V.  Physician Compensation

Physician compensation plays an important role in the functioning of the physician labor market.  Not only do earnings reward physicians for their training and services, but earnings provide both a signal and an economic incentive that affects the decision to enter medicine as a profession, choice of medical specialty, choice of practice location, and level of work effort.

This chapter focuses on four compensation issues: (1) What are the historical trends in physician compensation?  (2) What are the key determinants of compensation?  (3) What are the expected future trends in physician compensation?  (4) How will trends in compensation affect the physician supply?  In Chapters VI and VII , respectively, compensation issues are discussed as related to physician gender and race/ethnicity.

A.   Historical Trends in Physician Compensation

AMA statistics show a steady growth in real, mean income between 1982 and 1993 for physician in most specialties, but that during the mid-to-late-1990s mean earnings tended to decline (Exhibit 54).  Unfortunately, AMA no longer publishes information on mean earnings, and earnings data from other sources are not necessarily a representative sample of physicians. During the 1980s and 1990s the higher-income specialties (e.g., surgery, obstetrics/gynecology, and radiology) tended to see the largest volatility in mean earnings, while the lower-income specialties (e.g., psychiatry, pediatrics and general/family practice tended to have relatively constant mean earnings.

Mean income tended to be highest for physicians practicing in metropolitan areas with less than 1 million in population, followed by physicians practicing in metropolitan areas with greater than 1 million in population (Exhibit 55).  Physicians practicing in non-metropolitan areas tended to have consistently lower mean income.  These differences in mean income do not control for differences in specialty composition, cost of living, and other determinants of income (e.g., patient care hours worked).

Mean income tended to be systematically correlated with physician age (Exhibit 56).  Mean income increases with physician age up through age 55, at which time mean income for the age group falls.  Much of this decline for older physicians is attributed to reduced work hours.

Exhibit 54. Mean Net Income After Expenses and Before Taxes: by Specialty

[D]

Source: AMA Physician Socioeconomic Statistics (various years).

Exhibit 55. Mean Net Income After Expenses and Before Taxes: by Metropolitan Location

[D]

Source: AMA Physician Socioeconomic Statistics (various years).

Exhibit 56. Mean Income After Expenses and Before Taxes: by Physician Age

[D]

Source: AMA Physician Socioeconomic Statistics (various years).

A comparison of mean physician earnings to Bureau of Labor Statistics (BLS) employment cost indices for the health services sector, the civilian sector, and executive positions finds that during the 1990s physician earnings fell by 10 to 15 percent relative to earnings in other occupations (Exhibit 57).  During the 1990s, mean physicians’ earnings fell by 5 percent in inflation-adjusted terms.  During this time mean earnings in the health services sector tended to rise slightly, while mean earnings of executives (an occupation group that is perhaps comparable with physicians in terms of attracting bright and talented people) increased by approximately 10 percent in real terms.

Exhibit 57. Change in Mean Income Compared to Employment Cost Indices: 1990 to 2000

[D]

Sources: AMA Physician Socioeconomic Statistics. Employment Cost Index data obtained from the Bureau of Labor Statistics.

While market factors such as increased managed care in the 1990s can explain part of the decline in mean earnings for physicians, changes in the demographic and specialty composition of the physician workforce also help explain this trend.  Trends in physician earnings can be better understood by analyzing the determinants of physician compensation.

B.  Determinants of Physician Compensation

Payment rates for physician services reflect the pecuniary value that society places on physician services.  Rates reflect the value of physician time and expertise, so physicians in specialties that require more years of training and for which there are few substitutes (e.g., surgical specialties) tend to earn more than physicians in primary care specialties.  An analysis of AMA’s 1998 SMS data finds substantial variation in physician pay (Exhibit 58).  Annual earnings is positively (and statistically significantly) associated with working more hours, being male, being a U.S. medical graduate, being in the middle of one’s career, working in a multi-physician practice versus solo practice, and being board certified.  There is also substantial regional variation in earnings.

Exhibit 58. Regression Analysis of Physician Earnings

Source: Analysis of the 1998 AMA Socioeconomic Monitoring System File.

C.  Expected Future Trends in Physician Compensation

Three trends will likely affect physician compensation over the next couple of decades: (1) the aging of the population and the greater reliance on Medicare as a source of revenue, (2) increasing competition from NPCs, and (3) efforts by insurers to make patients and physicians more cost conscious.

Population Aging and Medicare

The aging of the population means that a larger proportion of physician revenue will come from Medicare, so physician earnings and practice patterns will become increasingly tied to Medicare policies and reimbursement formulas.  AMA’s (1999) Physician Socioeconomic Statistics reports that in 1998 an estimated 29 percent of revenue received by non-Federal physicians came from Medicare, 12 percent came from Medicaid, 43 percent came from private insurance, 12 percent came from patient out-of-pocket payments, and 4 percent came from other sources.  BHPr (2003) estimates that between 2000 and 2020, the proportion of physician services that are provided to patients age 65 and older will grow from approximately 32 percent to 39 percent.

In an era of increasing Federal debt and large, projected increases in the number of Medicare and Social Security beneficiaries, there will likely be strong pressure to hold down medical costs—potentially affecting physicians through downward pressures on reimbursement rates and utilization of some services. 

Exhibit 59. Average Percent of Non-Federal Physician Revenue by Payment Source 1998

Source: Physician Socioeconomic Statistics, 1999-2000 Edition (AMA, 1999).

Exhibit 60. Estimated Percentage of Physician Time Spent Providing Care to Patients Age 65 and Older

Source: BHPr (2003).

Competition from Non-Physician Providers

As discussed in Section III.C the increased supply of NPCs could reduce the demand for physicians to provide some services (especially primary care services).  Because NPCs can provide some services currently offered by physicians but at a lower cost, there exists an economic incentive for health care groups to use NPCs to provide services within their legal scope of practice.

Increased Cost Consciousness

Health practitioners, patients, policymakers, researchers and others have long understood that insurance markets can create inefficiencies in the health care system because patients and physicians are partially shielded from the true cost of their health care decisions.  The increased use of managed care principles in the 1990s (e. g, the use of gatekeepers, exclusive networks, utilization review and capitation) were driven largely by the belief that the health care system could be more efficient if physicians bore a greater burden of the costs of their decisions, either through financial incentives or constraints on prescription of services.

The backlash against the most restrictive forms of managed care has forced payers to find other ways to control costs, and in recent years the trend has been to make patients more cost-conscious by shifting a greater portion of the cost of care onto patients.  This is occurring through increasing deductibles and co-payments, as well as efforts to inform patients of the true cost of their health care utilization.  Greater cost-consciousness on the part of patients could result in decreased utilization of physician services which could have a negative impact on physician earnings.

D.  Supply Implications of Trends in Physician Compensation

As discussed in Chapter II , research shows that physicians respond to financial incentives regarding (1) the decision to enter medicine as a profession, (2) choice of medical specialty, (3) practice location, (4) number of hours working to provide patient care, and (5) decision when to retire.  Research suggests that physicians also react to market forces that affect earnings by changing the way they practice medicine.

Practice Patterns

Farber and Murray (2001) report findings from a 2000 survey of the economic situation of physicians.  Over half the survey respondents indicated that they have taken or plan to take one or more of the following actions to counter lagging income (Exhibit 61).  Cutting practice expenses was the most popular choice (53 percent), followed by expanding services (34 percent), renegotiating contracts (32 percent) and hiring NPs or PAs (19 percent).  While 15 percent of respondents indicated they planed to cut back on their HMO practice, 7 percent indicated they plan to increase their HMO practice.  The number of respondents indicating they plan to change from solo to group practice (6 percent) equaled the number indicating they plan to change from group to solo practice (6 percent).  Likewise, the number indicating they would seek to affiliate with a hospital (6 percent) equaled the number indicating they would like to end an existing hospital affiliation (6 percent).  Relocation was an option being considered by 14 percent of survey respondents, with 3 percent of physicians considering a change in specialty.

Exhibit 61. Percentage of Physicians Indicating They Have or Plan to Take the Following Actions

[D]

Source: Medical Economics 2000 Financial Survey, Farber and Murray (2001).

Hours Worked and Income Tax Rates

Discussions of ways to reduce the growing national debt often include the topic of income taxes.  Most physicians are in the highest marginal income tax brackets so changes in income tax rates can have a significant impact on after-tax earnings.  Economic theory suggests that a rise in income tax rates could have conflicting implications for physician supply.  The substitution effect suggests that raising tax rates will reduce the benefits of working additional hours, so raising income tax rates could have a negative impact on physician supply by reducing the incentive to work long hours.  Countering this is the income effect--some physicians might choose to work longer hours to make up for income lost to higher taxes.

Showalter and Thurston (1997) find that higher marginal tax rates reduce the number of hours physicians work, on average.  The reduction in hours occurs mainly among self-employed physicians, which suggests that changes in physician hours worked could become less sensitive to income tax policy over time if a greater proportion of physicians become employees versus self-employed.  Using data from the 1983-1985 Physicians’ Practice Costs and Income Survey and State variation in income tax rates, these authors estimate a 17 hour reduction in annual hours worked for each 1 percentage point increase in a State’s top marginal tax rate.  Thus, a 10 percentage point increase in the top tax bracket would reduce average annual hours worked by approximately 170 hours per year (or approximately 7.5 percent of average hours spent in direct patient care).  Presumably, physicians’ labor supply response to changes in income tax rates would be similar for both changes in State and Federal tax rates (although changes in State tax rates could result in some cross-State migration).

Specialty Choice

Systematic differences across specialties in mean earnings and work behavior reflect differences in training costs (e.g., length of residency), nature of the work (e.g., time commitments, job stress), market conditions (e.g., demand for services), and perhaps differences in physician ability.  An important question for physician workforce modeling is “How responsive is specialty choice to expected future earnings?”

To determine the extent to which market forces and economic factors affect specialty choice, we investigated whether there are “specialty premiums.” We define specialty premiums as systematic differences across specialties in physician income after controlling for expected returns to training and differences in the characteristics or practice decisions of physicians (e.g., number of hours worked).  An earnings premium in a particular specialty provides a financial incentive for new medical graduates to enter that specialty.  Theory suggests that while there might be short-run premiums for a particular specialty, in a market economy with good information on physician earnings the long-run premiums should be similar across specialties.

Earlier studies of the returns to medical training (e.g., Sloan and Feldman, 1978; Burnstein and Cromwell, 1985) estimate the returns to training for physicians compared to other occupations. We modify their approach to estimate returns to training in a particular medical specialty versus the returns to becoming a physician in general.  Training costs and future earnings are compared starting with the first year of residency and ending at age 65 (the assumed age of retirement).  We use general and family practitioners as the reference category for determining the opportunity cost associated with training in another specialty and estimate the return on training for nine specialty groups.

The expected returns to training are determined by the amount and timing of expected future earnings and training costs.  To compute the returns to specialty training costs and future expected earnings are discounted to estimate the present value (PV) of net earnings for a particular specialty. 

Any differences across specialties in returns to training are determined by differences in the internal rate of return, and the internal rate of return is that rate which makes the present value in the above equation equal to zero.  To estimate present value of expected future earnings, current actual mean earnings are used for physicians in different age groups.  The returns to specialty training both with and without adjusting for differences are estimated across specialties in average hours worked per year, but only the adjusted numbers are reported (Exhibit 62).

Although physicians in surgical specialties earn substantially more than general and family practitioners, surgical specialties require additional years of training and thus a delay in significant earnings.  Consequently, taking into account the additional costs of training (such as foregone earnings during residency) and differences in hours worked the returns to training in surgery are approximately 3 to 5 percent.  The returns to training for radiologists and anesthesiologists are the highest—ranging from 4 to 7 percent.  The returns to OBGYN training declined during the 1990s from 5 percent down to 2 percent.  The returns to pathology and internal medicine were in the 1 to 2 percent range, and specialization in psychiatry is associated with a negative rate of return compared to entering general and family practice. 

These estimated premiums are not adjusted for physician geographic location, and the fact that some specialists predominantly locate in higher-cost, metropolitan areas results in an overestimate of the earnings premium for those specialties.   Also, the estimates do not account for differences across professions in attributes that could require compensating differentials (e.g., job stress, requirements to be on-call).

AMA’s (1985) estimates of returns to training (adjusted here to reflect differences relative to becoming a general/family practitioner) are much higher—e.g., 7 percent for general internal medicine, 14 percent for surgery, 0 percent for pediatrics, 17 percent for obstetrics/gynecology, 18 percent for radiology, 5 percent for psychiatry, 22 percent for anesthesiology, and 14 percent for pathology.

Exhibit 62 Return to Training by Specialty (Relative to GP/FP)

[D]

Understanding how monetary incentives motivate physicians to work more or fewer hours in patient care, choice of specialty and practice location, practice patterns, and retirement behavior helps build more reliable projections of future physician supply.  The paucity of information in the literature on the relationship between compensation and physician behavior is due, in part, to the lack of good data on physician earnings.

In summary, during the next couple of decades there will likely be competing factors placing upward and downward pressures on physician earnings.  The projected growing shortfall of physicians—and in particular specialists—will have an elevating effect on earnings.  The increasing reliance on Medicare as a source of income, combined with pressures by the Federal Government to control rising Medicare and Social Security expenditures, will have a depressing effect on physician earnings.  Physicians in some specialties will encounter increasing competition from NPCs, as well as from other physicians in other specialties expanding their scope of practice, and increased competition could negatively impact compensation.  Insurance practices that increase patients’ cost awareness could also affect demand for services and thus negatively impact physician earnings.