1975, p. 176). Reinhardt stated that physicians generally employ less than the (financially) optimal number of aides in their practices. He indicated that the representative practice in his sample could profitably employ four aides per physician -- a figure more than double the sample average of 1.96 aides per physician. A recent report (Abt, 1976, p. 31) listed four possible reasons for the sub-optimal employment of aides by physicians: (a) Physicians may have little talent for entrepreneurial- managerial functions. (b) Training new personnel is quite costly to the physician, -60- given the opportunity costs of his time. (c) Physicians may have an aversion to personnel management. (d) Indivisibilities may prevent physicians from hiring an optimal number of aides. Similar reasons also appear in Lave et al. (1974) and in Reinhardt (1972). «B]= VI. ECONOMETRIC STUDIES OF PRICES IN THE PHYSICIAN SERVICES MARKET Much of the empirical evidence with respect to the price of physician services is obtained from the analysis of price equations. In these equa- tions, prices (often disaggregated by procedure and physician specialty) are regressed against a "reasonable" set of explanatory variables. At times, mo explicit structural model of price setting behavior is specified (e.g., Cantwell, 1976a; Steinwald and Sloan, 1974; Newhouse and Phelps, 1974). More often, either a competitive or monopolistic market is assumed and physicians are described as profit maximizers. A. PRICE EQUATIONS DERIVED FROM COMPETITIVE ASSUMPTIONS Under competitive conditions, prices are determined by the interaction of supply and demand forces. When (and if) a competitive equilibrim is achieved, the quantity of the commodity produced is equal to the aggregate demand for that commodity at the market determined price. Assuming that competitive conditions prevail, a price equation can be constructed as follows (see Huang 1976): (quantity) demand = Qd = 8 ® + B x Rg <0 2 1 (quantity) supply = Qs =a p +o X a >0 1 2 2 1 Qd = Qs In the equations above, x and x, are exogeneous demand and supply shift variables while p is the market clearing price. These equations can be -62- solved to yield a "reduced form" price equation: B a am p= [—2)x - [2 )« a - 8 1 o = B 2 1 1 1 1 This is a linear equation in demand and supply variables which may be rewritten as: 1b = + (1b) P, = CX, + CX Such price equations appeared in Feldstein (1970), Newhouse (1970), and Kehrer and Knowles (1974). B. PRICE EQUATIONS DERIVED FROM MONOPOLY CONDITIONS Some researchers have developed price equations based upon monopolistic market structures in which physicians act as profit maximizers. Under these assumptions, a price equation can be derived as follows: cost =C=vy +yQ+ vy Q? 0 1 2 where Q is the quantity of services sold by the physician. Further, let the quantity demanded at a price (p) be defined by: demand £ Q = Bp + B R, <0 1 2 The inverse demand function is defined as: p = ~1/8 (8x7) = aX + aQ 1 2 1 11 “fF Define profit (m) as revenue minus cost: Profit = mw = (p*Q) = C Profit maximization occurs whenever: an/aQ = (p + aQ) - 3C/3Q = 0 If cost (C) is linear (see Sloan, 1976 and Newhouse, 1970) profit maximi- zation requires: an/aQ=p+aQ-y =0 1 and the reduced form price equation is: (2a) p = (a /2)X + (1/2)y 2 1 1 1 which may be rewritten as: (2b) p =CX + (1/2)y 2 11 1 If cost (C) is quadratic, then marginal cost is linear -- e.g., Frech and Ginsburg (1972). If so, profit maximization requires: an/aQ = aQ + (aQ +a QX ) -y -2y yQ=0 1 1 1 2 The reduced form price equation is: 2 Gy - a) 3a sf X + 2 7 £2) Ps 2(a = vy) 1 2(a - Y,) l 2 As before, the price equation can be expressed as: (3b) =CX +¢C bp, 11 21 -64- In each of the three cases above, price (®,) is a linear function of demand variables x) and parameters of the cost function (y ). Sloan 1 (1976) defined these (y ) parameters to be linear combinations of various 1 t supply variables [e-8057, - a, + (4 +weges) + (d rents). In this way, 2 the three price equations reduce to linear combinations of demand and supply variables. C. IDENTIFICATION AND THE PRICE EQUATION As the three price equations stand, the parameters (a's, B's, y's) of the structural (i.e., demand, supply and cost) equations cannot be recovered from estimates of the reduced form parameters (C's). However, various econometrit¢ techniques such as indirect least squares and two stage least squares regression can be employed to obtain estimates of structural equa- tion parameters (see Fuchs and Kramer, 1972, for example). Such techniques usually require the imposition of various restrictions upon the model's structural equations. Generally, these restrictions are such that certain exogenous variables are excluded from the structural equations in order to "identify" these equations. Often, economic (as opposed to econometric) considerations limit the inferences which researchers can make from an analysis of reduced form price equations. For example, price equations (la) and (2a) above were developed under assumed competitive and monopolistic market conditions respectively. Yet, due to restrictions on the signs of variables in the structural demand, supply and cost equations, the coefficients of the x variable should have the same sign in both price equations. Hence, an examination of the regression of prices against 2 will not enable researchers - 5 to determine whether competition or monopoly is a more appropriate model of the physician's services market. Reinhardt, in his 1977 critique of Sloan and Feldman, similarly showed how a neoclassical model and a provider-induced demand model of the physicians' services market responded similarly to shifts in the physician-population ratio. That is, an observed decrease in price levels following an increase in the supply of physicians would not be incon- sistent with either the neoclassical or provider induced demand models. For other examples of the difficulty inherent in using reduced form price equations to differentiate between alternative market models, see Frech and Ginsburg (1972) and Newhouse and Sloan (1972). Even though reduced form price equations have not been particularly useful tools in attempts to discriminate among alternative market struc- tures, such equations are often of interest because of the insights they provide. From such equations, policy makers and researchers can observe the responses of physicians' prices to simultaneous changes in characteristics of physicians, patients and market conditions. It should be remembered, however, that cause-effect relationships usually cannot be readily inferred from these reduced form price equations. These equations only reflect the correlation between fees and market conditions. As the foregoing showed, the underlying market structure may not be uniquely determined from reduced form results. D. EMPIRICAL FINDINGS Many econometric studies of the price of physicians' services are based on the assumption that physicians maximize profits and that prices serve to clear the market. Hence, price equations in the literature are B= essentially similar in structure. Independent variables in these equations may be broadly classified as: (a) indicators of insurance coverage; (b) measures of physician density; (c) characteristics of physicians and their practices; (d) socio-economic characteristics of relevant populations; (e) measures of the availability of hospital services. Dependent variables in price equations are often measures of physicians’ usual fees or average revenue per visit. Neither is an entirely satisfactory measure of price. Usual fees, for example are reported by physicians for particular procedures. However, the bundle of services which compose a procedure may vary across physicians. Likewise, average revenue does not hold the mix and/or number of services performed per visit constant across physicians. It is also difficult to construct an accurate average revenue per visit series because precise measures of visits per physician are hard to obtain (see Sloan, 1976, for a more detailed discussion of problems associated with usual fee and average revenue measures). The empirical findings from various studies of the prices of physicians' services are summarized below. 1. Insurance Coverage Many studies of physician fees and revenues have employed measures of insurance coverage as explanatory variables. In general, insurance coverage variables display a strong positive statistical association with physicians' prices. A variety of insurance measures have appeared in the literature. They broadly may be classified as measures of (a) coinsurance rate, (b) Medicaid-Medicare, and (c) major medical coverage. -67- a. Coinsurance rate In a series of studies (1974, 1976) Newhouse and Phelps estimated price and utilization equations for various medical services. In their 1976 study they reported the elasticity of physician visit prices with respect to the coinsurance rate to be in the range -.21 to -.15. In their view, the latter result indicated that (p. 208) "an increase from no coverage to full coverage increases the price of the physician selected by 18 percent." b. Medicaid-Medicare Of these two, Medicare m2asures are most consistently positive and significant in price regressions. Using data from various AMA periodic surveys, Kehrer and Knowles (1974), Cantwell (1976) and Sloan (1976) generally found measures of Medicaid coverage insignificatnt in price equa- tions. Sloan, however, found significant, positive signs on a Medicare measure (Medicare expenditure per capita) in nine of eleven price equations. He reported elasticities of fees with respect to Medicare expenditures of around .10, while elasticities of physician average revenue with respect to Medicare were in the .60 to .75 range. Holahan et al. (1978) regressed physicians' billed charges per standard (CRVS) unit of physicians' services against a set of explanatory variables which included the Medicare and Medicaid reasonable charge levels. Coefficients on both reasonable charge variables were positive and significant, although the Medicaid fee coeffi- cients were smaller than those of the Medicare reasonable charge variable. Holahan et al.also reported a significant relationship between billed charges and a measure of the lagged rate of change of physicians' Medicare reasonable fees. This finding supported their theory that physicians have an understanding of the operations of the Medicare reimbursement system Be and its dependence upon past fee levels. Hence the Medicare system may have an inherent inflationary bias. One reason why Medicare coverage dis- plays a stronger statistical relationship to prices than does Medicaid coverage is that Medicare may present the physician with a greater opportunity to charge higher prices. As Gabel et al. (1976) state: '"Medicaid fees in many areas are substantially less than those paid by Medicare. Since physicians must accept the Medicaid payment as full payment for the service, there is no incentive for physicians to treate Medicaid patients as full pay patients in those areas." Cs Major Medical Coverage Steinwald and Sloan (1974) and later Sloan (1976) found significant, positive relationships between physicians' prices and the percentage of state population covered by major medical insurance. Sloan, for example, found that major medical coverage had a significant impact on general practitioners' office visit fees; the elasticity (evaluated at the mean) was 0.17. Steinwald and Sloan found positive and significant associa- tions between major medical coverage and fees charged by general practi- tioners, general surgeons and internists (but not pediatricians or obstetri- cian-gynecologists). In the significant cases, elasticities were in the .19 to .30 range. In Sloan's study, physician average revenue was much more responsive to major medical coverage than fees had been. He found major medical coverage elasticities of average revenue for G.P.'s and internists to be .48 and .52 respectively. Redisch et al. (1977) found that surgical specialists' gross incomes per encounter were negatively and significantly affected by the percent of the specialists' patients who were not covered by insurance. Surgical specialists’ hospital visit follow- up fees, however, were not significantly related to insurance coverage. «69 Redisch et al.reported similar results with respect to primary care physi- cians' usual office visit fees. In this case, fees were not affected by patients' insurance coverage. However, the gross incomes of primary care physicians were negatively (and significantly) related to the percent of the physicians' patients not covered by insurance. The consistently strong relationship between physicians’ average revenue and the extent of insurance coverage is worth noting, especially since revenue was much more responsive to insurance coverage than was the case for fees. A possible interpretation (suggested by Sloan, 1976) is that insurance coverage may encourage physicians to dis- aggregate such general services as "office visits' into component parts. Each component is then billed to the insurer separately. In this event, fees for particular services may remain stable while practice revenues increase. Separate billing for components also puts upward pressure upon measures of the quantity of physicians' services provided in the market. The aggregate revenue from insured practice may be as important as the extent of insurance coverage in determining physician fees. Kehrer and Knowles used the percentage of firm revenues derived from insured, fee-for-service practice as an independent variable in several price equa- tions. They found significant, positive coefficients on the revenue-from- insured service variable in regressions on general practitioner fees and fee indexes. For other specialty types the coefficient was generally posi- tive but not significant in fee regressions. Finally, in a study using national aggregate data, Feldstein (1970) reported a long run elasticity of price with respect to insurance coverage of .36. Lave (1978) lists three possible reasons for the positive effect of insurance on price: -70- (1) Price may increase because insurance increases demand in the face of a rising supply curve. (2) Price may increase because physicians who face excess demand now believe it is more reasonable for them to raise their fees. (3) Price may increase because of (insured) consumer- patients searching less for lower priced physicians. Each reason corresponds to a set of policy options which could dampen the price increases triggered by expanding insurance coverage. Unfortunately, data currently available do not enable policy makers to determine the causal mechanism behind the statistical association between price and insurance coverage. 2. Physician-Population Ratio The controversial positive relationship between physician density and fees was documented in an earlier section of this review. Newhouse (1970) reported (controlling for per capita income) a positive partial correlation (0.55) between the number of physicians per capita in cities and general practitioners' officevisit fees. In multivariate price equa- tions, however, the association between the physician-population ratio and physicians' prices is inconsistent. Physicians produce services which can be complements of (or sub- stitutes for) the services of other physicians practicing in the same region. For instance, two internists located in the same medical complex are likely to provide similar (perhaps substitutable) services. The members of a surgical team, however, are more likely to provide services complementary to each other. One might expect physicians' fees to be affected by a change in the wT local availability of physicians of the same (own) specialty type. Physi- cians' fees might be affected in a different way by a shift in the number of physicians in "other" specialty types practicing in the same region. Newhouse and Phelps (1974) reported "one of the most striking findings" was the positive elasticity (.40) between office visit prices and physicians per capita. Similarly, Dyckman (1978) estimated significant, positive coefficients for physicians-per-capita variables in several regression equations in which the dependent variable was a measure of metropolitan area surgeons' fees. When a similar set of regressions was run on surgeons' fees deflated by the cost of living, the coefficients of the physicians per 1,000 population variables were generally negative but never statistically significant. The Institute of Medicine (1976) also compared regression results from equations in which county Medicare fee indexes were adjusted for cost-of-living differences with those which were not adjusted for cost-of-living. In both cases, fee levels were signifi- cantly and positively related to physicians per capita. The elasticity of the unadjusted Medicare fee index with respect to physicans per 100,000 population was only .13, however. In several econometric studies, physician density is expressed in terms of "own-type'" or 'cross-type' physicians per capita. In other words, the dependent variable measuring fees of physicians in a particular specialty is regressed against the number of physicians per capita in the same (own) or different (cross) specialty. Sloan (1976) for example found a significant negative association between state aggregate fee levels (and revenues) of general practitioners and own type physicians per 10,000 population. He found that similar results held with respect to surgeons' appendectomy fees. Sloan's elasticity measures were small, however, (less than -0.2) except in J De G.P. average revenue equations where elasticities were around unity. Steinwald and Sloan (1974) found mixed associations between fee levels and own-type physicians per capita. A negative associatiion was found in fee equations for general practitioners and general surgeons. A positive association appeared in their analyses of internist, OB/GYN and pediatri- cian fees. The results of Holahan et al. (1978) support those of Sloan, and of Sloan and Steinwald. That is, Holahan et al.found a significant, negative association between G.P.'s billed charges and own-specialty physi- cians per capita. They found no significant relationship between billed charges and own-type physicians per capita in their analyses of internists or general surgeons. Finally, Redisch et al. (1977) reported that physicians’ fees and (especially) gross incomes per encounter were positively related to own type physicians per capita. Their reported elasticities of gross income per encounter (with respect to own-type physician density) were .9 for surgical specialties and .33 for primary care physicians. Similarly, fees of primary care physicians displayed an elasticity of .25 with respect to the number of primary care physicians per 100,000 population. Redisch et al.present their findings as evidence that physicians can maintain target income levels (in the face of rising physician-population ratio) via increased revenues per encounter resulting from changes in the mix of services they provide or via outright increases in fee levels. Steinwald and Sloan (1974) and Sloan (1976) included a measure of physicians in "other" fields per capita as an independent variable in their price equations. In both studies, surgeons' appendectomy fees were signifi- cantly and positively related to the density of "other" physician types within the region. In both studies, internists' fees and revenues were negatively related to the density of cross-type physicians. Steinwald and «7 Ge Sloan found G.P.'s office and hospital visit fees to be positively and significantly related to the local density of "other" physician types. On the other hand, Holahan et al. (1978) found a significant, negative coefficient on the "other physicians per capita" variable in their equation explaining G.P.'s billed charges per unit of service. Again, they found no significant relationship between "other physicians per capita" and the level of billed charges of internists or general surgeons. The Canadian experience with nationwide health insurance coverage was studied by Berry et al. (1978). In their report, they noted that rising physician-population ratios in Quebec did not appear to depress average gross payments per physician even though fee levels remained constant. While this phenomenon is consistent with the theory of physician-induced demand, Berry et al. also observed that it is consistent with "the hypothesis that Quebec Medicare, which provides first dollar coverage for physician services, led to large increases in patient demand for medical care, and that the increased utilization was merely the result of physicians fulfilling that demand." The foregoing results are especially significant in view of the increase in the physician-population ratio projected for the last quarter of this century. Apparently, the physician stock has a significant impact upon price and revenue levels, but the magnitude and direction of this impact is far from clear. Sloan's (1976) statement that "it would be useful to conduct additional tests on the influence of the physician-population ratio on fees" is a suggestion with considerable merit. 3. Characteristics of Physicians and Practices Physicians' fees may vary systematically with a host of character- istics reflecting the quality of physician services, the costs of operation and organization of the practice. i 7 a. Quality Measure Like other professionals, physicians are heterogeneous with respect to the skill and knowledge they possess. Hence, there exists a quality dimension to many physicians' services. To the extent that quality differentials are apparent to clients, doctors' fees may vary systematically with the perceived quality or skill level of the physician. Sloan and Feldman (1977) go so far as to suggest that the quality (and level of amenities) associated with physicians' practices varies directly with the physician-population ratio. Furthermore, they believe that fee levels reflect these quality-amenity variations. Hence, the often cited positive relationship between physicians' fees and physicians per capita is, in Sloan and Feldman's opinion, really a manifestation of the relationship between fees and quality-amenity levles. Unfortunately, direct measures of physicians' quality are not readily available. Hence, proxy variables must suffice as rough quality indicators. Measures such as specialty board certification, medical school faculty status, vintage (year license was received) and foreign medical training may correlate with the quality of a physician's services. Some (or all) of these measures were used by Steinwald and Sloan (1974), Sloan (1976), Kehrer and Knowles (1974) and Redisch et al. (1977) in studies of price setting. Specialty board certification frequently is positively associated with fee levels, although the association usually is weak. Faculty status, graduation from a foreign medical school, or recent licensure do not display consistent, significant associations with fee levels. b. Costs of Practice Operation Recent patterns of physicians' fee inflation reflect, in wg Gp part, the rising practice expenses which confronted physicians during the 1970's. As revealed by the data in Table III, practice expenses generally grew faster than practice total revenues between 1970 and 1974. In order to protect their income levels, physicians may have raised fees and/or altered the mix and volume of the services they provide. Dyckman (1978, p. 89) reported that in 1976 the wages paid to staff were the largest (43 percent) component of physicians' practice expenses. The next largest single item of expense was housing (10 percent), followed by private transportation (5 percent). If fees reflect expenses, then it is not unreasonable to expect fees to vary with personnel costs. Research consistently reveals a strong, positive association between staff wage levels and physicians' fees. Feldstein (1970), for example, found consistently positive elasticities of price with respect to an aggregate input cost measure. He felt this finding was counter to the belief "that an increased use of para- medical personnel and other inputs would restrain the increase in physicians’ fees." Steinwald and Sloan (1974) and Sloan (1976) included an estimate of wage levels for non-physician personnel as an explanatory variable in their price equations. Steinwald and Sloan found that "of all explanatory vari- ables, the ancillary personnel wage has the most definite impact on fees, judged both in terms of significance levels and elasticities associated with the parametric estimates." They reported elasticities of fees with respect to ancillary wages in the .45 to 1.16 range. Sloan (1976) found similar (.66 to .92) elasticities in his study of state average fee levels. Sloan also found elasticities on the order of .63 and .68 in physician's average revenue estimates. Kehrer and Knowles (1974) found no consistent relation- ship between fees and wages paid to allied health personnel. Cantwell (1976) Fe reported a positive relationship between practice expenses per visit and physicians' fees in metropolitan areas. Conversely, he found a negative relationship between expenses per visit and fees of physicians practicing in rural counties. Finally, Feldman (1975) found the number of waiting rooms per visit (a possible proxity for practice expenses) was positively related to the price of office visits. c. Organization of the Practice Physicians' fees may vary with the organizational form of physicians' practices. For example, group practices may be able to exploit scale economies and in this way decrease practice expenses per visit. If so, group practices would be in a position to charge lower fees per visit than are charged by doctors in solo practice. Conversely, cost sharing arrangements within physicians' groups may act as a disincentive with respect to efficient utilization of shared resources. To the extent that this is true, the costs of group practice (and hence fees) might increase. Steinwald and Sloan (1974) included a dummy variable in their fee equations to indicate that 50 percent (or more) of a physician's medical income came from practice arrangements in which revenues were shared with other physicians. In seven out of eleven fee equations, shared revenues were associated with lower fees. Redisch et al. (1977) reported that primary care physicians (belonging to groups which share expenses equally) and surgical specialists (practicing in groups which share expenses and revenues equally) both had significantly lower gross incomes per encounter than did solo practitioners. The same study showed that practicing in groups (which share incomes and expenses equally) had a significant, positive influence upon surgical specialists' hospital visit fees but had no significant effect upon primary care physicians' office «J Fe visit fees. Sloan (1976) developed explanatory variables which measured the percent of physicians in a state who were members of physician groups of particular sizes. The association between these variables and state average fee levels was mixed. Kehrer and Knowles (1974) found negative and signifi- cant relationships between fee levels of individual practices and the percent of local physicians who were members of group practice. However, they also reported that the size of the average group practice in a county relative to the size of the average practice in that county was consistently posi- tive and significantly associated with individual practices' fee levels. 4. Socioeconomic Characteristics A variety of economic and demographic characteristics of physicians and their clients appear as explanatory variables in price equations. These include various income measures, as well as the ages of physicians and their potential patients. A measure of patients' health status is usually absent, probably due to insufficient data. a. Income of Consumers A priori, one should expect a positive association between consumers' incomes and physicians' fees for two reasons. First, we know from demand studies that the income elasticity of demand for physicians’ services is usually positive and often significant. Hence, as incomes rise, the effective demand for higher priced physicians' services also rises. Second, it has been alleged that doctors frequently set charges using a fee scale based upon their patients' incomes (see Kessel, 1958; Rayack, 1971; Backman, 1953). Empirical estimates of the income elasticity of the price of physicians' services support this hypothesis that fees and incomes are related, as shown below in Table X. Positive and (usually) significant «7B TABLE X. INCOME ELASTICITY OF PHYSICIANS' FEES Source Elasticity Income Measure Newhouse (1970) «7 to .9 Income per Capita (cities) Feldstein (1970) .09 to .21 Median per Capita Income (U.S.) Phelps and Newhouse (1976) .072 to .075 Individuals' Reported Wages/Weeks Sloan (1976) .6 Per Capita Income (State) coefficients on regional per capita income measures also were reported by Cantwell (1976a) and by Kehrer and Knowles (1974) in their price equation estimations. Fees vary positively with the income levels of physicians’ patients as well. In several studies of physicians' fees, a measure of the percentage of the physician's patients with annual incomes over $15,000 was included as an independent variable. Cantwell (1976a) and Kehrer and Knowles (1974) reported consistently positive and usually significant coefficients on their "rich patients" variable. Steinwald and Sloan (1974) found positive coefficients on their "rich patients' measures in all but two (out of eleven) price equations. They suggested that the two negative coefficients indicate "that general practitioner-performed appendectomies and obstetrical deliveries may be inferior goods"; i.e., as their incomes rise, individuals will purchase fewer of these services from G.P.'s. Steinwald and Sloan also included a measure of the percentage of physicians’ patients with incomes under $5,000 per year in their fee regressions. This "poor patient" variable had a negative coefficient in all regressions and usually was significant. Finally, Steinwald and Sloan noted substantial -79- inter-specialty variation in the response of price to patients' incomes. They found that general practitioners' fees were less responsive to patients’ incomes than were the fees charged by other specialty types. b. Ages of Physicians and Patients The physician's age is used as an explanatory variable in many studies of fee levels (Kehrer and Knowles, 1974; Steinwald and Sloan, 1974; Sloan, 1976; Cantwell, 1976a). Fees tend to be negatively associated with the physician's age, but there is sufficient variability in the signs and significance of coefficients across studies to make the association tentative. Wise and Zook (1977) found the number of years elapsed since the physician graduated from medical school to be negatively and significantly related to the price of an office visit. Cantwell (1976a), Steinwald and Sloan (1974) and Dyckman (1978) included a measure of patients' ages in their fee analyses. Cantwell's study indicated a negative association between fees and the percentage of a physician's patients over 65 or under 15 years of age. Conversely, Steinwald and Sloan, as well as Dyckman,reported a positive association between fees and patients over 65 years of age. The affect (if any) of patients’ and physicians' ages upon fee levels is unclear. Age is undoubtedly correlated with levels of insurance coverage, personal income, health status, and the type of physician chosen, as well as with the treatment sought. Hence, a failure to adequately account for some of these correlated measures in price equations may be the reasons for incon- sistent signs on the '"age' variables. 5. Relationship Between the Hospital Sector and Physicians' Fees Some portion of the work-week for most physicians is spent within a hospital setting. While the extent of this in-hospital component of practice may vary across physician specialties and geographic areas, it is reasonable to -80- assume that aspects of the local hospital care system determine, in part, the character of the physician's practice. The precise nature of this rela- tionship, especially as it affects price setting in the physicians' services market, is not clear. As reported earlier, there is some evidence to suggest that physicians may take the availability of hospital facilities into consid- eration when making their choice of practice location. Dyckman reported (in his study of surgical fees) that physicians per population and surgeons per population were correlated, .75 and .71 respectively, with local hospital beds available per capita. While the local availability of hospital facilities may be an important factor in determining a region's supply of physicians, the relationship between availability of hospital facilities and physicians’ fees is not clear. That is, while a strong regional hospital sector may serve to induce in-migration of physicians, the relationship between the local physician-population ratio and local fee levels is uncertain. As the size and complexity of the local hospital system increases, it seems reasonable to assume that the total supply of hospital services available to local physicians ought to increase. If expanding the supply of complementary hospital services enables physicians to increase their patient load, then the average revenue which must be extracted from each patient to meet the physician's income target (assuming one exists) should decline. However, the average charge and other aspects of the physician's hospital based practice depend upon more factors than simply the availability of hospital services. Redisch (1978) suggests that three administrative aspects of the hospital sector affect physicians' practices. These features are: (a) the separation of physician and hospital bills for jointly produced health care; wl le (b) the proration of basic hospital service costs over all patients; and (c) the pervasiveness of insurance for hospital services. Redisch believes these features cause physicians to underestimate the true (social) cost of marginal units of hospital care. Hence physicians may employ economically inefficient quantities of hospital services in their practices. It is still uncertain, however, how such potentially inefficient input combinations are related to physicians' fees and revenues. Most econometric studies of physicians' fees have not attempted to assess the effect which the local hospital sector may have upon fees. In those studies where the effect of the hospital sector is considered, this sector usually is represented by an indicator of the local availability of hospital beds per capita (see Dyckman, 1978; Redisch et al., 1977; Wise and Zook, 1977). This variable (beds per capita) has at least one unsatis- factory characteristic. As Dyckman showed, if beds-per-capita is highly correlated with physicians-per-capita, then inclusion of both variables in regression analyses may result in unstable coefficient estimates. In Dyckman's analysis of average surgical fee levels, significant and negative coefficients were estimated for the beds-per-capita variable whenever city-size and regional indicator variables were excluded. When these latter variables were included, however, coefficients on hospital beds per capita were no longer significant. Furthermore, when surgical fees were deflated by a cost of living adjustment, and regional dummy variables were included, the coefficients of the beds per capita became positive, but were not statistically significant. Redisch et al, (1977) found little statistical association between various aspects of primary care physicians' practices and the availability of hospital beds. They -82- did report a significant, negative relationship between beds per capita and the gross income per encounter for surgical specialists, the relevant elasticity being -.76. They had expected, however, that greater hospital bed capacity would result in an increased volume of surgery which, in turn, would yield increased revenues per encounter. Hence, the significant negative coefficient on the beds-per-capita measure was unexpected. Redisch et al also reported a positive elasticity (p. 18) of surgical specialists’ gross incomes with respect to hospital beds per capita. Finally, in Wise and Zook's reduced form equations, a beds-per-capita variable was positively but not significantly related to physician's charges for routine office visits deflated by cost of living. At least two other studies employed other measures of the hospital sector in an attempt to explain the pricing behavior of physicians. Muller and Otelsberg (1978) found the number of ambulatory visits per hospital bed to have a positive and significant coefficient in regression analyses of fee indexes and expenditures by insured patients. On the other hand, in their study of fees, Kehrer and Knowles found generally insignificant coefficients on the ratio of the number of practice visits produced in the hospital to the number of visits produced in the office by physician firms. E. GENERAL CRITICISMS OF ECONOMETRIC PRICING MODELS Some general criticisms can be made of the empirical research described above. First, most of the models cannot describe the dynamic reaction of the market to disequilibrating forces. Even if the market does move toward equilibrium after exogenous shifts in (for example) the physician population ratio, policy makers may be as much interested in how a new equilibrium is reached as in what the new equilibrium will look like. Second, measurement ~83~ problems associated with limitations on the data available to researchers seriously limit the confidence with which one may draw conclusions with respect to the physician services market. In most studies, actual fee and procedure data are not used. Rather, reported (hence, possibly biased) "average" fee levels from surveys of physicians are used by researchers. Also, the use of aggregate measures of such variables as physicians per population, and levels of insurance coverage may produce results which are not supported when these measures are replaced by their disaggregate counter- parts. For example, when Steinwald and Sloan disaggregated the physician- population ratio into "within field" and "across field" components, they found different effects for general practitioners than for specialists in their fee equations. Finally, quality differences often are alluded to by researchers as a possibly significant factor in fee variation. However, no reliable measures of physician quality are available as yet. -84- VII. CONCLUSIONS In light of the foregoing survey of the current literature relevant to the price of physician services, several inferences may be drawn with respect to future levels of physicians' fees. It is likely, ceteris paribus, that doctors' prices will continue to increase at rates exceeding the rates of increases in both the prices of other services and prices overall. This assertion reflects not only recent trends in physicians' fees but also the apparent insensitivity of demand, in this market, to price increases. American medical care will be affected by two significant forces over the next quarter century. These are: (a) the expected growth in the supply of physicians per capita, and (b) the (potential) further expansion of medical insurance coverage under some form of national health insurance. While current models of the physician services market are inadequate to predict the precise effect these forces will have upon price levels, empirial evidence does exist with which one can make some conjectures. The market for physicians' services does not seem particularly respon- sive to '"mormal" supply and demand pressures; hence, it is not clear (in theory) how the projected increases in physicians per capita will affect prices. Some empirical evidence suggests that output per physician decreases as the local supply of doctors increases. Also, some data show that physicians' fees may increase when the local supply of physicians increases. Physicians' fees also seem positively related to the level of insurance coverage. This is especially true when the insurance coverage permits -85- physicians to charge patients for that portion of fees not paid for by the insurance carrier. Hence, the potential exists for an inflationary trend in physicians' fees under national health insurance unless some control measures to mitigate fee inflation are included in the plan. Alternative forms of medical care delivery may relieve inflationary pressures on doctors' prices, but the extent to which this would be true in practice is not clear. Available data indicate that physician producti- vity can be significantly increased if more aides (on the average) are employed in the physician firm. However, such a condition also seems posi- tively related to higher fee levels. Physicians belonging to a revenue sharing group apparently are able or willing to charge lower prices for their services. However, group practice was not always associated with lower fee levels. In general, not enough is known about the way physicians set prices to infer what will happen to fee levels if a substantially larger proportion of physicians adopts this form of practice. It is apparent from the literature that much work remains to be done with respect to modeling the physicians' services market. Although there exists a growing body of empirical work relating physicians' fees (and revenues) to a variety of explanatory variables, these results are only preliminary. As richer data sets become available, it should be possible to test alternative simultaneous equation models of the physician services market. With these more complex models will hopefully come increased insight into the mechanism by which physicians' prices are set. ~B6~ APPENDIX RATES OF CHANGE IN UTILIZATION DERIVED FROM CHANGES IN EXPENDITURES AND PRICES DEFINITIONS F = CPI physicians' fee index in year t. F_ -F 1 AP = tt tl proportional change in fee index between F.1 years t-1 and t. bv, = utilization of physicians' services in year t. Ug =~ Uy AU = 9 = proportional change in utilization between t t-1 years t-1 and t. E = expenditures on physicians' services in year t. E -E 1 AE = —t—t = proportional change in expenditures on t t-1 physicians' services between years t-1 and t. DERIVATION OF RATE OF CHANGE IN UTILIZATION Since expenditures reflect a volume of services purchased at a given price: BF (2) (Uy; + Fp_p)* (AU) (AF) + U_;+(8F)= Fy, + Fa ’ 03 00) * Fea ’ Up-1’ Equation (2) can be expressed (substituting E_ for u,2,) as: E, = E._; (BU*OF) +E; (OF) +E; (AU) +E, which can be reformulated as t t-1 (3) = AE (8U, + AF) + (AF) + AU, The proportional change in utilization of physicians' services between years t-1 and t therefore can be inferred from reported changes in price levels and expenditures: AE, - AF, AU, 1+ AF.) or AE_ - AF (4) ay, = —t___ Et 1+ AF, -88- REFERENCES Abt Associates Inc. "Physician Survey on Administrative Costs and Medicaid Participation: Literature Review,'" Cambridge, Massachusetts: Abt Associates Inc. 1976. Mimeo. Acton, Jan P. "Demand for Health Care Among the Urban Poor, with Special Emphasis on the Role of Time," in The Role of Health Insurance in the Health Services Sector, R.N. Rosset (ed.). New York: National Bureau of Economic Research, 1976, 165-208. Acton, Jan P. ''Nonmonetary Factors in the Demand for Medical Services: Some Empirical Evidence," Journal of Political Economy, 83:3 (1975), 595-614, American Medical Association. Reference Data on Profile of Medical Practice, Chicago: American Medical Association, 1973, 1976. Andersen, Ronald and Benham, Lee. 'Factors Affecting the Relationship Between Family Income and Medical Care Consumption," in Empirical Studies in Health Economics, H.C. Klarman (ed.). Baltimore: The John Hopkins Press, 1970. Arrow, Kenneth J. "Uncertainty and the Welfare Economics of Medical Care," American Economic Review, 53 (December 1963): 941-973. Backman, Jules. "Professional Fees: Factors Affecting Fee Setting in the Several Professions,'" Journal of Accountancy, 95 (May 1953), 554-66. Barzel, Yoram. "Productivity and the Price of Medical Services," Journal of Political Economy, 77 (November-December 1969), 1014-1027. Baumol, W.J. Economic Theory and Operations Analysis, Englewood Cliffs, New Jersey: Prentice Hall, Inc., 1965. 311-337. Benham, L., Maurizi, A., and Reder, M.W. "Location and Migration of Medics: Physicians and Dentists,'" Review of Economics and Statistics, 50 (August 1968), 332-347. Berry, C., Brewster, J.A., Held, P., Kehrer, B. Manheim, L, and Reinhardt, U. Study of the Responses of Canadian Physicians to the Introduction of Universal Medical Care Insurance: The First Five Years in Quebec, Princeton, New Jersey: Mathematica Policy Research Inc., February 28, 1978. Brown, Douglas M. and Lapan, Harvey E. "The Rising Price of Physicians’ Services: A Comment ,'" The Review of Economics and Statistics, LIV (February 1972), 101-105. Cantwell, J.R. "A Cross-Section Econometric Analysis of Physician Location and Pricing," a paper presented at the Medical Economics Session, Midwest Economics Association Meetings, April 3, 1976a. Cantwell, J.R. "Physician Fee Variation and Reimbursement in California and Georgia," a paper presented at the Medical Economics Session, Western Economic Association 1977 meetings, Anaheim, California: June 22, 1977. «80 Cantwell, J.R. "Physician Fee Variation in Illinois," a paper delivered at the Rockford School of Medicine First Annual Illinois Health Care Research Symposium, November 11, 1976b. David, K.A. and Reynolds, R. "The Impact of Medicare and Medicaid on Access to Medical Care," in The Role of Health Insurance in the Health Services Sector, R.N. Rosett (ed.) New York: National Bureau of Economic Research, 1976. 391-425. David, K.A. and Russell, L.B, "The Substitution of Hospital Outpatient Care for Inpatient Care,’ The Review of Economics and Statistics 54:2 (May 1972), 109«120. Dunlop, D. and Zubkoff, M. "Inflation and Consumer Behavior in the Health Care Sector,'" in Health: A Victim or Cause of Inflation, M. Zubkoff (ed.) New York: Milbank Memorial Fund, 1976, 84-114. Dyckman, Zachary, Y. A Study of Physicians' Fees, Staff report prepared by the Council of Wage and Price Stability, Executive Office of the President, Washington, D.C.: March 1978. Eisenberg, Barry, "Trends in Physicians' Incomes, Expenses and Fees, 1969-74," in Reference Data on Profile of Medical Practice 75-76, J.A. Cantwell (ed.). Chicago: The American Medical Association, 1976. Eisenberg, B. and Cantwell, J. "Policies to Influence the Spatial Distribution of Physicians: A Conceptual Review of Selected Programs and Empirical Evidence," Medical Care 14:6,(June 1976) Evans, Robert G. Price Formation in the Market for Physicians' Services in Canada, 1957-1969. Study prepared for the Prices and Incomes Commission, Canada, 1972. Ottawa, Canada: Information Canada, 1973. Evans, R. G. "Supplier-Induced Demand: Some Empirical Evidence and Implica=- tions," in The Economics of Health and Medical Care, M. Perlman (ed.). New York: John Wiley and Sons, 1974. 162-173. Evans, R. G. (a review of) The Economics of Health and Medical Care, Mark Perlman (ed.) in The Canadian Journal of Economics, 9:3 (August 1976), 532-537. Feldman, R. "Price and Quality Differences in the Physicians' Services Market." Paper presented at the Dallas meetings of the Allied Social Science Association, December 29, 1975, Mimeo. Feldstein, Martin S. "The Rising Price of Physicians' Services,'" The Review of Economics and Statistics, LII (May 1970), 121-33. Feldstein, Martin S. "The Rising Price of Physicians' Services: A Reply," The Review of Economics and Statistics, LIV (February 1972), 105-107. -90- Feldstein, P.J. Health Associations and the Demand for Legislation: The Political Economy of Health Cambridge, Massachusetts: Ballinger Publishing Co. 1977. Ferguson, C.E. Microeconomic Theory, Homewood, Illinois: R.D. Irwin Inc., 1972. Frech III, H.E. and Ginsburg, Paul B. "Physician Pricing: Monopolistic or Competitive: Comment," The Southern Economic Journal, 38 (April 1972), 573-77. Fuchs, Victor R. and Kramer, Marcia J. Determinants of Expenditures for Physicians' Services in the United States, National Center for Health Services Research and Development. Washington, 1972. Gabel, J., Blaxall, M.0., Burney, I. and Schieber, G., "Paying the Physician: Some Lessons from the Medicare-Medicaid Experience," paper presented at the American Public Health Association Annual Meeting, Miami Beach, Florida, October 17-21, 1976. Mimeo Garbarino, J.W. "Price Behavior and Productivity in the Medical Market," Industrial and Labor Relations Review 13 (October 1959), 3-15. Gibson, R.M, and Mueller, M.S. "National Health Expenditures, Fiscal Year 1976," Social Security Bulletin (April 1977). Grossman, Michael, "The Demand for Health: A Theoretical and Empirical Investigation,’ New York: Columbia University Press, 1972. Haddock, Douglass, "Set Proper Charges on the Basis of Your Costs," Medical Economics, (October 28, 1968), 75-80. Hadley, Jack and Lee, Robert. Physicians' Price and Output Decisions: Theory and Evidence, Working paper 998-8, The Urban Institute, Washington, D.C., April, 1978. Holahan, J., Hadley, J., Scanlon, W., Lee, R. and Bluck, J.,Physician Pricing in California: Executive Summary working paper 998-10, The Urban Institute, Washington, D.C., April, 1978. Huang, Lien-fu in The Role of Health Insurance in the Health Services Sector, R.N. Rosett (ed.). New York: National Bureau of Economic Research, 1976, 354-359. Institute of Medicine, Medicare-Medicaid Reimbursement Policies: Part III, study supported by U.S.D.H.E.W. contract number SSA-PMB-74-250. National Academy of Sciences, Washington, D.C., July 1976. Kehrer, Barbara H. and Knowles, James C. "An Econometric Analysis of Prices Charges for Medical Services," Center for Health Services Research and Development, American Medical Association, September 1974. Mimeo. = Kessel, Reuben A. "Price Discrimination in Medicine," Journal of Law and Economics, 1 (October 1958), 20-53 Kessel, Reuben A. "The A.M.A. and the Supply of Physicians," Law and Contemporary Problems, XXXV (Spring 1970), 267-83. Lave, J.R. "Forecasts of Physician Prices and Expenditures 1978," paper prepared for Health Care of American Economy: Issues and Forecasts, a conference sponsored by the Blue Cross and Blue Shield Association in Miami, Florida: January 15-18, 1978. Mimeo. Lave, J.R. and Lave, L.B. and Morton T.E. "Sources and Uses of Paramedical Personnel," in Incentives and Planning in Social Policy, B. Stein and S. Miller (ed.). Chicago: Aldine, 1973. 37-70. Leveson, I. "The Demand for Neighborhood Medical Care," Inquiry, 7:4 (December 1970). Masson, Robert T. and Wu, S. "Price Discrimination for Physicians' Services," Journal of Human Resources, IX (Winter)1974), 63-79. Monsma, George N., Jr. "Marginal Revenue and the Demand for Physicians’ Services," in Empirical Studies in Health Economics. Herbert K. Klarman, (ed.) with the assistance of Helen H. Jaszi, Proceedings of the Second Conference on the Economics of Health, Baltimore: The Johns Hopkins Press, 1970. Newhouse, Joseph P. "A Model of Physician Pricing," The Southern Economic Journal, 37 (October 1970), 174-83. Newhouse, J.P. and Phelps, C.E. "New Estimates of Price and Income Elasticities of Medical Care Services," in The Role of Health Insurance in the Health Services Sector, R.N. Rosett (ed.). New York: National Bureau of Economic Research, 1976. 261-313. Newhouse, J.P. and Phelps, C.E. "Price and Income Elasticities for Medical Care Services," in The Economics of Health and Medical Care, Mark Perlman (ed.).New York: John Wiley and Sons, 1974, 139-161. Newhouse, Joseph and Sloan, F. "Physician Pricing: Monopolistic or Competi- tive: Reply," The Southern Economic Journal, 38 (April 1972), 577-80. Pauly, Mark '"Is Medical Care Different?" paper presented at Federal Trade Commission Conference, January 1-2, 1977. Mimeo. Pauly, Mark V. "The Role of Demand Creation in the Provision of Health Services," a paper presented at the Annual Meeting of the American Economic Association, Dallas, Texas: December 30, 1975. Mimeo. Phelps, C.E. and Newhouse, J.P. "Effect of Coinsurance: A Multivariate Analysis," Social Security Bulletin, (June 1972), 20-28. "om Rayack, Elton, "The Supply of Physicians' Services," Industrial and Labor Relations Review, 17 (January 1964), 331-237. Rayack, Elton, 'The Physicians' Service Industry," in The Structure of American Industry, New York: Macmillan Co., 1971, 419-56. 4th Edition, Walter Adams (ed.). Reder, M.W, "Some Problems in the Measurement of Productivity in the Medical Care Industry," in Production and Productivity in the Service Industries, V.R. Fuchs (ed.). National Bureau of Economic Research, Studies in Income and Wealth, Vol. 24, New York: Columbia University Press, 1969. Redisch, M., Gabel, J. and Blaxall, M. "Physician Pricing, Costs and Income," paper prepared for Western Economic Association Meetings, Anaheim California: June 20, 1977. Mimeo. Reinhardt, U.E. "A:Production Function for Physicians' Services," Review of Economics and Statistics 54 (February 1972a) 55-56. Reinhardt, U.E. "Health Manpower Policy in the United States," paper presented to the Bicentennial Conference on Health Policy, University of Pennsylvania: November 11-12, 1976. Mimeo. Reinhardt, U.E. "Parkinson's Law and the Demand for Physicians' Services," comment on 'Monopolistic Elements in the Market for Physician Services," a paper presented by Frank Sloan and Roger Feldman to the Federal Trade Commission's Conference on Competition in the Health Care Sector, Washington, D.C.: June 1-2, 1977. Mimeo. Reinhardt, U.E. Physician Productivity and the Demand for Health Manpower, Cambridge, Massachusetts, Ballinger Publishing Co., 1975. Ruffin, R.L. and Leigh, D.E. "Charity, Competition, and the Pricing of Doctors' Services,'" Journal of Human Resources, 8 (Spring 1973), 212-222, Sloan, F.A. "A Microanalysis of Physicians' House of Work Decisions, in The Economics of Health and Medical Care, Mark Perlman (ed.), New York: John Wiley and Sons, 1974. Sloan, F.A. "Physician Fee Inflation: Evidence from the Late 1960's," in The Role of Health Insurance in the Health Services Sector, R. N. Rossett (ed.).New York: National Bureau of Economic Research, 1976, 321-354, Sloan, F.A. "Physician Supply Behavior in the Short Run," Industrial and Labor Relations Review, 28:4 (July 1975), 549-569. Sloan, Frank and Feldman, Roger 'Monopolistic Elements in the Market for Physicians' Services," paper prepared for presentation at a conference on "Competition in the Health Care Sector: Past, Present, and Future," Washington, D.C.: June 1-2, 1977. Steinwald, Bruce, and Sloan, Frank A. "Determinants of Physicians' Fees," Journal of Business (October 1974), 493-511. -93- Tharp, Ralph "Set Your Office Fees by a Time Clock? We Do:" Medical Economics (September 1, 1975), 74-76. Theodore, C.N. and Warner, J.S. ""Physicians' Fees and the Source of Their Increase," in Health: A Victim or Cause of Inflation?, Michael Zubkoff (ed.). New York: Milbank Memorial Fund, 1976, 149-163. Vahovick, S.G. "Physicians' Supply Decisions by Specialty: 2SLS Model," Industrial Relations, 16:1 (February 1977), 51-60. Walter, A.A. "Production and Cost Functions: An Econometric Survey," Econometrica 31 (January-April 1963), 1-66. Wise, D.A., and Zook, C. "Physician Responses to the Demand of Their Services: The Characteristics of Office Practice,' report performed under “contract number HRA-231-75-0023 to Mathematics Policy Research from the Health Resources Administration, U.S.D.H.E.W. (August 1977). 7 - # U. S. GOVERNMENT PRINTING OFFICE : 1979 281-200/506 Win (029133956