downThe Bureau of Labor Statistics’ survey of households has for the last year shown far weaker job growth than its larger and more closely followed survey of payrolls, even when the two define jobs the same way.

In December, nonfarm payrolls rose 18,000 while household employment plunged 436,000. But such monthly changes are hard to compare because the two surveys define employment differently. For example, the household survey counts the self-employed, while the payroll survey doesn’t. The payroll survey counts someone with two jobs twice, while the household survey counts him once. Moreover, the household sample is far smaller and thus more volatile. By design, its raw data is never revised, which imparts a false sense of reliability.

Still, once those adjustments are made, the picture remains the same. Household employment has risen just 100,000 since December 2006 and when the definition of employment is changed to match that of the payroll survey, the increase is just 375,000, according to the BLS. The increase in nonfarm payrolls was 1,270,000 in the same period. Some of that latter increase will be trimmed during the BLS’ annual benchmark revisions to be released next month. It has previously estimated payrolls would be revised down for the 12 months through last March by 297,000, which, assuming an average monthly reduction of 25,000, would reduce the 12-month increase through December by just 75,000. But the downward revisions could be larger given the arrival of more comprehensive quarterly state unemployment records for subsequent periods. –Greg Ip

---