Today the Labor Department reported employers added only 18,000 jobs last month -- a disappointing total not far from recession, in which the nation sheds jobs instead of adding them. This is only an estimate however, and in coming months the government will refine the figure and eventually pin it to reality by counting unemployment-insurance accounts. Many economists complain that the actual job-growth figures are often not very close to the initial estimates. The estimates are at special risk of being wrong at turning points in the economy -- beginnings or ends of recessions.

So I was curious about how good a job the Labor Department did at counting jobs during the last recession, in 2001. The answer is, pretty good. A comparison of the preliminary estimates and the final figures shows substantail revisions but not much change in the overall trend, which was steady job loss through the year. Conclusion: We should take the preliminary estimates quite seriously, as the stock and bond markets are today.

2001 Recession: March--November

Initial job growth estimate / Final number

January: +269,000/ -13,000
February: +135,000/ +80,000 (New York Times headline: "Job Growth Unexpectedly Strong, Easing Recession Fears")
March: -86,000/-47,000
April: -223,000/-295,000
May: -19,000/-32,000
June: -119,000/-130,000
July: - 42,000/ -117,000
August: –113,000/ -134,000
September: -199,000/ -255,000
October: -415,000/ -330,000
November: -331,000/ -308,000
December: -124,000/ -162,000