Tracking Health Care Costs:
Inflation Returns
Despite rising trends, the double-digit cost increases of the 1980s are not
likely to return.
b y Ch r i s t o p h er Ho g a n , P au l B . Gi n s b u r g , a n d Jo n R . Ga b e l
T
r e n ds i n health care spending that underlie private health insurance pre-miums occupy center stage in any de-bate over the affordability of health care and the future of employer-sponsored health in-surance. Premium trends affect employers’ decisions to offer insurance, the types of plans offered, the level of copayments and other out-of-pocket costs, and employees’ share of premiums. Ultimately, increases in premiums affect not only employers’ willingness to offer insurance but also employees’ ability to pur-chase coverage when it is offered.The period 1994–1998 was a time of record-low rates of growth in health insur-ance premiums and in the underlying medical expenses that are covered. A robust economy and low premium growth enabled the pre-mium for a typical private health insurance plan to grow more slowly than did gross do-mestic product (GDP) per capita.
This is the fifth in a series of annual analy-ses that synthesize trends relevant to private health insurance premiums.1 Here we argue that a return to higher rates of premium growth was probably inevitable. In hindsight, the respite from rapid premium increases dur-ing 1994–1998 was the result of uniquely fa-vorable conditions: low growth in underlying spending combined with insurers’ willing-ness to absorb cost increases to gain market share. In this paper we first review recent
trends in medical spending that are relevant to private insurance premiums and then dis-cuss how the relationship between trends in spending and premiums reflects the health in-surance underwriting cycle.
Study Methods
Health care spending can be measured in vari-ous ways: (1) using data from insurers and from consumers on payments to providers of care; (2) using data from providers on reve-nues; or (3) using data from providers on costs incurred in the delivery of care, with labor costs being the most important component. Because we have placed a premium on timeli-ness in order to discuss the most recent trends, this analysis makes the greatest use of data on provider revenues and on labor costs incurred by health care establishments.
We use the Milliman and Robertson (M&R) Health Cost Index to reflect expendi-ture increases underlying private health in-surance premiums.2This index, which is in-tended to assist insurers in forecasting their claims payments and comparing them with those of others, is based on provider revenues (a proxy for spending on services) gathered from a variety of sources, some widely avail-able and some proprietary.3 It is limited to health services that tend to be insured: in-patient and outin-patient hospital services, phy-sician services, and prescription drugs. Since
Christopher Hogan is president of Direct Research, LLC, a small health services research consulting firm in Vienna, Virginia, and a consultant to the Center for Studying Health System Change (HSC) in Washington, D.C. Paul Ginsburg is president of HSC. Jon Gabel is vice-president of health systems studies at the Health Research and Educational Trust, also in Washington.
provider revenue data tend to cover all pa-tients, M&R actuaries subtract data on Medi-care payments to providers, to arrive at a se-ries that more closely reflects the population with private health insurance coverage.4
We use this index because (1) the actuarial work to remove revenues from Medicare is valuable, since that component of revenues has followed trends that are distinct from those of private insurance; and (2) because it is available with a very short time lag. We have compared historical data from an ex-panded version of this series (which includes Medicare spending) with the national health spending accounts maintained by the Health Care Financing Administration (HCFA)— generally recognized as the “gold standard” for tracking expenditures—and found that they track fairly closely.5
We look at data on payroll costs for non-supervisory workers in health services estab-lishments (SIC 8000) compiled by the Bureau of Labor Statistics (BLS), to gain insight into providers’ costs. The sample includes both private and public employers but excludes nonsalaried health professionals. These data also are available with a very short time lag. Trends in provider revenues from M&R and payroll costs from the BLS are reported on a per capita basis. This is the most relevant measure for policymakers and is directly com-parable to trends in premiums.
Data on premiums in employment-based health insurance come from the Kaiser/Health Research and Educational Trust (HRET) Sur-vey of Employers and its predecessor surSur-veys. The 2000 Kaiser/HRET survey used a strati-fied random sample of 1,887 private and public firms employing three or more workers. The sample frame is Dun and Bradstreet’s listing of businesses that have entered the credit market. Data are from telephone interviews with employee benefit managers conducted from January to May 2000. The survey contin-ues the health benefits survey first conducted by the Health Insurance Association of Amer-ica (HIAA) from 1987 to 1991 and then by KPMG Peat Marwick from 1991 to 1998. The core questions in these surveys—premium
in-creases, the monthly cost of coverage, em-ployee contributions, self-insurance status, and plan enrollments—are virtually identical. For the years 1991, 1992, 1994, and 1997, KPMG sampled only firms with 200 or more workers. We use published data from the U.S. De-partment of Labor’s Consumer Expenditure Survey (CES) to gain additional insight into how spending trends are passing through to consumers. The CES surveys approximately 5,000 households about their spending by category of goods and services over the past quarter. The survey includes separate ques-tions for health insurance premiums, medical services, drugs, and medical supplies.
Underlying Spending Trends
Provider revenues per privately insured per-son increased 6.6 percent in 1999, compared with 5.1 percent in 1998 and 3.3 percent in 1997, according to the M&R Health Cost In-dex (Exhibit 1).6Components of the 1999 in-crease followed the same pattern as prior years, with sharply higher drug spending and hospital outpatient costs, but small increases or even decreases in hospital inpatient costs.Drug spending accounted for about 44 per-cent of the 1999 increase, slightly higher than in 1998.7About one-third of the 1999 increase in drug spending was attributable to higher drug prices as measured by the Consumer Price Index (CPI); the remainder, to new drugs and increases in use of existing drugs. A recent analysis of claims for 1996–1999 of a major pharmacy benefits manager (PBM) by the Schneider Institute of Health Policy, Brandeis University, attributed the increases in spending per enrollee to the following cate-gories: 32 percent, higher costs per day; 15 per-cent, more days per prescription; 40 perper-cent, more prescriptions per user; and 13 percent, more users per enrollee.8
Hospital inpatient revenues, by contrast, accounted for only 3 percent of the 1999 in-crease. This continues the pattern of the last half of the 1990s, during which hospital reve-nues per privately insured person fell more often than they rose and balanced the effect of other components to restrain overall health
care spending increases. The spending slow-down underlying private health insurance during the 1990s was largely attributable to the behavior of hospital inpatient spending (as measured by revenues). Inpatient care is a large component of services covered by pri-vate insurance, so the years of declines in in-patient spending were also years of slow growth in total spending. If inpatient reve-nues were excluded from the calculation, the average rate of spending growth for the dec-ade would nearly double, and the slowdown of the mid-1990s would virtually disappear.
Spending for physician services accounted for 32 percent of the 1999 increase. The trend for this component also dipped in mid-decade, but by 1999 it had nearly returned to its earlier rate of increase. The mid-1990s were a period of substantial deflation in physician fees paid by private insurers, with many in-surers bringing their rates down toward or below Medicare’s payment level.9By the end of the decade the pullback from intensively managed care and the broadening of insurers’ physician networks may have limited the op-portunities for continued fee reductions.
Spending for hospital outpatient services accounted for 21 percent of the 1999 increase. This category grew at a consistently high rate throughout the decade. Since 1992 annual per capita revenue increases for this sector aver-aged about 8.5 percent. The rate in any year
never deviated more than one percentage point from the average.
Rates of increased spending for the first quarter of 2000 are very similar to those for the first quarter of 1999 (Exhibit 1). The only difference of importance is a slightly lower rate of increase for drug spending.
BLS data on payrolls in health services es-tablishments provide timely insight into the largest cost factor that providers face. (Unlike the M&R index, this series applies to reve-nues from all payers, including Medicare.) Payrolls increased only 3.3 percent per capita in 1999 (Exhibit 2). Payroll per capita rose at an annual rate of nearly 5 percent in the first seven months of 2000, returning to the trend seen in the mid-1990s but well below the pace seen earlier in the decade.
Insurance Premium Trends
The sharp increase in premiums for 2000—8.3 percent, compared with 4.8 percent in 1999—was widely anticipated, given the re-cent discrepancies between cost and pre-mium growth and the behavior of the health insurance underwriting cycle (Exhibit 3). If the trend in spending underlying health in-surance premiums in 2000 is similar to that in 1999 (less than 7 percent), this will mark the first year since 1994 that premium increases exceeded underlying spending.The early 1990s were characterized by an EXHIBIT 1
Trends In Provider Revenues From Non-Medicare Patients, By Component, 1991–2000
1991 1992 1993 6.9% 6.6 5.0 3.5% 2.8 4.8 16.8% 13.9 8.9 5.4% 5.9 3.3 12.4% 11.7 7.1 1994 1995 1996 2.1 2.2 2.0 –2.0 –3.5 –4.4 8.7 7.9 7.7 1.7 1.9 1.6 5.2 10.6 11.0 1997 1998 1999 2000a 3.3 5.1 6.6 6.5 –5.3 –0.9 0.6 1.0 9.5 7.8 8.4 8.2 3.4 4.7 5.2 5.2 11.5 14.1 18.4 17.2
SOURCE: Milliman and Robertson Health Cost Index ($0 deductible).
aData through March 2000, compared with corresponding months in 1999.
unexpected decline in health care spending growth, resulting in higher underwriting profits and thereby attracting capital into the health insurance industry. Thus, insurers cut premiums to enter new markets and gain mar-ket share; as a result, premiums rose more slowly than did spending for services covered by health insurance. This, in turn, led to
underwriting losses and insurers’ withdraw-ing from selected markets. While employers’ resistance to premium increases may have de-layed the onset of the stage in which premium increases exceed trends in underlying spend-ing for services, this stage of the cycle prob-ably accounts for a portion of the sharp in-creases for 2000.
EXHIBIT 2
Changes In Employment, Hours, And Earnings In Health Services Establishments, 1991–2000 1991 1992 1993 9.0% 7.3 5.5 3.5% 3.2 2.0 5.3% 3.9 3.4 3.1% 2.4 2.5 1994 1995 1996 4.4 4.5 4.8 1.6 1.6 1.5 2.7 2.9 3.2 2.7 2.8 3.4 1997 1998 1999 2000c 5.7 4.3 3.3 4.9 2.5 0.8 –0.2 1.2 3.2 3.5 3.5 3.7 3.9 4.1 3.6 3.7
SOURCE: U.S. Department of Labor, Bureau of Labor Statistics, Employment, Hours, and Earnings database.
aProduct of second and third columns.
bProduct of number of production workers (excludes executives and managers) and average hours per week, adjusted for
changes in U.S. population.
cData through July 2000, compared with corresponding months in 1999.
EXHIBIT 3
Annual Increase In Employer-Based Insurance Premiums And Underlying Spending, 1991–2000 1991 1992 1993 11.5% 10.9 8.0 –b –b 8.5% 6.9% 6.6 5.0 1994 1995 1996 4.8 2.1 0.5 –b 2.3 0.8 2.1 2.2 2.0 1997 1998 1999 2000c 2.1 3.3 4.1 7.5 –b 3.7 4.8 8.3 3.3 5.1 6.6 6.5
SOURCE: Spending data are from the Milliman and Robertson Health Cost Index ($0 deductible). Premiums are from the Kaiser/Health Research and Educational Trust (HRET) survey of employer-based health plans for 1998–2000 and from the KPMG survey for 1991–1998.
aFirms with 200 or more workers. bNot available.
cData through March 2000, compared with corresponding months in 1999.
Data on underwriting gains and losses by Blue Cross and Blue Shield plans support the notion that insurers may just be entering the next phase of the underwriting cycle in 2000 (Exhibit 4). On average, Blues plans essen-tially broke even on health insurance under-writing in 1999, after having losses in the prior four years. If the 2000 premium increase out-strips cost growth, the industry may move into the initial portion of the profitable phase of the underwriting cycle this year.
Another indication that we are entering a new stage of the underwriting cycle comes from a comparison of premium increases for fully insured and self-insured plans, which are identified in the Kaiser/HRET survey. “Premi-ums” for self-insured plans reflect Consolidated Omnibus Budget Reconciliation Act (COBRA)– mandated rates filed by employers with the De-partment of Labor, which are based on employ-ers’ projections of claims costs. In contrast to premiums for fully insured plans, they are un-likely to reflect underwriting practices of set-ting premium increases above or below
pro-jected cost increases. The premium increase for 2000 was 9.6 percent for fully insured plans and 7.1 percent for self-insured plans. In contrast, premium increases for fully insured plans were two percentage points lower in 1995 than were those for self-insured plans.
Within plan types, differences in premium increases were equally dramatic. For example, among health maintenance organizations (HMOs), premiums increased 9.4 percent for fully insured plans and 4.5 percent for self-insured plans. Among preferred provider or-ganizations (PPOs), premiums increased 10.9 percent among fully insured plans and 7.4 per-cent among self-insured plans.
The recent cyclical phenomenon of larger increases in premiums for fully insured than for self-insured plans has encouraged some employers to self-insure. The percentage of employees enrolled in self-insured plans in-creased from 48 percent in 1999 to 51 percent in 2000. This has also stimulated a shift to PPO plans, which are primarily self-insured plans, and away from HMO and
point-of-EX H IB I T 4
Blue Cross And Blue Shield Plans, Underwriting Gains And Losses, As Percentage Of Revenue, 1990–1999
SOURCE: Authors’ analysis of data provided by the Blue Cross and Blue Shield Association. 1990 Percent 1.0 0.0 –0.5 –1.0 –1.5 1992 1994 1996 1.5 2.0 1998 0.5 1991 1993 1995 1997 1999 221
service (POS) plans, which are largely fully insured. We expect these trends to continue into 2001 and 2002.
Implications For Consumers
Persons with employer-sponsored coverage have largely been insulated from health care cost growth since the middle of the decade. According to the CES, consumers spent 5.5 percent of after-tax income on health care in 1993. By 1998 (the most recentyear available), that amount had fallen below 5 percent. The only health care item com-manding a larger share of in-come in 1998 than in 1993 was prescription drugs, mainly as a result of an 8 percent increase in out-of-pocket spending in 1998. Earlier in the decade, ex-panding coverage for prescrip-tion drugs, related to rising en-rollment in managed care plans, had sharply limited the total out-of-pocket spending increase. For employees’ share of premiums, the Kaiser/
HRET survey shows distinctive patterns for employee-only and family coverage. For the former, employers have been willing to pick up an increasing share of costs in recent years. For example, from 1996 to 2000 employees’ share of the premium declined from 21 per-cent to 14 perper-cent. However, the employee share of the premium for family coverage was virtually unchanged, declining from 28 per-cent to 27 perper-cent.10
The Future
The most disturbing development over the past two years has been the steep growth in spending for covered medical services. Over the entire underwriting cycle, average pre-mium growth largely reflects this trend. Al-though some of the jump in premiums reflects the current phase of the underwriting cycle, an important segment reflects the trend in un-derlying spending.
Several factors indicate that future
spend-ing increases will continue to be higher than in recent years. (1) In response to the managed care backlash and increased regulation of plans at the state and federal levels, managed care plans are becoming less restrictive and consumers are shifting toward more loosely managed plan types, such as PPOs. (2) Expen-sive new technologies, especially pharmaceu-ticals, continue to enter medical practice. (3) Continuing provider consolidation could lead to higher rates for services. Moreover, even without this trend, the sharp reductions in provider fees ne-gotiated during the mid- or late 1990s appear as a one-time realignment that is now be-hind us. (4) Reductions in spending for inpatient hospi-tal services of the magnitude seen in the 1990s may no longer be feasible and, in any case, will have a smaller im-pact on the trend of total spending.
In the short run, several in-dicators suggest that costs and premiums will continue to in-crease sharply. For the first seven months of the year, health establishments’ per capita payrolls increased at nearly a 5 percent annual rate, faster than the 1999 rate of increase. For the first quarter of 2000, the Health Cost In-dex increased at essentially the same rate as in 1999. In addition, the California Public Em-ployees Retirement System (CalPERS), one of the nation’s largest public purchasers of pri-vate health insurance, announced an average premium increase of 9.2 percent for its HMO contracts for 2001, prior to any increases in copayment or deductible requirements.11
Despite these pressures, the double-digit increases of the late 1980s and early 1990s are not likely to return. Today’s health care mar-kets are far more competitive than they were in the past. Price pressure on providers is likely to lead to continuing efforts on their part to cut their costs. However, continued growth in spending—even if the magnitude of the increase is less than historical rates—has
“The sharp
reductions in
provider fees
negotiated during
the mid- or late
1990s appear as a
one-time
realignment that
is now behind us.”
implications for the economy as a whole, for state and federal budgets, and for the unin-sured. Substantial premium increases mean less consumer spending on other goods and services, lower wage increases, and lower profits. Increased growth in outlays for Medi-care and Medicaid would mar the rosy fiscal picture that governments now enjoy. Finally, increases in insurance premiums could lead to an increasing number of uninsured persons. A recent study shows that 20 percent of the un-insured have access to employer-sponsored coverage but have declined that coverage, mostly because of costs.12
T
he i nevi ta bl e con clusion is that U.S. businesses, households, and gov-ernments face difficult choices in the immediate future. During the past few years managed care plans and employers have been retreating from “managed care heavy.” The tightest labor market in three decades has shielded employees from the cost conse-quences. Whenever the economy softens, em-ployees will be pressed to decide whether to reconsider tightly managed care or to accept more responsibility for health care costs, either at the point of service or in premium payments.The authors thank Jay Thayakaran of Milliman and Robertson for permission to use the Health Cost Index. They gratefully acknowledge the Robert Wood Johnson Foundation and the Commonwealth Fund for their financial support.
NOTES
1. P.B. Ginsburg and J.D. Pickreign, “Tracking Health Care Costs,” Health Affairs (Fall 1996): 140–149; P.B. Ginsburg and J.D. Pickreign, “Tracking Health Care Costs: An Update,” Health
Affairs (July/Aug 1997): 151–155; P.B. Ginsburg and
J.R. Gabel, “Tracking Health Care Costs: What’s New in 1998?” Health Affairs (Sep/Oct 1998): 141–146; and P.B. Ginsburg, Tracking Health Care
Costs: Long-Predicted Upturn Appears, Issue Brief no.
23 (Washington: Center for Studying Health System Change, November 1999).
2. The term “underlying private health insurance premiums” is used broadly. Premiums may fol-low a different trend than underlying spending does because of the insurance underwriting cy-cle (discussed at length in this paper) and be-cause of changes in benefit structure. We are
focusing on how premiums would change prior to these two factors playing a role in determining the actual premium trend.
3. The index that M&R provides to clients simu-lates trends in claims for a “standard” private health insurance policy with a $250 deductible. The trend in such an index would slightly over-state the trend in spending underlying private insurance because the standard policy would pay for a slightly higher proportion of expendi-tures each year. M&R has provided us with a version of the index that reflects a hypothetical policy with no deductible.
4. Ideally, revenues from Medicaid and from unin-sured patients should be excluded as well, but data limitations have precluded this.
5. In an earlier paper on tracking health care costs, Ginsburg and Pickreign compared the M&R Health Cost Index ($0 deductible), which they expanded to include Medicare, with data from the National Health Accounts (NHA) from 1975 through 1994. The expanded Health Cost Index predicted the NHA series quite well. The mean annual percentage change differed by only 0.4 percentage points, although the expanded Health Cost Index had a larger variance. The mean absolute difference in annual percentage changes in spending was 2.1 percentage points. See Ginsburg and Pickreign, “Tracking Health Care Costs” (1996).
6. These data are for the privately insured popula-tion only and exclude revenues from Medicare, which generally has a distinct trend.
7. Each category’s contribution to overall spending growth was calculated as the change in total spending within category as a percentage of the change overall. This accounts for both the size of the category (share of costs) and the rate of growth of spending within the category. 8. S.S. Wallack et al., “Sources of Growth of
Phar-maceutical Expenditures” (Presentation at Ac-cess to Pharmaceuticals: The Seventh Princeton Conference, sponsored by the Council on the Economic Impact of Health System Change, Princeton, New Jersey, 11–13 May 2000). 9. Physician Payment Review Commission, Annual
Report to the Congress, 1996 (Washington: PPRC,
1996), 218.
10. J. Gabel et al., “Job-Based Health Insurance in 2000: Premiums Rise Sharply while Coverage Grows,” Health Affairs (Sep/Oct 2000): 144–151. 11. B. Branch and P. Macht, “CalPERS Approves
Co-Pay Changes, Rates for 2001,” California Public Employees Retirement System press release (Sacramento: CalPERS, 17 May 2000).
12. P.J. Cunningham, E. Schaefer, and C. Hogan, Who
Declines Employer-Sponsored Health Insurance and Is Uninsured? Issue Brief no. 22 (Washington: HSC,
1999).
