The February jobs report was reasonably good. It wasn't great; we created an average of 250,000 jobs a month for four years at the end of the 90s. Coming off a severe downturn we should be seeing jobs growth of 400,000 a month, as we did following the 81-82 recession and the 74-75 recession, but 227,000 jobs is definitely an improvement over what we had been seeing.

But the NYT got a bit carried away. It told readers:

"Household survey respondents indicated that 879,000 more people were working in February than in January. Though it is not unusual for the two surveys to differ, it is unusual for the growth in the household survey to be so much greater."

No, that isn't quite right. The survey showed a rise in employment of 428,000 jobs. That's good, but not 879,000.



The NYT is in fact right on this. The article was referring to a series that BLS constructs that adjusts the household survey for differences in concept with the establishment survey. This measure excludes self-employed workers and adjusts for workers with multiple jobs. This series did in fact show a gain in employment of 879,000 in February.

The biggest reason for the difference between the two series was a sharp drop in the number of workers reported as self-employed. This reduced the gain in employment shown in the published data, however it would not affect the establishment series.

Thanks Zee for calling this to my attention.