Categories
Employment and Earnings

Explanatory notes and estimates of error

Explanatory notes and estimates of error

Introduction

The statistics in this periodical are compiled from two major sources: (1) household interviews, and (2) reports from employers.

Data based on household interviews are obtained from the Current Population Survey (CPS), a sample survey of the population 16 years of age and over. The survey is conducted each month by the U.S. Census Bureau for the Bureau of Labor Statistics and provides comprehensive data on the labor force, the employed, and the unemployed, classified by such characteristics as age, sex, race, family relationship, marital status, occupation, and industry attachment. The survey also provides data on the characteristics and past work experience of those not in the labor force. The information is collected by trained interviewers from a sample of about 60,000 households (beginning with July 2001 data) located in 754 sample areas. These areas are chosen to represent all counties and independent cities in the United States, with coverage in 50 States and the District of Columbia. The data collected are based on the activity or status reported for the calendar week including the 12th of the month.

Data based on establishment records are compiled each month from mail questionnaires and telephone interviews by the Bureau of Labor Statistics, in cooperation with State agencies. The Current Employment Statistics (CES) survey is designed to provide industry information on nonfarm wage and salary employment, average weekly hours, average hourly earnings, and average weekly earnings for the Nation, States, and metropolitan areas. The employment, hours, and earnings series are based on payroll reports from a sample of about 350,000 establishments employing about 39 million nonfarm wage and salary workers. The data relate to all workers, full or part time, who receive pay during the payroll period that includes the 12th of the month.

RELATIONSHIP BETWEEN THE HOUSEHOLD AND ESTABLISHMENT SERIES

The household and establishment data complement one another, each providing significant types of information that the other cannot suitably supply. Population characteristics, for example, are obtained only from the household survey, whereas detailed industrial classifications are much more reliably derived from establishment reports.

Data from these two sources differ from each other because of variations in definitions and coverage, source of information, methods of collection, and estimating procedures. Sampling variability and response errors are additional reasons for discrepancies. The major factors that have a differential effect on the levels and trends of the two data series are as follows.

Employment

Coverage. The household survey definition of employment comprises wage and salary workers (including domestics and other private household workers), self-employed persons, and unpaid workers who worked 15 hours or more during the reference week in family-operated enterprises. Employment in both agricultural and nonagricultural industries is included. The payroll survey covers only wage and salary employees on the payrolls of nonfarm establishments.

Multiple jobholding. The household survey provides information on the work status of the population without duplication, because each person is classified as employed, unemployed, or not in the labor force. Employed persons holding more than one job are counted only once. In the figures based on establishment reports, persons who worked in more than one establishment during the reporting period are counted each time their names appear on payrolls.

Unpaid absences from jobs. The household survey includes among the employed all civilians who had jobs but were not at work during the reference week–that is, were not working but had jobs from which they were temporarily absent because of illness, vacation, bad weather, childcare problems, or labor-management disputes, or because they were taking time off for various other reasons, even if they were not paid by their employers for the time off. In the figures based on payroll reports, persons on leave paid for by the company are included, but those on leave without pay for the entire payroll period are not.

Hours of work

The household survey measures hours worked for all workers, whereas the payroll survey measures hours for private production or nonsupervisory workers paid for by employers. In the household survey, all persons with a job but not at work are excluded from the hours distributions and the computations of average hours at work. In the payroll survey, production or nonsupervisory employees on paid vacation, paid holiday, or paid sick leave are included and assigned the number of hours for which they were paid during the reporting period.

Earnings

The household survey measures the earnings of wage and salary workers in all occupations and industries in both the private and public sectors. Data refer to the usual earnings received from the worker’s sole or primary job. Data from the establishment survey generally refer to average earnings of production and related workers in mining and manufacturing, construction workers in construction, and nonsupervisory employees in private service-producing industries. For a comprehensive discussion of the various earnings series available from the household and establishment surveys, see BLS Measures of Compensation, Bulletin 2239 (Bureau of Labor Statistics, 1986).

COMPARABILITY OF HOUSEHOLD DATA WITH OTHER SERIES

Unemployment insurance data. The unemployed total from the household survey includes all persons who did not have a job during the reference week, were currently available for a job, and were looking for work or were waiting to be called back to a job from which they had been laid off, whether or not they were eligible for unemployment insurance. Figures on unemployment insurance claims, prepared by the Employment and Training Administration of the U.S. Department of Labor, exclude, in addition to otherwise ineligible persons who do not file claims for benefits, persons who have exhausted their benefit rights, new workers who have not earned rights to unemployment insurance, and persons losing jobs not covered by unemployment insurance systems (some workers in agriculture, domestic services, and religious organizations, and self-employed and unpaid family workers).

In addition, the qualifications for drawing unemployment compensation differ from the definition of unemployment used in the household survey. For example, persons with a job but not at work and persons working only a few hours during the week are sometimes eligible for unemployment compensation but are classified as employed, rather than unemployed, in the household survey.

Agricultural employment estimates of the U.S. Department of Agriculture. The principal differences in coverage are the inclusion of persons under 16 in the National Agricultural Statistics Service series and the treatment of dual jobholders, who are counted more than once if they work on more than one farm during the reporting period. There also are wide differences in sampling techniques and data collecting and estimating methods, which cannot be readily measured in terms of their impact on differences in the levels and trends of the two series.

COMPARABILITY OF PAYROLL EMPLOYMENT DATA WITH OTHER SERIES

Statistics on manufacturers and business, U.S. Census Bureau. BLS establishment statistics on employment differ from employment counts derived by the U.S. Census Bureau from its censuses or sample surveys of manufacturing and business establishments. The major reasons for noncomparability are different treatment of business units considered parts of an establishment, such as central administrative offices and auxiliary units; the industrial classification of establishments; and different reporting patterns by multiunit companies. There also are differences in the scope of the industries covered–for example, the Census of Business excludes professional services, public utilities, and financial establishments, whereas these are included in the BLS statistics.

County Business Patterns, U.S. Census Bureau. Data in County Business Patterns (CBP) differ from BLS establishment statistics in the treatment of central administrative offices and auxiliary units. Differences also may arise because of industrial classification and reporting practices. In addition, CBP excludes interstate railroads and most of government, and coverage is incomplete for some of the nonprofit agencies.

Employment covered by State unemployment insurance programs. Most nonfarm wage and salary workers are covered by the unemployment insurance programs. However, some employees, such as those working in parochial schools and churches, are not covered by unemployment insurance, whereas they are included in the BLS establishment statistics.

Household Data (“A” tables, monthly; “D” tables, quarterly)

COLLECTION AND COVERAGE

Statistics on the employment status of the population and related data are compiled by BLS using data from the Current Population Survey (CPS). This monthly survey of households is conducted for BLS by the U.S. Census Bureau through a scientifically selected sample designed to represent the civilian noninstitutional population. Respondents are interviewed to obtain information about the employment status of each member of the household 16 years of age and older. The inquiry relates to activity or status during the calendar week, Sunday through Saturday, that includes the 12th day of the month. This is known as the “reference week.” Actual field interviewing is conducted in the following week, referred to as the “survey week.”

Each month, about 60,000 occupied units are eligible for interview. Some 4,500 of these households are contacted but interviews are not obtained because the occupants are not at home after repeated calls or are unavailable for other reasons. This represents a noninterview rate for the survey that ranges between 7 and 8 percent. In addition to the 60,000 occupied units, there are about 12,000 sample units in an average month that are visited but found to be vacant or otherwise not eligible for enumeration. Part of the sample is changed each month. The rotation plan, as will be explained later, provides for three-fourths of the sample to be common from one month to the next, and one-half to be common with the same month a year earlier.

CONCEPTS AND DEFINITIONS

The concepts and definitions underlying labor force data have been modified, but not substantially altered, since the inception of the survey in 1940; those in use as of January 1994 are as follows:

Civilian noninstitutional population. Included are persons 16 years of age and older residing in the 50 States and the District of Columbia who are not inmates of institutions (for example, penal and mental facilities, homes for the aged), and who are not on active duty in the Armed Forces.

Employed persons. All persons who, during the reference week, (a) did any work at all (at least 1 hour) as paid employees, worked in their own business, profession, or on their own farm, or worked 15 hours or more as unpaid workers in an enterprise operated by a member of the family, and (b) all those who were not working but who had jobs or businesses from which they were temporarily absent because of vacation, illness, bad weather, childcare problems, maternity or paternity leave, labor-management dispute, job training, or other family or personal reasons, whether or not they were paid for the time off or were seeking other jobs.

Each employed person is counted only once, even if he or she holds more than one job. For purposes of occupation and industry classification, multiple jobholders are counted in the job at which they worked the greatest number of hours during the reference week.

Included in the total are employed citizens of foreign countries who are temporarily in the United States but not living on the premises of an embassy. Excluded are persons whose only activity consisted of work around their own house (painting, repairing, or own home housework) or volunteer work for religious, charitable, and other organizations.

Unemployed persons. All persons who had no employment during the reference week, were available for work, except for temporary illness, and had made specific efforts to find employment sometime during the 4-week period ending with the reference week. Persons who were waiting to be recalled to a job from which they had been laid off need not have been looking for work to be classified as unemployed.

Duration of unemployment. This represents the length of time (through the current reference week) that persons classified as unemployed had been looking for work. For persons on layoff, duration of unemployment represents the number of full weeks they had been on layoff. Mean duration is the arithmetic average computed from single weeks of unemployment; median duration is the midpoint of a distribution of weeks of unemployment.

Reason for unemployment. Unemployment also is categorized according to the status of individuals at the time they began to look for work. The reasons for unemployment are divided into five major groups: (1) Job losers, comprising (a) persons on temporary layoff, who have been given a date to return to work or who expect to return within 6 months (persons on layoff need not be looking for work to qualify as unemployed), and (b) permanent job losers, whose employment ended involuntarily and who began looking for work; (2) Job leavers, persons who quit or otherwise terminated their employment voluntarily and immediately began looking for work; (3) Persons who completed temporary jobs, who began looking for work after the jobs ended; (4) Reentrants, persons who previously worked but who were out of the labor force prior to beginning their job search; and (5) New entrants, persons who had never worked. Each of these five categories of the unemployed can be expressed as a proportion of the entire civilian labor force; the sum of the four rates thus equals the unemployment rate for all civilian workers. (For statistical presentation purposes, “job losers” and “persons who completed temporary jobs” are combined into a single category until seasonal adjustments can be developed for the separate categories.)

Jobseekers. All unemployed persons who made specific efforts to find a job sometime during the 4-week period preceding the survey week are classified as jobseekers. Jobseekers do not include persons classified as on temporary layoff, who, although often looking for work, are not required to do so to be classified as unemployed. Jobseekers are grouped by the methods used to seek work. Only active methods–which have the potential to result in a job offer without further action on the part of the jobseeker—qualify as job search. Examples include going to an employer directly or to a public or private employment agency, seeking assistance from friends or relatives, placing or answering ads, or using some other active method. Examples of the “other” category include being on a union or professional register, obtaining assistance from a community organization, or waiting at a designated labor pickup point. Passive methods, which do not qualify as job search, include reading (as opposed to answering or placing) “help wanted” ads and taking a job training course.

Labor force. This group comprises all persons classified as employed or unemployed in accordance with the criteria described above.

Unemployment rate. The unemployment rate represents the number unemployed as a percent of the labor force.

Participation rate. This represents the proportion of the population that is in the labor force.

Employment-population ratio. This represents the proportion of the population that is employed.

Not in the labor force. Included in this group are all persons in the civilian noninstitutional population who are neither employed nor unemployed. Information is collected on their desire for and availability to take a job at the time of the CPS interview, job search activity in the prior year, and reason for not looking in the 4-week period prior to the survey week. This group includes discouraged workers, defined as persons not in the labor force who want and are available for a job and who have looked for work sometime in the past 12 months (or since the end of their last job if they held one within the past 12 months), but who are not currently looking because they believe there are no jobs available or there are none for which they would qualify.

Persons classified as not in the labor force who are in the sample for either their fourth or eighth month are asked additional questions relating to job history and workseeking intentions. These latter data are available on a quarterly basis.

Occupation, industry, and class of worker. This information for the employed applies to the job held in the reference week. Persons with two or more jobs are classified in the job at which they worked the greatest number of hours. The unemployed are classified according to their last job. The occupational and industrial classification of CPS data is based on the coding systems used in the 1990 census.

The class-of-worker breakdown assigns workers to the following categories: Private and government wage and salary workers, self-employed workers, and unpaid family workers. Wage and salary workers receive wages, salary, commissions, tips, or pay in kind from a private employer or from a government unit. Self-employed persons are those who work for profit or fees in their own business, profession, trade, or farm. Only the unincorporated self-employed are included in the self-employed category in the class-of-worker typology. Self-employed persons who respond that their businesses are incorporated are included among wage and salary workers because, technically, they are paid employees of a corporation. Unpaid family workers are persons working without pay for 15 hours a week or more on a farm or in a business operated by a member of the household to whom they are related by birth or marriage.

Multiple jobholders. These are employed persons who, during the reference week, either had two or more jobs as a wage and salary worker, were self-employed and also held a wage and salary job, or worked as an unpaid family worker and also held a wage and salary job. Excluded are self-employed persons with multiple businesses and persons with multiple jobs as unpaid family workers.

Hours of work. These statistics relate to the actual number of hours worked during the reference week. For example, persons who normally work 40 hours a week but were off on the Columbus Day holiday would be reported as working 32 hours, even though they were paid for the holiday. For persons working in more than one job, the published figures relate to the number of hours worked in all jobs during the week; all the hours are credited to the major job. Unpublished data are available for the hours worked in each job and for usual hours.

At work part time for economic reasons. Sometimes referred to as involuntary part time, this category refers to individuals who gave an economic reason for working 1 to 34 hours during the reference week. Economic reasons include slack work or unfavorable business conditions, inability to find full-time work, and seasonal declines in demand. Those who usually work part time must also indicate that they want and are available for full-time work to be classified as on part time for economic reasons.

At work part time for noneconomic reasons. This group includes those persons who usually work part time and were at work 1 to 34 hours during the reference week for a noneconomic reason. Noneconomic reasons include, for example: Illness or other medical limitations, childcare problems or other family or personal obligations, school or training, retirement or Social Security limits on earnings, and being in a job where full-time work is less than 35 hours. The group also includes those who gave an economic reason for usually working 1 to 34 hours but said they do not want to work full time or are unavailable for such work.

Usual full- or part-time status. Data on persons “at work” exclude persons who were temporarily absent from a job and therefore classified in the zero-hours-worked category, “with a job but not at work.” These are persons who were absent from their jobs for the entire week for such reasons as bad weather, vacation, illness, or involvement in a labor dispute. In order to differentiate a person’s normal schedule from his or her activity during the reference week, persons also are classified according to their usual full- or part-time status. In this context, full-time workers are those who usually worked 35 hours or more (at all jobs combined). This group will include some individuals who worked less than 35 hours in the reference week for either economic or noneconomic reasons and those who are temporarily absent from work. Similarly, part-time workers are those who usually work less than 35 hours per week (at all jobs), regardless of the number of hours worked in the reference week. This may include some individuals who actually worked more than 34 hours in the reference week, as well as those who are temporarily absent from work. The full-time labor force includes all employed persons who usually work full time and unemployed persons who are either looking for full-time work or are on layoff from full-time jobs. The part-time labor force consists of employed persons who usually work part time and unemployed persons who are seeking or are on layoff from part-time jobs. Unemployment rates for full- and part-time workers are calculated using the concepts of the full- and part-time labor force.

White, black, and other. These are terms used to describe the race of persons. Included in the “other” group are American Indians, Alaskan Natives, and Asians and Pacific Islanders. Because of the relatively small sample size, data for “other” races are not published. In the enumeration process, race is determined by the household respondent.

Hispanic origin. This refers to persons who identified themselves in the enumeration process as Mexican, Puerto Rican, Cuban, Central or South American, or of other Hispanic origin or descent. Persons of Hispanic origin may be of any race; thus, they are included in both the white and black population groups.

Vietnam-era veterans. These are persons who served in the Armed Forces of the United States between August 5, 1964, and May 7, 1975. Published data are limited to men in the civilian noninstitutional population; that is, veterans in institutions and women are excluded. Nonveterans are persons who never served in the Armed Forces.

Usual weekly earnings. Data represent earnings before taxes and other deductions, and include any overtime pay, commissions, or tips usually received (at the main job, in the case of multiple jobholders). Earnings reported on a basis other than weekly (for example, annual, monthly, hourly) are converted to weekly. The term “usual” is as perceived by the respondent. If the respondent asks for a definition of usual, interviewers are instructed to define the term as more than half the weeks worked during the past 4 or 5 months. Data refer to wage and salary workers (excluding all self employed persons regardless of whether their businesses were incorporated) who usually work full time on their sole or primary job.

Median earnings. These figures indicate the value that divides the earnings distribution into two equal parts, one part having values above the median and the other having values below the median. The medians shown in this publication are calculated by linear interpolation of the $50 centered interval within which each median falls. Data expressed in constant dollars are deflated by the Consumer Price Index for All Urban Consumers (CPI-U).

Single, never married; married, spouse present; and other marital status. These are the terms used to define the marital status of individuals at the time of interview. Married, spouse present, applies to husband and wife if both were living in the same household, even though one may be temporarily absent on business, on vacation, on a visit, in a hospital, etc. Other marital status applies to persons who are married, spouse absent; widowed; or divorced. Married, spouse absent relates to persons who are separated due to marital problems, as well as to husbands and wives who are living apart because one or the other was employed elsewhere or was on duty with the Armed Forces, or for any other reasons.

Household. A household consists of all persons–related family members and all unrelated persons–who occupy a housing unit and have no other usual address. A house, an apartment, a group of rooms, or a single room is regarded as a housing unit when occupied or intended for occupancy as separate living quarters. A householder is the person (or one of the persons) in whose name the housing unit is owned or rented. The term is never applied to either husbands or wives in married-couple families but relates only to persons in families maintained by either men or women without a spouse.

Family. A family is defined as a group of two or more persons residing together who are related by birth, marriage, or adoption; all such persons are considered as members of one family. Families are classified either as married-couple families or as families maintained by women or men without spouses. A family maintained by a woman or a man is one in which the householder is either single, widowed, divorced, or married, spouse absent.

HISTORICAL COMPARABILITY

Changes in concepts and methods

While current survey concepts and methods are very similar to those introduced at the inception of the survey in 1940, a number of changes have been made over the years to improve the accuracy and usefulness of the data. Some of the most important changes include:

* In 1945, the questionnaire was radically changed with the introduction of four basic employment questions. Prior to that time, the survey did not contain specific question wording, but, rather, relied on a complicated scheme of activity prioritization.

* In 1953, the current 4-8-4 rotation system was adopted, whereby households are interviewed for 4 consecutive months, leave the sample for 8 months, and then return to the sample for the same 4 months of the following year. Before this system was introduced, households were interviewed for 6 consecutive months and then replaced. The new system provided some year-to-year overlap in the sample, thereby improving measurement over time.

* In 1955, the survey reference week was changed to the calendar week including the 12th day of the month, for greater consistency with the reference period used for other labor-related statistics. Previously, the calendar week containing the 8th day of the month had been used as the reference week.

* In 1957, the employment definition was modified slightly as a result of a comprehensive interagency review of labor force concepts and methods. Two relatively small groups of persons classified as employed, under “with a job but not at work,” were assigned to different classifications. Persons on layoff with definite instructions to return to work within 30 days of the layoff date, and persons volunteering that they were waiting to start a new wage and salary job within 30 days of interview, were, for the most part, reassigned to the unemployed classification. The only exception was the small subgroup in school during the reference week but waiting to start new jobs, which was transferred to not in the labor force.

* In 1967, more substantive changes were made as a result of the recommendations of the President’s Committee to Appraise Employment and Unemployment Statistics (the Gordon Committee). The principal improvements were as follows:

a) A 4-week job search period and specific questions on jobseeking activity were introduced. Previously, the questionnaire was ambiguous as to the period for jobseeking, and there were no specific questions concerning job search methods.

b) An availability test was introduced whereby a person must be currently available for work in order to be classified as unemployed. Previously, there was no such requirement. This revision to the concept mainly affected students, who, for example, may begin to look for summer jobs in the spring although they will not be available until June or July. Such persons, until 1967, had been classified as unemployed but since have been assigned to the “not in the labor force” category.

c) Persons “with a job but not at work” because of strikes, bad weather, etc., who volunteered that they were looking for work were shifted from unemployed status to employed.

d) The lower age limit for official statistics on employment, unemployment, and other labor force concepts was raised from 14 to 16 years. Historical data for most major series have been revised to provide consistent information based on the new minimum age limit.

e) New questions were added to obtain additional information on persons not in the labor force, including those referred to as “discouraged workers,” defined as persons who indicate that they want a job but are not currently looking because they believe there are no jobs available or none for which they would qualify.

f) New “probing” questions were added to the questionnaire in order to increase the reliability of information on hours of work, duration of unemployment, and self-employment.

* In 1994, major changes to the Current Population Survey (CPS) were introduced, which included a complete redesign of the questionnaire and the use of computer-assisted interviewing for the entire survey. In addition, there were revisions to some of the labor force concepts and definitions, including the implementation of some changes recommended in 1979 by the National Commission on Employment and Unemployment Statistics (NCEUS, also known as the Levitan Commission). Some of the major changes to the survey were:

a) The introduction of a redesigned and automated questionnaire. The CPS questionnaire was totally redesigned in order to obtain more accurate, comprehensive, and relevant information, and to take advantage of state-of-the-art computer interviewing techniques.

b) The addition of two, more objective, criteria to the definition of discouraged workers. Prior to 1994, to be classified as a discouraged worker, a person must have wanted a job and been reported as not currently looking because of a belief that no jobs were available or that there were none for which he or she would qualify. Beginning in 1994, persons classified as discouraged must also have looked for a job within the past year (or since their last job, if they worked during the year), and must have been available for work during the reference week (a direct question on availability was added in 1994; prior to 1994, availability had been inferred from responses to other questions). These changes were made because the NCEUS and others felt that the previous definition of discouraged workers was too subjective, relying mainly on an individual’s stated desire for a job and not on prior testing of the labor market.

c) Similarly, the identification of persons employed part time for economic reasons (working less than 35 hours in the reference week because of poor business conditions or because of an inability to find full-time work) was tightened by adding two new criteria for persons who usually work part time: They must want and be available for fulltime work. Previously, such information was inferred. (Persons who usually work full time but worked part time for an economic reason during the reference week are assumed to meet these criteria.)

d) Specific questions were added about the expectation of recall for persons who indicate that they are on layoff. To be classified as “on temporary layoff,” persons must expect to be recalled to their jobs. Previously, the questionnaire did not include explicit questions about the expectation of recall.

e) Persons volunteering that they were waiting to start a new job within 30 days must have looked for work in the 4 weeks prior to the survey in order to be classified as unemployed. Previously, such persons did not have to meet the job search requirement in order to be included among the unemployed.

For additional information on changes in CPS concepts and methods, see “The Current Population Survey: Design and Methodology,” Technical Paper 63 (Washington, U.S. Census Bureau and Bureau of Labor Statistics, March 2000), available on the Internet at www.bls.census.gov/cps/tp/tp63.htm; “Overhauling the Current Population Survey–Why is it Necessary to Change?,” “Redesigning the Questionnaire,” and “Evaluating Changes in the Estimates,” Monthly Labor Review, September 1993; and “Revisions in the Current Population Survey Effective January 1994,” in the February 1994 issue of this publication.

Noncomparability of labor force levels

In addition to the refinements in concepts, definitions, and methods made over the years, other changes also have affected the comparability of the labor force data.

* Beginning in 1953, as a result of introducing data from the 1950 census into the estimating procedures, population levels were raised by about 600,000; labor force, total employment, and agricultural employment were increased by about 350,000, primarily affecting the figures for totals and for men; other categories were relatively unaffected.

* Beginning in 1960, the inclusion of Alaska and Hawaii resulted in increases of about 500,000 in the population and about 300,000 in the labor force. Four-fifths of the labor force increase was in nonagricultural employment; other labor force categories were not appreciably affected.

* Beginning in 1962, the introduction of data from the 1960 census reduced the population by about 50,000 and labor force and employment by about 200,000; unemployment totals were virtually unchanged.

* Beginning in 1972, information from the 1970 census was introduced into the estimation procedures, increasing the population by about 800,000; labor force and employment totals were raised by a little more than 300,000; unemployment levels and rates were essentially unchanged.

* In March 1973, a subsequent population adjustment based on the 1970 census was introduced. This adjustment, which affected the white and black-and-other groups but had little effect on totals, resulted in the reduction of nearly 300,000 in the white population and an increase of the same magnitude in the black-and-other population. Civilian labor force and total employment figures were affected to a lesser degree; the white labor force was reduced by 150,000, and the black-and-other labor force rose by about 210,000. Unemployment levels and rates were not significantly affected.

* Beginning in January 1974, the method used to prepare independent estimates of the civilian noninstitutional population was modified to an “inflation-deflation” approach. This change in the derivation of the estimates had its greatest impact on estimates of 20- to 24-year-old men–particularly those in the black-and-other population–but had little effect on estimates of the total population 16 years and over. Additional information on the adjustment procedure appears in “CPS Population Controls Derived from Inflation-Deflation Method of Estimation,” in the February 1974 issue of this publication.

* Effective in July 1975, as a result of the large inflow of Vietnamese refugees to the United States, the total and black-and-other independent population controls for persons 16 years and over were adjusted upward by 76,000–30,000 men and 46,000 women. The addition of the refugees increased the black-and-other population by less than 1 percent in any age-sex group, with all of the changes being confined to the “other” component of the population.

* Beginning in January 1978, the introduction of an expansion in the sample and revisions in the estimation procedures resulted in an increase of about 250,000 in the civilian labor force and employment totals; unemployment levels and rates were essentially unchanged. An explanation of the procedural changes and an indication of the differences appear in “Revisions in the Current Population Survey in January 1978” in the February 1978 issue of this publication.

* Beginning in October 1978, the race of the individual was determined by the household respondent for the incoming rotation group households, rather than by the interviewer as before. The purpose of this change was to provide more accurate estimates of characteristics by race. Thus, in October 1978, one-eighth of the sample households had race determined by the household respondent and seven-eighths of the sample households had race determined by interviewer observation. It was not until January 1980 that the entire sample had race determined by the household respondent. The new procedure had no significant effect on the estimates.

* Beginning in January 1979, the first-stage ratio adjustment method was changed in the CPS estimation procedure. Differences between the old and new procedures existed only for metropolitan and nonmetropolitan area estimates, not for the total United States. The reasoning behind the change and an indication of the differences appear in “Revisions in the Current Population Survey in January 1979” in the February 1979 issue of this publication.

* Beginning in January 1982, the second-stage ratio adjustment method was changed. The rationale for the change and an indication of its effect on national estimates of labor force characteristics appear in “Revisions in the Current Population Survey Beginning in January 1982” in the February 1982 issue of this publication. In addition, current population estimates used in the second-stage estimation procedure were derived from information obtained from the 1980 census, rather than the 1970 census. This change caused substantial increases in the total population and in the estimates of persons in all labor force categories. Rates for labor force characteristics, however, remained virtually unchanged. Some 30,000 labor force series were adjusted back to 1970 to avoid major breaks in series. The adjustment procedure used also is described in the February 1982 article cited above. The revisions did not, however, smooth out the breaks in series occurring between 1972 and 1979 (described above), and data users should consider them when comparing estimates from different periods.

* Beginning in January 1983, the first-stage ratio adjustment method was updated to incorporate data from the 1980 census. The rationale for the change and an indication of its effect on national estimates for labor force characteristics appear in “Revisions in the Current Population Survey Beginning in January 1983” in the February 1983 issue of this publication. There were only slight differences between the old and new procedures in estimates of levels for the various labor force characteristics and virtually no differences in estimates of participation rates.

* Beginning in January 1985, most of the steps of the CPS estimation procedure–the noninterview adjustment, the first- and second-stage ratio adjustments, and the composite estimator–were revised. These procedures are described in the Estimating Methods section. A description of the changes and an indication of their effect on national estimates of labor force characteristics appear in “Changes in the Estimation Procedure in the Current Population Survey Beginning in January 1985” in the February 1985 issue of this publication. Overall, the revisions had only a slight effect on most estimates. The greatest impact was on estimates of persons of Hispanic origin. Major estimates were revised back to January 1980.

* Beginning in January 1986, the population controls used in the second-stage ratio adjustment method were revised to reflect an explicit estimate of the number of undocumented immigrants (largely Hispanic) since 1980 and an improved estimate of the number of emigrants among legal foreign-born residents for the same period. As a result, the total civilian population and labor force estimates were raised by nearly 400,000; civilian employment was increased by about 350,000. The Hispanic-origin population and labor force estimates were raised by about 425,000 and 305,000, respectively, and Hispanic employment was increased by 270,000. Overall and subgroup unemployment levels and rates were not significantly affected. Because of the magnitude of the adjustments for Hispanics, data were revised back to January 1980 to the extent possible. An explanation of the changes and an indication of their effect on estimates of labor force characteristics appear in “Changes in the Estimation Procedure in the Current Population Survey Beginning in January 1986” in the February 1986 issue of this publication.

* Beginning in August 1989, the second-stage ratio estimation procedures were changed slightly to decrease the chance of very small cells occurring and to be more consistent with published age, sex, race cells. This change had virtually no effect on national estimates.

* Beginning in January 1994, 1990 census-based population controls, adjusted for the estimated undercount, were introduced into the second-stage estimation procedure. This change resulted in substantial increases in total population and in all major labor force categories. Effective February 1996, these controls were introduced into the estimates for 1990-93. Under the new population controls, the civilian noninstitutional population for 1990 increased by about 1.1 million, employment by about 880,000, and unemployment by approximately 175,000. The overall unemployment rate rose by about 0.1 percentage point. For further information, see “Revisions in the Current Population Survey Effective January 1994,” and “Revisions in Household Survey Data Effective February 1996” in the February 1994 and March 1996 issues, respectively, of this publication.

Additionally, for the period January through May 1994, the composite estimation procedure was suspended for technical and logistical reasons.

* Beginning in January 1997, the population controls used in the second-stage ratio adjustment method were revised to reflect updated information on the demographic characteristics of immigrants to, and emigrants from, the United States. As a result, the civilian noninstitutional population 16 years and over was raised by about 470,000. The labor force and employment levels were increased by about 320,000 and 290,000, respectively. The Hispanic-origin population and labor force estimates were raised by about 450,000 and 250,000, respectively, and Hispanic employment was increased by 325,000. Overall and subgroup unemployment rates and other percentages of labor market participation were not affected. An explanation of the changes and an indication of their effect on national labor force estimates appear in “Revisions in the Current Population Survey Effective January 1997” in the February 1997 issue of this publication.

* Beginning in January 1998, new composite estimation procedures and minor revisions in the population controls were introduced into the household survey. The new composite estimation procedures simplify processing of the monthly labor force data at BLS, allow users of the survey microdata to more easily replicate the official estimates released by BLS, and increase the reliability of the employment and labor force estimates. The new procedures also produce somewhat lower estimates of the civilian labor force and employment and slightly higher estimates of unemployment. For example, based on 1997 annual average data, the differences resulting from the use of old and new composite weights were as follows: Civilian labor force (-229,000), total employed (-256,000), and total unemployed (+27,000). Unemployment rates were not significantly affected.

Also beginning in January 1998, the population controls used in the survey were revised to reflect new estimates of legal immigration to the United States and a change in the method for projecting the emigration of foreign-born legal residents. As a result, the Hispanic-origin population was raised by about 57,000; however, the total civilian noninstitutional population 16 years and over was essentially unchanged. More detailed information on these changes and their effect on the estimates of labor force change and composition appear in “Revisions in the Current Population Survey Effective January 1998,” in the February 1998 issue of this publication.

* Beginning in January 1999, the population controls used in the survey were revised to reflect newly updated information on immigration. As a result, the civilian noninstitutional population 16 years and over was raised by about 310,000. The impact of the changes varied for different demographic groups. The civilian noninstitutional population for men 16 years and over was lowered by about 185,000, while that for women was increased by about 490,000. The Hispanic-origin population was lowered by about 165,000 while that of persons of non-Hispanic origin was raised by about 470,000. Overall labor force and employment levels were increased by about 60,000 each, while the Hispanic labor force and employment estimates were reduced by about 225,000 and 215,000, respectively. The changes had only a small impact on overall and subgroup unemployment rates and other percentages of labor market participation. An explanation of the changes and an indication of their effect on national labor force estimates appear in “Revisions in the Current Population Survey Effective January 1999” in the February 1999 issue of this publication.

* Beginning in January 2000, the population controls used in the survey were revised to reflect newly updated information on immigration and an upward revision in the number of deaths. As a result, the civilian noninstitutional population 16 years and over was lowered by about 215,000. The labor force and employment levels were decreased by about 125,000 and 120,000, respectively. Overall and subgroup unemployment rates and other percentages of labor market participation were not significantly affected. An explanation of the changes and an indication of their effect on national labor force estimates appear in “Revisions in the Current Population Survey Effective January 2000” in the February 2000 issue of this publication.

Changes in the occupational and industrial classification systems

Beginning in 1971, the comparability of occupational employment data was affected as a result of changes in the occupational classification system for the 1970 census that were introduced into the CPS. Comparability was further affected in December 1971, when a question relating to major activity or duties was added to the monthly CPS questionnaire in order to more precisely determine the occupational classification of individuals. As a result of these changes, meaningful comparisons of occupational employment levels could not be made between 1971-72 and prior years nor between those 2 years. Unemployment rates were not significantly affected. For a further explanation of the changes in the occupational classification system, see “Revisions in Occupational Classifications for 1971” and “Revisions in the Current Population Survey” in the February 1971 and February 1972 issues, respectively, of this publication.

Beginning in January 1983, the occupational and industrial classification systems used in the 1980 census were introduced into the CPS. The 1980 census occupational classification system evolved from the Standard Occupational Classification (SOC) system and was so radically different in concepts and nomenclature from the 1970 system that comparisons of historical data are not possible without major adjustments. For example, the 1980 major group “sales occupations” is substantially larger than the 1970 category “sales workers.” Major additions include “cashiers” from “clerical workers” and some self-employed proprietors in retail trade establishments from “managers and administrators, except farm.”

The industrial classification system used in the 1980 census was based on the 1972 Standard Industrial Classification (SIC) system, as modified in 1977. The adoption of the new system had much less of an adverse effect on historical comparability than did the new occupational system. The most notable changes from the 1970 system were the transfer of farm equipment stores from “retail” to “wholesale” trade and of postal service from “public administration” to “transportation,” and some interchange between “professional and related services” and “public administration.” Additional information on the 1980 census occupational and industrial classification systems appears in “Revisions in the Current Population Survey Beginning in January 1983” in the February 1983 issue of this publication.

Beginning in January 1992, the occupational and industrial classification systems used in the 1990 census were introduced into the CPS. (These systems were based largely on the 1980 Standard Occupational Classification (SOC) and 1987 Standard Industrial Classification (SIC) systems, respectively.) There were a few breaks in comparability between the 1980 and 1990 census-based systems, particularly within the “technical, sales, and administrative support” categories. The most notable changes in industry classification were the shift of several industries from “business services” to “professional services” and the splitting of some industries into smaller, more detailed categories. A number of industry titles were changed as well, with no change in content.

Sampling

Since the inception of the survey, there have been various changes in the design of the CPS sample. The sample traditionally is redesigned and a new sample selected after each decennial census. Also, the number of sample areas and the number of sample persons are changed occasionally. Most of these changes are made to improve the efficiency of the sample design, increase the reliability of the sample estimates, or control cost.

Changes in this regard since 1960 are as follows: When Alaska and Hawaii received statehood in 1959 and 1960, respectively, three sample areas were added to the existing sample to account for the population of these States. In January 1978, a supplemental sample of 9,000 housing units, selected in 24 States and the District of Columbia, was designed to provide more reliable annual average estimates for States. In October 1978, a coverage improvement sample of approximately 450 sample household units representing 237,000 occupied mobile homes and 600,000 new construction housing units was added. In January 1980, another supplemental sample of 9,000 households selected in 32 States and the District of Columbia was added. A sample reduction of about 6,000 units was implemented in May 1981. In January 1982, the sample was expanded by 100 households to provide additional coverage in counties added to the Standard Metropolitan Statistical Areas (SMSAs), which were redefined in 1973. In January 1985, a new State-based CPS sample was selected based on 1980 census information. A sample reduction of about 4,000 households was implemented in April 1988; the households were reinstated during the 8-month period, April-November 1989. A redesigned CPS sample based on the 1990 decennial census was selected for use during the 1990s. Households from this new sample were phased into the CPS between April 1994 and July 1995. The July 1995 sample was the first monthly sample based entirely on the 1990 census. For further information on the 1990 sample redesign, see “Redesign of the Sample for the Current Population Survey” in the May 1994 issue of this publication.

The original 1990 census-based sample design included about 66,000 housing units per month located in 792 selected geographic areas called primary sampling units (PSUs). The sample initially was selected to meet specific reliability criteria for the Nation, for each of the 50 States and the District of Columbia, and for the sub-State areas of New York City and the Los Angeles-Long Beach metropolitan area. In 1996, the original sample design reliability criteria were modified to reduce costs. In July 2001, the CPS sample was expanded to support the State Children’s Health Insurance Program. For further information on the sample expansion, see “Expansion of the Current Population Survey Sample Effective July 2001” in the August 2001 issue of this publication. The current criteria, given below, are based on the coefficient of variation (CV) of the unemployment level, where the CV is defined as the standard error of the estimate divided by the estimate, expressed as a percentage. These CV controls assume a 6-percent unemployment rate to establish a consistent specification of sampling error.

The current sample design, introduced in July 2001, includes about 72,000 “assigned” households from 754 sample areas. Sufficient sample is allocated to maintain, at most, a 1.9-percent CV on national monthly estimates of unemployment level, assuming a 6-percent unemployment rate. This translates into a change of 0.2 percentage point in the unemployment rate being significant at a 90-percent confidence level. For each of the 50 States and for the District of Columbia, the design maintains a CV of at most 8 percent on the annual average estimate of unemployment level, assuming a 6-percent unemployment rate. About 60,000 assigned households are required in order to meet the national and State reliability criteria. Due to the national reliability criterion, estimates for several large States are substantially more reliable than the State design criterion requires. Annual average unemployment estimates for California, Florida, New York, and Texas, for example, carry a CV of less than 4 percent. In support of the State Children’s Health Insurance Program, about 12,000 additional households are allocated to the District of Columbia and 31 States. (These are generally the States with the smallest samples after the 60,000 households are allocated to satisfy the national and State reliability criteria.)

In the first stage of sampling, the 754 sample areas are chosen. In the second stage, ultimate sampling unit clusters composed of about four housing units each are selected. Each month, about 72,000 housing units are assigned for data collection, of which about 60,000 are occupied and thus eligible for interview. The remainder are units found to be destroyed, vacant, converted to nonresidential use, containing persons whose usual place of residence is elsewhere, or ineligible for other reasons. Of the 60,000 housing units, about 7.5 percent are not interviewed in a given month due to temporary absence (vacation, etc.), other failures to make contact after repeated attempts, inability of persons contacted to respond, unavailability for other reasons, and refusals to cooperate (about half of the noninterviews). Information is obtained each month for about 112,000 persons 16 years of age or older.

Selection of sample areas. The entire area of the United States, consisting of 3,141 counties and independent cities, is divided into 2,007 sample units (PSUs). In most States, a PSU consists of a county or a number of contiguous counties. In New England and Hawaii, minor civil divisions are used instead of counties.

Metropolitan areas within a State are used as a basis for forming PSUs. Outside of metropolitan areas, counties normally are combined except when the geographic area of an individual county is too large. Combining counties to form PSUs provides greater heterogeneity; a typical PSU includes urban and rural residents of both high and low economic levels and encompasses, to the extent feasible, diverse occupations and industries. Another important consideration is that the PSU be sufficiently compact so that, with a small sample spread throughout, it can be efficiently canvassed without undue travel cost.

The 2,007 PSUs are grouped into strata within each State. Then, one PSU is selected from each stratum with the probability of selection proportional to the population of the PSU. Nationally, there are a total of 428 PSUs in strata by themselves. These strata are self-representing and are generally the most populous PSUs in each State. The 326 remaining strata are formed by combining PSUs that are similar in such characteristics as unemployment, proportion of housing units with three or more persons, number of persons employed in various industries, and average monthly wages for various industries. The single PSU randomly selected from each of these strata is nonself-representing because it represents not only itself but the entire stratum. The probability of selecting a particular PSU in a nonself-representing stratum is proportional to its 1990 population. For example, within a stratum, the chance that a PSU with a population of 50,000 would be selected for the sample is twice that for a PSU having a population of 25,000.

Selection of sample households. Because the sample design is State based, the sampling ratio differs by State and depends on State population size as well as both national and State reliability requirements. The State sampling ratios range roughly from 1 in every 100 households to 1 in every 3,000 households. The sampling ratio occasionally is modified slightly to hold the size of the sample relatively constant given the overall growth of the population. The sampling ratio used within a sample PSU depends on the probability of selection of the PSU and the sampling ratio for the State. In a sample PSU with a probability of selection of 1 in 10 and a State sampling ratio of 3,000, a within PSU sampling ratio of 1 in 300 achieves the desired ratio of 1 in 3,000 for the stratum.

The 1990 within-PSU sample design was developed using block-level data from the 1990 census. (The 1990 census was the first decennial census that produced data at the block level for the entire country.) Normally, census blocks are bounded by streets and other prominent physical features such as rivers or railroad tracks. County, minor civil division, and census place limits also serve as block boundaries. In cities, blocks can be bounded by four streets and be quite small in land area. In rural areas, blocks can be several square miles in size.

For the purpose of sample selection, census blocks were grouped into three strata: Unit, group quarters, and area. (Occasionally, units within a block were split between the unit and group-quarters strata.) The unit stratum contained regular housing units with addresses that were easy to locate (for example, most single-family homes, townhouses, condominiums, apartment units, and mobile homes). The group-quarters stratum contained housing units in which residents shared common facilities or received formal or authorized care or custody. Unit and group-quarters blocks exist primarily in urban areas. The area stratum contains blocks with addresses that are more difficult to locate. Area blocks exist primarily in rural areas.

To reduce the variability of the survey estimates and to ensure that the within-PSU sample would reflect the demographic and socioeconomic characteristics of the PSU, blocks within the unit, group-quarters, and area strata were sorted using geographic and block-level data from the census. Examples of the census variables used for sorting include proportion of minority renter-occupied housing units, proportion of housing units with female householders, and proportion of owner-occupied housing units. The specific sorting variables used differed by type of PSU (urban or rural) and stratum.

Within each block, housing units were sorted geographically and grouped into clusters of approximately four units. A systematic sample of these clusters was then selected independently from each stratum using the appropriate within PSU sampling ratio. The geographic clustering of the sample units reduces field representative travel costs. Prior to interviewing, special listing procedures are used to locate the particular sample addresses in the group-quarters and area blocks.

Units in the three strata described above all existed at the time of the 1990 decennial census. Through a series of additional procedures, a sample of building permits is included in the CPS to represent housing units built after the decennial census. Adding these newly built units keeps the sample up-to-date and representative of the population. It also helps to keep the sample size stable: Over the life of the sample, the addition of newly built housing units compensates for the loss of “old” units that may be abandoned, demolished, or converted to nonresidential use.

Rotation of sample. Part of the sample is changed each month. Each monthly sample is divided into eight representative subsamples or rotation groups. A given rotation group is interviewed for a total of 8 months, divided into two equal periods. It is in the sample for 4 consecutive months, leaves the sample during the following 8 months, and then returns for another 4 consecutive months. In each monthly sample, one of the eight rotation groups is in the first month of enumeration, another rotation group is in the second month, and so on. Under this system, 75 percent of the sample is common from month to month, and 50 percent is common from year to year for the same month. This procedure provides a substantial amount of month-to-month and year-to-year overlap in the sample, thus providing better estimates of change and reducing discontinuities in the data series without burdening any specific group of households with an unduly long period of inquiry.

CPS sample, 1947 to present. Table 1-A provides a description of some aspects of the CPS sample designs in use since 1947. A more detailed account of the history of the CPS sample design appears in “The Current Population Survey: Design and Methodology,” Technical Paper 63, (Washington, U.S. Census Bureau and Bureau of Labor Statistics, March 2000), available on the Internet at www.bls.census.gov/cps/tp/tp63.htm. A description of the 1990 census-based sample design appears in “Redesign of the Sample for the Current Population Survey,” in the May 1994 issue of this publication. A description of the sample expansion in support of the State Children’s Health Insurance Program appears in “Expansion of the Current Population Survey Sample Effective July 2001” in the August 2001 issue of this publication. A section describing the allocation of the additional sample will be added to the Internet version of Technical Paper 63.

ESTIMATING METHODS

Under the estimating methods used in the CPS, all of the results for a given month become available simultaneously and are based on returns from the entire panel of respondents. The estimation procedure involves weighting the data from each sample person by the inverse of the probability of the person being in the sample. This gives a rough measure of the number of actual persons that the sample person represents. Since 1985, most sample persons within the same State have had the same probability of selection. Some selection probabilities may differ within a State due to the sample design or for operational reasons. Field subsampling, for example, which is carried out when areas selected for the sample are found to contain many more households than expected, may cause probabilities of selection to differ for some sample areas within a State. Through a series of estimation steps (outlined below), the selection probabilities are adjusted for noninterviews and survey undercoverage; data from previous months are incorporated into the estimates through the composite estimation procedure.

1. Noninterview adjustment. The weights for all interviewed households are adjusted to account for occupied sample households for which no information was obtained because of absence, impassable roads, refusals, or unavailability of the respondents for other reasons. This noninterview adjustment is made separately for clusters of similar sample areas that are usually, but not necessarily, contained within a State. Similarity of sample areas is based on Metropolitan Statistical Area (MSA) status and size. Within each cluster, there is a further breakdown by residence. Each MSA cluster is split by “central city” and “balance of the MSA.” Each non

MSA cluster is split by “urban” and “rural” residence categories. The proportion of sample households not interviewed varies from 7 to 8 percent, depending on weather, vacation, etc.

2. Ratio estimates. The distribution of the population selected for the sample may differ somewhat, by chance, from that of the population as a whole in such characteristics as age, race, sex, and State of residence. Because these characteristics are closely correlated with labor force participation and other principal measurements made from the sample, the survey estimates can be substantially improved when weighted appropriately by the known distribution of these population characteristics. This is accomplished through two stages of ratio adjustment, as follows:

a. First-stage ratio estimation. The purpose of the first stage ratio adjustment is to reduce the contribution to variance that results from selecting a sample of PSUs rather than drawing sample households from every PSU in the Nation. This adjustment is made to the CPS weights in two race cells: Black and nonblack; it is applied only to PSUs that are not self-representing and for those States that have a substantial number of black households. The procedure corrects for differences that existed in each State cell at the time of the 1990 census between 1) the race distribution of the population in sample PSUs and 2) the race distribution of all PSUs. (Both 1 and 2 exclude self-representing PSUs.)

b. Second-stage ratio estimation. This procedure substantially reduces the variability of estimates and corrects, to some extent, for CPS undercoverage. The CPS sample weights are adjusted to ensure that sample-based estimates of population match independent population controls. Three sets of controls are used:

1) 51 State controls of the civilian noninstitutional population 16 years of age and older,

2) National civilian noninstitutional population controls for 14 Hispanic and 5 non-Hispanic age-sex categories,

3) National civilian noninstitutional population controls for 66 white, 42 black, and 10 “other” age-sex categories.

The independent population controls are prepared by projecting forward the resident population as enumerated on April 1, 1990. The projections are derived by updating demographic census data with information from a variety of other data sources that account for births, deaths, and net migration. Estimated numbers of resident Armed Forces personnel and institutionalized persons reduce the resident population to the civilian noninstitutional population. Estimates of net census undercount, determined from the Post Enumeration Survey, are added to the population projections. Prior to January 1994, the projections were based on earlier censuses, and there was no correction for census undercount. A summary of the current procedures used to make population projections is given in “Revisions in the Current Population Survey Effective January 1994,” appearing in the February 1994 issue of this publication.

3. Composite estimation procedure. The last step in the preparation of most CPS estimates makes use of a composite estimation procedure. The composite estimate consists of a weighted average of two factors: The two-stage ratio estimate based on the entire sample from the current month and the composite estimate for the previous month, plus an estimate of the month-to-month change based on the six rotation groups common to both months. In addition, a bias adjustment term is added to the weighted average to account for relative bias associated with month-in-sample estimates. This month-in-sample bias is exhibited by unemployment estimates for persons in their first and fifth months in the CPS being generally higher than estimates obtained for the other months.

The composite estimate results in a reduction in the sampling error beyond that which is achieved after the two stages of ratio adjustment. For some items, the reduction is substantial. The resultant gains in reliability are greatest in estimates of month-to-month change, although gains usually are also obtained for estimates of level in a given month, change from year to year, and change over other intervals of time.

Rounding of estimates

The sums of individual items may not always equal the totals shown in the same tables because of independent rounding of totals and components to the nearest thousand. Similarly, sums of percent distributions may not always equal 100 percent because of rounding. Differences, however, are insignificant.

Reliability of the estimates

An estimate based on a sample survey has two types of error — sampling error and nonsampling error. The estimated standard errors provided in this publication are approximations of the true sampling errors. They incorporate the effect of some nonsampling errors in response and enumeration, but do not account for any systematic biases in the data.

Nonsampling error. The full extent of nonsampling error is unknown, but special studies have been conducted to quantify some sources of nonsampling error in the CPS. The effect of nonsampling error is small on estimates of relative change, such as month-to-month change; estimates of monthly levels tend to be affected to a greater degree.

Nonsampling errors in surveys can be attributed to many sources, for example, the inability to obtain information about all persons in the sample; differences in the interpretation of questions; inability or unwillingness of respondents to provide correct information; inability of respondents to recall information; errors made in collecting and processing the data; errors made in estimating values for missing data; and failure to represent all sample households and all persons within sample households (undercoverage).

Nonsampling errors occurring in the interview phase of the survey are studied by means of a reinterview program. This program is used to estimate various sources of error, as well as to evaluate and control the work of the interviewers. A random sample of each interviewer’s work is inspected through reinterview at regular intervals. The results indicate, among other things, that the data published from the CPS are subject to moderate systematic biases. A description of the CPS reinterview program and some results may be found in “The Current Population Survey Reinterview Program, January 1961 through December 1966,” Technical Paper No. 19 (Washington, U.S. Census Bureau, 1968).

The effects of some components of nonsampling error in the CPS data can be examined as a result of the rotation plan used for the sample, because the level of the estimates varies by rotation group. A description appears in Barbara A. Bailar, “The Effects of Rotation Group Bias on Estimates from Panel Surveys,” Journal of the American Statistical Association, March 1975, pp. 23-30.

Undercoverage in the CPS results from missed housing units and missed persons within sample households. The CPS covers about 92 percent of the decennial census population (adjusted for census undercount). It is known that the CPS undercoverage varies with age, sex, race, and Hispanic origin. Generally, undercoverage is larger for men than for women and is larger for blacks, Hispanics, and other races than for whites. Ratio adjustment to independent age-sex-race-origin population controls, as described previously, partially corrects for the biases due to survey undercoverage. However, biases exist in the estimates to the extent that missed persons in missed households or missed persons in interviewed households have characteristics different from those of interviewed persons in the same age-sex-race-origin group.

Additional information on nonsampling error in the CPS appears in Camilla Brooks and Barbara Bailar, “An Error Profile: Employment as Measured by the Current Population Survey,” Statistical Policy Working Paper 3 (Washington, U.S. Department of Commerce, Office of Federal Statistical Policy and Standards, September 1978); Marvin Thompson and Gary Shapiro, “The Current Population Survey: An Overview,” Annals of Economic and Social Measurement, Vol. 2, April 1973; and “The Current Population Survey: Design and Methodology,” Technical Paper 63 (Washington, U.S. Census Bureau and Bureau of Labor Statistics, March 2000), available on the Internet at www.bls.census.gov/eps/tp/tp63.htm. The last document includes a comprehensive discussion of various sources of errors and describes attempts to measure them in the CPS.

Sampling error. When a sample, rather than the entire population, is surveyed, estimates differ from the true population values that they represent. This difference, or sampling error, occurs by chance, and its variability is measured by the standard error of the estimate. Sample estimates from a given survey design are unbiased when an average of the estimates from all possible samples would yield, hypothetically, the true population value. In this case, the sample estimate and its standard error can be used to construct approximate confidence intervals, or ranges of values that include the true population value with known probabilities. If the process of selecting a sample from the population were repeated many times, an estimate made from each sample, and a suitable estimate of its standard error calculated for each sample, then:

1. Approximately 68 percent of the intervals from one standard error below the estimate to one standard error above the estimate would include the true population value.

2. Approximately 90 percent of the intervals from 1.645 standard errors below the estimate to 1.645 standard errors above the estimate would include the true population value.

3. Approximately 95 percent of the intervals from 1.96 standard errors below the estimate to 1.96 standard errors above the estimate would include the true population value.

These confidence interval statements are approximately true for the CPS. Although the estimating methods used in the CPS do not produce unbiased estimates, biases for most estimates are believed to be small. Methods for estimating standard errors reflect not only sampling errors but also some kinds of nonsampling error. Although both the estimates and the estimated standard errors depart from the theoretical ideal, the departures are minor and have little impact on the confidence interval statements. When clarity is needed, an estimated confidence interval is specified to be “approximate,” as is the estimated standard error used in the computation.

Tables 1-B through 1-D are provided so that approximate standard errors of estimates can be easily obtained. Tables 1-B and 1-C give approximate standard errors for estimated monthly levels and rates for selected employment status characteristics; the tables also provide approximate standard errors for consecutive month-to-month changes in the estimates. It is impractical to show approximate standard errors for all CPS estimates in this publication, so table 1-D provides parameters and factors that allow the user to calculate approximate standard errors for a wide range of estimated levels, rates, and percentages, and also changes over time. The parameters and factors are used in formulas that are commonly called generalized variance functions.

The approximate standard errors provided in this publication are based on the sample design and estimation procedures as of 1996, and reflect the population levels and sample size as of that year. Standard errors for years prior to 1996 may be roughly approximated by applying these adjustments to the standard errors presented here. (More accurate standard error estimates for historical CPS data may be found in previous issues of this publication.)

1. For the years 1967 through 1995, multiply the standard errors by 0.96.

2. For the years 1956 through 1966, multiply the standard errors by 1.17.

3. For years prior to 1956, multiply the standard errors by 1.44.

Use of tables 1-B and 1-C. These tables provide a quick reference for standard errors of major characteristics. Table 1-B gives approximate standard errors for estimates of monthly levels and consecutive month-to-month changes in levels for major employment status categories. Table 1-C gives approximate standard errors for estimates of monthly unemployment rates and consecutive month-to-month changes in unemployment rates for some demographic, occupational, and industrial categories. For characteristics not given in tables 1-B and 1-C, refer to table 1-D.

Illustration. Suppose that, for a given month, the number of women age 20 years and over in the civilian labor force is estimated to be 60,000,000. For this characteristic, the approximate standard error of 245,000 is given in table 1-B in the row “Women, 20 years and over; Civilian labor force.” To calculate an approximate 90-percent confidence interval, multiply the standard error of 245,000 by the factor 1.645 to obtain 403,000. This number is subtracted from and then added to 60,000,000 to obtain an approximate 90-percent confidence interval: 59,597,000 to 60,403,000. Concluding that the true civilian labor force level lies within an interval calculated in this way would be correct for roughly 90 percent of all possible samples that could have been selected for the CPS.

Use of table 1-D. This table gives a and b parameters that can be used with formulas to calculate approximate monthly standard errors for a wide range of estimated levels, proportions, and rates. Factors are provided to convert monthly measures into approximate standard errors of estimates for other periods (quarterly and yearly averages) and approximate standard errors for changes over time (consecutive monthly changes, changes in consecutive quarterly and yearly averages, and changes in monthly estimates 1 year apart).

The standard errors for estimated changes in level from one month to the next, one year to the next, etc., depend more on the monthly levels for characteristics than on the size of the changes. Likewise, the standard errors for changes in rates (or percentages) depend more on the monthly rates (or percentages) than on the size of the changes. Accordingly, the factors presented in table 1-D are applied to the monthly standard error approximations for levels, percentages, or rates; the magnitudes of the changes do not come into play. Factors are not given for estimated changes between nonconsecutive months (except for changes of monthly estimates 1 year apart); however, the standard errors may be assumed to be higher than the standard errors for consecutive monthly changes.

Standard errors of estimated levels using table 1-D. The approximate standard error se(x) of x, an estimated monthly level, can be obtained using the formula below, where a and b are the parameters from table 1-D associated with a particular characteristic.

se(x) = [square root of a[x.sup.2]] + bx

Illustration. Assume that, in a given a month, there are an estimated 3 million unemployed men. Obtain the appropriate a and b parameters from table 1-D (Total or white; Men; Unemployed). Use the formula for se(x) to compute an approximate standard error on the estimate of x = 3,000,000.

a = -0.0000348 b = 2927.43

se(3,000,000) = [square root of -0.0000348(3,000,000)[sup.2]] + 2927.43(3,000,000)] [approximately equal to] 92,000

Procedure for using table 1-D factors for levels. Table 1 -D gives factors that can be used to compute approximate standard errors of levels for other periods or for changes over time. For each characteristic, factors f are given for:

Consecutive month-to-month changes

Changes in monthly estimates 1 year apart

Quarterly averages

Changes in consecutive quarterly averages

Yearly averages

Changes in consecutive yearly averages

For a given characteristic, the table 1-D factor is used in the following formula, which also uses the a and b parameters from the same line of the table. A three-step procedure for using the formula is given. The fin the formula is frequently called an adjustment factor, because it appears to adjust a monthly standard error se(x). However, the x in the formula is not a monthly level, but an average of several monthly levels (see examples listed under Step 1, below).

se(x, f) = f * se(x) = f * [square root of (a[x.sup.2] + bx)]

where x is an average of monthly levels over a designated period.

Step 1. Average monthly levels appropriately in order to obtain x. Levels for 3 months are averaged for quarterly averages, and those for 12 months are averaged for yearly averages. For changes in consecutive averages, average over the 2 months, 2 quarters, or 2 years involved. For changes in monthly estimates 1 year apart, average the 2 months involved.

Step 2. Calculate an approximate standard error se(x), treating the average x from step 1 as if it were an estimate of level for a single month. Obtain parameters a and b from table 1-D. (Note that, for some characteristics, an approximate standard error of level could instead be obtained from table 1-B and used in place of se(x) in the formula.)

Step 3. Determine the standard error se (x, f) on the average level or on the change in level. Multiply the result from step 2 by the appropriate factor f. The a and b parameters used in step 2 and the factor f used in this step come from the same line in table 1-D.

Illustration of a standard error computation for consecutive month change in level. Continuing the previous example, suppose that in the next month the estimated number of unemployed men increases by 150,000, from 3,000,000 to 3,150,000.

Step 1. The average of the two monthly levels is x = 3,075,000.

Step 2. Apply the a and b parameters from table 1-D (Total or white; Men; Unemployed) to the average x, treating it like an estimate for a single month.

a = -0.0000348 b = 2927.43

se(3,075,000) = [square root of -0.0000348(3,075,000)[sup.2] + 2927.43(3,075,000)] [approximately equal to] 93,000

Step 3. Obtain f = 1.27 from the same row of table 1-D in the column “Consecutive month-to-month change,” and multiply the factor by the result from step 2.

se(150,000) = f * se(3,075,000) = 1.27 * 93,000 [approximately equal to] 118,000

For an approximate 90-percent confidence interval, compute 1.645 * 118,000 [approximately equal to] 194,000. Subtract the number from and add the number to 150,000 to obtain an interval of-44,000 to 344,000. This is an approximate 90-percent confidence interval for the true change, and since this interval includes zero, one cannot assert at this level of confidence that any real change has occurred in the unemployment level. The result also can be expressed by saying that the apparent change of 150,000 is not significant at a 90 percent confidence level.

Illustration of a standard error computation for quarterly average level. Suppose that an approximate standard error is desired for a quarterly average of the black employment level. Suppose that the estimated employment levels for the 3 months making up the quarter are 14,900,000, 15,000,000, and 15,100,000.

Step 1. The average of the three monthly levels is x = 15,000,000.

Step 2. Apply the a and b parameters from table 1-D (Black; Total; Civilian labor force, employed, and not in labor force) to the average x, treating it like an estimate for a single month.

a = -0.0001541 b = 3295.99

se(15,000,000) = [square root of -0.0001541(15,000,000)[sup.2] +

3295.99(15,000,000)] [approximately equal to] 122,000

Step 3. Obtain f = .86 from the same row of table 1-D in the column “Quarterly averages,” and multiply the factor by the result from step 2.

se(15,000,000) = .86 * 122,000 [approximately equal to] 105,000

Illustration of a standard error computation for change in quarterly level. Continuing the example, suppose that, in the next quarter, the estimated average employment level for blacks is 15,400,000, based on monthly levels of 15,300,000, 15,400,000, and 15,500,000. This is an estimated increase of 400,000 over the previous quarter.

Step 1. The average of the two quarterly levels is x = 15,200,000.

Step 2. Apply the a and b parameters from table 1-D (Black; Total; Civilian labor force, employed, and not in labor force) to the average x, treating it like an estimate for a single month.

a = -0.0001541 b = 3295.99

se(15,200,000) = [square root of 4-0.0001541(15,200,000)[sup.2] + 3295.99(15,200,000)] [approximately equal to] 120,000

Step 3. Obtain f = .78 from the same row of table 1-D in the column “Change in consecutive quarterly averages,” and multiply the factor by the result from step 2.

se(400,000) = .78 * se(15,200,000) = .78 * 120,000 [approximately equal to] 94,000

For an approximate 95-percent confidence interval, compute 1.96 * 94,000 [approximately equal to] 184,000. Subtract the number from and add the number to 400,000 to obtain an interval of 216,000 to 584,000. The interval excludes zero. Another way of stating this is to observe that the estimated change of 400,000 clearly exceeds 1.96 standard errors, or 184,000. One can conclude from these data that the change in quarterly averages is significant at a 95-percent confidence level.

Standard errors of estimated rates and percentages using table 1-D. As shown in the formula below, the approximate standard error se(p, y) of an estimated rate or percentage p depends, in part, upon the number of persons y in its base or denominator. Generally, rates and percentages are not published unless the monthly base is greater than 75,000 persons, the quarterly average base is greater than 60,000 persons, or the yearly average base is greater than 35,000 persons. The b parameter is obtained from table 1-D. When the base y and the numerator of p are from different categories within the table, use the b parameter from table 1-D relevant to the numerator of the rate or percentage.

se(p, y) = [square root of b/y p(100-p)]

Note that se(p, y) is in percent.

Illustration. For a given month, suppose y = 6,200,000 women 20 to 24 years of age are estimated to be employed. Of this total, 2,000,000, or p = 32 percent, are classified as part-time workers. Obtain the parameter b = 3005.06 from the table 1-D row (Employment; Part-time workers) that is relevant to the numerator of the percentage. Apply the formula to obtain:

se(p, y) = [square root of 6,200,000/3005.06(32)(100 – 32) [approximately equal to] 1.0 percent

For an approximate 95-percent confidence interval, compute 1.96 * 1.0 percent, and round the result to 2 percent. Subtract this from and add this to the estimate of p = 32 percent to obtain an interval of 30 percent to 34 percent.

Procedure for using table 1-D factors for rates and percentages. Table 1-D factors can be used to compute approximate standard errors on rates and percentages for other periods or for changes over time. As for levels, there are three steps in the procedure for using the formula.

se(p, y, f) = f * se(p, y) = f * [square root of b/y p(100-p)]

where p and y are averages of monthly estimates over a designated period. Note that se (p, y, f) is in percent.

Step 1. Appropriately average estimates of monthly rates or percentages to obtain p, and also average estimates of monthly levels to obtain y. Rates for 3 months are averaged for quarterly averages, and those for 12 months are averaged for yearly averages. For changes in consecutive averages, average over the 2 months, 2 quarters, or 2 years involved. For changes in monthly estimates 1 year apart, average the 2 months involved.

Step 2. Calculate an approximate standard error se (p, y), treating the averages p and y from step 1 as if they were estimates for a single month. Obtain the b parameter from the table 1-D row that describes the numerator of the rate or percentage. (Note that, for some characteristics, an approximate standard error could instead be obtained from table 1-C and used in place of se (p, y) in the formula.)

Step 3. Determine the standard error se (p, y, j) on the average level or on the change in level. Multiply the result from step 2 by the appropriate factor f. The b parameter used in step 2 and the factor fused in this step come from the same line in table 1-D.

Illustration of a standard error computation for consecutive month change in percentage. Continuing the previous example, suppose that, in the next month, 6,300,000 women 20 to 24 years of age are reported employed, and that

2,150,000, or 34 percent, are part-time workers.

Step 1. The month-to-month change is 2 percent = 34 percent – 32 percent. The average of the two monthly percentages of 32 percent and 34 percent is needed (p = 33 percent), as is the average of the two bases of 6,200,000 and 6,300,000 (y = 6,250,000).

Step 2. Apply the b = 3005.06 parameter from table 1-D (Employment; Part-time workers) to the averaged p and y, treating the averages like estimates for a single month.

se(p, y) = [square root of 6,250,000/3005.06(33)(100 – 33)] [approximately equal] to 1.0 percent

Step 3. Obtain f = .65 from the same row of table 1-D in the column “Consecutive month-to-month change,” and multiply the factor by the result from step 2.

se(2%) = .65 * 1.0 percent = .65 percent

For an approximate 95-percent confidence interval, compute 1.96 * .65 percent, and round the result to 1.3 percent. Subtract this from and add this to the 2-percent estimate of change to obtain an interval of 0.7 percent to 3.3 percent. Because this interval excludes zero, it can be concluded at a 95-percent confidence level that the change is significant.

Establishment Data (“B” tables)

DATA COLLECTION

BLS cooperates with State Employment Security Agencies in the Current Employment Statistics (CES) or establishment survey to collect data each month on employment, hours, and earnings from a sample of nonfarm establishments (including government). This sample includes about 350,000 reporting units. From these data, a large number of employment, hours, and earnings series in considerable industry and geographic detail are prepared and published each month. Historical statistics are available at http://www.bls.gov, the BLS Internet site.

Each month, BLS and the State agencies collect data on employment, payrolls, and paid hours from a sample of establishments. Data are collected by touchtone data entry (TDE) from most respondents. Under the TDE system, the respondent uses a touchtone telephone to call a toll-free number and activate an interview session. The questionnaire resides on the computer in the form of prerecorded questions that are read to the respondent. The respondent enters numeric responses by pressing the touchtone phone buttons. Each answer is read back for respondent verification.

For establishments that do not use TDE, data are collected mostly by mail, FAX, or Electronic Data Interchange (EDI), or on magnetic tape or computer diskette. Computer-assisted telephone interviewing (CATI) is used for a small number of respondents (5 percent). BLS is also pilot testing reporting via the World Wide Web. Chart 1 shows the percentages of the establishments using different data collection methods.

All reports are edited by the State agencies each month to make sure that the data are correctly reported and that they are consistent with the data reported by the establishment in earlier months. The State agencies forward the data to BLS-Washington. They also use the data to develop State and area estimates of employment, hours, and earnings. At BLS, the data are edited again by computer to detect processing and reporting errors that may have been missed in the initial State editing; the edited data are used to prepare national estimates.

CONCEPTS

Industrial classification

Establishments reporting on Form BLS 790 are classified into industries on the basis of their principal product or activity, as determined from information on annual sales volume. Since January 1980, this information has been collected on a supplement to the quarterly unemployment insurance tax reports filed by employers. For an establishment making more than one product or engaging in more than one activity, the entire employment of the establishment is included under the industry indicated by the principal product or activity.

All data on employment, hours, and earnings for the Nation (beginning with August 1990 data) and for States and areas (beginning with January 1990 data) are classified in accordance with the 1987 Standard Industrial Classification Manual (SIC), U.S. Office of Management and Budget.

Industry employment

Employment data, except those for the Federal Government, refer to persons on establishment payrolls who received pay for any part of the pay period that includes the 12th day of the month. For Federal Government establishments, employment figures represent the number of persons who occupied positions, either full- or part-time, on the last day of the calendar month or the last day of the last full pay period of the calendar month. Intermittent Federal Government workers are counted if they performed any service during the month. Agencies are required to consistently report employment data on either a calendar month basis or pay period basis. The only exception to this rule occurs at the end of the fiscal year when all agencies are required to report data as of September 30th.

The data exclude proprietors, the self-employed, unpaid volunteer or family workers, farmworkers, and domestic workers. Salaried officers of corporations are included. Government employment covers only civilian employees; military personnel are excluded. Employees of the Central Intelligence Agency, the Defense Intelligence Agency, and the National Security Agency, also are excluded.

Persons on establishment payrolls who are on paid sick leave (for cases in which pay is received directly from the firm), on paid holiday, or on paid vacation, or who work during a part of the pay period even though they are unemployed or on strike during the rest of the period are counted as employed. Not counted as employed are persons who are on layoff, on leave without pay, or on strike for the entire period, or who were hired but have not yet reported during the period.

Indexes of diffusion of employment change. These indexes measure the dispersion among industries of the change in employment over the specified timespan. The overall indexes are calculated from 353 seasonally adjusted employment series (3-digit industries) covering all nonfarm payroll employment in the private sector. The manufacturing diffusion indexes are based on 136 3-digit industries.

To derive the indexes, each component industry is assigned a value of 0, 50, or 100 percent, depending on whether its employment showed a decrease, no change, or an increase, respectively, over the timespan. The average value (mean) is then calculated, and this percent is the diffusion index number.

The reference point for diffusion analysis is 50 percent, the value indicating that the same number of component industries had increased as had decreased. Index numbers above 50 show that more industries had increasing employment and values below 50 indicate that more had decreasing employment. The margin between the percent that increased and the percent that decreased is equal to the difference between the index and its complement–that is, 100 minus the index. For example, an index of 65 percent means that 30 percent more industries had increasing employment than had decreasing employment (65-(100-65) = 30). However, for dispersion analysis, the distance of the index number from the 50-percent reference point is the most significant observation.

Although diffusion indexes commonly are interpreted as showing the percent of components that increased over the timespan, it should be remembered that the index reflects half of the unchanged components as well. (This is the effect of assigning a value of 50 percent to the unchanged components when computing the index.)

Industry hours and earnings

Average hours and earnings data are derived from reports of payrolls and hours for production and related workers in manufacturing and mining, construction workers in construction, and nonsupervisory employees in private service-producing industries.

Production and related workers. This category includes working supervisors and all nonsupervisory workers (including group leaders and trainees) engaged in fabricating, processing, assembling, inspecting, receiving, storing, handling, packing, warehousing, shipping, trucking, hauling, maintenance, repair, janitorial, guard services, product development, auxiliary production for plant’s own use (for example, power plant), recordkeeping, and other services closely associated with the above production operations.

Construction workers. This group includes the following employees in the construction division: Working supervisors, qualified craft workers, mechanics, apprentices, helpers, laborers, and so forth, engaged in new work, alterations, demolition, repair, maintenance, and the like, whether working at the site of construction or in shops or yards at jobs (such as precutting and preassembling) ordinarily performed by members of the construction trades.

Nonsupervisory employees. These are employees (not above the working-supervisor level) such as office and clerical workers, repairers, salespersons, operators, drivers, physicians, lawyers, accountants, nurses, social workers, research aides, teachers, drafters, photographers, beauticians, musicians, restaurant workers, custodial workers, attendants, line installers and repairers, laborers, janitors, guards, and other employees at similar occupational levels whose services are closely associated with those of the employees listed.

Payroll. This refers to the payroll for full- and part-time production, construction, or nonsupervisory workers who received pay for any part of the pay period that includes the 12th day of the month. The payroll is reported before deductions of any kind, such as those for old-age and unemployment insurance, group insurance, withholding tax, bonds, or union dues; also included is pay for overtime, holidays, and vacation, and for sick leave paid directly by the firm. Bonuses (unless earned and paid regularly each pay period); other pay not earned in the pay period reported (such as retroactive pay); tips; and the value of free rent, fuel, meals, or other payment in kind are excluded. Employee benefits (such as health and other types of insurance, contributions to retirement, and so forth, paid by the employer) also are excluded.

Hours. These are the hours paid for during the pay period that includes the 12th of the month for production, construction, or nonsupervisory workers. Included are hours paid for holidays and vacations, and for sick leave when pay is received directly from the firm.

Overtime hours. These are hours worked by production or related workers for which overtime premiums were paid because the hours were in excess of the number of hours of either the straight-time workday or the workweek during the pay period that included the 12th of the month. Weekend and holiday hours are included only if overtime premiums were paid. Hours for which only shift differential, hazard, incentive, or other similar types of premiums were paid are excluded.

Average weekly hours. The workweek information relates to the average hours for which pay was received and is different from standard or scheduled hours. Such factors as unpaid absenteeism, labor turnover, part-time work, and stoppages cause average weekly hours to be lower than scheduled hours of work for an establishment. Group averages further reflect changes in the workweek of component industries.

Indexes of aggregate weekly hours. The indexes of aggregate weekly hours are prepared by dividing the current month’s aggregate by the average of the 12 monthly figures for 1982. For basic industries, the hours aggregates are the product of average weekly hours and production worker or nonsupervisory worker employment. At all higher levels of industry aggregation, hours aggregates are the sum of the component aggregates.

Average overtime hours. Overtime hours represent that portion of average weekly hours that exceeded regular hours and for which overtime premiums were paid. If an employee were to work on a paid holiday at regular rates, receiving as total compensation his or her holiday pay plus straight-time pay for hours worked that day, no overtime hours would be reported.

Because overtime hours are premium hours by definition, weekly hours and overtime hours do not necessarily move in the same direction from month to month. Such factors as work stoppages, absenteeism, and labor turnover may not have the same influence on overtime hours as on average hours. Diverse trends at the industry group level also may be caused by a marked change in hours for a component industry in which little or no overtime was worked in both the previous and current months.

Average hourly earnings. Average hourly earnings are on a “gross” basis. They reflect not only changes in basic hourly and incentive wage rates, but also such variable factors as premium pay for overtime and late-shift work and changes in output of workers paid on an incentive plan. They also reflect shifts in the number of employees between relatively high-paid and low-paid work and changes in workers’ earnings in individual establishments. Averages for groups and divisions further reflect changes in average hourly earnings for individual industries.

Averages of hourly earnings differ from wage rates. Earnings are the actual return to the worker for a stated period; rates are the amount stipulated for a given unit of work or time. The earnings series do not measure the level of total labor costs on the part of the employer because the following are excluded: Irregular bonuses, retroactive items, payments of various welfare benefits, payroll taxes paid by employers, and earnings for those employees not covered under production worker, construction worker, or nonsupervisory employee definitions.

Average hourly earnings, excluding overtime. Average hourly earnings, excluding overtime-premium pay, are computed by dividing the total production worker payroll for the industry group by the sum of total production worker hours and one-half of total overtime hours. No adjustments are made for other premium payment provisions, such as holiday pay, late-shift premiums, and overtime rates other than time and one-half.

Railroad hours and earnings. The figures for Class I railroads plus Amtrak (excluding switching and terminal companies) are based on monthly data summarized in the M-300 report of the Interstate Commerce Commission, and relate to all employees except executives, officials, and staff assistants (ICC group I) who received pay during the month. Average hourly earnings are computed by dividing total compensation by total hours paid for. Average weekly hours are obtained by dividing the total number of hours paid for, reduced to a weekly basis, by the number of employees. Multiplying average weekly hours by average hourly earnings yields average weekly earnings.

Average weekly earnings. These estimates are derived by multiplying average weekly hours estimates by average hourly earnings estimates. Therefore, weekly earnings are affected not only by changes in average hourly earnings but also by changes in the length of the workweek. Monthly variations in such factors as the proportion of part-time workers, stoppages for varying reasons, labor turnover during the survey period, and absenteeism for which employees are not paid may cause the average workweek to fluctuate.

Long-term trends of average weekly earnings can be affected by structural changes in the makeup of the workforce. For example, persistent long-term increases in the proportion of part-time workers in retail trade and many of the services industries have reduced average workweeks in these industries and have affected the average weekly earnings series.

Real earnings. These earnings are in constant dollars and are calculated from the earnings averages for the current month using a deflator derived from the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W). The reference year for these series is 1982.

ESTIMATING METHODS

[NOTE: This section and the next apply to all industries except those in the mining, construction, manufacturing, and wholesale trade major industry divisions. (See the section on CES sample redesign for information on those industries.)]

The Current Employment Statistics (CES) or establishment survey estimates of employment are generated through an annual benchmark and monthly sample link procedure. Annual universe counts or benchmark levels are generated primarily from administrative records on employees covered by unemployment insurance (UI) tax laws. These annual benchmarks, established for March of each year, are projected forward for each subsequent month based on the trend of the sample employment, using an estimation procedure called the link relative. Benchmarks and sample link relatives are computed for each basic estimating cell and summed to create aggregate-level employment estimates.

Benchmarks

For the establishment survey, annual benchmarks are constructed in order to realign the sample-based employment totals for March of each year with the UI-based population counts for March. These population counts are much less timely than sample-based estimates; however, they provide an annual point-in-time census for employment.

Population counts are derived from the administrative file of employees covered by UI. All employers covered by UI laws are required to report employment and wage information to the appropriate State Employment Security Agency four times a year. Approximately 99 percent of private employment within the scope of the establishment survey is covered by UI. A benchmark for the remaining 1 percent is constructed from alternate sources, primarily records from the Interstate Commerce Commission and the Social Security Administration. The full benchmark developed for March replaces the March sample-based estimate for each basic cell. The monthly sample-based estimates for the year preceding and the year following the benchmark are also then subject to revision.

Monthly estimates for the year preceding the March benchmark are readjusted using a “wedge-back” procedure. The difference between the final benchmark level and the previously published March sample estimate is calculated and spread back across the previous 11 months. The wedge is linear; eleven-twelfths of the March difference is added to the February estimate, ten-twelfths to the January estimate, and so on, back to the previous April estimate, which receives one-twelfth of the March difference. This assumes that the total estimation error since the last benchmark accumulated at a steady rate throughout the current benchmark year.

Estimates for the 11 months following the March benchmark also are recalculated each year. These post-benchmark estimates reflect the application of sample-based monthly changes to new benchmark levels for March, and the recomputation of bias adjustment factors for each month. Bias factors are updated to take into account the most recent experience of the estimates generated by the monthly sample versus the full universe counts derived from the UI.

Following the revision of basic employment estimates, all other derivative series (such as number of production workers and average hourly earnings) also are recalculated. New seasonal adjustment factors are calculated and all data series for the previous 5 years are re-seasonally adjusted before full publication of all revised data in June of each year.

Monthly estimation

Estimates are derived from a sample of approximately 350,000 business establishments nationwide. A current month’s estimate is derived as the product of the previous month’s estimate and a sample link relative for the current month. A bias adjustment factor is then applied to this result, primarily to account for new business births during the month.

Stratification. The sample is stratified into basic estimating cells for purposes of computing national employment, hours, and earnings estimates. Cells are defined primarily by detailed industry, and secondarily by size, for a majority of cells. In a few industries, mostly within the construction division, geographic stratification also is used. Industry classification is in accordance with the 1987 Standard Industrial Classification Manual (SIC); most estimation cells are defined at the 4-digit SIC level.

This detailed stratification pattern allows for the production and publication of estimates in considerable industry detail. Sub-industry stratification by size is important because major statistics that the survey measures, particularly employment change and average earnings, often vary significantly between establishments of different size. Stratification reduces the variance of the published industry-level estimates.

Link relative technique. A ratio of the previous to the current month’s employment is computed from a sample of establishments reporting for both months–this ratio is called a “link relative.” For each basic cell, a link relative is computed and applied to the previous month’s employment estimate to derive the current month’s estimate. Thus, a March benchmark is moved forward to the next March benchmark through application of monthly link relatives. Basic cell estimates created through the link relative technique are aggregated to form published industry level estimates for employment, as described in table 2-A. Basic estimation and aggregation methods for the hours and earnings data also are shown in table 2-A.

Model-based adjustment. Except for the goods-producing and wholesale trade divisions, bias adjustment factors are computed at the 3-digit SIC level and applied each month at the basic cell level, as part of the standard estimation procedures. The main purpose of bias adjustment is to reduce a primary source of nonsampling error in the survey–the inability to capture, on a timely basis, employment generated by new firm births. There is a lag of several months between an establishment’s opening for business and its appearing on the UI universe frame and being available for sampling. Nonsampling methods must be used to capture the portion of employment growth accounted for by new firms; otherwise, substantial underestimation of total employment levels would occur. Formal bias adjustment procedures have been used in the establishment survey since the late 1960s. Prior to the 1983 benchmark, bias adjustments were derived from a simple mean error model, which averaged undercount errors for the previous 3 years to arrive at bias projections for the coming year. The undercount errors were measured as the difference between sample-based estimate results and benchmark levels.

This procedure eventually proved inadequate during periods of rapidly changing employment trends, and the bias adjustment methodology was revised. Research done in the early 1980s indicated that bias requirements were strongly correlated with current employment growth or decline. Based on this research, a revised method was developed that uses the sample data on employment growth over the most recent two quarters, and a regression-derived coefficient for the significance of that change, to adjust the mean error model results. This change in methodology provided a more cyclically sensitive bias model. The regression-adjusted mean error model has been used for the production of national estimates since 1983.

The current model still has limitations on its ability to react to changing economic conditions or changing error structure relationships between the sample-based estimates and the UI universe counts. A principal limitation is the inability to incorporate UI universe counts as they become available on an ongoing basis, with a 6- to 9-month lag from the reference period. For this reason, the current quarterly outputs from the model are subject to intervention analysis and adjustments can be made to model results prior to the establishment of final bias levels for a quarter. Review for purposes of intervention analysis is done primarily in terms of detection of outlier (abnormally high or low) values, and by comparison of CES sample and bias trends with the most recent quarterly observations of UI universe counts.

Although the primary function of bias adjustment is to account for employment resulting from new business formations, it also adjusts for other elements of nonsampling error in the survey, because the primary input to the modeling procedure is total estimation error. Significant among these nonsampling error sources is a business death bias. When a sampled firm closes down, mostoftenit simply does not respond to the survey that month, rather than reporting zero employment. Follow-up with nonrespondents may reveal an out-of-business firm, but this information often is received too late to incorporate into monthly estimates, and the firm is simply treated as a nonrespondent for that month.

Because the bias adjustments incorporated into the estimates represent a composite of a birth bias, a death bias, and a number of other differences between the sample-based estimates and the population counts, the monthly bias adjustment levels have no specific economic meaning in and of themselves.

Table 2-B summarizes the total model-based adjustments for the past decade. The table displays the average monthly “model adjustment added” and the average monthly “model adjustment required” with the benchmark revisions for each year. Model adjustment added shows the average amount of model adjustment that was added each month over the course of an interbenchmark period. Prior to 2000, the model adjustment was the bias adjustment. Beginning with 2000, the model adjustment included a net birth/death total in addition to the bias. For example, the bias added for 2000 is listed as 153,000; this represents the average of the bias and the net birth/death adjustment made each month over the period April 1999 through March 2000. (See the section on “Redesign methodology” for more information.)

Model adjustment required is computed retrospectively, after the March benchmark for a given year is known. Adjustment required figures are calculated by taking the difference between a March estimate derived purely from the sample (that is, a series calculated without bias adjustment) and the March benchmark. Dividing this figure by 12 gives the average monthly model adjustment required figure. The adjustment required is thus defined as the amount of model adjustment that would have achieved a zero benchmark error. The difference between the total model adjustment required and the total model adjustment added is then, by definition, approximately the benchmark revision amount, for any given year. Also provided in table 2-B are the March-to-March changes. As discussed above, the over-the-year changes indicate correlation with the model adjustment added and model adjustment required figures.

THE SAMPLE

Design

The emphasis in the establishment survey is on producing timely data at minimum cost. Therefore, the primary goal of its design is to sample a large enough segment of the universe to provide reliable estimates that can be published both promptly and regularly. The present sample allows BLS to produce preliminary total nonfarm employment estimates for each month, including some limited industry detail, within 3 weeks after the reference period, and data in considerably more detail with an additional 1-month lag.

The CES survey, which was begun over 50 years ago, predates the introduction of probability sampling methods and has operated as a quota sample since its inception. Quota sampling is different from probability sampling in that it requires a fixed number of units, but they need not have been drawn in a random selection process.

The sampling plan used in the establishment survey is a form of sampling with probability proportionate to size, known as “sampling proportionate to average size of establishment.” This design results in an optimum allocation of the sample among strata because sampling variance is proportional to the average size of establishments. The universe of establishment employment is highly skewed, with a large percentage of total employment concentrated in relatively few establishments. Because variance on a population total estimate is a function of percentage universe coverage achieved by the sample, it is efficient to sample larger establishments at a higher rate than smaller establishments, assuming the cost per sample unit is fairly constant across size classes.

Under the establishment survey design, large establishments fall into certainty strata for sample selection. The size of the sample for the various industries is determined empirically based on experience and cost considerations. For example, in a manufacturing industry with a high proportion of total employment concentrated in a small number of establishments, a larger percent of total employment is included in the sample. Consequently, the sample design for such industries provides for a complete census of the large establishments, with a relatively few chosen from among the smaller establishments. For an industry in which a large proportion of total employment is accounted for by small establishments, the sample design again calls for inclusion of all large establishments but also for a more substantial number of smaller ones. Many industries in the trade and services divisions fall into this category. To keep the sample to a size that can be handled with available resources, these industries are sampled with a smaller proportion of total universe coverage than is the case for most manufacturing industries.

Coverage

Table 2-C shows the latest benchmark employment levels and the approximate proportion of total universe employment coverage at the total nonfarm and major industry division levels. The coverage for individual industries within the divisions may vary from the proportions shown.

Reliability

The establishment survey, like other sample surveys, is subject to two types of error–sampling and nonsampling. The magnitude of sampling error, or variance, is directly related to the size of the sample and the percentage of universe coverage achieved by the sample. The establishment survey sample covers nearly one-third of total universe employment; this yields a very small variance on the total nonfarm estimates. Measurements of error associated with sample estimates are provided in tables 2-D and 2-E.

Benchmark revision as a measure of survey error. The sum of sampling and nonsampling error can be considered total survey error. Unlike most sample surveys, for which only sampling error can be estimated, the CES yields an annual approximation of total error, on a lagged basis, because of the availability of the independently derived universe data. While the benchmark error is used as a measure of total error for the CES survey estimate, it actually represents the difference between two independent estimates derived from separate survey processes (specifically, the CES sample process and the UI universe process), and thus reflects the errors present in each program. Historically, the benchmark revision has been very small for total nonfarm employment. Over the past decade, percentage benchmark error has averaged 0.3 percent, with absolute revisions ranging from less than 0.05 percent to 0.7 percent. Table 2-D shows the most cur rent benchmark revisions, along with 10-year mean revisions and mean absolute revisions for major industries. Mean revisions give an indication of bias in the estimates; unbiased estimates have a mean revision close to zero, as over- and under-estimations cancel out over time. Mean absolute revisions give an overall indication of the accuracy of the estimates; the larger the value, the further the estimate was from the final benchmark level.

Revisions between preliminary and final data. First preliminary estimates of employment, hours, and earnings, based on less than the total sample, are published immediately following the reference month. Final revised sample-based estimates are published 2 months later, when nearly all the reports in the sample have been received. Table 2-E presents the root-mean-square error, the mean percent, and the mean absolute percent revision that may be expected between the preliminary and final employment estimates.

Revisions of preliminary hours and earnings estimates are normally not greater than 0.1 hour for weekly hours and 1 cent for hourly earnings at the total private nonfarm level, and may be slightly larger for the more detailed industry groupings.

CES sample redesign

In June 1995, BLS announced plans for a comprehensive sample redesign of its monthly payroll survey. The initial research phase for the CES sample redesign was completed in 1997, and BLS launched a production test of the new sample design at that time. The production test phase concluded in June 2000, when the first estimates from the new design, for the wholesale trade industry, were published with the 1999 benchmark revisions. With the 2000 benchmark revisions, estimates for the mining, construction, and manufacturing industries were published under the new design for the first time. Redesigned samples for the remaining industry divisions will be phased in with the next two benchmark releases.

Original sample design limitations. The original CES survey is based on a quota sample, the inception of which, over 50 years ago, predated the introduction of probability sampling as the internationally recognized standard for sample surveys. Quota samples are known to be at risk for potentially significant biases. Introducing a probability-based sample for CES ensures a proper representation of the universe of nonfarm business establishments through randomized selection techniques and the regular rotation of sample members.

In addition, the CES sample redesign addresses a second critical limitation of the current CES sample, which is a lack of timely sample-based representation of employment from new business births. Procedures have been developed for regular sample updates that will ensure better representation of new units in the CES sample. Time series modeling techniques are being used to estimate the residual portion of birth employment not accounted for through the improved sampling techniques. Introduction of a probability-based sample for the CES survey allows for the publication of sampling errors and confidence intervals, standard survey accuracy measures not directly applicable to the current nonprobability design. Overall accuracy of the survey employment estimates, however, is still best measured by the magnitude of annual benchmark revisions, as they encompass the total estimation error associated with the CES employment series.

The new CES sample design. The new design is a stratified, simple random sample of worksites, clustered by UI account number. The UI account number is a major identifier on the BLS longitudinal database of employer records, which serves as both the sampling frame and the benchmark source for the CES employment estimates. The sample strata, or subpopulations, are defined by State, industry, and employment size, yielding a State-based design. The sampling rates for each stratum are determined through a method known as optimum allocation, which distributes a fixed number of sample units across a set of strata to minimize the overall variance, or sampling error, on the primary estimate of interest. The total nonfarm employment level is the primary estimate of interest, and the new design gives top priority to measuring it as precisely as possible, or, in other words, minimizing the statistical error around the statewide total nonfarm employment estimates.

For the CES redesign, the number of sample units drawn was fixed to the approximate size of the original CES sample, which is the sample size supported by current program resources. This sample size makes possible the publication of considerable industry and geographic detail within a State, and provides for highly reliable national CES estimates at the total nonfarm and detailed industry levels.

Frame and sample selection. The Longitudinal Data Base (LDB) is the universe from which BLS draws the CES sample. The LDB contains data on approximately 7.5 million U.S. business establishments, representing nearly all nonfarm elements of the U.S. economy. The ES-202 program collects these data from employers, on a quarterly basis, in cooperation with State Employment Security Agencies (SESAs). The LDB contains employment and wage information from employers, as well as name, address, and location information. It also contains identification information such as Unemployment Insurance (UI) Account Number, Reporting Unit Number, and LDB Number.

The LDB consists of all employers covered under the Unemployment Insurance Tax System. That system covers 97 percent of all employers in the 50 States, the District of Columbia, Puerto Rico, and the Virgin Islands. There are a few sections of the economy that are not covered, including the self-employed, small family businesses, railroads, charitable organizations, small agricultural employers, and elected officials. Data for employers generally are reported at the worksite level. Employers who have multiple establishments within a State usually report data for each individual establishment. The LDB tracks establishments over time and links them from quarter to quarter.

Permanent Random Numbers (PRNs) have been assigned to all UI accounts on the sampling frame. As new units appear on the frame, random numbers are assigned to those units as well. As records are linked across time, the PRN is carried forward in the linkage.

The probability sample is stratified by State, industry, and size. Stratification groups population members together for the purpose of sample allocation and selection. The strata, or groups, are composed of homogeneous units. With 11 industries and 8 size classes, there are 88 total allocation cells per State. The sampling rate for each stratum is determined through a method known as optimum allocation. Optimum allocation minimizes variance at a fixed cost or minimizes cost for a fixed variance. Under the CES probability design, a fixed number of sample units for each State is distributed across the allocation strata in such a way as to minimize the overall variance, or sampling error, of the total State employment level. The number of sample units in the CES probability sample is fixed to the approximate size of the existing nonprobability CES survey. The optimum allocation formula will place more sample in cells for which data cost less to collect, cells that have more units, and cells that have a larger variance. When compared with the quota sample, there are fewer units selected in manufacturing and more units selected in services.

During the first quarter of each year, a new sample is drawn from the LDB. Annual sample selection helps keep the CES survey current with respect to employment from business births and business deaths. In addition, the updated universe files provide the most recent information on industry, size, and metropolitan area designation.

After all out-of-scope records are removed, the sampling frame is sorted into allocation cells. Within each allocation cell, units are sorted by MSA and by the size of the MSA, which is the number of UI accounts in that MSA. As the sampling rate is uniform across the entire allocation cell, implicit stratification by MSA ensures that a proportional number of units are sampled from each MSA. Some MSAs may have too few UI accounts in the allocation cell; these MSAs are collapsed and treated as a single MSA. Within each selection cell, the units are sorted by PRN, and units are selected according to the specified sample selection rate. The number of units selected randomly from each selection cell is equal to the product of the sample selection rate and the number of eligible units in the cell, plus any carryover from the prior selection cell. The result is rounded to the nearest whole number. Carryover is defined as the amount that is rounded up or down to the nearest whole number.

Once the sample is drawn, sample selection weights are calculated based on the number of UI accounts actually selected within each allocation cell. The sample selection weight is approximately equal to the inverse of the probability of selection, or the inverse of the sampling rate. It is computed as:

Sample selection weight = [N.sub.h] / [n.sub.h]

where:

[N.sub.h] = the number of noncertainty UI accounts within the allocation cell that are eligible for sample selection

[n.sub.h] = the number of noncertainty UI accounts selected within the allocation cell

To further reduce enrollment workload caused by the annual update of the sample, BLS has established a “swapping” procedure in which sample members selected in the previous year are used in lieu of new sample members. As a result of the swap procedure, the amount of sample overlap from year to year is increased. A sample is selected from the first-quarter frame using the random sampling procedures. If a new sample member is selected during random sampling, a check is made for a previously selected unit that was not selected in the new sample. The previously selected unit must be within the same State, industry, and size class and must have the same PRN date as the originally selected unit. Newly selected units are replaced until all suitable replacements are exhausted. The units are generally available for swapping due to changes in the MSA, SIC, and size of units.

As a result of the swap procedure, approximately 90 percent of the Current Employment Statistics Sample Redesign (CES-R) sample overlaps from one year to the next. Before the swap procedure was implemented, approximately 35,000 new UI accounts were selected each year during the annual update. With the swap procedure, this number is reduced by as much as 40 percent, or 15,000 units.

Due to the dynamic economy, there is a constant cycle of business births and deaths. A semiannual update is performed during the third quarter of each year. This update selects units from the population of births and other units not previously eligible for selection, and includes them as part of the sample. Updated location, contact, and administrative information is provided for all establishments that were selected in the annual sample selection.

Sample enrollment activities. The primary enrollment of new establishments for the CES-R is taking place in BLS Data Collection Centers (DCCs) located in Atlanta, Kansas City, and Dallas, and in the Electronic Data Interchange (EDI) Center in Chicago. Once the sample has been sent to the DCCs, interviewers enroll the selected establishments. While the UI account represents the sample unit, interviewers are responsible for tracking and collecting the data for the individual establishments, regardless of the current UI configuration associated with the establishments.

In the case of large, multiple-worksite UI accounts, it is sometimes necessary to subsample employers. This occurs when:

— the company cannot report for all worksites from a central location;

— the company cannot provide an aggregate report for the entire UI account;

— there are too many individual worksites to make it practical to contact each of them.

With subsampling of a smaller number of worksites, both interviewer workload and respondent burden are reduced without significantly reducing the accuracy of the estimates, but this technique will result in a small increase in variance. In the event that a UI account is subsampled, weight adjustments are made to reflect each of the worksites’ probability of selection.

Estimation. Under the new methodology, CES uses a matched sample concept and weighted link relative estimator to produce employment, hours, and earnings estimates. Consistent with the historical CES definition, a matched sample is defined to be all sample members that have reported data for the reference month and the month prior. A slight adjustment to the above matched definition is made to exclude from the matched sample any sample unit that reports that it is out-of-business. The reasoning behind this handling is described later in the section on estimation of business births and deaths.

The estimator for employment and that for hours and earnings uses the sample trend in the cell to move the previous level or ratio to the current-month estimated level or ratio. In the case of all employees, an additive model-based component is applied as well. This component also is described in the business birth and death estimation section.

The basic formula for estimating employment is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where:

i = matched sample unit;

[w.sub.i] = weight associated with the CES report;

[ae.sub.c,i] = current-month reported all employees;

[ae.sub.p,i] = previous-month reported all employees;

[AE.sub.c] = current-month estimated all employees; and

[AE.sub.p] = previous-month estimated all employees.

The basic form for the estimator used to develop the current-month production workers series is:

[PW.sub.c] = ([AE.sub.c] x [PWRATIO.sub.c]), and

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where:

i = matched sample unit;

[w.sub.i] = weight associated with the CES report;

[PW.sub.c] = current-month estimated production workers;

[PWRATIO.sub.c] = current-month production-worker-to-all-employee ratio;

[PWRATIO.sub.p] = previous-month production-worker-to-all-employee ratio;

[pw.sub.c,i] = current-month reported production workers;

[pw.sub.p,i] = previous-month reported production workers;

[ae.sub.c,i] = current-month reported all employees;

[ae.sub.p,i] = previous-month reported all employees; and

[AE.sub.c] = current-month estimated all employees.

Estimation of the series for women workers is identical to that described for production workers, with the appropriate substitution of women worker values for the production worker values in the previous formulas.

The same basic form of the estimator holds for all data types. The basic estimators of average weekly hours and average hourly earnings are:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

and

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where:

i = matched sample unit;

[w.sub.i] = weight associated with the CES report;

[AWH.sub.c] = current-month estimated average weekly hours;

[AWH.sub.p] = previous-month estimated average weekly hours;

[wh.sub.c,i] = current-month reported weekly hours;

[wh.sub.p,i] = previous-month reported weekly hours;

[pw.sub.c,i] = current-month reported production workers;

[pw.sub.p,i] = previous-month reported production workers;

[AHE.sub.c] = current-month estimated average hourly earnings;

[AHE.sub.p] = previous-month estimated average hourly earnings;

[WH.sub.c] = current-month estimated weekly man hours;

[WH.sub.p] = previous-month estimated average man hours;

[pr.sub.c,i] = current-month reported weekly payroll; and

[pr.sub.p,i] = previous-month reported weekly payroll.

Estimation of overtime hours is identical to that described for weekly hours, with the appropriate substitution of overtime hours values for the weekly hours values in the previous formula.

Benchmarking. Annual benchmark adjustment that revises 2 years of data continues under the redesign, but with slight modification to the process. Under the original CES procedures, when national series are benchmarked, sample links derived from the final (or third) set of monthly estimates are applied to the March benchmark level to re-estimate 1 year forward from the new benchmark levels. The year prior to the benchmark is adjusted by a simple wedge-back procedure that distributes the benchmark error in equal increments across the 11 months preceding the March benchmark.

For initial implementation of the redesign estimates for mining, manufacturing, and wholesale trade, the estimates for both the year prior to and the year following the March benchmark month were revised to incorporate sample-based estimates calculated from the new sample and estimators. Thus, there is more revision in the benchmark period under the redesign than experienced previously for all data types. In particular, basic cell-level hours and earnings estimates, which have no benchmark revision under current procedures, are subject to change.

The construction series are revised for the year following the benchmark. The year prior to the benchmark was revised using the quota sample estimate. As sample enrollment for the construction industries was not completed until the end of the second quarter, it was not feasible to use the new metholology for the wedge period.

Business birth and death estimation. In a dynamic economy, firms are continually going out-of-business while, at the same time, new businesses are opening. These two normal occurrences offset each other to some extent. That is, firms that are born replace firms that die. CES uses this fact to account for a large proportion of the employment associated with business births. This is accomplished by excluding such units from the matched sample definition. Effectively, business deaths are not included in the sample-based link portion of the estimate, and the implicit imputation of their previous month’s employment is assumed to offset a portion of the employment associated with births.

There is an operational advantage associated with this approach as well. Most firms will not report that they have gone out-of-business; rather, they simply cease reporting and are excluded from the link, as are all other nonrespondents. As a result, extensive follow-up with monthly nonrespondents to determine whether a company is out-of-business or simply did not respond is not required.

Employment associated with business births will not exactly equal that associated with business deaths. The amount by which it differs varies by month and by industry. As a result, the residual component of the birth/death offset must be accounted for by using a model-based approach.

With any model-based approach, it is desirable to have 5 or more years of history to use in developing the models. Due to the absence of reliable counts of monthly business births and deaths, development of an appropriate birth/death residual series assumed the following form:

Birth/death residual = Population – Sample-based estimate + Error

Simulated monthly probability estimates over a 7-year period were created and compared with population employment levels. Moving from a simulated benchmark, the differences between the series across time represent a cumulative birth/death component. Those residuals are converted to month-to-month differences and used as input series to the modeling process.

Models are fit using X-12 ARIMA (Auto-Regressive Integrated Moving Average). Outliers, level shifts, and temporary ramps are automatically identified. Seven models are tested, and the model exhibiting the lowest average forecast error is selected for each series.

Difference between the birth/death model and bias adjustment. Table 2-F compares the level of bias adjustment applied in the previously published CES series with the net birth/death adjustment used in the redesign series in mining, construction, and manufacturing. Over the course of the “post-benchmark year” from April 2000 to March 2001, the cumulative bias adjustment added 246,000 to the mining, construction, and manufacturing employment level, while the net birth/ death model added 154,000 overall. Note that the latter model has greater variability from month to month, including months with a negative adjustment. This mainly reflects the seasonal pattern of the net birth/death series observed in the historical UI universe data series.

The net birth/death models will replace the bias adjustment modeling currently used for the CES program as estimates for each major industry division are phased in for official publication. The ARIMA model component is updated and reviewed on a quarterly basis, as are the current bias adjustments. However, the net birth/death model component figures are unique to each month, unlike the bias adjustments, which are identical for all 3 months of a given quarter.

An important conceptual and empirical distinction between current bias adjustment and new net birth/death models involves the elements that the models are designed to identify. Although the primary purpose of the existing bias adjustment process is to account for new business birth employment, it also adjusts for other elements of nonsampling error, or bias, in the current CES estimate because the primary input to the model is total estimation error. Sampling bias can be significant in the existing sample because of its quota design, and the bias component is therefore relatively large. In contrast, the net birth/death models estimate only the residual component not measurable by the sample; the models do not attempt to correct for deficiencies in sample design. Therefore, the net birth/death model component in the redesign series is expected to be significantly smaller than the bias adjustment component in the current CES estimates.

The most significant potential drawback to a model-based approach is that time series modeling assumes a predictable continuation of historical patterns and relationships. Therefore, a model-based approach is likely to have some difficulty producing reliable estimates at economic turning points or during periods in which there are sudden changes in trend. In sum, accurate estimation of the business birth component of total nonfarm employment will continue to be the most difficult issue in CES employment estimation.

Variance estimation for the CE$ redesign estimates. A probability-based sample allows for the calculation and publication of sampling variances and confidence intervals–standard survey accuracy measures not directly applicable to the current nonprobability design. The estimation of sample variance for the survey is accomplished through use of the method of Balanced Half Samples (BHS). This replication technique uses half samples of the original sample and calculates estimates using those subsamples. The sample variance is calculated by measuring the variability of the subsample estimates. The weighted link estimator is used to calculate both estimates and variances. The sample units in each cell–where a cell is based on State, industry, and size classification–are divided into two random groups. The basic BHS method is applied to both groups. The subdivision of the cells is done systematically, in the same order as the initial sample selection. Weights for units in the half sample are multiplied by a factor of 1 + [gamma] where weights for units not in the half sample are multiplied by a factor of 1 – [gamma]. Estimates from these sub-groups are calculated using the estimation formula described previously.

The formula used to calculate CES variances is as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where:

[[theta].sup.+.sub.a] = [theta]([Y.sup.+.sub.a], [X.sup.+.sub.a], …..) is the half-sample estimator;

[gamma] = 1/2;

k = number of half-samples; and

[theta] = original full sample estimates

Appropriate uses of sampling variances in CES. Variance statistics are useful for comparison purposes, but they do have some limitations. Variances reflect the error component of the estimates that is due to surveying only a subset of the population, rather than conducting a complete count of the entire population. However, they do not reflect nonsampling error, such as response errors, and bias due to nonresponse. The overall performance of the program (calculating all-employee estimates) will still be measured in terms of the benchmark revisions. Variances for items not benchmarked–that is, average hourly earnings and average weekly hours–can serve as a more meaningful measure of their error now with a representative probability sample. The variances of the over-the-month change estimates are very useful in determining when changes are significant at some level of confidence.

Sampling errors for probability-based industries. The sampling errors shown for the goods-producing and wholesale trade industries have been calculated for estimates that follow the benchmark employment revision by a period of 12 to 24 months. Since the error estimates generally increase as a function of time after the month of benchmark revision, this period was determined to be the period of greatest interest for the estimates. For example, the May 2001 estimates follow the benchmark revision (March 2000) by 14 months. The errors are presented as median values of the observed error estimates. These estimates have been estimated using the method of Balanced Half Samples with the probability sample data and sample weights assigned at the time of sample selection.

Illustration of the use of table 2-G. Table 2-G provides a reference for relative standard errors of three major series developed from the CES–estimates of the numbers of all employees (AE), of average hourly earnings (AHE), and of average weekly hours (AWH) within the same industry. The standard errors of differences between estimates in two non-overlapping industries are calculated as:

S difference = [square root of [s.sup.2.sub.1] + [s.sup.2.sub.2]]

since the two estimates are independent.

The errors are presented as relative standard errors (standard error divided by the estimate and expressed as a percent). Multiplying the relative standard error by its estimated value gives the estimate of the standard error.

Suppose that the level of all employees for wholesale trade in a given month is estimated at 7,054,000. The approximate relative standard error of this estimate (0.54 percent) is provided in table 2-G. A 90-percent confidence interval would then be the interval:

7,054,000 +/- (1.645 * .0054 * 7,054,000)

= 7,054,000 +/- 62,660

= 7,116,660 to 6,991,340

Illustration of the use of table 2-H. Table 2-H provides a reference for the standard errors of 1-, 3-, and 12-month changes in AE, AHE, and AWH. The errors are presented as standard errors of the changes.

Suppose that the over-the-month change in AHE from January to February for the stone, clay, and glass products industry within manufacturing is $0.11. The standard error for a 1-month change for this industry from the table is $0.06. The interval estimate of the over-the-month change in AHE that will include the true over-the-month change with 90-percent confidence is calculated:

$0.11 +/- (1.645 * $0.06)

= $0.11 +/- $0.10

= $0.01 to $0.21

The true value of the over-the-month change is in the interval $0.01 to $0.21. Because this interval does not include $0.00 (no change), the change of $0.11 shown is significant at the 90-percent confidence level. Alternatively, the estimated change of $0.11 exceeds $0.10 (1.645 * $0.06); therefore, one could conclude from these data that the change is significant at the 90-percent confidence level.

STATISTICS FOR STATES AND AREAS (Tables B-7, B-14, and B-18)

As explained earlier, State agencies in cooperation with BLS collect and prepare State and area employment, hours, and earnings data. These statistics are based on the same establishment reports used by BLS. However, BLS uses the full CES sample to produce monthly national employment estimates, while each State agency uses its portion of the sample to independently develop a State employment estimate.

The CES area statistics relate to metropolitan areas. Definitions for all areas are published each year in the issue of Employment and Earnings that contains State and area annual averages (usually the May issue). Changes in definitions are noted as they occur. Additional industry detail may be obtained from the State agencies listed on the inside back cover of each issue.

Caution in aggregating State data. The national estimation procedures used by BLS are designed to produce accurate national data by detailed industry; correspondingly, the State estimation procedures are designed to produce accurate data for each individual State. State estimates are not forced to sum to national totals or vice versa. Because each State series is subject to larger sampling and nonsampling errors than is the national series, summing them cumulates individual State-level errors and can cause distortions at an aggregate level. This has been a particular problem at turning points in the U.S. economy, when the majority of the individual State errors tend to be in the same direction. Due to these statistical limitations, the Bureau does not compile or publish a “sum-of-States” employment series. Additionally, BLS cautions users that such a series is subject to a relatively large and volatile error structure, particularly at turning points.

Region, State, and Area Labor Force Data (“C” Tables)

FEDERAL-STATE COOPERATIVE PROGRAM

Labor force and unemployment estimates for States, labor market areas (LMAs), and other areas covered under Federal assistance programs are developed by State employment security agencies under a Federal-State cooperative program. The local unemployment estimates which derive from standardized procedures developed by BLS are the basis for determining eligibility of an area for benefits under Federal programs such as the Job Training Partnership Act.

Annual average data for the States and 337 areas shown in table C-3 are published in Employment and Earnings (usually the May issue). For regions, States, selected metropolitan areas, and central cities, annual average data classified by selected demographic, social, and economic characteristics are published in the BLS bulletin, Geographic Profile of Employment and Unemployment.

Labor force estimates for counties, cities, and other small areas have been prepared for administration of various Federal economic assistance programs and may be ordered from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. The report “Unemployment in States and Local Areas” is published monthly through GPO and is available in microfiche form only, on a subscription basis.

ESTIMATING METHODS

Monthly labor force, employment, and unemployment estimates are prepared for the 50 States, the District of Columbia, and over 6,500 areas, including nearly 2,400 LMAs, counties, and cities with a population of 25,000 or more. Regional aggregations are derived by summing the State estimates. The estimation methods are described below for States (and the District of Columbia) and for substate areas. At the sub-LMA (county and city) level, estimates are prepared using disaggregation techniques based on decennial and annual population estimates and current unemployment insurance data. A more detailed description of the estimation procedure is contained in the BLS document, Manual for Developing Local Area Unemployment Statistics.

Estimates for States

Current monthly estimates. Effective January 1996, civilian labor force and unemployment estimates for all States and the District of Columbia are produced using models based on a “signal-plus-noise” approach. The model of the signal is a time series model of the true labor force which consists of three components: A variable coefficient regression, a flexible trend, and a flexible seasonal component. The regression techniques are based on historical and current relationships found within each State’s economy as reflected in the different sources of data that are available for each State–the Current Population Survey (CPS), the Current Employment Statistics (CES) survey, and the unemployment insurance (UI) system. The noise component of the models explicitly accounts for auto correlation in the CPS sampling error and changes in the average magnitude of the error. In addition, the models can identify and remove the effects of outliers in the historical CPS series. While all the State models have important components in common, they differ somewhat from one another to better reflect individual State characteristics.

Two models–one for the employment-to-population ratio and one for the unemployment rate–are used for each State. The employment-to-population ratio, rather than the employment level, and the unemployment rate, rather than the unemployment level, are estimated primarily because these ratios are usually more meaningful for economic analysis.

The employment-to-population ratio models use the relationship between the State’s monthly employment from the CES and the CPS. The models also include trend and seasonal components to account for movements in the CPS not captured by the CES series. The seasonal component accounts for the seasonality in the CPS not explained by the CES, while the trend component adjusts for long-run systematic differences between the two series.

The unemployment rate models use the relationship between the State’s monthly unemployment insurance (UI) claims data and the CPS unemployment rate, along with trend and seasonal components.

In both the employment-to-population ratio and unemployment rate models, an important feature is the use of a technique that allows the equations to adjust automatically to structural changes that occur. The regression portion of the model includes a built-in tuning mechanism, known as the Kalman Filter, which revises a model’s coefficients when the new data that become available each month indicate that changes in the data relationships have taken place. Once the estimates are developed from the models, levels are calculated for employment, unemployment, and labor force.

Benchmark correction procedures. Once each year, monthly estimates for all States and the District of Columbia are adjusted, or benchmarked, by BLS to the annual average CPS estimates. The benchmarking technique employs a procedure (called the Denton method) which adjusts the annual average of the models to equal the CPS annual average, while preserving, as much as possible, the original monthly seasonal pattern of the model estimates.

Estimates for substate areas

Monthly labor force, employment, and unemployment estimates for two large substate areas–New York City and the Los Angeles-Long Beach metropolitan area–are obtained using the same modeling approach as for states. Estimates for the nearly 2,400 remaining LMAs, are prepared through indirect estimation techniques, described below.

Preliminary estimate–employment. The total civilian employment estimates are based largely on CES data. These “place-of-work” estimates must be adjusted to refer to place of residence as used in the CPS. Factors for adjusting from place of work to place of residence have been developed on the basis of employment relationships at the time of the 1990 decennial census. These factors are applied to the CES estimates for the current period to obtain adjusted employment estimates, to which are added estimates for employment not represented in the CES–agricultural employees, nonagricultural self-employed and unpaid family workers, and private household workers.

Preliminary estimate–unemployment. In the current month, the estimate of unemployment is an aggregate of the estimates for each of two categories: (1) Persons who were previously employed in industries covered by State UI laws; and (2) those who were entering the civilian labor force for the first time or reentering after a period of separation.

Substate adjustment for additivity. Estimates of employment and unemployment are prepared for the State and all LMAs within the State. The LMA estimates geographically exhaust the entire State. Thus, a proportional adjustment is applied to all substate preliminary LMA estimates to ensure that they add to the independently estimated State totals for employment and unemployment. For California and New York, the proportional adjustment is applied to all LMAs other than the two modeled areas, to ensure that the LMA estimates sum to an independent model-based estimate for the balance of State.

Benchmark correction. At the end of each year, substate estimates are revised. The revisions incorporate any changes in the inputs, such as revisions in the CES-based employment figures, corrections in UI claims counts, and updated historical relationships. The updated estimates are then readjusted to add to the revised (benchmarked) State estimates of employment and unemployment.

Seasonal Adjustment

Over the course of a year, the size of the Nation’s labor force, the levels of employment and unemployment, and other measures of labor market activity undergo sharp fluctuations due to such seasonal events as changes in weather, reduced or expanded production, harvests, major holidays, and the opening and closing of schools. Because these seasonal events follow a more or less regular pattern each year, their influence on statistical trends can be eliminated by adjusting the statistics from month to month. These adjustments make it easier to observe the cyclical and other nonseasonal movements in the series. In evaluating changes in a seasonally adjusted series, it is important to note that seasonal adjustment is merely an approximation based on past experience. Seasonally adjusted estimates have a broader margin of possible error than the original data on which they are based, because they are subject not only to sampling and other errors but are also affected by the uncertainties of the seasonal adjustment process itself. Seasonally adjusted series for selected labor force and establishment-based data are published monthly in Employment and Earnings.

Household data

Since January 1980, national labor force data have been seasonally adjusted with a procedure called X-II ARIMA (Auto-Regressive Integrated Moving Average), which was developed at Statistics Canada as an extension of the standard X-11 method. A detailed description of the procedure appears in The X-II ARIMA Seasonal Adjustment Method by Estela Bee Dagum, Statistics Canada Catalogue No. 12564E, January 1983. BLS uses an extension of X-11 ARIMA to allow it to adjust more adequately for the effects of the presence or absence of religious holidays in the April survey reference period and of Labor Day in the September reference period. This extension was applied for the first time at the end of 1989 to three persons-at-work labor force series which tested as having significant and well-defined effects in their April data associated with the timing of Easter.

At the beginning of each calendar year, projected seasonal adjustment factors are calculated for use during the January-June period. In July of each year, BLS calculates and publishes in Employment and Earnings projected seasonal adjustment factors for use in the second half, based on the experience through June. Revisions of historical data, usually for the most recent 5 years, are made only at the beginning of each calendar year. However, as a result of the revisions to the estimates for 1970-81 based on 1980 census population counts, revisions to seasonally adjusted series in early 1982 were carried back to 1970. In 1994, data were revised only for that year because of the major redesign and 1990 census-based population controls, adjusted for the estimated undercount, introduced into the Current Population Survey. In 1996, 1990-93 data also were revised to incorporate these 1990 census-based population controls and seasonally adjusted series were revised back to 1990. Subsequent revisions were carried back only to 1994 through 1998, when the standard 5-year revision period was reinstated.

All labor force and unemployment rate statistics, as well as the major employment and unemployment estimates, are computed by aggregating independently adjusted series. For example, for each of the three major labor force components-agricultural employment, nonagricultural employment, and unemployment–data for four sex-age groups (men and women under and over 20 years of age) are separately adjusted for seasonal variation and are then added to derive seasonally adjusted total figures. The seasonally adjusted figure for the labor force is a sum of eight seasonally adjusted civilian employment components and four seasonally adjusted unemployment components. The total for unemployment is the sum of the four unemployment components, and the unemployment rate is derived by dividing the resulting estimate of total unemployment by the estimate of the labor force. Because of the independent seasonal adjustment of various series, components will not necessarily add to totals.

In each January issue (March issue in 1996), Employment and Earnings publishes revised seasonally adjusted data for selected labor force series based on the experience through December, new seasonal adjustment factors to be used to calculate the civilian unemployment estimate for the first 6 months of the following year, and a description of the current seasonal adjustment procedure.

Establishment data

Effective in June 1996, with the release of the March 1995 benchmark revisions, BLS began using an updated version of the X-12 ARIMA software developed by the Bureau of the Census to seasonally adjust national establishment-based employment, hours, and earnings series.

The conversion to X-12 ARIMA allows BLS to refine its seasonal adjustment procedures to control for survey interval variations, sometime referred to as the 4- vs. 5-week effect. While the CES survey is referenced to a consistent concept, the pay period including the 12th day of the month, inconsistencies arise because there are variations of 4 or 5 weeks between the week of the 12th in any given pair of months. In highly seasonal months and industries, this variation can be an important determinant of the magnitude of seasonal hires or layoffs that have occurred at the time the survey is taken, thereby complicating seasonal adjustment. The interval effect adjustment is accomplished through the REGARIMA (regression with auto-correlated errors) option in the X-12 software. This process combines standard regression analysis, which measures correlations between two or more variables, with ARIMA modeling, which describes and predicts the behavior of a data series based on its own past history. In this application, the correlations of interest are those between employment levels in individual calendar months and the length of the survey intervals for those months. The REGARIMA models estimate and remove the variation in employment levels attributable to 11 separate survey intervals, one specified for each month, except March. March is excluded because this month has a 5-week interval between the February and March surveys only every 29 years.

Effective with the release of the March 1997 benchmark, seasonally adjusted series for hours and earnings of production or nonsupervisory workers from 1989 forward incorporate refinements to the seasonal adjustment process to correct for distortions related to the method of accounting for the varying length of payroll periods across months–a calendar effect.

REGARIMA modeling also is used to identify, measure, and remove this calendar effect for the publication level seasonally adjusted hours and earnings series.

Projected seasonal factors for the establishment-based series are calculated and published twice a year, paralleling the procedure used for the household series. Revisions to historical data (usually the most recent 5 years) are made once a year, coincident with benchmark revisions. All series are seasonally adjusted using multiplicative models in X- 12. Seasonal adjustment factors are computed and applied at component levels. For employment series, these are generally the 2-digit SIC levels. Seasonally adjusted totals are arithmetic aggregations for employment series and weighted averages of the seasonally adjusted data for hours and earnings series.

Seasonally adjusted average weekly earnings are the product of seasonally adjusted average hourly earnings and average weekly hours. Average weekly earnings in constant dollars, seasonally adjusted, are obtained by dividing the average weekly earnings series by the seasonally adjusted Consumer Price Index for Urban Wage Eamers and Clerical Workers (CPI-W), and multiplying by 100. Indexes of aggregate weekly hours, seasonally adjusted, are obtained by multiplying average weekly hours by production or nonsupervisory workers and dividing by the 1982 annual average base. For total private, total goods-producing, total private service-producing, and major industry divisions, the indexes of aggregate weekly hours, seasonally adjusted, are obtained by summing the aggregate weekly hours for the appropriate component industries and dividing by the 1982 annual average base.

Seasonally adjusted data are not published for a number of series characterized by small seasonal components relative to their trend-cycle and/or irregular components. These series, however, are used in the aggregation to higher level seasonally adjusted series.

Seasonal adjustment factors for Federal Government employment are derived from unadjusted data which include Christmas temporary workers employed by the Postal Service. The number of temporary census workers for the decennial census, however, is removed prior to the calculation of seasonal adjustment factors.

The standard procedure for seasonal adjustment for the local education employment series was improved with the 1997 benchmark. In the past, the seasonal factors for this industry were derived using the standard seasonal adjustment procedure of a logarithmic transformation of the data as input for the multiplicative decomposition of the series. However, in recent years, the forecasted seasonal factors have failed to adequately reflect the changing behavior of this industry in the summer months. The factors for this industry are now derived using a square-root transformation of the data as input for an additive decomposition of the series. These modifications produce seasonal factors that better reflect current industry seasonal patterns. However, the annual averages of seasonally adjusted and unadjusted series will not be equal.

BLS also makes special adjustments for floating holidays for the establishment-based series on average weekly hours and manufacturing overtime hours. From 1988 forward, these adjustments are now accomplished as part of the X-12 ARIMA/REGARIMA modeling process. The special adjustment made in November each year to adjust for the effect of poll workers in the local government employment series also is incorporated into the X-12 process from 1988 forward.

Revised seasonally adjusted national establishment-based series based on the experience through March 2001, new seasonal adjustment factors for March-October 2001, and a description of the current seasonal adjustment procedure appear in the June 2001 issue of Employment and Earnings. Revised factors for the September 2001-April 2002 period will appear in the December 2001 issue.

Beginning in 1993, BLS introduced publication of seasonally adjusted nonfarm payroll employment data by major industry for all States and the District of Columbia (table B-7). Seasonal adjustment factors are applied directly to the employment estimates at the division level (component series for manufacturing and trade) and then aggregated to the State totals. The recomputation of seasonal factors and historical revisions are made coincident with the annual benchmark adjustments. State estimation procedures are designed to produce accurate (unadjusted and seasonally adjusted) data for each State. BLS independently develops a national employment series; State estimates are not forced to sum to national totals. Because each State series is subject to larger sampling and nonsampling errors than the national series, summing them cumulates individual State level errors and can cause significant distortions at an aggregate level. Due to these statistical limitations, BLS does not compile a “sum-of-States” employment series, and cautions users that such a series is subject to a relatively large and volatile error structure.

Region and State labor force data

Beginning in 1992, BLS introduced publication of seasonally adjusted labor force data for the census regions and divisions, the 50 States, and the District of Columbia (tables C-1 and C-2). Beginning in 1998, regional aggregations are derived by summing the State estimates. Using the X-11 ARIMA procedure, seasonal adjustment factors are computed and applied independently to the component employment and unemployment levels and then aggregated to regional or State totals. Current seasonal adjustment factors are produced for 6-month periods twice a year. Historical revisions usually are made at the beginning of each calendar year. Because of the separate processing procedures, totals for the Nation, as a whole, differ from the results obtained by aggregating regional or State data.

Table 1-A. Characteristics of the CPS sample, 1947 to present

Households eligible

Number of

sample Not

Period areas Interviewed interviewed

Aug. 1947 to Jan. 1954 68 21,000 500-1,000

Feb. 1954 to Apr. 1956 230 21,000 500-1,000

May 1956 to Dec. 1959 (1) 330 33,500 1,500

Jan. 1960 to Feb. 1963 (2) 333 33,500 1,500

Mar. 1963 to Dec. 1966 357 33,500 1,500

Jan. 1967 to July 1971 449 48,000 2,000

Aug. 1971 to July 1972 449 45,000 2,000

Aug. 1972 to Dec. 1977 461 45,000 2,000

Jan. 1978 to Dec. 1979 614 53,500 2,500

Jan. 1980 to Apr. 1981 629 62,200 2,800

May 1981 to Dec. 1984 629 57,800 2,500

Jan. 1985 to Mar. 1988 729 57,000 2,500

Apr. 1988 to Mar. 1989 729 53,200 2,600

Apr. 1989 to Oct. 1994 (3) 729 57,400 2,600

Nov. 1994 to Aug. 1995 (4) 792 54,500 3,500

Sept. 1995 to Dec. 1995 792 52,900 3,400

Jan. 1996 to June 2001 754 46,250 3,750

July 2001 to present (5) 754 55,500 4,500

Households

visited

but not

Period eligible

Aug. 1947 to Jan. 1954 3,000-3,500

Feb. 1954 to Apr. 1956 3,000-3,500

May 1956 to Dec. 1959 6,000

Jan. 1960 to Feb. 1963 6,000

Mar. 1963 to Dec. 1966 6,000

Jan. 1967 to July 1971 8,500

Aug. 1971 to July 1972 8,000

Aug. 1972 to Dec. 1977 8,000

Jan. 1978 to Dec. 1979 10,000

Jan. 1980 to Apr. 1981 12,000

May 1981 to Dec. 1984 11,000

Jan. 1985 to Mar. 1988 11,000

Apr. 1988 to Mar. 1989 11,500

Apr. 1989 to Oct. 1994 (3) 11,800

Nov. 1994 to Aug. 1995 (4) 10,000

Sept. 1995 to Dec. 1995 9,700

Jan. 1996 to June 2001 10,000

July 2001 to present (5) 12,000

(1) Beginning in May 1956, these areas were chosen to provide coverage

in each State and the District of Columbia.

(2) Three sample areas were added in 1960 to represent Alaska and

Hawaii after statehood.

(3) The sample was increased incrementally during the 8-month period,

April-November 1989.

(4) Includes 2,000 additional assigned housing units from Georgia and

Virginia that were gradually phased in during the 10-month period,

October 1994-August 1995.

(5) Includes 12,000 assigned housing units in support of the State

Children’s Health Insurance Program.

Table 1-B. Approximate standard errors for major employment

status categories

(In thousands)

Consecutive

Characteristic Monthly month-to-

level month change

Total

Total, 16 years and over:

Civilian labor force 267 174

Employed 273 177

Unemployed 131 166

Men, 20 years and over:

Civilian labor force 184 120

Employed 196 128

Unemployed 83 106

Women, 20 years and over:

Civilian labor force 209 136

Employed 215 140

Unemployed 77 98

Both sexes, 16 to 19 years:

Civilian labor force 90 87

Employed 95 91

Unemployed 56 93

Black

Total, 16 years and over:

Civilian labor force 113 73

Employed 121 79

Unemployed 64 81

Men, 20 years and over:

Civilian labor force 81 53

Employed 85 55

Unemployed 39 50

Women, 20 years and over:

Civilian labor force 72 47

Employed 77 50

Unemployed 40 50

Both sexes, 16 to 19 years:

Civilian labor force 42 40

Employed 39 38

Unemployed 28 46

Hispanic origin

Total, 16 years and over:

Civilian labor force 90 59

Employed 100 65

Unemployed 54 69

Table 1-C. Approximate standard errors for unemployment rates

by major characteristics

(In percent)

Consecutive

Monthly month-to-

Characteristic rate month change

Total 0.09 0.12

Men .12 .16

Men, 20 years and over .12 .15

Women .13 .17

Women, 20 years and over .13 .16

Both sexes, 16 to 19 years .66 1.08

White .10 .12

Black .39 .49

Hispanic origin .37 .47

Married men, spouse present .12 .15

Married women, spouse present .14 .18

Women who maintain families .43 .54

Occupation

Managerial and professional specialty .12 .15

Executive, administrative,

and managerial .17 .21

Professional specialty .16 .21

Technical, sales, and administrative

support .16 .21

Technicians and related support .39 .49

Sales occupations .27 .34

Administrative support, including

clerical .23 .29

Service occupations .29 .37

Private household 1.51 1.92

Protective service .58 .74

Service, except private household and

protective .33 .42

Precision production, craft, and repair .28 .35

Mechanics and repairers .40 .50

Construction trades .50 .64

Other precision production, craft,

and repair .50 .63

Operators, fabricators, and laborers .30 .38

Machine operators, assemblers,

and inspectors .45 .57

Transportation and material moving

occupations .45 .58

Handlers, equipment cleaners, helpers,

and laborers .66 .84

Construction laborers 1.80 2.29

Other handlers, equipment cleaners,

helpers, and laborers .69 .88

Farming, forestry, and fishing .72 .91

Industry

Nonagricultural private wage and salary

workers .11 .14

Goods-producing industries .22 .27

Mining 1.67 2.12

Construction .51 .65

Manufacturing .23 .29

Durable goods .29 .36

Nondurable goods .38 .48

Service-producing industries .12 .16

Transportation, communications, and

public utilities .34 .43

Wholesale and retail trade .23 .30

Finance, insurance, and real estate .29 .37

Services .18 .23

Government workers .18 .23

Agricultural wage and salary workers 1.07 1.36

Table 1-D. Parameters and factors for computation of approximate

standard errors for estimates of monthly levels

Parameters

Characheristic a b

Total or white

Total:

Civilian labor force, employed,

and not in labor force -0.0000077 1586.29

Unemployed – .0000174 3005.06

Men:

Civilian labor force, employed,

and not in labor force – .0000348 2927.43

Unemployed – .0000348 2927.43

Women:

Civilian labor force, employed,

and not in labor force – .0000325 2693.27

Unemployed – .0000325 2693.27

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force – .0002436 3005.06

Unemployed – .0002436 3005.06

Black

Total:

Civilian labor force, employed,

and not in labor force – .0001541 3295.99

Unemployed – .0001541 3295.99

Men:

Civilian labor force, employed,

and not in labor force – .0003361 3332.28

Unemployed – .0003361 3332.28

Women:

Civilian labor force, employed,

and not in labor force – .0002821 2944.26

Unemployed – .0002821 2944.26

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force – .0015306 3295.99

Unemployed – .0015306 3295.99

Hispanic origin

Total:

Civilian labor force, employed,

and not in labor force – .0001868 3295.99

Unemployed – .0001868 3295.99

Men:

Civilian labor force, employed,

and not in labor force – .0003630 3332.28

Unemployed – .0003630 3332.28

Women:

Civilian labor force, employed,

and not in labor force – .0003800 2944.26

Unemployed – .0003800 2944.26

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force – .0018224 3295.99

Unemployed – .0018224 3295.99

Employment

Educational attainment -0.0000174 3005.06

Marital status, men – .0000348 2927.43

Marital status, women – .0000325 2693.27

Women who maintain families – .0000325 2693.27

Mining and manufacturing – .0000174 3005.06

Other industries and

occupations – .0000174 3005.06

Agriculture:

Total – .0013447 2989.22

Wage and salary workers – .0013447 2989.22

Self-employed workers – .0013447 2989.22

Unpaid family workers – .0013447 2989.22

Nonagricultural industries:

Total – .0000174 3005.06

Wage and salary workers – .0000174 3005.06

Self-employed workers – .0000174 3005.06

Unpaid family workers – .0000174 3005.06

Full-time workers – .0000174 3005.06

Part-time workers – .0000174 3005.06

Multiple jobholders – .0000174 3005.06

At work

Total and nonagricultural

industries:

Total – .0000174 3005.06

1 to 4 and 5 to 14 hours – .0000174 3005.06

15 to 29 hours – .0000174 3005.06

30 to 34 or 35 to 39 hours – .0000174 3005.06

1 to 34 or 40 hours – .0000174 3005.06

41 to 48 or 49 to 59 hours – .0000174 3005.06

35+, 41+, or 60+ hours – .0000174 3005.06

Part time for economic reasons – .0000174 3005.06

Part time for noneconomic

reasons – .0000174 3005.06

Unemployment

Educational attainment – .0000174 3005.06

Marital status, men – .0000348 2927.43

Marital status, women – .0000325 2693.27

Women who maintain families – .0000325 2693.27

Industries and occupations – .0000174 3005.06

Full-time workers – .0000174 3005.06

Part-time workers – .0000174 3005.06

Less than 5 weeks – .0000174 3005.06

5 to 14 weeks – .0000174 3005.06

15 to 26 weeks – .0000174 3005.06

15+ or 27+ weeks – .0000174 3005.06

All reasons for unemployment,

except temporary layoff – .0000174 3005.06

On temporary layoff – .0000174 3005.06

Not in the labor force

Total – .0000077 1586.29

Persons who currently want

a job and discouraged

workers – .0000174 3005.06

Factors

Characheristic Consecutive Year-to-year

month-to- change

month of monthly

change estimates

Total or white

Total:

Civilian labor force, employed,

and not in labor force 0.65 1.22

Unemployed 1.27 1.38

Men:

Civilian labor force, employed,

and not in labor force .65 1.23

Unemployed 1.27 1.39

Women:

Civilian labor force, employed,

and not in labor force .65 1.22

Unemployed 1.27 1.39

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .96 1.32

Unemployed 1.65 1.37

Black

Total:

Civilian labor force, employed,

and not in labor force .65 1.22

Unemployed 1.28 1.38

Men:

Civilian labor force, employed,

and not in labor force .65 1.25

Unemployed 1.27 1.37

Women:

Civilian labor force, employed,

and not in labor force .65 1.27

Unemployed 1.27 1.39

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .96 1.33

Unemployed 1.65 1.37

Hispanic origin

Total:

Civilian labor force, employed,

and not in labor force .65 1.20

Unemployed 1.28 1.38

Men:

Civilian labor force, employed,

and not in labor force .65 1.26

Unemployed 1.29 1.38

Women:

Civilian labor force, employed,

and not in labor force .65 1.21

Unemployed 1.27 1.38

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .96 1.34

Unemployed 1.65 1.42

Employment

Educational attainment 0.65 1.11

Marital status, men .65 1.15

Marital status, women .65 1.18

Women who maintain families .65 1.18

Mining and manufacturing .37 .98

Other industries and

occupations .65 1.25

Agriculture:

Total .62 1.22

Wage and salary workers .62 1.22

Self-employed workers .65 .92

Unpaid family workers .65 1.21

Nonagricultural industries:

Total .65 1.15

Wage and salary workers .65 1.13

Self-employed workers .65 1.15

Unpaid family workers .65 1.26

Full-time workers .65 1.17

Part-time workers .65 1.27

Multiple jobholders 1.27 1.29

At work

Total and nonagricultural

industries:

Total .65 1.21

1 to 4 and 5 to 14 hours 1.65 1.36

15 to 29 hours 1.27 1.33

30 to 34 or 35 to 39 hours 1.65 1.34

1 to 34 or 40 hours 1.27 1.30

41 to 48 or 49 to 59 hours 1.65 1.34

35+, 41+, or 60+ hours 1.27 1.25

Part time for economic reasons 1.47 1.37

Part time for noneconomic

reasons 1.27 1.29

Unemployment

Educational attainment 1.27 1.38

Marital status, men 1.27 1.39

Marital status, women 1.27 1.39

Women who maintain families 1.27 1.39

Industries and occupations 1.27 1.38

Full-time workers 1.27 1.38

Part-time workers 1.65 1.40

Less than 5 weeks 1.27 1.38

5 to 14 weeks 1.65 1.37

15 to 26 weeks 1.65 1.39

15+ or 27+ weeks 1.27 1.42

All reasons for unemployment,

except temporary layoff 1.27 1.38

On temporary layoff 1.65 1.35

Not in the labor force

Total .65 1.22

Persons who currently want

a job and discouraged

workers 1.65 1.41

Factors

Characheristic Change in

Quarterly consecutive

averages quarterly

averages

Total or white

Total:

Civilian labor force, employed,

and not in labor force 0.87 0.77

Unemployed .72 .91

Men:

Civilian labor force, employed,

and not in labor force .86 .79

Unemployed .72 .91

Women:

Civilian labor force, employed,

and not in labor force .87 .78

Unemployed .71 .90

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .81 .87

Unemployed .68 .88

Black

Total:

Civilian labor force, employed,

and not in labor force .86 .78

Unemployed .73 .90

Men:

Civilian labor force, employed,

and not in labor force .84 .82

Unemployed .73 .91

Women:

Civilian labor force, employed,

and not in labor force .84 .80

Unemployed .71 .90

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .80 .85

Unemployed .68 .86

Hispanic origin

Total:

Civilian labor force, employed,

and not in labor force .86 .82

Unemployed .71 .90

Men:

Civilian labor force, employed,

and not in labor force .84 .82

Unemployed .71 .90

Women:

Civilian labor force, employed,

and not in labor force .86 .84

Unemployed .71 .89

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .81 .84

Unemployed .70 .89

Employment

Educational attainment 0.87 0.92

Marital status, men .86 .93

Marital status, women .85 .94

Women who maintain families .85 .94

Mining and manufacturing .91 .78

Other industries and

occupations .85 .97

Agriculture:

Total .84 .91

Wage and salary workers .84 .91

Self-employed workers .91 .80

Unpaid family workers .80 .96

Nonagricultural industries:

Total .88 .75

Wage and salary workers .88 .84

Self-employed workers .87 .96

Unpaid family workers .81 .95

Full-time workers .85 .92

Part-time workers .81 .89

Multiple jobholders .78 .91

At work

Total and nonagricultural

industries:

Total .84 .77

1 to 4 and 5 to 14 hours .67 .86

15 to 29 hours .73 .88

30 to 34 or 35 to 39 hours .67 .86

1 to 34 or 40 hours .76 .87

41 to 48 or 49 to 59 hours .71 .86

35+, 41+, or 60+ hours .78 .86

Part time for economic reasons .67 .87

Part time for noneconomic

reasons .74 .85

Unemployment

Educational attainment .72 .91

Marital status, men .72 .91

Marital status, women .71 .90

Women who maintain families .71 .90

Industries and occupations .72 .91

Full-time workers .72 .91

Part-time workers .69 .88

Less than 5 weeks .72 .91

5 to 14 weeks .66 .88

15 to 26 weeks .67 .89

15+ or 27+ weeks .75 .93

All reasons for unemployment,

except temporary layoff .72 .91

On temporary layoff .68 .87

Not in the labor force

Total .87 .77

Persons who currently want

a job and discouraged

workers .63 .83

Factors

Characheristic Change in

Yearly consecutive

averages yearly

averages

Total or white

Total:

Civilian labor force, employed,

and not in labor force 0.68 0.81

Unemployed .42 .57

Men:

Civilian labor force, employed,

and not in labor force .66 .80

Unemployed .43 .57

Women:

Civilian labor force, employed,

and not in labor force .67 .81

Unemployed .41 .55

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .55 .71

Unemployed .40 .53

Black

Total:

Civilian labor force, employed,

and not in labor force .66 .80

Unemployed .43 .58

Men:

Civilian labor force, employed,

and not in labor force .62 .76

Unemployed .43 .58

Women:

Civilian labor force, employed,

and not in labor force .64 .78

Unemployed .41 .56

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .56 .70

Unemployed .41 .52

Hispanic origin

Total:

Civilian labor force, employed,

and not in labor force .65 .78

Unemployed .42 .56

Men:

Civilian labor force, employed,

and not in labor force .62 .76

Unemployed .41 .55

Women:

Civilian labor force, employed,

and not in labor force .63 .76

Unemployed .41 .55

Both sexes, 16 to 19 years:

Civilian labor force, employed,

and not in labor force .58 .73

Unemployed .41 .55

Employment

Educational attainment 0.61 0.74

Marital status, men .59 .72

Marital status, women .57 .72

Women who maintain families .57 .72

Mining and manufacturing .74 .84

Other industries and

occupations .55 .70

Agriculture:

Total .57 .72

Wage and salary workers .57 .72

Self-employed workers .73 .82

Unpaid family workers .49 .61

Nonagricultural industries:

Total .71 .83

Wage and salary workers .67 .79

Self-employed workers .58 .71

Unpaid family workers .50 .65

Full-time workers .59 .72

Part-time workers .55 .69

Multiple jobholders .50 .64

At work

Total and nonagricultural

industries:

Total .66 .79

1 to 4 and 5 to 14 hours .38 .51

15 to 29 hours .45 .58

30 to 34 or 35 to 39 hours .39 .51

1 to 34 or 40 hours .51 .64

41 to 48 or 49 to 59 hours .45 .57

35+, 41+, or 60+ hours .53 .65

Part time for economic reasons .39 .52

Part time for noneconomic

reasons .49 .62

Unemployment

Educational attainment .42 .57

Marital status, men .43 .57

Marital status, women .41 .55

Women who maintain families .41 .55

Industries and occupations .42 .57

Full-time workers .42 .57

Part-time workers .40 .53

Less than 5 weeks .42 .57

5 to 14 weeks .35 .50

15 to 26 weeks .36 .50

15+ or 27+ weeks .44 .60

All reasons for unemployment,

except temporary layoff .42 .57

On temporary layoff .40 .53

Not in the labor force

Total .68 .81

Persons who currently want

a job and discouraged

workers .36 .48

Table 2-A. Summary of methods for computing industry statistics

on employment, hours, and earnings for the non-probability-based

and the probability-based sample estimates

Non-probability sample Probability sample

Employment, Basic estimating cell Basic estimating cell

hours, and (industry, region, size, (industry, 4-digit

earnings or region/size cell) published level)

All employees All-employee estimate for All-employee estimate for

previous month multiplied previous month multiplied

by ratio of all employees by weighted ratio of all

in current month to all employees in current month

employees in previous to all employees in

month, for sample estab- previous month, for sample

lishments that reported establishments, which

for both months. (1) reported for both

months. (2)

Production of All-employee estimate for All-employee estimate for

nonsupervisory current month multiplied current month multiplied

workers, women by (1) ratio of production by (1) the ratio of the

employees or nonsupervisory workers sum of the weighted

to all employees in sample production or nonsupervi-

establishments for current sory workers and the sum

month, (2) estimated ratio of the weighted all

of women to all employees for the current

employees. (3) month and the sum of the

weighted production or

nonsupervisory workers and

the sum of the weighted

all employees for the

previous month that is

applied to the previous

month’s production or

nonsupervisory worker

ratio, (2) the ratio of

the sum of the weighted

women workers and the sum

of the weighted all

employees for the current

month and the sum of the

weighted women workers and

the sum of the weighted

all employees for the

previous month that is

applied to the previous

month’s women worker

ratio.

Average weekly Production or nonsupervi- Production or nonsupervi-

hours sory worker hours divided sory worker hours divided

by number of production or by number of production or

nonsupervisory nonsupervisory

workers. (3) workers. (4)

Average weekly Production worker overtime Production worker overtime

overtime hours hours divided by number of hours divided by number of

production workers. (3) production workers. (4)

Average hourly Total production or non- Total production or non-

earnings supervisory worker payroll supervisory worker payroll

divided by total produc- divided by total produc-

tion or nonsupervisory tion or nonsupervisory

worker hours. (3) worker hours. (4)

Average weekly Product of average weekly Product of average weekly

earnings hours and average hourly hours and average hourly

earnings. earnings.

Both samples

Employment, Aggregate industry level

hours, and (division and, where Annual average data

earnings stratified, industry)

All employees Sum of all-employee esti- Sum of monthly esti-

mates for component cells. mates divided by 12.

Production of Sum of production or Sum of monthly esti-

nonsupervisory nonsupervisory worker es- mates divided by 12.

workers, women timates, or estimates of

employees women employees, for

component cells.

Average weekly Average, weighted by pro- Annual total of aggregate

hours duction or nonsupervisory hours (production or

worker employment, of the nonsupervisory worker

average weekly hours for employment multiplied by

component cells. average weekly hours)

divided by annual sum of

employment.

Average weekly Average, weighted by pro- Annual total of aggregate

overtime hours duction worker employ- overtime hours (production

ment, of the average or nonsupervisory

weekly overtime hours for worker employment

component cells. multiplied by average weekly

overtime hours) divided

by annual sum of

employment.

Average hourly Average, weighted by ag- Annual total of aggregate

earnings gregate hours, of the av- payrolls (production or

erage hourly earnings for nonsupervisory worker

component cells. employment multiplied by

weekly hours and hourly

earnings) divided by

annual aggregate hours.

Average weekly Product of average weekly Product of average weekly

earnings hours and average hourly hours and average hourly

earnings. earnings.

(1) The estimates are computed by multiplying the above product by

bias adjustment factors that compensate for the underrepresentation

of newly formed enterprises and other sources of bias in the sample.

(2) The estimates are computed by applying a unique monthly birth/

death model component that estimates the residual net birth/death

employment not accounted for by the sample,

(3) The sample production-worker ratio, women-worker ratio, average

weekly hours, average overtime hours, and average hourly earnings

are modified by a wedging technique designed to compensate

for changes in the sample arising mainly from the voluntary

characteristics of the reporting. The wedging procedure accepts the

advantage of continuity from the use of the matched sample and, at the

same time, tapers or wedges the estimate toward the level of the

latest sample average.

(4) A weighted link relative estimator is used to move average weekly

hours, average overtime hours, and average hourly earnings forward

from the point at which the probability-based sample estimates

are introduced. For average weekly hours, this ratio is weighted hours

divided by weighted production/nonsupervisory workers. For average

hourly earnings, this ratio is weighted payroll divided by weighted

hours. This will effectively preserve the true month-to-month sample

movement if the new probability sample has different levels than the

current sample.

Table 2-B. March employment benchmarks and model adjustments for

total private industries, March 1990-2000

(In thousands)

Average monthly

Benchmark model adjustment Over-the-year

employment

Year Employment Revision Added (3) Required change (5)

(1) (2) (4)

1990 90,546 -261 85 63 1,531

1991 88,790 -583 61 12 -1,756

1992 88,347 -130 33 22 -443

1993 89,790 288 83 107 1,443

1994 92,730 688 115 171 2,940

1995 96,175 511 144 187 3,445

1996 98,158 72 129 135 1,983

1997 101,040 518 130 173 2,882

1998 103,965 85 150 157 2,925

1999 106,627 242 150 170 2,662

2000 (6) 109,432 352 153 183 2,805

(1) Universe counts for March of each year are use to make

annual benchmark adjustments to the employment estimates.

About 97 percent of the benchmark employment is from unemployment

insurance administrative records, and the remaining 3

percent is from alternate sources. Data represent benchmark

levels as originally computed.

(2) Difference between the final March sample-based estimate and

the benchmark level for total private employment.

(3) The average amount of mode/adjustment each month over the

course of an inter-benchmark period, that is, from April of the prior

year through March of the given year.

(4) The difference between the March benchmark and the March

estimate derived solely from the sample without model adjustment,

converted to a monthly amount by dividing by 12.

(5) March-to-March changes in the benchmark employment

evel.

(6) Wholesale trade uses the net birth/death model.

NOTE: Data in this table exclude government employment because

there is no bias adjustment for this sector.

Table 2-C. Employment benchmarks and approximate coverage

of BLS employment and payrolls sample, March 2000

Sample coverage

Employ-

ment Employees

Industry bench- Number

marks of Percent

(thou- establish- Number of

sands) ments (1) (thou- bench-

sands) marks

Total 130,492 242,854 38,925 30

Mining 525 1,229 127 24

Construction 6,325 23,023 1,024 16

Manufacturing 18,441 22,069 5,801 32

Transportation and

public utilities 6,929 (2) 14,259 2,041 29

Wholesale trade 6,960 8,540 517 7

Retail trade 22,829 54,341 4,867 21

Finance, insurance

and real estate 7,528 19,514 1,858 25

Services 39,895 65,402 7,430 19

Government

Federal 2,808 (3) 7,077 2,808 100

State 4,902 7,545 3,775 77

Local 13,350 19,855 8,677 65

(1) Counts reflect reports used in final estimates. Because not all

establishments report payroll and hours information, hours and

earnings estimates are based on a smaller sample than employment

estimates.

(2) The Interstate Commerce Commission provides a complete

count of employment for Class I railroads plus Amtrak. A small

sample is used to estimate hours and earnings data.

(3) Total Federal employment counts by agency for use in national

estimates are provided to BLS by the U.S. Office of Personnel

Management. Detailed industry estimates for the Executive Branch, as

well as State and area estimates of Federal employment, are based

on a sample of reports covering about 60 percent of employment in

Federal establishments.

Table 2-D. Current (March 2000) and historical benchmark revisions

(Numbers in thousands)

March 2000 Ten-year average

benchmark mean percent

revision revision

Industry Level Percent Actual Absolute

Total 468 0.4 0.2 0.3

Total private 352 .3 .2 .4

Goods-producing 70 .3 .5 .7

Mining 0 0 .6 1.1

Metal mining -4 -10.0 -2.0 3.2

Coal mining -1 -1.3 0 2.3

Oil and gas extraction 2 .7 1.3 1.8

Nonmetallic minerals, except

fuels 2 1.9 .6 1.6

Construction 37 .6 .4 1.2

General building contractors 24 1.6 .2 2.2

Heavy construction, except

building 16 2.0 1.6 1.8

Special trade contractors -2 (1) .2 1.1

Manufacturing 33 .2 .5 .6

Durable goods 32 .3 .6 .8

Lumber and wood products 6 .7 .6 1.5

Furniture and fixtures 2 .4 .8 1.2

Stone, clay, and glass products 15 2.6 .5 1.0

Primary metal industries 3 .4 .3 .8

Blast furnaces and basic steel

products 0 0 .5 .9

Fabricated metal products 6 .4 .5 .8

Industrial machinery and

equipment -10 -.5 .6 .9

Computer and office equipment -4 -1.1 .6 1.5

Electronic and other electrical

equipment 9 .5 .4 .7

Electronic components and

accessories 13 2.0 .7 1.3

Transportation equipment 7 .4 1.0 1.1

Motor vehicles and equipment 0 0 1.2 1.3

Aircraft and parts 5 1.1 .6 1.1

Instruments and related products -3 -.4 .6 1.3

Miscellaneous manufacturing

industries -3 -.8 .9 1.4

Nondurable goods 1 (1) .3 .5

Food and kindred products 19 1.1 .2 .9

Tobacco products 1 2.9 .5 2.4

Textile mill products -11 -2.1 (1) 1.0

Apparel and other textile

products -12 -1.8 .3 1.3

Paper and allied products -2 -.3 .4 .8

Printing and publishing -2 -.1 .1 .5

Chemicals and allied products 7 .7 .2 .8

Petroleum and coal products -3 -2.4 .3 1.7

Rubber and miscellaneous

plastics products 7 .7 .6 .9

Leather and leather products -3 -4.2 0 2.2

Service-producing 398 .4 .1 .3

Transportation and public

utilities 28 .4 .1 .8

Transportation 6 .1 (1) 1.0

Railroad transportation 15 6.4 -.1 1.3

Local and interurban passenger

transit -22 -4.5 -.9 2.4

Trucking and warehousing 17 0.9 -1.1 2.6

Water transportation -5 -2.7 .8 3.6

Transportation by air -1 -.1 2.5 4.0

Pipelines, except natural gas 2 14.3 1.2 5.0

Transportation services -1 -.2 -.3 2.1

Communications and public

utilities 22 .9 .2 1.1

Communications 27 1.7 .4 1.7

Electric, gas, and sanitary

services -6 -.7 -.1 .7

Wholesale trade -41 -.6 -.3 .9

Durable goods -7 -.2 -.1 .9

Nondurable goods -34 -1.2 -.6 1.1

Retail trade 247 1.1 .5 .7

Building materials and garden

supplies -4 -.4 -.4 1.1

General merchandise stores 85 3.1 1.7 2.7

Department stores 89 3.7 1.9 3.1

Food stores 6 .2 (1) .5

Automotive dealers and service

stations -2 -.1 -.9 .9

New and used car dealers 4 .4 .8 .9

Apparel and accessory stores -8 -.7 .4 1.3

Furniture and home furnishings

stores 17 1.5 -.7 1.3

Eating and drinking places 122 1.5 1.1 1.4

Miscellaneous retail

establishments 29 1.0 .1 1.0

Finance, insurance, and real estate -43 -.6 -.1 1.1

Finance -8 -.2 -.5 1.2

Depository institutions -17 -.8 -.9 1.3

Commercial banks -26 -1.8 -.7 1.1

Savings institutions 9 3.6 -2.8 6.1

Nondepository institutions -4 -.6 1.7 2.8

Mortgage bankers and brokers -13 -4.1 1.5 5.5

Security and commodity brokers 3 .4 .4 1.0

Holding and other investment

offices 8 3.2 -3.8 5.1

Insurance -11 -.5 .4 1.3

Insurance carriers -2 -.1 .6 1.4

Insurance agents, brokers, and

services -10 -1.3 .1 1.1

Real estate -24 -1.6 -.3 1.3

Services (2) 91 .2 .1 .5

Agricultural services 2 .3 1.0 1.2

Hotels and other lodging places 1 .1 .7 1.2

Personal services -24 -1.8 .5 1.3

Business services 107 1.1 .4 1.5

Services to buildings -6 -.6 .1 1.1

Personnel supply services 48 1.3 1.0 2.5

Help supply services 64 1.9 1.7 2.5

Computer and data processing

services 143 6.9 2.3 3.3

Auto repair, services, and parking 40 3.2 -.8 1.8

Miscellaneous repair services -18 -4.9 -2.8 5.2

Motion pictures -35 -5.9 -2.3 4.0

Amusement and recreation services -27 -1.7 -.2 3.1

Health services -41 -.4 -.3 .5

Offices and clinics of medical

doctors -5 -.3 -.3 1.0

Nursing and personal care

facilities 5 .3 (1) .7

Hospitals -29 -.7 -.5 .6

Home health care services 2 .3 1.0 2.3

Legal services -2 -.2 -.7 .8

Educational services -29 -1.2 .6 2.1

Social services -54 -1.9 -.3 1.4

Child day care services -47 -6.5 -1.0 5.5

Residential care -14 -1.8 -.5 1.4

Museums and botanical and

zoological gardens 5 5.0 1.8 2.3

Membership organizations 34 1.4 1.9 2.4

Engineering and management

services 5 .1 -1.0 1.5

Engineering and architectural

services 13 1.3 -.2 1.2

Management and public relations -14 -1.3 -2.2 3.1

Services, nec -2 -3.9 -.2 3.8

Government 116 .6 (1) .3

Federal 0 0 0 0

Federal, except Postal Service 0 0 0 0

State 43 .9 .1 .6

Education 48 2.2 .2 1.2

Other State government -6 -.2 (1) .5

Local 73 .5 (1) .3

Education 53 .7 (1) .4

Other local government 19 .3 .1 .4

(1) Less than 0.05 percent.

(2) Includes other industries, not shown separately.

NOTE: Nec is an abbreviation for “not elsewhere classified” and

designates broad categories of industries that cannot be more

specifically identified.

Table 2-E. Errors of preliminary employment estimates

Mean percent

Root-mean-square revision

error of monthly

Industry level (1) Actual Absolute

Total 42,300 0 0

Total private 35,400 0 0

Goods-producing 10,400 0 0

Mining 1,700 0 .3

Metal mining 400 -.1 .6

Coal mining 700 .1 .6

Oil and gas extraction 1,400 0 .4

Nonmetallic minerals, except

fuels 400 .1 .3

Construction 6,600 0 .1

General building contractors 3,400 .1 .2

Heavy construction, except

building 3,200 .1 .3

Special trade contractors 4,200 0 .1

Manufacturing 10,600 0 0

Durable goods 7,500 0 0

Lumber and wood products 1,600 0 .2

Furniture and fixtures 1,000 0 .2

Stone, clay, and glass products 1,200 0 .2

Primary metal industries 1,600 0 .2

Blast furnaces and basic steel

products 1,200 -.1 .4

Fabricated metal products 2,000 0 .1

Industrial machinery and

equipment 2,700 0 .1

Computer and office equipment 2,000 .3 .4

Electronic and other electrical

equipment 2,100 0 .1

Electronic components and

accessories 1,600 0 .2

Transportation equipment 5,700 0 .2

Motor vehicles and equipment 4,600 0 .3

Aircraft and parts 1,600 -.1 .2

Instruments and related products 1,300 0 .1

Miscellaneous manufacturing 800 0 .2

Nondurable goods 4,900 0 .1

Food and kindred products 2,900 0 .1

Tobacco products 600 .9 1.2

Textile mill products 1,100 0 .2

Apparel and other textile

products 2,600 .2 .3

Paper and allied products 1,200 0 .1

Printing and publishing 1,500 0 .1

Chemicals and allied products 1,600 -.1 .1

Petroleum and coal products 800 -.1 .4

Rubber and miscellaneous

plastics products 1,200 0 .1

Leather and leather products 400 0 .3

Service-producing 49,000 0 0

Transportation and public

utilities 8,700 0 .1

Transportation 8,300 -.1 .1

Railroad transportation 2,100 -.2 .7

Local and interurban passenger

transit 2,600 -.2 .4

Trucking and warehousing 4,900 -.1 .2

Water transportation 1,500 -.1 .7

Transportation by air 6,800 0 .4

Pipelines, except natural gas 100 -.3 .7

Transportation services 1,400 -.1 .2

Communications and public

utilities 3,700 .1 .1

Communications 3,200 .1 .2

Electric, gas, and sanitary

services 1,300 0 .1

Wholesale trade 7,200 .1 .1

Durable goods 4,400 .1 .1

Nondurable goods 4,700 0 .1

Retail trade 27,600 0 .1

Building materials and garden

supplies 2,800 .1 .2

General merchandise stores 19,200 0 .5

Department stores 18,900 -.1 .6

Food stores 5,300 0 .1

Automotive dealers and service

stations 2,900 -.1 .1

New and used car dealers 1,100 -.1 .1

Apparel and accessory stores 5,200 .2 .4

Furniture and home furnishings

stores 2,300 0 .2

Eating and drinking places 10,000 0 .1

Miscellaneous retail

establishments 8,200 .2 .2

Finance, insurance, and real estate 5,700 0 .1

Finance 4,500 0 .1

Depository institutions 3,100 -.1 .1

Commercial banks 2,800 -.1 .1

Savings institutions 700 -.1 .2

Nondepository institutions 2,000 0 .2

Mortgage bankers and brokers 1,500 0 .4

Security and commodity brokers 1,100 0 .1

Holding and other investment

offices 1,700 -.1 .6

Insurance 2,600 0 .1

Insurance carriers 2,300 0 .1

Insurance agents, brokers, and

service 1,300 .1 .1

Real estate 2,300 0 .1

Services (2) 30,200 0 .1

Agricultural services 3,400 .1 .3

Hotels and other lodging places 6,300 0 .3

Personal services 6,100 -.1 .3

Business services 14,700 0 .1

Services to buildings 2,500 0 .2

Personnel supply services 11,700 0 .3

Help supply services 11,100 0 .3

Computer and data processing

services 3,100 0 .2

Auto repair, services, and parking 1,900 0 .1

Miscellaneous repair services 1,000 0 .2

Motion pictures 5,800 .2 .8

Amusement and recreation services 9,200 .1 .4

Health services 5,100 0 0

Offices and clinics of medical

doctors 2,300 0 .1

Nursing and personal care

facilities 1,500 0 .1

Hospitals 3,300 0 .1

Home health care services 1,800 .1 .2

Legal services 1,400 0 .1

Educational services 12,400 .1 .5

Social services 9,200 .1 .2

Child day care services 4,300 .2 .5

Residential care 1,300 0 .1

Museums and botanical and

zoological gardens 500 0 .4

Membership organizations 3,300 0 .1

Engineering and management

services 5,100 0 .1

Engineering and architectural

services 2,000 -.1 .2

Management and public relations 3,500 .1 .3

Services, nec 500 -.1 .8

Government 22,100 0 .1

Federal 12,400 0 .3

Federal, except Postal Service 10,100 .1 .3

State 12,000 0 .2

Education 10,600 .1 .5

Other State government 4,500 0 .1

Local 16,900 0 .1

Education 14,700 0 .2

Other local government 8,700 .1 .1

(1) The root-mean-square error is the square root of the mean

squared error. The mean squared error is the square of the difference

between the final and preliminary estimates averaged across a series of

monthly observations.

(2) Includes other industries, not shown separately.

NOTE: Nec is an abbreviation for “not elsewhere classified” and

designates broad categories of industries that cannot be more

specifically identified. Errors are based on differences from January

1996 through December 2000.

Table 2-F. Bias adjustment effects for published series versus

net birth/death model effects for the mining, construction,

and manufacturing industries

(In thousands)

Mining Construction

Net birth/death

Bias adjustment Bias

Year and month adjustment for the adjustment

for published post- for published

series benchmark series

period

Monthly amount

2000:

April 0 0 12

May 0 1 12

June 0 0 12

July 0 0 12

August 0 1 12

September 0 1 12

October 0 0 14

November 0 0 14

December 0 0 14

2001:

January 0 -7 13

February 0 0 13

March 0 0 13

Cumulative total 0 -4 153

Construction Manufacturing

Net birth/death Net birth/death

adjustment Bias adjustment

Year and month for the adjustment for the

post- for published post-

benchmark series benchmark

period period

Monthly amount

2000:

April 44 8 1

May 46 8 17

June 32 8 11

July 14 7 1

August 17 7 12

September 11 7 8

October 10 9 -4

November -13 9 3

December -16 9 3

2001:

January -85 7 -22

February 13 7 10

March 31 7 14

Cumulative total 104 93 54

Table 2-G. Relative standard error for estimates of employment,

hours, and earnings in selected industries

(In percent)

Relative standard error

Average

Average hourly

Industry All employees weekly hours earnings

Mining 1.90 2.50 2.06

Metal mining 3.56 3.74 3.06

Coal mining 3.79 3.45 2.54

Oil and gas extraction 2.37 3.96 3.61

Nonmetallic minerals,

except fuels 3.06 1.75 1.98

Construction .63 .74 .65

General building

contractors 1.13 1.31 1.26

Heavy construction, except

building 1.66 1.70 1.31

Special trade contractors .82 1.11 .85

Manufacturing .24 .27 .24

Durable goods .32 .37 .30

Lumber and wood products .95 1.28 .71

Furniture and fixtures .95 1.49 1.13

Stone, clay, and glass

products 1.08 2.06 1.22

Primary metal industries .87 1.49 .93

Blast furnaces and

basic steel products 1.30 3.03 1.68

Fabricated metal

products .73 1.00 .76

Industrial machinery and

equipment .62 .80 .70

Computer and office

equipment 1.91 5.69 3.82

Electronic and other

electrical equipment .81 1.05 1.08

Electronic components

and accessories 1.30 1.09 2.20

Transportation equipment 1.12 .98 .87

Motor vehicles and

equipment 1.75 1.32 1.32

Aircraft and parts 1.42 1.49 1.81

Instruments and related

products 1.06 1.52 .89

Miscellaneous

manufacturing 1.47 1.72 1.79

Nondurable goods .39 .48 .38

Food and kindred products .92 .92 .91

Tobacco products 3.19 2.87 3.82

Textile mill products 1.11 1.79 1.26

Apparel and other textile

products 1.88 1.95 1.34

Paper and allied products .87 1.03 .76

Printing and publishing .72 1.03 1.21

Chemicals and allied

products .85 1.18 1.40

Petroleum and coal products 1.82 4.73 2.62

Rubber and miscellaneous

plastics products .68 1.08 .70

Leather and leather

products 3.82 3.03 1.46

Wholesale trade .54 .73 .80

Durable goods .55 .70 .95

Nondurable goods .92 1.40 1.37

Table 2-H. Standard error for change in levels estimates of employment,

hours, and earnings in selected Industries

Standard error

1-month change

Industry

All Average Average

em- weekly hourly

ployees hours earnings

Mining 2,375 0.36 0.11

Metal mining 326 .39 .10

Coal mining 476 .44 .13

Oil and gas extraction 2,110 .61 .20

Nonmetallic minerals, except fuels 754 .40 .09

Construction 11,731 .10 .04

General building contractors 5,745 .20 .08

Heavy construction, except building 4,347 .29 .09

Special trade contractors 9,671 .13 .05

Manufacturing 13,086 .05 .02

Durable goods 10,138 .07 .02

Lumber and wood products 2,599 .21 .04

Furniture and fixtures 1,823 .22 .04

Stone, clay, and glass products 1,882 .33 .06

Primary metal industries 1,642 .22 .07

Blast furnaces and basic steel

products 831 .36 .18

Fabricated metal products 2,875 .14 .04

Industrial machinery and equipment 3,728 .13 .03

Computer and office equipment 1,659 .43 .13

Electronic and other electrical

equipment 3,224 .17 .05

Electronic components and

accessories 1,662 .29 .11

Transportation equipment 5,823 .23 .08

Motor vehicles and equipment 5,877 .32 .11

Aircraft and parts 992 .29 .09

Instruments and related products 1,731 .25 .06

Miscellaneous manufacturing 1,629 .26 .07

Nondurable goods 7,290 .08 .02

Food and kindred products 4,972 .18 .04

Tobacco product 514 .48 .25

Textile mill products 1,425 .21 .04

Apparel and other textile products 2,687 .25 .04

Paper and allied products 1,375 .21 .05

Printing and publishing 2,889 .17 .05

Chemicals and allied products 2,013 .22 .08

Petroleum and coal products 878 .79 .22

Rubber and miscellaneous plastics

products 2,166 .17 .04

Leather and leather products 627 .41 .08

Wholesale trade 9,548 .10 .04

Durable goods 6,629 .11 .06

Nondurable goods 6,211 .15 .05

Standard error

3-month change

Industry

All Average Average

em- weekly hourly

ployees hours earnings

Mining 4,320 0.48 0.15

Metal mining 619 .75 .16

Coal mining 833 .63 .18

Oil and gas extraction 3,805 .79 .26

Nonmetallic minerals, except fuels 1,689 .52 .14

Construction 19,046 .15 .06

General building contractors 9,008 .25 .10

Heavy construction, except building 6,941 .38 .14

Special trade contractors 16,126 .18 .07

Manufacturing 18,795 .07 .02

Durable goods 14,892 .09 .03

Lumber and wood products 4,316 .27 .05

Furniture and fixtures 2,800 .28 .06

Stone, clay, and glass products 3,296 .40 .08

Primary metal industries 2,703 .26 .09

Blast furnaces and basic steel

products 1,325 .45 .23

Fabricated metal products 4,744 .19 .05

Industrial machinery and equipment 5,928 .18 .05

Computer and office equipment 3,295 .56 .27

Electronic and other electrical

equipment 5,161 .23 .07

Electronic components and

accessories 3,692 .33 .15

Transportation equipment 7,914 .34 .12

Motor vehicles and equipment 7,812 .53 .18

Aircraft and parts 1,893 .35 .12

Instruments and related products 3,492 .35 .07

Miscellaneous manufacturing 2,593 .31 .11

Nondurable goods 11,710 .10 .03

Food and kindred products 8,401 .23 .05

Tobacco product 1,064 .44 .54

Textile mill products 2,092 .29 .04

Apparel and other textile products 4,684 .34 .06

Paper and allied products 2,327 .24 .07

Printing and publishing 4,591 .23 .07

Chemicals and allied products 4,040 .30 .11

Petroleum and coal products 1,403 1.33 .32

Rubber and miscellaneous plastics

products 3,594 .22 .05

Leather and leather products 1,167 .58 .12

Wholesale trade 15,513 .13 .06

Durable goods 11,247 .13 .08

Nondurable goods 9,929 .21 .07

Standard error

12-month change

Industry

All Average Average

em- weekly hourly

ployees hours earnings

Mining 6,972 0.76 0.28

Metal mining 1,098 1.31 .34

Coal mining 1,961 1.27 .42

Oil and gas extraction 5,718 1.17 .47

Nonmetallic minerals, except fuels 2,608 .65 .24

Construction 29,060 .27 .10

General building contractors 14,865 .35 .17

Heavy construction, except building 9,848 .57 .23

Special trade contractors 24,619 .38 .12

Manufacturing 38,622 .10 .03

Durable goods 28,805 .12 .04

Lumber and wood products 6,432 .40 .07

Furniture and fixtures 4,432 .39 .10

Stone, clay, and glass products 5,213 .64 .15

Primary metal industries 5,288 .33 .12

Blast furnaces and basic steel

products 2,916 .70 .27

Fabricated metal products 9,727 .33 .08

Industrial machinery and equipment 9,460 .24 .10

Computer and office equipment 5,566 1.43 .66

Electronic and other electrical

equipment 11,287 .49 .12

Electronic components and

accessories 7,364 .51 .25

Transportation equipment 14,799 .37 .13

Motor vehicles and equipment 14,953 .54 .20

Aircraft and parts 4,850 .56 .27

Instruments and related products 6,795 .50 .11

Miscellaneous manufacturing 4,192 .44 .16

Nondurable goods 20,172 .17 .04

Food and kindred products 10,307 .33 .10

Tobacco product 1,505 .90 .45

Textile mill products 3,618 .50 .11

Apparel and other textile products 7,557 .58 .10

Paper and allied products 5,310 .38 .12

Printing and publishing 8,240 .34 .12

Chemicals and allied products 8,130 .49 .17

Petroleum and coal products 2,310 1.90 .53

Rubber and miscellaneous plastics

products 5,605 .34 .08

Leather and leather products 2,357 .90 .18

Wholesale trade 29,903 .23 .10

Durable goods 19,165 .23 .13

Nondurable goods 19,307 .41 .15

Chart 1: Distribution of CES sample by

collection mode

FAX/EDI/WEB 11%

Mail 16%

CATI 5%

TDE 61%

Tape/diskette 7%

Note: Table made from pie chart.

COPYRIGHT 2002 U.S. Department of Labor

COPYRIGHT 2004 Gale Group