Printer Friendly

New benchmarks and SIC codes for Establishment Survey.

New benchmarks and SIC codes for Establishment Survey With the release of data for August 1990, the Bureau of Labor Statistics introduced its annual revision of national estimates of employment, hours, and earnings from the monthly sample survey of nonfarm establishments. Each year, the sample estimates are adjusted to new benchmarks, which are comprehensive universe counts of employment based primarily on unemployment insurance reports filed by all employers with State employment security agencies.

Also effective with the August 1990 release, all industry series have been covverted to 1987 Standard Industrial Classfication (SIC) codes. (1) This new structure replaces the 1972 SIC coding structure previously in effect for the industry estimates.

All data from April 1988 forward have been revised to incorporate both the March 1989 benchmarks and the effects of the SIC revision. Historical (pre-1988) data for industry series affected by SIC redefinitions have been reconstructed where possible. Historical data for industry series unaffected by the SIC revision remain as previously published.

As is the usual practice with the introduction of new benchmarks, the Bureau has also revised all seasonally adjusted series for the previous 5-year period and has introduced new seasonal adjustment factors to be used to adjust data in the months ahead.

In addition, all published constant-dollar and indexed series have been recomputed on a 1982 base, replacing the previously published 1977-based data.

Conversion to 1987 SIC coding

The SIC codinig system is periodically updated to reflect structural and technological changes in the economy. The 1987 SIC revision marks the first full SIC restructuring since 1972; minor updates were made to the SIC system in 1977.

There were almost no changes in scope at the major industry division levels, with only very minor shifts between the finance, insurance, and real estate division and the services division. However, there were several significant redefinitions atothe two-digit level. In manufacturing, a substantial realignment took place between electronic and other electrical equipment (SIC 36) and instruments and related products (SIC 38). In services, a new two-digit code (SIC 87) was established for "engineering and management services." Most of the activities under this new heading had previously been classified as business services (SIC 73) or miscellaneous services (SIC 89). At the three- and four-digit SIC levels, changes in scope were both more prevalent and more substantial.

All restructured industries were reestimated using the 1987 SIC-coded sample data from January 1988 forward. Some aggregate-level industries, without scope changes, have also been affected by the retabulations because they are formed from the summations of restratified, reestimated component industries.

For industries with relatively minor scope changes, historical data were reconstructed back to the inception of the series wherever possible. The reconstruction of histrorical series was done by adjusting the existing 1972-based employement series for the percentage of employment lost or the percentage of employment gained from other industries, using ratios derived from first-quarter 1988 universe employment data. (2) Hours and earnings data for restructured series were derived by computing a weighted average of the component series they were derived from. The weights are the percentages of employment each told series contributed to the new series.

Effect of revisions

The net impact of the SIC restructuring and the adjustment to March 1989 benchmark levels on total nonfarm employment was an upward revision of only 9,000 from the previously published level. Table 1 presents, for March 1989, previously published estimates based on the 1972 SIC codes, retabulated estimates based on the 1987 SIC codes, and the newly published benchmark levels. It displays separately the revision effects due to SIC restructuring and those due to benchmarking and shows the net effect, which is the sum of the two.

For total nonfarm employment, the SIC revision effect, due entirely to restratification and not to any scope change, was 56,000, or less than 0.05 percent. At the detailed industry level, the largest effects of the SIC revision were in business services, instruments and related products, and electronic and other electrical equipment.

The benchmark effect shown in the table represents a comparison of March 1989 estimates retabulated under the 1987 SIC structure with the March 1989 benchmark levels. For total nonfarm employment, the benchmark level stands at 107,026,000. this represents a benchmark adjustment of -47,000, or less than 0.05 percent. There were, however, larger but essentially offsetting errors between the goods-producing and service-producing sectors. Benchmark revisions totaling -286,000 were spread across all the major industry divisions in the goods-producing sector, continuing the pattern of overestimation of these industries over the last several years. Offsets to this overestimation occurred in the service-producing industries, which were revised upward by a total of 239,000.

Revised estimates were computed each month from March 1989 forward (the postbenchmark period), based on the new benchmark levels. On a seasonally adjusted basis, the monthly revision increased from -77,000 in March 1989 to -153,000 by May 1990, with larger differences in some of the intervening months. These revisions reflects restratification effects from the SIC revision and a recomputation of both the bias adjustment and the seasonal adjustment factors. Table 2 shows the extent of the revisions for 1989 and 1990, in both level and change, through a comparison of seasonally adjusted monthly data as previously published and as revised.

Sources of differences

Differences between population benchmarks and sample-based estimates result from both sampling and nonsampling error. Sampling error occurs anytime a sample is used to make inferences about a population.

Both the benchmark levels and the sample-based estimates are subject to several sources of nonsampling error, cheef among which are (1) the inability to measure employment in new firms from the time of their inception, due to the time lag between the creation of new firms and their inclusion in the sample; (2) the procedures for handling changes in industrial classification; (3) the quality of the various source data used to derive the benchmark; (4) the inability to cover completely all firms in the target population; and (5) other sources of errors in coverage, response, processing, and collection.

Effect of revision on other series

As with the all-employee data, estimates were recomputed from sample data for women workers and production workers and for hours and earnings in industries affected by the SIC revisions, from January 1988 forward. At the total private level, hours and earnings were unchanged, and there were only minor changes in major division-level data.

Benchmarks are not available for the series on women, production and nonsupervisory workers, hours, and earnings. Women and production worker series are revised by applying the sample-derived ratio to the revised employment estimate at the basic cell level. These revisions are then summarized and incorporated into the broader industry groupings. Production and nonsupervisory worker employment estimates are used as weights in the estimation of hours and earnings at aggregate industry levels. Benchmark revisions to employment may cause shifts in these weights, with a minor effect of summary-level estimates of hours and earings.

Seasonal adjustment procedure

Each year, employment, hours, and earnings data from the new benchmark levels are incorporated into the calculation of new seasonal adjustment factors. The Bureau uses the X-11 ARIMA seasonal adjustment method, developed by Statistics Canada, (3) to seasonally adjust establishment-based employment, hours, and earnings data. The ARIMA option is used to project the unadjusted data forward for 1 year prior to seasonally adjusting the series. The use of ARIMA projections lessens the need for revisions of historical data in future seasonal adjustments.

All published seasonally adjusted series have been revised for the most recent 5 years (1985-90) for the incorporation of new seasonal factors, as usual. In addition, series affected by the SIC revision which were reconstructed for years prior to 1985 have again been seasonally adjusted, based on the 1987 SIC-based estimates.

Publication of revised data

Revised estimates for all series appear in the August 1990 issue of the Bureau's periodical, Employment and Earnings, along with a more complete explanation of benchmarking, SIC revision, bias factors, and the new seasonal adjustment factors.

Data for detailed industry categories of employment, hours, and earnings will be presented in the Bureau's historical bulletin, Employment, Hours, and Earnings, United States, 1909-90. This publication will contain all of the historical data that were revised as a result of the 1987 SIC revision, the March 1989 benchmarks, updated seasonal adjustment factors, and the rebasing of constant-dollar and indexed series, as well as prior data unaffected by these revisions. Estimates reflecting the new benchmarks appear in the "Current Labor Statistics" section of the Monthly Labor Review, beginning with September data in the November issue.


(1) As defined in the 1987 Standard Industrial Classification Manual, issued by the Executive Office of the President, Office of the Management and Budget.

(2) All ratios are based on first-quarter 1988 universe employment data. For additional information, see Employment Data under the New Standard Industrial Classifications, First Quarter 1988, Report 772, October 1989.

(3) A detailed description of the procedures appears in the X-11 ARIMA Seasonal Adjustment Method, by Estella Bee Dagum, Statistics Canada Catalague No. 12-564E, January 1983.

Patricia M. Getz is a supervisory economist in the Division of Montly Industry Employment Statistics, Bureau of Labor Statistics.
COPYRIGHT 1990 U.S. Bureau of Labor Statistics
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Technical notes; Standard Industrial Classification
Author:Getz, Patricia M.
Publication:Monthly Labor Review
Date:Nov 1, 1990
Previous Article:Contributions to savings and thrift plans.
Next Article:Quality adjustments for structural changes in the CPI housing sample.

Related Articles
Something new in SIC codes.
Developing statistics to meet society's needs.
Occupational staffing patterns at the four-digit SIC level.
Introducing the North American Industry Classification System.
The December review. (Labor Month in Review).
Implementing the North American Industry Classification System at BLS: this new classification system is a more viable way of classifying industries...
A new industrial classification system.
Transitional employment cost indexes for seasonal adjustment.
Industry net-output price indexes.
Industry net-output price indexes.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters