The ADP's monthly National Employment Report is published two days before the official BLS nonfarm payrolls report.
To many, the ADP report has been considered the best preview of private payroll stats from the BLS report.
“We’re going to be very accurate,” said Moody's Analytics' Mark Zandi at the time ADP rolled out its revised methodology months ago. "We’re going to nail this number,"
That doesn't seem to have happened, and economists are now protesting.
In a note Wednesday morning, Pantheon Macroeconomics' Ian Shepherdson explained the problem: ADP reports its figure not just from its own payrolls data, but also incorporates lagging data from the BLS itself.
And it appears they've now placed far too much emphasis on that lagging data, since they now show a strong correlation with the BLS' number from the prior month.
"This glaring flaw in the construction of the ADP report means we can no longer take the number seriously as an advance guide to the official data," he wrote.
He added that ADP "will not reveal its raw data, presumably because they do not correlate well with the official numbers."
ADP did not immediately respond to requests for comment. UPDATE: Moody's, the company that puts together the payrolls report with ADP, has responded, denying Shepherdson's assertion »
Bank of Tokyo-Mitsbushi's Chris Rupkey had an equally strident take on the issue today. In a note titled "ADP jobs data this morning...and the number is...anyone out there?" he wrote:
A lot of swings and misses for ADP when it comes to forecasting the monthly payroll jobs report. They thought jobs slowed further to 130K in October and instead jobs went sky-high to 212K. Actual jobs cooled to 87K in December after two months of big 200K gains yet they told us the number would be the biggest yet, up 238K. Wrong. Another swing, another miss. Why do we continue to take a look at this?"
Related story
Thomson Reuters managing economist Jeoff Hall tweeted the following chart showing that the average absolute miss between ADP and BLS has declined 33% since the change. But the absolute numbers remain far off, he said.
"For an indicator that's supposed to nail the number, I think they've been horrible at it," Hall told BI.
"When you get into the details, it’s even scarier," he added.
The revisions have come at the expense of accuracy in predicting industry-specific payrolls. Hall sent over the following table showing correlations have fallen off a cliff:
Finally, Hall noted just half the primary dealers from Reuters' economist poll now put out ADP previews. And three of those that do are simply restating their guess for BLS, he said.
However, not everyone agrees that we should just scrap the ADP report.
Mac Robertson, an independent portfolio manager and macro strategist, believes the ADP benefits from its immense direct access to corporate payroll data.
"I actually think ADP does a great job, and I with there were a lot more private sector economic reporting like this," he told BI.
He added that the focus on payrolls at all, especially in an expanding economy, is misplaced, and ends up being a lagging indicator.
"There's so much creative destruction, substitutions, and reclaiming that's going on," he said. "There's just massive cross-currents in employment."
UPDATE: Moody's, the company that puts together the payrolls report with ADP, has responded, denying Shepherdson's assertion »
