Where the Jobs Aren't -- New BLS Data Confirm Gloomy Jobs Picture in 2003 April 7, 2003
A new data set released yesterday by the Bureau of Labor Statistics (BLS) seems to indicate that drop in jobs in 2003 indicated by the widely cited monthly "payroll survey" was accurate. This diminishes the chance that the payroll survey figures will ultimately be revised higher, closer to the more optimistic employment outlook implied by the smaller "household survey."
The new BLS data are displayed in a publication showing trends in employment and wages in major U.S. counties between September 2002 and September 2003. It includes a county-by-county tables and a pair of summary "maps."
Economists have been puzzled by the fact that the surge in economic growth in 2003 was not accompanied by an increase in jobs. It's true that employment lags economic recovery. But in 2003, the mismatch between surging economic statistics and a stagnant labor market seemed especially bad.
Two things could have been happening:
1. Job growth was being underestimated
2. Economic growth was being overestimated
Household Survey vs. Payroll Survey. Centrists.Org has consistently noted the fact that payroll jobs figures and the (usually less reliable) household survey of employment were badly out of whack last year. We speculated that real jobs outlook in 2003 might not be as bleak as the payroll numbers said.
There were several reasons for that hope. First, the payroll survey doesn't track self-employment where no formal payroll is involved. Second, we speculated that people might be losing 2nd or 3rd jobs, which would show up as a jobs reduction in the payroll survey even though they were still employed. Third, we wondered if the payroll survey was picking up expanded employment at very small or micro-sized firms.
But so far, revisions have been small, and the new BLS data do not provide a new reason to believe that the payroll data were wrong in 2003 and will be revised higher next year.
Moreover, the more optimistic household survey has come in for criticism -- namely that its population totals are growing too fast. (This has the effect of inflating its overall employment numbers.)
So on balance, it seems like the gloomy jobs figures from the payroll survey may have the best indicator after all.
Was Economic Growth Overestimated in 2003? Another theory is that the prices we pay for imports of goods and services are falling more rapidly than the official "GDP" accounts of the U.S. have assumed. If true, that would imply that we imported more "real" stuff in 2003 than was reported, and that therefore domestic "real" production was lower than the official GDP numbers indicate.
If U.S. output was actually lower in 2003, that would help explain the weak employment situation. However, it will take time before the GDP accounts are revised to reflect any discrepancies in import prices.
One side effect of rapid estimated growth in output in 2003 was ultra-high productivity estimates. Productivity in the non-farm business sector grew by a sizzling 4.4 percent in 2003; in manufacturing it was even higher: 5.1 percent. That is twice the long-term historical rate, and twice the projection for the next decade used by the Congressional Budget Office.
If real GDP in 2003 is revised down in future years, as better estimates of import prices become available, then productivity growth will probably be revised down toward a more normal growth rate as well.
Links:
Bureau of Labor Statistics County Employment and Wages: Third Quarter 2003 (April 6, 2004)
Bureau of Labor Statistics Productivity and Costs, Fourth Quarter and Annual Average 2003, Revised (March 4, 2004)
Centrists.Org The March 2004 Jobs Report -- A Big Improvement (April 2, 2004)
Centrists.Org A Closer Look at the Latest Jobs Figures -- Plenty of Stimulus, a Scarcity of Confidence (January 11, 2004)
Centrists.Org Interpreting the Diverging Employment Statistics (September 29, 2003)
