Revisions of state and local area labor force statistics.
The Local Area Unemployment Statistics program produces civilian Labor force data for all States, metropolitan areas, counties, and cities with a population of 25,000 or more. In addition to their variety of uses by private industry and individuals, the data constitute one of the bases for the allocation of Federal funds to States and local areas under a variety of programs administered by the Department of Housing and Urban Development, the Department of Labor, and others.
The underlying concepts and definitions of all Labor force data published from the Local Area Unemployment Statistics program are consistent with those of the Current Population Survey (cps), a survey of about 59,500 households conducted monthly by the Bureau of the Census for the Bureau of Labor Statistics. All annual average data for States are drawn directly from the cps. Monthly cps data are used directly as the official Labor force levels only for the 11 largest States and two large areas. These States and areas have a sufficiently large sample in the cps to yield monthly estimates that meet BLS standards of reliability.
For the remaining 39 States and the District of Columbia, as well as for all areas, monthly estimates are developed by State employment security agencies, using the prescribed methodology from the Local Area Unemployment Statistics program. This methodology, commonly referred to as the "Handbook" methodology, uses establishment data derived from the Current Employment Statistics program and administrative data on State unemployment insurance claimants to develop the Labor force estimates for these States and areas. These derived Handbook estimates are adjusted by incorporating statewide monthly cps data to arrive at the official preliminary estimates.
The annual revision process, also called benchmarking, adjusts preliminary monthly estimates and historical monthly and annual average data by taking into account updated population estimates, revised employment figures from the Current Employment Statistics program, newly available decennial census data, and changes which occur from time to time in the geographic definitions of statistical estimating areas or in the estimating methodology.
Revisions to area data are made annually for 2 years of historical data (the minimum required for legislative purposes). Therefore, at any given time, BLS will have newly benchmarked area data for the most recent 2 years, plus preliminary monthly estimates for the current production year. Historical Handbook estimates at the statewide level are revised as necessary for all years in the database. Consequently, the fully consistent series for all States extend back to at least 1976. Indeed, data are available as far back as 1970 for many States, as well as for New York City and the Los Angeles-Long Beach Metropolitan Statistical Area.
Incorporation of updated population estimates.
Cps annual average Labor force data are revised each year to reflect new annual statewide population estimates issued by the Bureau of the Census for the most recent year or years. These revisions are often described as adjustments for updated "population controls." Monthly statewide data for the revised year(s) are adjusted correspondingly to reflect the new annual average data. Therefore, each year BLS issues a revised monthly and annual average State Labor force data series for all affected years. Table l shows the effect of revisions to population controls by comparing, for 1985, the original cps annual average unemployment levels with the revised levels.
The nature of the methodology for estimating local area Labor force data suggests that any revisions to statewide estimates necessitate a change in area estimates to maintain consistency between the State and area data series. Independent estimates of the levels of employment and unemployment are created for geographically exhaustive Labor Market Areas within each State; estimates for counties and cities within those areas are disaggregated from the independent Labor Market Area estimates. For both employment and unemployment, the sums of all Labor Market Area estimates are then forced to add up to the adjusted statewide estimates, using a straight-line ratio adjustment. Therefore, because of this procedure of additivity adjustment, revisions to State estimates (such as those effected by population controls) will automatically require the revision of all area estimates.
Adjustment for revised employment inputs.
Prior to the annual revision process of the Local Area Unemployment Statistics, the Current Employment Statistics program conducts its own revision process, using statistics from a universe count of industry employment as its benchmark. The revised Current Employment Statistics data become the new employment inputs (or benchmark) on which revised Local Area Unemployment Statistics Handbook estimates for States and local areas are based. All independent estimates developed by using the Handbook methodology are affected, as are data for local areas disaggregated from the Labor Market Areas.
Incorporation of decennial census data.
Breaks in series, requiring historical revisions, are created when BLS incorporates new decennial census data for States and loczal areas into lasbor force estimates. Decennial census data, like cps and Local Area Unemployment Statistics estimates, are based on place of residence. however, employment data from the Current Employment Statistics program are based on employees' "place of work" because the data come from a survey of business establishments. To make these employment data consistent with the cps, "residency-adjustment ratios" are applied to the monthly employment data derived from the Current Employment Statistics program. These residency-adjustment ratios are developed by using decennial census data and Current Employment Statistics data for the census year. Several years after each decennial census, new residency-adjustment ratios are developed by incorporating updated census data and are used in creating the monthly Handbook estimates for the remainder of the decade. To provide for consistent historical data, previous years' estimates (within the decade) must be revised using the new ratios. For example, the new residency-adjustment ratios for the 1980's were developed in late 1985. All estimates since that time have been based on the new ratios, and data were revised historically back to 1980.
Changes in geographic definitions.
The U.S. Office of Management and Budget periodically changes the geographic definitions of Metropolitan Statistical Areas, and other Labor Market Areas under which all Federal data collection is conducted, to conform with changes in the distribution of population. Revisions to historical data are required to produce a consistent time series. For example, in March 1985, with the publication of January 1985 data, the geographic definitions of many metropolitan areas, which were designated as Labor Market Areas, were changed by the addition or deletion of component counties. Historical data for redefined Metropolitan Statistical Areas and Primary Metropolitan Statistical Areas were revised to reflect the new definitions used in current data estimation; moreover, because of the additivity adjustment procedure, the data for all other Labor Market Areas within each State were affected and were also revised.
Changes in estimating methodology.
Revisions are also necessary when a change or enhancement in the estimating methodology is approved and instituted. To ensure consistency in the entire data series, a new or changed methodology must be used not only in current estimates, but also in revising historical estimates. Such methodological changes, therefore, are only instituted at the beginning of a calendar year and are required to be included in the revisions of all historical data.
Each of these adjustments illustrates the significance of the benchmarking process and the reasons for the inconsistency between newly revised, benchmarked data and unrevised, unbenchmarked data. Tables 2 and 3 illustrate the impact of the benchmarking process on preliminary statewide unemployment estimates. In both tables, unpublished preliminary annual average data for 1986 (calculated using published monthly data) are compared with the official 1986 cps annual average data for each State. Table 2 presents the difference in the preliminary and revised unemployment levels, and table 3 compares civilian unemployment rates.
The magnitude of the effect of the annual revision process on the preliminary estimates varies by State and area, depending on the degree to which new inputs differ from the original inputs. The revised figures, however, are always considered to be a more accurate reflection of the actual Labor market situation.
|Printer friendly Cite/link Email Feedback|
|Author:||Laedlein, Valyrie K.|
|Publication:||Monthly Labor Review|
|Date:||Jul 1, 1987|
|Previous Article:||Primer on Employee Retirement Income Security Act.|
|Next Article:||The defense buildup, 1977-85: effects on production and employment.|