Purpose: To provide estimates of employment, unemployment, and other characteristics of the general labor force, of the population as a whole, and of various subgroups of the population. Monthly labor force data for the country are used by the Bureau of Labor Statistics (BLS) to determine the distribution of funds under the Job Training Partnership Act. These data are collected through combined computer-assisted personal interviewing (CAPI) and computer-assisted telephone interviewing (CATI). In addition to the labor force data, the CPS basic funding provides annual data on work experience, income, and migration from the March Annual Demographic Supplement and on school enrollment of the population from the October Supplement. Other supplements, some of which are sponsored by other agencies, are conducted biennially or intermittently. Survey design and sample size: Data from the 1990 decennial census and the new construction universe were used to select independent samples for the 50 states and the District of Columbia. We completed phase-in of the 1990-based samples in July 1995. The combined samples allow the CPS to produce reliable monthly estimates for the country, as well as reliable estimates for all states and selected metropolitan areas on an annual average basis. The total sample size is approximately 72,000 assigned households per month located in 754 primary sampling units.
Each month's sample is composed of eight panels that rotate on a schedule of 4 months in, 8 months out, 4 months in, so that only 25 percent of the households differ between consecutive months.
The regular March CPS sample is supplemented with households containing people of Hispanic origin from the prior November panel. This is done to provide more reliable data for this subgroup in the March Annual Demographic Supplement.
Type of respondent: A household respondent, who must be a knowledgeable household member 15 years old or over, provides information for each household member. Some supplemental inquiries require response by a designated sample person.
Sponsoring agencies and legal authorities: The U.S. Census Bureau and the BLS jointly sponsor the survey under the authorities of Title 13, United States Code, Section 182, and Title 29, United States Code, Sections 1-9.
Periodicity: A continuing survey with interviewing conducted during one week of each month.
Release of results: The first release of monthly employment data by the BLS occurs approximately two weeks after completion of data collection. The final report, Employment and Earnings, is published by the BLS approximately six weeks after data collection. On a quarterly basis, earnings data for people in the labor force are published in the form of a press release, and characteristics of people not in the labor force are published in Employment and Earnings. Advance reports on supplement data are usually released approximately 3 to 6 months after data collection; final reports for supplements are typically released within one year to 18 months. Public use microdata files are made available within six months to one year after data collection.
Historical background: The Census Bureau conducted this survey monthly since 1942 in response to a need that emerged in the late 1930s for reliable and up-to-date estimates of unemployment. During the 50+ years of survey operations, revisions were made in the sample design, questionnaire content, and data processing procedures. Such changes from 1942 through 1975 are discussed in Concepts and Methods Used in Labor Force Statistics Derived from the Current Population Survey, a joint BLS/Census Bureau publication.
Between FY 1976 and FY 1980, the sample was expanded in increments to a total of about 85,000 assigned households per month to improve the reliability of annual average employment and unemployment data for states, metropolitan statistical areas, and central cities. In late FY 1981, the sample size was reduced to 71,000 households due to the BLS budget constraints. Because the reduction was confined to samples recently added to improve the reliability of state and substate data, it had virtually no effect on the reliability of national level estimates. A state-based sample redesign using 1980 census information was introduced in April 1984 to improve the reliability of states and substate areas without requiring increases in the total sample size. This sample design continues with the new sample selected using 1990 census information. In January 1996, we reduced the number of assigned households to 59,500 per month due to the BLS budget constraints. The reduction had virtually no effect on the reliability of national level estimates. This was because the entire reduction was confined to states and metropolitan areas that previously contained sample households that provided reliable estimates on a monthly basis.
In September 2000, the CPS began an expansion of its sample to accommodate increased funding provided by the State Children's Health Insurance Program (SCHIP). Although focused primarily on the March sample, the SCHIP funding provided for an approximately 20 percent increase in the basic CPS sample. This expansion, which occurred in 31 small states and the District of Columbia, was completed in November 2000. Over the next several months, both the BLS and the Census Bureau monitored the impact of this expansion on the CPS labor force estimates, and the BLS decided to begin using the expanded sample in its official CPS estimates in July 2001. Since the focus of this expansion was on estimates of children's health insurance coverage, the March CPS sample was further increased by bringing back 3 rotation groups of households from the previous November and also asking the March Annual Demographic Supplement in the rotation groups in February and in April that are not part of the March sample. Interviewing in these rotation groups was limited to all minority households and White households with children under 18.
We implemented three major initiatives starting in January 1994 - a fully automated data collection methodology, a new labor force questionnaire, and a completely revamped processing system.
Special feature: CATI continues to be used to conduct a portion of the CPS sample cases. Sample cases from all regions are systematically introduced into CATI to provide a monthly work load of 7,200 cases. All cases conducted in the field use CAPI.
Future outlook: We will continue to monitor closely the effects the January 1994 changes and the SCHIP sample increase have on the data.
