Lapas attēli

Data Used to Estimate U.S. Companies'
Retiree Health Liabilities

Active Workers
Covered in 1988

In 1983, 66.2 million active workers in 1983 had employer-or unionsponsored group health insurance, according to the CPS. After subtracting local, state, and federal government workers with group health coverage, we estimated that 51.9 million private workers had this kind of insurance. From the HIAA survey, we estimated that about 60 percent of these private workers (31.4 million) in 1983 were in companysponsored group health plans that continue coverage for retirees. This is 42 percent of the 74.3 million nonagricultural workers in the U.S. private sector in 1983 reported by the CPS.

For computational ease, we grouped the 66.2 million workers in the CPS
in 1983 into 5-year age groups. Assuming that the estimated 31.4 million
with retiree health coverage were distributed similarly by age, we aged
these groups to 1988 and added new entrants at age 40 to the covered
population in each year. We ignored workers under age 40 in our esti-
mates of retiree health liabilities because studies have shown that, on
average, workers have about 25 years of service with their last
employer before retirement. From CPS age-specific data, we estimated
that in 1988 there were 13.4 million active workers ages 40 to 64 with
retiree health coverage. We also applied mortality rates to each group in
each year, as described below. The resulting covered population at ages
40 and over in 1988 appears in fig. IV.1.

In sum, our estimates of total and accrued liabilities for retiree health benefits in 1988 were based on two groups of people: (1) the estimated 6.6 million retirees already receiving health benefits and (2) the estimated 13.4 million active workers ages 40-64 with retiree health coverage.

Companies' Retiree
Health Costs

Future Insurance Costs

We estimated employers' retiree health costs for 1988 at $777 per
retiree age 65 and over and $2,602 per retiree under 65. Employers'
health expenditures are greater for workers who retire before age 65
because they are not yet eligible for Medicare. For retirees age 65 and
over, Medicare pays for most medical expenditures, thereby reducing
employer health expenditures. In addition, some early retirees may have
poorer health when they retire than workers of the same age who
remain on the job.

To estimate future health insurance costs, we started with Labor's estimate of $553 as the 1983 average annual insurance premium for employer-provided health benefits for retirees aged 65 and over. This

Data Used to Estimate U.S. Companies'
Retiree Health Liabilities

Future Medical Inflation

estimate was based on 1977 National Medical Care Expenditure Survey data, updated to 1983 by Labor. For retirees under age 65, we adjusted Labor's estimate and estimated $1,852 as the average 1983 annual insurance premium for employer-provided health benefits for under-65 retirees. Both Labor's and GAO'S cost estimates include an amount for spousal and dependent coverage.

We projected these average cost estimates to 1988, using the Consumer Price Index for Medical Care (CPIMC), which increased by about 40 percent from 1983 to 1988. This raised age 65-and-over costs to $777 and under-age-65 costs to $2,602 in 1988 (including an adjustment for higher costs for early retirees than the average for active workers from data provided by the Washington Business Group on Health). Using our medical inflation assumption (described below), based on historical rates, we projected the 1988 annual costs into the future. We did not consider any company savings from Medicare catastrophic medical insurance in this projection.

The historical rise in companies' average retiree health costs has resulted from increases in medical personnel and supply costs, new and expensive technology (e.g., CAT scans), and higher utilization (e.g., more frequent mammograms and more numerous lab tests). Because of these factors and expectations by some experts that Medicare will require recipients or employers to shoulder a greater share of their medical expenses in the future, employers' cost of providing health benefits to retirees is expected to continue to rise faster than general inflation.

The Consumer Price Index (CPI), which measures general inflation, is based on prices of several household budget items including food, transportation, housing, and medical care. The CPIMC measures medical inflation—it represents the price of a market basket of health care services. In each decade since 1960, the yearly average CPIMC has risen (see fig. IV.2), from 4.0 to 7.8 to 8.6 percent during the 1960s, 1970s, and 1980s, respectively. Likewise, medical inflation has exceeded general inflation since 1960. Over the period we examined (1960-87), the CPIMC averaged 1.6 percentage points per year more than the CPI. During the period 1970-87, it rose on average 1.7 percentage points faster per year than the CPI. Most recently (1980-87), the CPIMC averaged 3.5 percentage points more than the CPI (see fig. IV.3).

To estimate companies' future retiree health benefit payments, we used a graded medical inflation factor based on the expectation that the CPIMC

[blocks in formation]

will continue to outpace the CPI but that the difference between the rates will decline. We assumed that medical inflation will exceed the CPI by 3.5 percentage points a year through 2001, by 2.75 percentage points a year through 2015, and by 2 percentage points a year thereafter.

In making assumptions about future health care inflation, we considered
the relationship between health care expenses and the gross national
product (GNP). Health expenditures were 11.1 percent of GNP in 1987,
according to the Health Care Financing Administration. If GNP grows at
a 2-percent real rate and medical inflation were to rise 4 percentage
points faster than the CPI for each of the next 30 years, starting in 1988,
health expenditures would grow from 11.1 percent in 1987 to 19.7 per-
cent of GNP in 2018 (see fig. IV.4). If medical inflation were 2.5 percent-
age points greater than the CPI (the assumption used by Labor in its
estimate for 1983), health expenditures would rise to 12.8 percent of
GNP. Our graded medical inflation assumption is intermediate to these
extremes, raising the health share of GNP to 14.9 percent in 30 years and
remaining constant thereafter.

[blocks in formation]

In our sensitivity analyses, we changed the graded medical inflation assumption to a 4/3/2- and a 3/2.5/2-percentage point excess over the CPI for our high and low estimates, respectively. Under the high-cost scenario, in 30 years, the share of GNP going to health care expenditures would rise to 16.4 percent and under the low-cost scenario, to 13.5 percent.

Discount Rate

To obtain the present value of future retiree health benefits in 1988, we
used a 7-percent discount rate. In health benefit valuations, the discount
rate is less important than its relation to the rate of general and medical
inflation. For our model, we assumed a 5-percent future annual inflation
rate and a 7-percent discount rate, which means our assumed real rate
of return on investments is 2 percent. We netted out the 5-percent gen-
eral inflation from our assumed medical inflation to obtain constant
1988-dollar estimates.

[blocks in formation]

Mortality Assumption

Sensitivity Analysis

In estimating how long workers and retirees will live, we used a standard unisex mortality table, adjusted to reflect expected mortality improvement. We selected the UP-1984 mortality table because it (1) is commonly used by actuaries and is recognized by government agencies as an appropriate basis for valuing pension benefits and (2) can be adjusted for expected mortality improvements by lowering the expected probability of dying at each age.

The effects on total liabilities of changing the values of selected variables in the model singly and in combination are shown in table IV.1. By this analysis, estimated liabilities are sensitive to medical cost inflation and life expectancy. For example, by using a lower medical inflation rate, our baseline estimate of total liabilities fell by 6 percent. Slightly higher or lower mortality changed estimated liabilities by 4 percent. Other variables had less impact. For instance, very large changes in early retirement rates (10 percent in either direction) had a relatively

« iepriekšējāTurpināt »