by Calculated Risk on 1/08/2007 12:20:00 PM

The ADP report has taken quite a beating over the last few days. David Gaffen wrote at the WSJ blog on Friday: ADP in ICU, part II

... anyone who traded on the ADP numbers ahead of today’s report lost money. “There isn’t any hiding from that,” said ... Joel Prakken chairman of Macroeconomic Advisers, the economic consulting firm that partners with Automatic Data Processing to produce the private sector monthly jobs report.

And Professor Duy comments on ADP in another great Fed Watch (Recommended reading!): Like a Broken Record

"I do not pay special attention to the ADP report. I simply don’t trust that it provides any useful information above that offered by a host of Wall Street economists. Moreover, when it misses, it misses big."

On a slightly different note, Professor Hamilton discusses why he blends the ADP report with the BLS reports.

As I noted last week, I just ignore the ADP report. But I thought I'd take another look.

First, it's important to note that the ADP report is for private sector jobs only, and the headline BLS number includes government jobs. So, to compare the ADP report to the BLS payroll report, government jobs should be excluded from the BLS report.

Click on graph for larger image.

This graph shows total private sector BLS payroll jobs vs. ADP jobs. Special thanks to Paul Kasriel of Northern Trust.

Note that the scale starts at 105 million to show the differences between the reports.

In recent months, the ADP report showed a total of almost 500K more jobs than the BLS report. After the weak December ADP report (showing a monthly decline of 40K jobs) and the strong BLS report (an increase of 150K private sector jobs, 167K including government jobs), the ADP report is showing about 300K more jobs than the BLS report.

This difference led me (and others) to discount the ADP report.

But maybe the ADP report has some merit. In October the BLS announced the preliminary estimate for the 2006 revision to the establishment survey - and the revision appears to be a "larger-than-normal" increase in jobs - perhaps an upwards revision of 800K jobs. So it is very possible, once the actual BLS revision is announced in February that the ADP report was actually pretty close earlier this year.

For now it's probably best to continue to ignore the ADP report (or use Hamilton's 10% suggestion), but I'll take another look at the January numbers.