A funny thing was found when scouring through the provided ADP excel source data (excel link here [3]) which is used to backfill into the jobs data: nothing less than proof that ADP is goalseeking their final payrolls number.
Exhibit A:
[4]
As can be seen in screenshot of the ADP excel sheet, cell J157, which is supposed to represent the 160K service workers added in January, is in reality 159.561397994999, or somehow ADP, which is supposed to be able to poll specific workers with granularity to the individual worker level, saw the added "benefit" of 159,561 round number workers, and an additional .397994999 worker. This amounts to about one arm and half a leg.
The same can be seen in the next door neighbor cell, which indicates that 15,832 "round number" goods workers were added, and then added an additional 0.756476997 worker, or roughly two legs and one arm.
What does this mean? Simple - instead of actually using definitive numbers, ADP is goalseeking the final result based on a predetermined final outcome it wants to hit, and then varies for the growth rate or the monthly addition variable.
Our question is: why does ADP do this if it has any hopes of being credible (when we already know is merely mimics the NFP data with a one month lag), and in order to salvage some credibility, when will ADP finally start providing not seasonally adjusted data like the BLS' Nonfarm payrolls report?