by Calculated Risk on 5/06/2009 01:32:00 PM

I added a caution with the ADP employment report this morning: "The ADP employment report hasn't been very useful in predicting the BLS numbers ..."

Several readers have sent me graphs showing that ADP and BLS employment numbers track pretty well over time. That is true - after revisions.

However I think in real time the ADP report isn't that useful for forecasting the BLS numbers (although it might offer a suggestion).

Click on graph for larger image in new window.

This graph compares the ADP and BLS employment numbers (note: non-government) as originally released since last Summer.

Here is the original releases for ADP and for the BLS.

Although the two reports generally move together, there have been a number of significant misses; as examples: September, November and December of last year.

The consensus for BLS reported job losses is 630,000 (including government, to be announced Friday), and the ADP report suggests the losses might be a little lower.

However, initial weekly unemployment claims only declined slightly in April, and for the headline number (that includes government - not included in ADP) there have been a number of reports of local government layoffs.

John A. Challenger, chief executive officer of the placement company, said [this morning] ... “state and local governments, as well as school districts, are really feeling the impact of this downturn.”

I think the job losses could be less than the average of the last 5 months (averaged 660 thousand per month), but not much less.