Some people are wondering why there are two such divergent numbers on employment. The reason is that there are actually two job surveys. One is based on asking establishments how many people are on their payroll, which initially covers roughly a third of all payroll employment. The second is based on asking households how many people in their family are working; the sample covers less than 1% of the population. Normal statistical variation guarantees that the two typically produce different results, though this month's difference was larger than usual. In addition, the two define employment differently. If somebody works two jobs, he will be counted twice by the payroll survey but just once by the household survey. The household survey counts domestic workers and the self-employed; the establishment survey does not. When the two diverge, you should put more weight on the establishment survey because of its larger sample size. As I note above, the big gain in the household survey this month in part is payback for unusual declines in the prior two months. That said, variance between the two in the last year has been small. The BLS estimates that in the last 12 months, payroll employment is up 1.8m while household employment is up 2.6m. Most of that gap, roughly 700,000, is due to definitional differences. If you adjust household employment to follow the payroll definition of a job, it's only up 1.9m. So the gap between the two is a relatively slim 95,000 over that period.