Navigating a "Wile. E. Coyote" Market

Friday's employment report, showing an increase of 18,000 in non-farm payrolls and a jump in the unemployment rate to 9.2% was widely viewed as a "shocker." Frankly, I don't understand the surprise. Between February and April, weekly new claims for unemployment (4 week average) dipped below 400,000, which was associated with a few months of nice growth in non-farm payroll employment. Since then, weekly unemployment claims have moved higher, and have been running at an average near 425,000 new claims weekly. Historically, that's a level that's fairly well correlated to roughly zero growth in non-farm payrolls. The data is certainly very choppy from month-to-month, which is enough to produce surprises around that average, but the errors also tend to mean-revert, meaning that unexpected job growth figures one month tend to be corrected in the opposite direction (where the "expectation" is taken on the basis of the weekly claims figures).