The statistics corner: update on the BLS sample redesign for the current employment statistics survey - Bureau of Labor Statistics

Business Economics, July, 1997 by Patricia M. Getz

In June 1995, the Bureau of Labor Statistics (BLS) announced plans for a comprehensive sample redesign of its monthly payroll survey. The Bureau's plans called for a two-year research effort to develop the new sample design, followed by a production test of survey methods and procedures, with a phased-in implementation of the new design following thereafter. As scheduled, the two-year research phase for the Current Employment Statistics (CES) sample redesign is nearing completion, and the Bureau will launch a production test of the new sample design in July of this year. The Bureau's initial plan called for the production test to be run in five states for a one-year period and be followed by a multiyear phasein of the new design. BLS has expanded this initial plan now to include all states in the production test phase, and also to provide an overlap period of parallel estimates from the new design, allowing for a comprehensive evaluation of the performance of the new sample methods, systems, and procedures before they become operational for the production of published employment, hours, and earnings series.

The CES redesign program will be conducted as a parallel operation to the ongoing CES program with a multiyear phasein of the new design by major industry division. There will be no change in the methodology for the published employment, hours, and earnings series until June 1999, when the first estimates from the new design will be introduced with the 1998 benchmark revisions, for the wholesale trade industry. The remaining industry divisions will be phased in with subsequent years' benchmark releases.

This paper describes the major features of the new sample design and expected improvements from it, as well as plans for the production test and implementation. The research and development work described below has been conducted under the auspices of a joint BLS/state workgroup and in consultation with experts in survey design under contract to BLS from Westat Inc., the National Opinion Research Center at the University of Chicago, and the University of Michigan Survey Research Center.

BACKGROUND

The CES is a federal/state cooperative program that provides monthly estimates of nonfarm payroll jobs and the hours and earnings of workers derived from a sample of nearly 400,000 business establishments nationwide. These data are some of the most closely watched and widely used economic indicators among public and private policy makers alike. The CES offers several advantages to its users: timely release of data, an abundance of industry and geographic detail, and an annual benchmark to full population counts from state Unemployment Insurance (UI) tax records, which helps to maintain overall survey accuracy.

CURRENT DESIGN LIMITATIONS

Two limitations of the CES survey now hamper its ability to reflect fully current monthly employment trends: the lack of a probability-based sample design, and the absence of a method for directly measuring employment from new business births. Both of these limitations are now addressed in the CES survey indirectly, through use of a time-series modeling technique known as bias adjustment.

These limitations affect not only national but state and metropolitan area series as well and contribute to a recurring problem of differing employment trends reported for the national versus the sum-of-the-states CES estimates.

The existing CES is a quota sample whose inception more than fifty 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 will more effectively insure a proper representation of the universe of nonfarm business establishments, through randomized selection techniques and the regular rotation of sample members. It also will allow for the publication of sampling errors and confidence intervals - standard survey accuracy measures not directly applicable to the current nonprobability design.

Business Births

New business births occur every month and impact the accurate measurement of monthly employment change in two ways:

1. New businesses generally start up and contribute to the overall monthly employment levels prior to the time they first appear on the UI universe frame and are available for sampling; this makes measurement other than by a model-based approach very difficult.

2. New business births tend to have rapid employment growth during their initial years in business; this necessitates a sample design approach that provides for the introduction of these units into the design as soon as they first appear on the frame.

During the past two years, BLS has studied sample-based approaches for each of these measurement issues. To address the issue of start-up employment levels, BLS has conducted research into the feasibility of obtaining earlier access to the state UI files to explore the feasibility of monthly sample-based estimation for new business birth employment. In addition, the Bureau has done extensive profiling of business birth and death patterns and of prototype testing of model-based procedures for estimating these components, as an alternative to sample-based estimation. Several different methods for estimating business birth and death employment will continue to be explored during the production test phase of the CES redesign.