Printer Friendly

Establishment survey incorporates March 1987 employment benchmarks.

With the release of data for May 1988, the Bureau of Labor Statistics introduced its annual revision of national estimates of employment, hours, and earnings from the monthly survey of establishments. The revision uses employment counts for March 1987 as a benchmark. As part of the annual benchmarking process, the Bureau also revised seasonally adjusted series for the past 5 years, and computed new seasonal adjustment factors.

Adjustment procedure

Monthly estimates from the Current Employment Statistics (CES) survey are based on information collected from a sample of approximately 300,000 establishments. To help improve their accuracy, the Bureau adjusts CES estimates each year to new benchmarks. Benchmarks are comprehensive counts of employment based primarily on mandatory unemployment insurance reports filed by employers with the State employment security agencies.

The current revisions are based on March 1987 benchmarks and affect all unadjusted series from April 1986 forward. As is the usual practice with the introduction of new benchmarks, the Bureau has also revised the seasonally adjusted series from January 1983 forward and has published new seasonal adjustment factors to be used for the coming year.

Current revisions. The March 1987 benchmark level for total nonagricultural employment-100.4 million-was only 35,000, or 0.04 percent, below the corresponding sample-based estimate, resulting in one of the survey's smallest benchmark revisions. The pattern of revisions was mixed across industry divisions, with downward revisions in the goods-producing industries (-0.7 percent) being about offset by upward revisions in the serviceproducing industries (0.2 percent). Table 1 provides the revisions for March 1987 by industry division.

New estimates were computed for April 1987 forward, based on the new March 1987 benchmark levels and recomputed seasonal adjustment and bias factors. In addition, the sample was redistributed into estimating cells that reflect their March 1987 employment size, and sample reports were added that had been received since the original estimates were made. The combined effect of the new benchmark level, recomputed seasonal and bias factors, the resized sample, and added late reporters resulted in the new estimates generally showing larger over-the-month employment gains than previously reported. The cumulative effect on the survey estimate from April 1987 through February 1988 was the addition of 364,000 workers to the estimate of total nonagricultural employment. (See table 2.)

Sources of the differences. Differences between the benchmark totals and the sample-based estimates are caused by both sampling and nonsampling error. Sampling error occurs whenever inferences are drawn from a sample about its universe.

Nonsampling error in the CES survey has three major sources: (1) new firm bias, (2) procedures for handling changes in industrial classification, and (3) other errors of coverage, response, processing, and collection. New firm bias is attributed to the fact that the sample does not readily capture new firms, nor the employment growth from these firms. The survey's sample design also places a higher probability of selection on firms with greater employment. This too creates bias problems, because small, newer firms tend to be responsible for an important portion of the overall employment growth. Revisions to other data. Benchmarks are not available for the series on women, production or nonsupervisory workers, and hours and earnings. Women and production workers series are revised by applying the sample-derived ratio to the revised employment estimate at the basic cell level. These revisions are then summarized to 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, affecting summary level estimates of hours and earnings.

Seasonal adjustment. Each year, employment, hours, and earnings data from the new benchmark are incorporated into the calculation of new seasonal adjustment factors. These new seasonal factors are applied to the unadjusted monthly estimates to produce seasonally adjusted estimates for the period April 1988 through March 1989. The Bureau uses the x-11 ARIMA seasonal adjustment method, an adaptation of the standard ratioto-moving average method, which provides for "moving" adjustment factors to take changing seasonal patterns into account.

Publication of revised estimates. Revised estimates for all CES series appear in the June issue of Employment and Earnings, along with a more complete explanation of the benchmarking procedure and the new seasonal adjustment factors. Revisions to historical series appear in the Supplement to Employment and Earnings, published in August. Estimates reflecting the new benchmarks appear in the Current Labor Statistics section of the Monthly Labor Review beginning with the July issue.
COPYRIGHT 1988 U.S. Bureau of Labor Statistics
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1988 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Farrell, John B.
Publication:Monthly Labor Review
Date:Oct 1, 1988
Previous Article:Adjusting the CPI shelter index to compensate for effect of depreciation.
Next Article:White-collar pay in nonservice industries, March 1988.

Related Articles
Revisions of state and local area labor force statistics.
Labor force data in the next century.
Notes on Current Labor Statistics.
Notes on current labor statistics.
Notes on current Labor Statistics.
Notes on current labor statistics.
Notes on current labor statistics.
Notes on current labor statistics.
Notes on current labor statistics.
Notes on current labor statistics.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters