Wall Street is revolting against ADP’s national employment report, the report whose data is supposed to be able to predict the first read of the BLS’ monthly payrolls survey.
The charge was led by Pantheon Macro’s Ian Shepherdson who said the accuracy of ADP payrolls, the means for measuring which was revised in October 2012, in predicting the BLS’ payroll survey had not improved.
Instead, it seems to be correlating with lagging BLS data, he said. Shepherdson surmised this was because ADP was placing undue emphasis on the BLS’ revised payroll from recent months.
But in an interview with BI, Sophia Koropeckyj, a managing director at Moody’s, which collaborates with ADP on the report, said that was not true.
The three main ingredients that figure into ADP’s survey, she said, were ADP figures, BLS data, and a business cycle indicator from the Philadelphia Fed. And the ADP’s survey receives the most weight.
“It is not dominated by the lagged, previous months’ data,” she told BI by phone.
ThomsonReuters economist Jeoff Hall told BI that while the October 2012 changes had improved the survey’s accuracy somewhat, the absolute numbers remain disappointingly off. Here was the chart he came up with:
Koropeckyj acknowledged that severe weather had knocked the recent accuracy of the report off. Prior to this period, she said, the survey was off by an average of just
30,000 payrolls every month.
Hall also pointed out that the correlation on ADP’s payroll measures for specific job sectors were even worse than the headline figure. Koropeckyj disputed Hall’s assertion, saying that only the correlation in professional business services had deteriorated.
However, she declined to provide data.
Today’s ADP came in at 191,000; Bloomberg consensus expectations for BLS’ figure for the same period is 200,000. We’ll see how this all shakes out Friday.