So, is this better accountability or more political fog? - KWC

 

U.S. Job Statistics Revamped
New Methods Adjust, Add Industry Categories

By John M. Berry, Washington Post Staff Writer, Thursday, June 5, 2003; Page E01

Normally, investors, analysts and government policymakers all would be scrutinizing the May payroll employment figures due out Friday for clues as to where the sluggish American economy is headed. This week, they may be struggling more to understand the numbers themselves.

The Bureau of Labor Statistics, which produces the payroll figures, will make several major revisions this week in the methods it uses to calculate the number of jobs on U.S. payrolls from a monthly survey of more than 300,000 business establishments. Some of the changes will make it particularly difficult to compare last month's figures with those analysts have followed in recent months, all of which will be revised in a number of ways.

"This will not be a run-of-the-mill employment release," said economist Ray Stone of Stone & McCarthy Research Associates, a financial markets research firm. "In fact, the scope of these definitional and methodological revisions is the most profound of the entire history of the establishment survey."

None of the changes will affect the monthly survey of about 60,000 households on which the national unemployment rate is based.  Over the past year or so, however, the results of the two surveys have diverged. The payroll survey, for example, has shown the number of jobs declining while the household survey has shown a large gain in jobs -- even after taking some updated population figures into account.

One of the bureau's changes in the payroll survey, a "benchmark" revision, occurs each year at this time. Using information from federal unemployment insurance returns that are filed by almost all firms, the bureau annually updates the data from its sample surveys. The agency announced earlier that its estimate of payroll employment for March 2002, which was 130.7 million jobs, would be lowered by 313,000. But because of some of the other changes, it is not clear how much payrolls for recent months will be reduced.

The most sweeping change involves an overhaul of the industry categories for which payroll figures are published, with roughly 25 more industries added to the list. For instance, employees of The Washington Post, heretofore a manufacturing firm, will be classified in a new publishing industries category, which in turn is part of a new information "supersector."

At the same time, the headquarters staff at General Motors Corp., also formerly classified as manufacturing workers, will be shifted to a management listing under professional and business services. While a few additional workers will be moved into the manufacturing category, many more will be listed elsewhere, likely reducing that industry's payroll count noticeably, analysts said

The new industry listings are part of the North American Industry Classification System, or NAICS, which has been designed to more clearly reflect the nature of today's economy than did the Standard Industrial Classification system, in place since 1987. Several other federal statistics, such as the Commerce Department's monthly retail sales figures, have already been converted to the new system.

With the May figures, the bureau will also complete the conversion to a new method of picking the firms that are included in the monthly payroll survey. According to the agency, the new sample is "more representative" than the old one and therefore statistically more accurate. Moreover, the new approach allows direct estimation of how many firms go out of business and how many are created. With that data, a less exact procedure called a "bias adjustment" will be dropped.

Yet another change to be introduced with this week's number is a different seasonal adjustment procedure. Until now, the bureau has adjusted the raw numbers for recurring seasonal patterns going back for 10 years and updated the adjustment factors twice a year. The 10-year period will still be used, but the adjustment factors will incorporate the latest month's data, so they are updated constantly.

http://www.washingtonpost.com/wp-dyn/articles/A15094-2003Jun4.html?nav=hptoc_b

 

 

Reply via email to