Earnings and Other Compensation Data at BLS: What Users Seek and What We Offer
by Natalie Kramer
Bureau of Labor Statistics

Originally Posted: February 26, 2003

Each day--by telephone, voice mail, and electronic mail--the Bureau of Labor Statistics receives dozens of requests for information on earnings and other forms of compensation. Why are these data in such high demand? A primary reason is that individuals want to compare their own earnings and compensation levels with those of others working in similar occupations and fields. Another is that human resource departments and businesses often use these kinds of data to guide them in establishing pay scales. In addition, unions and management employ these data in their wage negotiations, and academic and other researchers use compensation data for developing models or to support other aspects of their research. Requests have even come from divorce lawyers seeking compensation data to help them determine the appropriate levels of child support for their clients.

Many view the Department of Labor as the primary source of information on earnings and other compensation. They assume, for example, that a BLS economist can answer a question such as the following: "I am a welder in a private, 13-employee company in Stockton County, Mississippi. I have a certificate and 6½ years on the job. I am not a union member. What’s the ‘going rate’ for a guy like me?" In addition, economists at BLS often have to correct the assumption on the part of the public that employers are required to report wages paid and benefits offered to their employees to BLS. In fact, with the exception of those on the safety and health of workers, BLS surveys generally are voluntary.

The BLS programs producing compensation data are constantly seeking to improve their ways of collecting and publishing data; new programs emerge while others cease to exist or merge with others. For example, the Area Wage Surveys and the Industry Wage Surveys¹ evolved into Occupational Compensation Surveys (OCS),² which in turn were replaced by the National Compensation Survey (NCS). The OCS program provided data on pay and, periodically, rates of coverage for major employee benefits for the same preselected list of occupations in all the areas surveyed. In order to reflect the occupational mix typical of the establishments in the areas studied, occupations in the NCS are selected using probability sampling techniques.

BLS produces earnings and other compensation data in several survey programs. (See appendix A.) In the Office of Employment and Unemployment Statistics, four programs produce earnings data along with employment data. The Occupational Outlook Handbook is a comprehensive source of occupational descriptions, earnings data, and other occupation-specific information.

This article describes the earnings and other compensation data available from BLS, highlighting the differences in the various programs. First, it examines data from the Current Population Survey (CPS), the most comprehensive earnings series available by demographic characteristics. A description of earnings data collected as a part of establishment surveys follows. The article concludes with an analysis of the data on earnings and other compensation produced in the Office of Compensation and Working Conditions.

Earnings data collected as part of employment surveys

The Current Population Survey (CPS). The CPS is a household survey dating back to the 1940s. Mass unemployment during the 1930s forced government agencies to develop a way to measure accurately numbers of the unemployed. While the primary goal of the CPS is to assess the employment status of persons by various demographic characteristics, earnings data are also collected. Household members answer interviewers’ questions to the best of their ability and often on behalf of other household members not present at the time of the interview. Even though this approach might have shortcomings--the information may be less precise than that collected from establishments, for example--the CPS data can be tabulated and presented by numerous demographic characteristics, a valuable feature for many data users. The CPS presents data on weekly and hourly earnings by demographic group and full- and part-time status, as well as by occupation and industry. The demographic characteristics include age, sex, race, Hispanic origin, and marital status.

Data are collected on usual weekly earnings, which are defined as wages and salaries before taxes, and include any overtime pay, commissions, or tips usually received. The earnings are considered “usual” if they are perceived as such by the respondent. If earnings are reported in annual, monthly, biweekly, or hourly form, they are converted to weekly data by BLS.³

The Current Employment Statistics Survey (CES). The CES is an establishment-based survey, and while it focuses primarily on industry employment data, it also provides both an hourly and a weekly earnings series. These data are aggregated by detailed industry division--at two-, three-, and selected four-digit levels of the Standard Industrial Classification (SIC) coding system.⁴ CES earnings data are available for production or nonsupervisory workers only; these data are not available by occupation.

The CES program originated in 1915 and was at first limited to employment and payroll figures in four manufacturing industries. Every few years, new industries were added, and by 1932 the survey covered 91 manufacturing and 15 nonmanufacturing industries. In response to the deepening economic crisis of the 1930s, the CES was expanded to include average hourly earnings and average weekly hours. Nonmanufacturing industries were also later added to the scope of the survey. By 1940, employment estimates were available for 48 States and the District of Columbia. Employment, hours, and earnings estimates have been produced since 1949.

As mentioned previously, CES earnings data are developed for production and nonsupervisory workers only.⁵ Data by State and for many metropolitan areas are available for manufacturing industries. The earnings data are based on reports of aggregate payrolls, which include pay before deductions for Social Security, unemployment insurance, group insurance, withholding tax, salary reduction plans, bonds, and union dues. The payroll figures include supplemental pay, such as shift differentials and overtime, but exclude tips, production bonuses, commissions, and lump-sum payments unless these are paid regularly, which is often the case in wholesale trade and some other industries.

Covered Employment and Wages (ES-202). This program originated in 1935, when Congress authorized collection of data to determine the States’ compliance with the newly enacted Social Security Act. The program is a virtual census of payroll employment,⁶ resulting in comprehensive employment and wage data on workers covered by State unemployment insurance laws and the Unemployment Compensation for Federal Employees (UCFE) program.

In the ES-202 program, total annual wages, annual wages per employee, and average weekly wages are tabulated by industry, type of ownership, and establishment employment size (not by occupation, however). They are produced for metropolitan statistical areas (MSAs),⁷ counties, States, and the Nation. Annual and quarterly average wages are also produced for the same categories.

Average annual pay estimates in private industry are based on total wage figures as they are reported by employers to the Department of Labor for unemployment insurance purposes. Total wages of Federal workers are the gross amount of all payrolls. Average annual pay is computed by dividing total annual pay by annual average employment. Average weekly wages are obtained by dividing average annual pay estimates by 52, the number of weeks in a calendar year.

Definitions of pay may vary from State to State. In most States, the definition of total annual pay includes bonuses, stock options, the cash value of meals and lodging, and tips and other gratuities. Employer contributions to deferred compensation plans such as 401(k)s are considered pay in some States. Employee contributions for old-age, survivors, and disability insurance (OASDI), health insurance, unemployment insurance, workers’ compensation, and private pensions are considered pay; employer contributions to these benefits, however, are not. Money withheld for income taxes and union dues is counted as pay.

Occupational Employment Statistics (OES). This program is a cooperative effort between BLS and State employment agencies. It produces series reports⁸ on occupational employment and earnings nationally, by State, and for approximately 330 metropolitan statistical areas. It began in 1971, when 50,000 questionnaires were sent to manufacturing establishments by the Department of Labor in cooperation with 15 States. The intent was to collect data on national, State, and area occupational employment in cooperating States. Nonmanufacturing and State and local government establishments were soon added to the scope of the survey. In 1992, agricultural services were included. National occupational wage estimates were produced for the first time for 1996 and released in 1997.

Wage estimates by occupation are now produced annually by industry at the two- and three-digit SIC levels. Wages in the OES survey are defined as pay before deductions, exclusive of premium pay. Included are the base rate, cost-of-living allowances, guaranteed pay (defined as minimum pay guaranteed to incentive-paid workers), hazardous duty pay, incentive pay including commissions and production bonuses, and on-call pay. Excluded are back pay, jury-duty pay, overtime and severance pay, shift differentials, nonproduction bonuses, and tuition reimbursements.

The Occupational Outlook Handbook

The Occupational Outlook Handbook, a biennial publication of the Office of Employment Projections, is another comprehensive resource for data users. The Handbook contains brief articles for each occupation describing the educational requirements, employment prospects, working conditions, job responsibilities, and expected earnings. The articles differ in the amount and kinds of information provided for different occupations. Some include information such as pay increments gained with additional education or licensure, pay differentials between the private and the public sectors, and earnings differences by length of service.

Additional sources of information are also listed for each occupation. Most of the earnings data published in the Handbook are from the Occupational Employment Statistics program, in addition to some other sources such as the National Association of Colleges. The most important advantage of the Handbook is that it offers information on a wide variety of narrowly defined occupations. The data, however, are available at the national level only.

Data from the National Compensation Survey

The Office of Compensation and Working Conditions (OCWC) conducts the National Compensation Survey (NCS), an extensive program that encompasses an array of compensation measures. This program publishes information on occupational earnings, compensation cost trends, and employee benefits. The precursors to some of the current data series are among the oldest in the United States.

The national and local area wage surveys under the NCS replace the Occupational Compensation Program (OCS), which throughout the early 1990s provided data for local areas and the Nation as a whole. Since the survey’s primary purpose was adjustments in locality pay of Federal government employees in white-collar jobs, the list of surveyed occupations was preselected, limited, and identical in all areas.

The pilot releases of the local NCS area publications became available in late 1996. A key difference between the NCS and the OCS is the probability-based sample of occupations.⁹ The consequence of this new approach to sample selection is that earnings data are published for a wider variety of occupations.

Mean hourly earnings and mean weekly hours are presented for all private industry employees, State and local government workers, and the total non-Federal, nonagricultural workforce. Additional occupational wage data are produced by full-time and part-time status, union and nonunion status, time and incentive-pay status, major industry division, and establishment employment size. These data are available nationally, for approximately 90 metropolitan areas and nonmetropolitan counties, and for nine census divisions.¹⁰

In addition to mean hourly earnings estimates, which do not offer a detailed picture of how earnings figures are distributed within the range, hourly wage percentiles for establishment jobs are also published. The 10th, 25th, 50th (median), 75th, and 90th percentiles are presented. The percentiles describe the distribution of an occupation’s employment by the average wage rates for its jobs. For example, if at the 10th percentile, the wage for an occupation is $14.70, 10 percent of employees in this occupation earn the same as or less than an average of $14.70, and 90 percent earn $14.70 or more. Percentiles are calculated from average hourly wages for sampled establishment jobs within each occupation.¹¹

A unique feature that distinguishes the NCS wage surveys from other surveys that produce earnings series is the work-level data. Work levels are determined by assessing the duties and responsibilities of the occupation, using nine factors, including such factors as knowledge of the job, complexity, and supervision required.

Another useful feature of the NCS is pay relatives or ratios, which express an area’s average pay as a percent of national pay. They are calculated by dividing the figure for a particular area by the corresponding national figure and then multiplying by 100. For example, a pay relative of 106 percent for the Washington-Baltimore consolidated area means that pay for this locality is 6 percent above the national average. Pay relatives data are particularly sought by employees considering moving from one area to another or by employers planning to relocate or expand their businesses.

NCS mean hourly wage data by area can be retrieved through the Create Customized Tables tool. This system allows users to access a specific occupation (and work level) and obtain a wage estimate without scanning the entire online document. Estimates for multiple areas as well as occupations and levels can be produced in one inquiry.

Earnings are defined in the National Compensation Survey as straight-time hourly wages or salaries, including incentive pay, cost-of-living adjustments, and hazard pay. Vacations, holidays, nonproduction bonuses, premium pay for overtime and tips are excluded.

Since both the National Compensation Survey and the Occupational Employment Statistics program publish occupational wage estimates by major metropolitan area and for the Nation, a question might arise as to whether the two programs overlap. The answer is that there are a number of differences between them. (See appendix B.)

Employment Cost Index (ECI). The ECI is a principal Federal economic indicator and a widely used data series. The quarterly index is a measure of change in labor costs, which include wages and salaries, employer costs of benefits, and the two combined, total compensation.

The ECI originated in 1975, in the beginning covering only 3-month changes in wages and salaries in private nonfarm establishments in the 48 contiguous States. In 1978, 13 new series (union/nonunion status, manufacturing/nonmanufacturing, for example) were added and the survey was expanded to include Alaska and Hawaii. Total compensation figures, which are the sum of wages and salaries and benefits, were added in 1980. State and local governments were added in 1981, after which the survey covered the total civilian economy with the exception of the Federal work force, agricultural establishments, private households, and the self-employed. In 1991, seasonally adjusted indices for several major series were published.¹²

The ECI is a national index; however, regional¹³ indexes are available for broad data series. National data are presented for private industry and State and local government workers, and the aggregate of the two, all civilian workers. Separate series are available by selected worker and establishment characteristics, such as occupational and industrial group, establishment employment size, and union affiliation. The Employment Cost Index has been integrated into the National Compensation Survey and in the future may include local area compensation indices.

When first collected, the wage data are expressed in a variety of forms, such as hourly rates or weekly or annual salary figures. They are later converted to average wages per hour worked.¹⁴ To show cost changes free of the influences of employment shifts between occupations and industries, fixed employment weights are used to calculate the index. These industry and occupational weights are fixed over a period of approximately 10 years.¹⁵

Employer Costs for Employee Compensation (ECEC). This series shows average employer costs per hour worked of wages and salaries, benefits, and total compensation. Additionally, it shows the percent of compensation that each component represents. The ECEC series were first published in 1987, with the data initially limited to private industry. The data are presented quarterly. In 1988, data by major geographic region were added, as well as more detailed data by major industry division. Over time, new series were added. The ECEC now includes data on employer costs by various employee and establishment characteristics, such as major occupation and industry group, full- and part-time status, union affiliation, and establishment employment size.

Benefits data and plans for the future

When data users inquire about compensation, they often at the same time seek information about employee benefit costs, incidence, and provisions. The Employee Benefits Survey (EBS) has been publishing data on incidence and detailed provisions of employee benefit plans since 1979.¹⁶ In the last decade, the EBS has been conducted on three samples: medium and large private establishments on odd years, small private establishments and State and local governments on even years. With the exception of broad incidence data, which are produced by major region, all of the EBS data are national. They are presented for three broad occupational groupings: professional, technical, and related; clerical and sales; and blue-collar and service employees.¹⁷ Broad incidence data are also available by industry, union affiliation, and full- and part-time status.

Data users often express regrets that benefits data are not available by such characteristics as occupation, industry, local area, union affiliation, and nonprofit status. The Office of Compensation and Working Conditions is redesigning its approach to publishing benefits data to accommodate some of these concerns.

The last EBS bulletin in its old format contains 1998 data on State and local governments. Since then, benefits data have been published as part of the National Compensation Survey.¹⁸ Several major changes are now being envisioned. For example, there are plans to cross tabulate and publish side by side data on access to benefits and on their average costs. Data on total cost of benefits as well as the percentage born by employees (premiums and contributions) will be shown.

For the next several years, all private industry establishments--regardless of size--will be studied each year; eventually data on all civilian workers will be produced every year. Incidence and key provisions for all benefit plans and detailed plan provisions for health care and retirement plans will be studied each year; the plan is to make these data available by additional occupational and industry detail. Benefits data by such establishment and employee characteristics as establishment size, industry group, and occupational group will be added. Additionally, detailed benefits data for some of the largest metropolitan areas and nine census regions, as well as by such characteristics as profit-versus-nonprofit status and time-versus-incentive status may become available.

Finally, measures of occupational pay, employee benefit levels, and benefit costs will become more integrated in the National Compensation Program. Figures reflecting average per hour cost of benefits to employers will be published side by side with figures on corresponding benefit provisions, a novel and unique approach developed by the NCS.

Conclusion

In the age of changing information technology, the public is becoming more and more demanding of timely, accurate, and varied data on earnings and other forms of compensation. Users want to be able to obtain occupational data on wage rates, employee benefit costs, benefit availability, and detailed benefit provisions. Upon its full implementation, the National Compensation Survey will provide users with integrated, comprehensive statistics on employee compensation.

Appendix A. Earnings data series produced by the Bureau of Labor Statistics.

Appendix D. Definitions of earnings under different BLS survey programs

Appendix E. Reference information for BLS programs producing earnings data

Natalie Kramer
Economist, Division of Compensation Data Analysis and Planning, Bureau of Labor Statistics.
Telephone: (202) 691-6262, E-mail: Kramer_N@bls.gov

End Notes

1 Industry Wage Surveys’ predecessors are over a century old. The role of the initial studies of wages by industry was to measure the effects of tariffs on wages and prices. Area Wage Surveys were initiated in the 1940s to meet the need for data on pay of clerical personnel and workers performing manual jobs.

2 The OCS’ predecessor surveys were developed as a result of several statutes that called for comparability on a nationwide basis of pay in Federal government and the non-Federal sector; these were enacted in the 1960’s and 1970’s. In 1990, the BLS was commissioned to restructure these surveys to put more emphasis on locality surveys. The OCS continued to publish until 1997.

3 Definitions are from BLS Handbook of Methods, U.S. Department of Labor, Bureau of Labor Statistics, April 1997, Bulletin 2490.

4 See Standard Industrial Classification Manual, 1987 (Office of Management and Budget, 1987). BLS is expected to finish converting all of its programs to the North American Industry Classification System (NAICS) in 2005. For more on NAICS, see North American Industry Classification System: United States, 2002 (Office of Management and Budget, 2002); also, see the BLS NAICS page.

5 Data are published for production workers in manufacturing, where they are defined as employees up through the level of working supervisors, who are directly engaged in the manufacture of the company’s product. In mining and construction, similar concepts apply. In other private industries, data are published for nonsupervisory workers, who include most employees not in top executive and managerial positions.

6 Payroll employment for an establishment is defined as the number of workers on the payroll during the pay period including the 12th day of the month.

7 Metropolitan Statistical Area (MSA), as defined by the Office of Management and Budget, is a free-standing urban area with 50,000 or more inhabitants in its urbanized part and a total metropolitan population of at least 100,000 (75,000 in New England).

8 Series reports are published in economic news releases and summaries as well as online. OES does not produce hardbound publications.

9 For further information about the National Compensation Survey, please see Harriet G. Weinstein, "Overview of the NCS," Compensation and Working Conditions, Summer, 1998, pp. 41-44.

10 The census divisions are as follows: New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific.

11 Percentiles are calculated using average hourly wage rates for sampled establishment jobs within each occupation. Establishments report either individual worker earnings or average wage rates for each sampled job. If individual-worker earnings are provided, an average hourly wage rate is computed for the job and used in the calculation of the percentile estimates. The average hourly wages for each sampled job are then weighted and arrayed from lowest to highest.

12 Seasonally adjusted data show movements in labor costs that are free of seasonal influences.

13 In the ECI, the regions are as follows: Northeast, South, Midwest, and West. Plans for the future include making data available for the same 9 census divisions as in the NCS.

14 Benefit costs are also computed on an average cost per hour worked basis.

15 For more information on ECI construction and calculation, see Appendix A, Employment Cost Indexes, 1975-99, Bulletin 2532, September 2000, pp. 122-124.

16 Data on benefit costs to employers have always been published separately as a part of the Employer Costs for Employee Compensation series.

17 In State and local governments, the broad occupational groups are as follows: white-collar employees, except teachers; teachers; and blue-collar and service employees.

18 For further information on EBS and NCS, see Allan P. Blostin, "An Overview of the EBS and the NCS," in Compensation and Working Conditions, Spring 1999, pp. 2-5.

19 Data presented for all industries, private industries, and State and local government.

20 Data are published for approximately 80 metropolitan areas. The 9 census divisions are New England, Middle Atlantic, East North Central, West North Central, South Atlantic, East South Central, West South Central, Mountain, and Pacific.

21 Approximately 130 metropolitan and virtually all nonmetropolitan counties are surveyed. Although not published, nonmetropolitan area data are used to calculate Statewide and national estimates.

22 The ECI, from which Employer Costs for Employee Compensation is derived, measures changes in wages and salaries over time. It does not have wage levels data.

23 ECEC also publishes data on average benefit costs per hour worked.

24 Census regions are Northeast, South, Midwest, West.

25 Beginning with 1999 publications of national and regional data, establishments employing fewer than 50 workers have been included.

26 Legally required benefits in the preceding definition include old age, survivors, and disability insurance (OASDI), unemployment insurance, and worker’s compensation.