(February 1, 2008) – The Department of Labor released their January employment report, The Employment Situation: January 2008, this morning, reporting that nonfarm payroll employment and the unemployment rate were essentially unchanged in January at 138.1 million nonfarm payrolls–17,000 fewer than in December–and only a 0.1% decrease in the unemployment rate to 4.9%. The change in nonfarm payroll employment (-17,000) reflected declines in construction and manufacturing and job growth in health care. The number of unemployed persons (7.6 million) and the unemployment rate (4.9%) were essentially unchanged in January; average hourly earnings rose by 4 cents, or 0.2%.
While we should always be cautious about small changes in monthly data–the 138.1 million total nonfarm employment number reported this month is probably only accurate to within ±200,000 jobs–this report is especially confusing. The headline number, -17,000 jobs for the month, indicates a slowing economy with 51,000 job losses in goods industries not quite offset by 34,000 new jobs in the service sector. And on the surface, the household data show a 42,000 decrease in jobs from 153,866,000 in December to 153,824,000 in January. Both numbers are in sharp contrast to Wednesday’s ADP National Employment Report, which reported a 130,000 job increase in January, made up of -11,000 in the goods producing sector and +141,000 in the service sector.
The answer lies in the footnotes. On page 5, the BLS report describes significant revisions in the January establishment data which lie behind the -17,000 figure. This was an especially meaningful revision to reflect new payroll job benchmarks. In addition, establishment survey data were updated to the new March 2007 North American Industry Classifications System (NAICS) from the 2002 NAICS basis. They also revised post-January 1990 data using new seasonal adjustment factors. (Confused yet?) On net, these changes revised the March 2007 total nonfarm employment figure downward by 293,000 (284,000 seasonally adjusted). Using these revised figures, December employment increased by 82,000, more than four times the previously reported +18,000.
The BLS also revised the population estimates used to produce the numbers from the Household Survey. These revisions reduced estimated population in December by 745,000; the estimated labor force by 637,000; employment by 598,000; and unemployed by 40,000. They did not, however, revise the 2007 monthly numbers to reflect these changes; i.e. the December 2007/January 2008 comparison (-42,000) in the report has no meaning whatsoever. In a footnote on page 7, the BLS calculated the effect of these revisions on the monthly comparisons, which suggest that January employment actually increased by an incredible 635,000 jobs.
Although the BLS report receives most of the attention in the press, it is based on limited survey data. Household data are collected from a survey of 60,000 households; payroll data are collected from payroll records for 160,000 businesses and government agencies covering 400,000 individual work sites. The ADP estimates reported Wednesday are drawn from actual payroll data from 500,000 payrolls covering over 24 million workers.
Bottom line: the Labor Department has absolutely no idea what happened in the labor market in January. My sense is that total employment is increasing at or below the roughly 100,000 job per month increase we need to absorb increases in the labor force. This is consistent with GDP growth of about 2%.