How Good was the Sep NFP Release

On the one hand July was revised up 25K, August up 93K and Sept came in at 110K for a total of 228K more jobs than we would have thought the day before. On the other hand the March 2007 Benchmark revision slashed 257K jobs - leaving the overall level of jobs 17K less than we would have thought.

On net it is roughly a wash.

However, I would suggest that the sizable downward benchmark revision is not the best sign. The BLS methodology produces considerable inertia in the estimates. The birth death model creates or destroys jobs based on how well the monthly data performed in relation the benchmark last time around.

The upshot is that the NFP report will tend to miss turning points. Indeed, if the turning points are close together in time the the methodology should significatly amplify the peaks and troughs.

If the benchmark March 2007 was already revising jobs downward then that implies to me that the turning point was likely sometime before that. Thus all of our estimates until the new benchmarking comes into effect will be overshoots.

Moreover, there is reason to think the economy has weakened since
March 2007 and that our overshoots are actually accelerating. It will be interesting to see how the unemployment rate fairs by the beginning of 2008.