Printer-friendly version (HTML)

Retiree Health Care Benefits: Data Collection Issues
by William J. Wiatrowski
Bureau of Labor Statistics

Originally Posted: July 29, 2003

The aging of the baby boom generation and their likely expanded use of health care services are well documented.¹ Equally well reported is the debate over the obligation of employers to continue health care benefits after an employee retires.² Interested parties--employers, employee groups, policymakers, and others--need reliable information on issues related to retiree health care benefits so that informed decisions can be made. The Bureau of Labor Statistics (BLS) has provided data on retiree health care benefits for the past two decades, and in earlier years provided ad hoc information on the topic. Yet the capture and tabulation of these data pose unique problems, such as the following:

• Employers at a local unit of a large corporation providing data to BLS on pay and benefit practices may be unfamiliar with retiree health care issues that are handled through the corporate office.



• Due perhaps to the uncertainty of providing benefits into the future, employers may be reluctant to document retiree health care practices.³



• A tabulation of retiree health care practices offered to today’s employees may not reflect what those workers actually receive when they retire many years in the future.



• Retirement age is no longer a static concept; indeed retirement may not be defined exactly as it was in the past, as many individuals continue to work after the traditional retirement age.



• Legal obligations, such as the requirement that employers offer continued health care benefits coverage to certain former employees, make it difficult to define exactly what constitutes retiree health care benefits.



• The health care plans offered to active employees--for example, a choice of several health maintenance organizations (HMOs)--may be unrelated to those plans offered to retirees, making it difficult to assess the change in benefit level that occurs at retirement.



• The availability of Medicare and new Medicare Choice alternatives makes it difficult to assess the actual level of benefit provided by an employer plan.

This article examines the data collection and tabulation difficulties related to retiree health care benefits. Policy issues that affect employer decisions to provide benefits will be identified. Changes to survey questions and tabulations over the past 20 years will be explored, along with the rationale behind those changes. Future data collection and tabulation plans also are discussed. To begin, some facts about older Americans, health care needs, and employee benefits introduce the debate.

Americans are older, healthier

At the beginning of the 21st century, the latest Census reports that there are 281 million Americans; 21 percent are age 55 or older and 13 percent are age 65 and older. The number of older Americans is projected to grow rapidly over the next few decades. (See chart 1.) This trend, like many at the turn of the century, is driven by the baby boom generation, those post-World War II babies the oldest of whom are now in their late 50s.⁴

The fact that older Americans make up a growing portion of the total population is not lost on those who debate the future of the nation’s social programs designed for the elderly--Social Security and Medicare. Both programs began when there were many more working-age contributors than retirement-age recipients. Now both programs face financial difficulties, as the number of recipients gets closer to the number of contributors.⁵

Not only are there a growing number of older Americans, but they are living longer and healthier lives, benefiting from greater improvements in health care. When Social Security was introduced in 1935, the average life expectancy for a 65-year-old was about 12 years. When Medicare was introduced in 1965, the average life expectancy for a 65-year-old was about 14 years. Today a 65-year-old can expect to live another 18 years.⁶ Health benefits for older Americans--either Medicare, employer plans for retirees, individually purchased coverage, or some combination of these--will be needed for longer periods of time than ever before.

Older Americans spend more on health care services than the population as a whole. In 1999, Americans age 65 and older spent 11.4 percent of their total household expenditures on health care services, compared with 5.3 percent for all consumers.⁷ At the same time, the price of health care goods and services is rising at a faster pace than the cost of other products; the Consumer Price Index for medical care rose 4.1 percent in 2000, above the 3.4-percent increase for all goods and services.⁸

While older Americans spend a greater proportion of their income on health care than do younger Americans, older Americans also increasingly are reporting that they are in better health. Between 1991 and 1998, the percent of those age 55 to 64 who self-reported themselves in fair or poor health declined from 21 to 18 percent. During the same period, those age 65 and older reporting fair or poor health dropped from 29 percent to 27 percent.⁹

This increased use of health care services and improved health outcomes does not come without a cost. In 1999, health care spending accounted for 13.0 percent of the gross domestic product of the United States--just over $1.2 trillion. This is up from 5.1 percent and $27 billion in 1960.¹⁰

One result of expanded health care services and improved health outcomes may be the ability of workers to stay at their job longer. The concept of "retirement" has changed considerably from what it was in the 1970s--complete stoppage of work at age 65 with a fixed pension payable for life. Workers have new types of retirement plans and investment choices. Retirement income benefits come from many different plans and from many different employers. Older Americans must navigate a maze of inconsistent benefit eligibility ages (for example employer pension at age 60, Social Security at age 67, and individual retirement arrangements at age 70½). Finally, workers may need income from continued work after retirement age, or workers may derive satisfaction from a new career that starts after their "primary" job is completed.¹¹

Phased retirement, a relatively new phenomenon that has recently gained attention, may offer an added complexity in understanding retiree health care benefits. Under a phased retirement plan, an older worker may work a limited number of hours or may only work during certain peak periods. Pay is reduced to reflect these reduced hours, with partial pension benefits available to supplement income. The status of health care benefit plans during this period of phased retirement is critical. Issues that may arise include whether the phased retiree is considered an active employee or a retiree when determining the choice of health care plans, the amount of required employee contributions, or the coordination with other health care benefits, such as Medicare.¹²

Added to all this confusion is the need for health care benefits during retirement, especially for those not yet eligible for Medicare (age 65). Spouse and dependent health care benefit coverage is important as well, especially for those older Americans who may have young children. (The average age of women having their first child rose from age 21 in 1968 to age 25 in 1997. More dramatically, in 1972, 7 percent of first births occurred among women age 30 to 34; by 1997, women in this age group accounted for 24 percent of first births.¹³) Some older workers may feel "locked" in their job¹⁴ because of the availability of health care benefits; individual purchase of health insurance, especially for those between age 55 to 65, can be expensive.¹⁵

Employer-provided health care benefits: some policy issues

Health care benefit coverage for active employees and their families is widespread. Data from the Employee Benefits Survey for 1996-98, for example, show that about 60 percent of workers had such benefits from their employer.¹⁶ Similarly, data from the Current Population Survey show that in 1998 65 percent of nonelderly Americans had employment-based health care benefits, either as an employee or a dependent.¹⁷ For those age 65 and older, health care benefits are widespread, with Medicare being the most prevalent source of coverage. (See table 1.)

There are variations in the availability of health care benefits by establishment size, industry, full-time versus part-time status, unionization, and other employment-related variables. Similarly, such coverage may only be available after a waiting period, and some of the cost may be borne by the employee. The rising cost of health care benefits throughout the late 20th century (see table 2) led employers to look for ways to control costs, such as offering managed care plans (health maintenance organizations, for example), increasing cost sharing by employees and dependents (in the form of higher deductibles, for example), and limiting coverage for some services (such as requiring surgery be performed on an outpatient basis). While these cost-containment features have become widespread, there is less evidence that employers have dropped health care coverage for active employees. In the tight labor market of the late 1990s, the availability of such benefits may have been necessary for hiring and retention purposes.¹⁸

For retirees, there has been a slight decline in employer coverage and a shift away from complete employer funding of such coverage. Among the factors that may influence employer decisions in this area are recent legislation and litigation. Two laws, while not specifically targeted at retiree coverage, nonetheless place upon employers certain obligations when employees covered by health care plans leave their job. These laws generally are known by their acronyms, COBRA and HIPAA.

The Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA) established continuation of coverage benefits for employees who leave a job with health care benefits. Individuals may continue coverage under their former employer’s plan for up to 18 months following separation; continuation of coverage benefits are also available to dependents of covered employees. Eligibility requirements must be met and certain restrictions apply. In addition, employers may charge those receiving continued coverage up to 102 percent of the total premium. Thus, while coverage is available at group insurance rates and without proof of insurability, the employer no longer finances the premiums. Early retirees (prior to age 65) may look upon the availability of COBRA coverage at the employer’s group rates as a substitute for retiree health care benefits. Employers may also consider their obligation under COBRA as a substitute for retiree health care benefits. COBRA rights end when individuals become eligible for Medicare--typically age 65.¹⁹

In addition to COBRA coverage, the Health Insurance Portability and Accountability Act of 1996 (HIPAA) guarantees formerly covered employees and dependents access to an individually purchased health insurance plan without regard to health status. Insurers are not restricted in the premiums they can charge for such plans, however; such premiums may be related to the age or health status of the insured. Once again, employers may consider the availability of individually-purchased health insurance plans to be a substitute for employer-sponsored health care benefits for early retirees.

While the availability of Medicare and coverage continuation under COBRA and HIPAA may influence an employer’s decision to offer retiree health care benefits, accounting rules that came into effect in the early 1990s may have provided an additional disincentive for publicly traded firms to provide retiree health benefits. Financial Accounting Standard 106 (FAS 106) requires firms to record on their financial statements the unfunded liability represented by their retiree health care plan. According to Paul Fronstin of the Employee Benefit Research Institute, "While FAS 106 did not require funds to be actually moved or reallocated, the recognition of these liabilities in financial statements dramatically impacts a company’s calculation of its profits and losses and thereby creates a strong incentive for financial managers to limit expenses."

Legal debates surrounding retiree health care benefits often involve the obligation of the employer. Several court decisions on this topic have considered employer’s rights to modify or terminate benefits, and whether or not plan documentation specifically allows such actions.²⁰ Lack of documentation adds to the challenges of capturing and reporting accurate data on these benefits.

Finally, retiree health care benefits are the subject of new legal debates, as retirees brought suit for age discrimination against a public employer that treated retirees under age 65 differently from those age 65 and older.²¹ This merely continues the uncertainty surrounding the legal and accounting context of retiree health care benefits, and consequently continues to make it difficult to collect data on such benefits. A review of collection challenges and the resulting data follows.

Capturing data on retiree health care benefits

BLS has published at least some data on employer-provided benefits for most of its nearly 120 years of existence. Early studies were often snapshots of benefit practices among a small number of employers or a particular industry. More consistent data were first captured around the middle of the 20th century, consisting of the percent of workers within various localities that were offered a short list of benefits (such as paid vacations, life insurance, and retirement plans).

Case studies of employer insurance and retirement plans were published throughout the 1960s and 1970s, a precursor to the first BLS comprehensive benefits survey, conducted in 1979. First called the Level of Benefits Survey and later the Employee Benefits Survey, this study of the percent of workers who have various benefits, and the features of those benefits, has now been subsumed into the National Compensation Survey (NCS). NCS is a comprehensive study of wages, benefit costs, benefit availability, and benefit provisions for all workers in the United States.²²

BLS first attempted to capture data on retiree health care within its benefits survey in 1979, when the survey was in its infancy. Efforts were made to answer an ambitious series of questions for each health care plan in the survey, including the following:

• May retirees participate in the group health care plan?



• Does the plan or employer pay all or a portion of the Medicare Part B²³ premium for retirees?



• How are an employee’s health care benefits affected by retirement prior to age 65? Indicate whether benefits continue, duration of benefits, change in benefit level (increase, decrease, unchanged), and change in employee premium (increase, decrease, unchanged).

Table 3 provides a complete list of survey questions and shows how the questions have changed over time. In the first few years of the survey, data on retiree health care benefits were not published. Issues to be resolved included the interpretation of some of the questions, the relationship between questions, and the availability of data from employers.

Data on retiree health care benefits were first published in 1983. In that year and the following year, the published data were limited to the percent of those full-time employees with health care benefits who had at least partially employer-paid health care benefits available to them after retirement. For most years since then, some data have been published on the extent of workers covered by retiree health benefits and certain features of those benefits. These data form a more or less consistent trend in the availability of retiree health benefits and show a decline in the percent of workers with such benefits. But this trend masks many difficulties that were encountered in conceptualizing, collecting, and tabulating the data, and many changes that were made throughout this period to improve the data.

Two basic survey concepts--employee and benefits--are at the nexus of some of the data collection difficulties. The benefits survey is a survey of active employees. While the unit of observation has changed somewhat over time--first a census of all employees in a surveyed establishment and then a sample of selected occupations--the survey has always been limited to active employees. The number of employees receiving each benefit is captured, as is detailed information on the provisions of those benefits. What is not captured is the number of retirees. Test collection of the number of retirees proved difficult, as such records are not always available from the typical survey respondent (often a human resources employee) or at the location of collection (typically an establishment at a single location).

Obtaining data on retiree health care was also hindered by the definition of a "benefit" used in the survey. To qualify as a benefit, the plan or policy would typically have to be at least partially funded by the employer. This is due in part to the fact that employer costs for benefits are captured and tabulated as part of the survey. (These data are published in the Employment Cost Index as rates of change in employer benefit costs and in the Employer Costs for Employee Compensation as dollars and cents per hour worked.)

In addition, because the concept behind the survey is to capture compensation provided by the employer as part of the employment relationship (or contract), benefits with no employer cost were excluded. This exclusion of benefits with no employer cost has a large effect on the collection of retiree health care benefits, because such benefits may be paid entirely by the retiree.

Both of these issues--the lack of data on retirees and the exclusion of benefits with no employer cost--had to be resolved before data could be tabulated on retiree health care benefits. Given the difficulties associated with collecting numbers of retirees, the concept evolved into the percent of active employees covered by an employer health care plan who could continue their health care benefits when they retired. While this concept is conceptually different from a count of retirees, it has the advantage of being easier to obtain from a surveyed employer.

As noted, BLS studies of employer-provide benefits were limited to those plans that had an employer cost. The lack of an employer cost for some retiree health care benefits was an additional obstacle to overcome. To do so, BLS incorporated an exception to the definition of a benefit--an exception that has since been implemented to cover additional items. While the presence of an employer cost was maintained as the primary criteria, benefits were also examined to determine if they could only be obtained as a result of the employer-employee relationship.

The idea is that such a plan is a benefit to the employee because, absent their employment, they would not be able to receive the plan. Because retiree-paid health care benefits are typically part of a group health care product obtained by the employer, the retiree benefits would not be available unless the employment relationship existed. Furthermore, as part of a group plan, the retiree enjoys certain cost savings that might not be available with an individually purchased plan.²⁴ This concept has subsequently been invoked as other items were added to the benefits survey, including retiree-paid life insurance, unpaid family leave, employee-financed reimbursement accounts, and salary reduction arrangements with no employer funding.

What is a "plan"?

Once BLS had overcome these initial conceptual hurdles, it was able to publish the first data on the availability of retiree health care benefits in 1983. Because data on active employee health benefit enrollment is captured by plan, the availability of retiree health care benefits was asked of employees in each of the employer’s plans. (The retiree health care questions have always been limited to medical plans; no data are captured on the existence of retiree plans covering dental care, vision care, or other specialty services.) Unfortunately, this "plan" concept also proved to be a difficult data collection issue. Employers might offer a choice of plans for active employees, but provide only one plan for retirees. While all active employees might have had the opportunity to obtain the retiree plan, each active employee plan was not available to retirees. This plan concept failed to facilitate the comparison of the level of benefits of the active employee plan with the retiree plan. If active employees could choose from several plans with different features, but retirees were limited to one plan, it was difficult to assess the change in benefits that occurred at retirement. Because of these difficulties, published data were limited to an indication of the percent of those with health care benefits who had benefits available after retirement.

The assessment of the change in benefits between active employees and retirees is also hampered by the availability of Medicare for those aged 65 and older. When compared with active employee coverage, Medicare may add or delete benefits. While the employer’s retiree plan may attempt to fill in the gaps between Medicare and active employee coverage, a clear comparison of benefit levels is still difficult. Finally, the addition in the late 1990s of Medicare Choice, allowing Medicare recipients to designate a health care source--such as an HMO--to provide their Medicare benefits, further complicates the comparison of pre- and post-retirement benefit levels.

Beginning in 1986, BLS began a three-fold process of refining the collection of retiree health care data. One refinement involved the streamlining of questions asked; a second involved a change in concept; and a third involved a change in the source of data.

By 1987 BLS had reduced the questions to the following:

• How are an employee’s health care benefits affected by retirement prior to age 65? Indicate whether benefits continue, duration of benefits, change in benefit level (increase, decrease, unchanged), and change in employee premium (increase, decrease, unchanged).



• How are an employee’s health care benefits affected by retirement at age 65 or later? Indicate whether benefits continue, duration of benefits, change in benefit level (increase, decrease, unchanged), and change in employee premium (increase, decrease, unchanged).

A question on eligibility requirements for retiree health care benefits was added in 1989. One way employers could limit their obligation to provide retiree health care benefits was to impose an eligibility requirement (typically years of service) that must be met for a retiree to obtain benefits. Unfortunately, this data item also proved elusive to collect, as many employers, especially those without a traditional defined-benefit pension plan, did not have specific eligibility requirements.

The conceptual change had to do with the unit of observation. BLS moved away from the concept of collecting data by active employee plans and started asking questions about the availability of retiree health care benefits for employees in surveyed occupations.²⁵ The revised concept resulted in tabulations of the percent of active employees who could get benefits when retired, regardless of their active health plan coverage or their choice of plan. This eliminated some of the conceptual problems and provided a clearer answer, but ultimately made any comparison of level of benefits between active employees and retirees impossible. Current practice is to capture the following information for each surveyed employee:

• Are retiree health care benefits made available for those who retire prior to age 65? If so, who pays the premium (employer only, retiree only, both employer and retiree)?



• Are retiree health care benefits made available for those who retire at age 65 or later? If so, who pays the premium (employer only, retiree only, both employer and retiree)?

Prior to 1986, all data on retiree health care benefits were obtained from plan documents collected from surveyed employers. This is the primary method used by the BLS benefits survey to obtain detailed information on plan provisions, such as coverage for selected procedures or the amount of deductible. As noted earlier, written documentation of retiree health care benefits is not always complete and has been the subject of litigation. To overcome this obstacle, BLS began asking employers about the availability and details of retiree health care benefits as part of the data collection interview. While this change improved the availability and quality of data, it also added to the burden placed upon employers. The number of questions asked about retiree health care benefits was reduced in an attempt to limit the employer burden.

Trends in data on retiree health care benefits

While data on the availability of retiree health care benefits are available starting in 1983, changes in questions and concepts limit the availability of data prior to 1988. From that point forward, the data show a slow decline in active employee health benefits coverage and a similar slow decline in coverage for retirees. The most dramatic shift has been away from full funding by employers. In 1988, employers paid the full cost for just over half of those offered retiree health care coverage. In 1997, less than one-third of those offered retiree coverage had benefits funded entirely by their employer.²⁶ (See charts 2 and 3.)

Data on the availability of retiree health care benefits and the financing arrangements of those benefits are limited to those cases where the employer finances at least part of the benefit. Beginning in 1993, the survey also captured data on retiree benefits available but fully financed by the retiree. Unfortunately, the distinction between retiree financed coverage, continuation of coverage under COBRA, and conversion to an individual plan under HIPAA is not always clear. Results from 1997 indicate only a small percentage of employees have fully retiree-paid coverage prior to age 65, while half of all employees have no such coverage. Further research and refinement of concepts is needed to eliminate any confusion between retiree benefits and coverage continuation.

Data on the level of benefits and worker eligibility proved elusive, but were published for a few years. For example, data from the 1991 survey indicated that 3 out of 4 workers with coverage after retirement would see no change in their level of benefit prior to age 65. In the same year, a slightly smaller percent (69 percent) of those with retiree health care coverage at age 65 or older would see no change in coverage after retirement, although their benefits would be coordinated with Medicare.²⁷ Also in 1991, data on eligibility requirements were tabulated. The results indicated the majority of active employees with health care benefits had to meet a service requirement or qualify for pension benefits (which also impose service requirements) to be eligible for retiree health care benefits. As can be seen in table 4, there was little variation for those under and over age 65.

More data to come

Collecting data on employee benefits requires constant refinement. BLS is currently building new data-capture vehicles to take advantage of changing technology. At the same time, all data elements and concepts are under review. Data will continue to be captured and published on the availability and financing of health care benefits for retirees under and over age 65; refinements to the data to eliminate any existing conceptual problems will be pursued.

New data will be available for selected establishment characteristics, such as major industry group. New tabulations are also being developed to estimate the percent of establishments offering retiree health care benefits, in addition to the percent of employees who can receive such benefits. The aging population, advances in medicine, and ongoing policy discussions about the future of Medicare make retiree health care benefits a topic of continued interest. BLS will continue to research and refine methods to provide the best possible data on this issue.

William J. Wiatrowski
Assistant Commissioner, Office of Safety, Health and Working Conditions, Bureau of Labor Statistics.
Telephone: (202) 691-6304, E-mail: Wiatrowski.William@bls.gov

Notes

1 See, for example, Retiree Health Benefits--Employer-Sponsored Benefits May be Vulnerable to Further Erosion, United States General Accounting Office, GAO-01-374, May 2001.

2 For a discussion of this topic, see Paul Fronstin, "Retiree Health Benefits: Trends and Outlook," Employee Benefit Research Institute, Issue Brief, Number 236, August 2001.

3 Most court cases on this subject have focused on the employer’s responsibility to indicate in writing that a benefit plan may be modified or terminated. In general, if a benefit plan includes a written clause allowing modification or termination, the courts have held that such actions are legal. See, for example, Curtiss-Wright Corp. v. Schoonejongen, 514 U.S. 73, 78 (1995).

4 Total United States population, and data by age, are available from the U.S. Census Bureau at http://factfinder.census.gov/servlet/BasicFactsServlet, visited August 27, 2001.

5 For information on the financial status of Social Security, see U.S. Department of Health and Human Services, Annual Report of the Board of Trustees, Federal Old Age, Survivor, and Disability Insurance. For information on the financial status of Medicare, see U.S. Department of Health and Human Services, Heath Care Financing Administration, Annual Report of the Board of Trustees of the Federal Hospital Insurance Trust Fund. Note: The Health Care Financing Administration has recently been renamed the Centers for Medicare and Medicare Services.

6 For life expectancy data, see U.S. Department of Health and Human Services, Health United States 1999, 1999, pp. 30-31.

7 Bureau of Labor Statistics, Consumer Expenditure Survey.

8 Bureau of Labor Statistics, Consumer Price Index.

9 Self-reported health status is only one of a variety of different means of assessing the outcome of use of health care goods and services. These data, and other information on health status, may be found in Health United States, 2000, National Center for Health Statistics, 2000.

10 Katharine R. Levit, et al., "National Health Expenditures in 1997: More Slow Growth," Health Affairs (November/December 1998), pp. 99-110. See also National Health Expenditures Tables at http://www.hcfa.gov/stat/nhe-oact/tables, visited October 26, 2001.

11 For information on changing retirement ages, see William J. Wiatrowski, "Changing Retirement Age: Ups and Downs," Monthly Labor Review, April 2001, pp. 3-12.

12 For a discussion of phased retirement, see David Rajnes, Phased Retirement: Leaving the Labor Force, Employee Benefits Research Institute, EBRI Notes, Volume 22, Number 9, September 2001.

13 Data on age of women having children are from the National Center for Health Statistics. See http://www.cdc.gov/nchs/datawh/statab/unpubd/natality/natab97.htm visited August 27, 2001.

14 For a discussion of barriers to job mobility due to the availability of health care benefits, a phenomenon referred to as "job lock," see "Health care and the workplace: Recent trends and current status," in Report on the American Workforce, U. S. Department of Labor, 1995, pp. 101-135.

15 For detailed information on the cost of individually purchased health insurance, see Deborah J. Chollet and Adele M. Kirk, "Understanding Individual Health Insurance Markets: Structure, Practices, and Products in Ten States," Kaiser Family Foundation, March 1998.

16 Data on participation in employer-provided health care benefits are from the BLS Employee Benefits Survey. The 1996-98 estimate covers full-time and part-time workers in private sector establishments and State and local governments. See Compensation and Working Conditions, Winter 2000, p. 58.

17 Data on the source of health care benefits are from the U.S. Bureau of Labor Statistics Current Population Survey. Tabulations were produced by the Employee Benefit Research Institute. See EBRI Health Benefits Databook, Employee Benefit Research Institute, 1999.

18 For information on barriers to employer-provided health insurance coverage, see William J. Wiatrowski, "Who Really has Access to Employer-Provided Health Benefits?," Monthly Labor Review, June 1995, pp. 36-44.

19 Disabled individuals may continue their COBRA coverage for an additional 11 months beyond the standard 18 months. This corresponds with the 29 month Medicare eligibility period for disabled individuals (5-month eligibility for Social Security disability benefits plus an additional 24-month eligibility period for Medicare benefits). For more information on COBRA, see U.S. Department of Labor, Pension and Welfare Benefits Administration, Health Benefits Under the Consolidated Omnibus Budget Reconciliation Act (COBRA), July 1999.

20 See, for example, Curtiss-Wright Corp. v. Schoonejongen, 514 U.S. 73, 78 (1995).

21 See Erie County Retirees Association v. Erie County Employees’ Retirement Board, 140 F.Supp. 2d 466 (April 16, 2001).

22 For more information on the National Compensation Survey, see William J. Wiatrowski, "The National Compensation Survey: Compensation Statistics for the 21st Century," Compensation and Working Conditions, Winter 2000, pp. 5-14.

23 Medicare Part B, known as Medical Insurance, generally covers medical care received on an outpatient basis, such as doctors visits and diagnostic testing. Individuals covered by Medicare must pay a monthly premium to receive Part B benefits. In some cases, an employer or former employer may pay that premium for the individual. Most Medicare recipients are not required to pay premiums for Part A coverage (Hospital Insurance, covering inpatient services).

24 See Deborah J. Chollet and Adele M. Kirk, "Understanding Individual Health Insurance Markets: Structure, Practices, and Products in Ten States."

25 Data are actually collected for those employees in a sample of occupations in surveyed establishments; the sample of occupations is selected using probability techniques and is designed to represent all employees in the establishment.

26 Data on retiree health care benefits are for full-time employees in establishments with 100 workers or more. Similar data are available for employees in small establishments and for those in State and local governments. See Employee Benefits in Small Private Establishments, 1996, U.S. Bureau of Labor Statistics, Bulletin 2507, April 1999, and Employee Benefits in State and Local Governments, 1994, U.S. Bureau of Labor Statistics, Bulletin 2477, May 1996. While the survey of State and local governments was conducted again in 1998, response rates for questions on retiree health care benefits were too low to support publication.

27 Medicare coordination of benefits rules determine whether Medicare or the employer plan pays benefits first. The rules vary depending upon several factors, including the age and work status of the employee and the size of the employer. For more information, see Medicare and Other Health Benefits: Your Guide to Who Pays First, Health Care Financing Administration (now the Centers for Medicare and Medicaid Services), March 2000.