Reclaiming Common Sense

A very solid ADP payroll number was released on Wednesday March 8, 2017 in advance of the March 10, 2017 February Jobs Report.Type your paragraph here. They reported a surge of 298,000 seasonally adjusted private sector jobs. This seasonally adjusted private sector number is the umber that former President Obama used to tout. This data was after the January ADP number was revised up from 246,000 to 261,000. This upward adjustment means that based on the original January ADP Number that we added 313,000 private sector workers. Can you imagine the headlines if this is reflected in the Jobs Report this Friday? The people at ADP have a disclaimer that, in part, states " No commercial reproduction, use or re-use of any kind may be made of the Information without the prior written consent of ADP, which may be withheld in its sole and absolute discretion in each instance. " This leaves this column with the opportunity to discuss the situations that lead this author to be skeptical regarding the ADP data.


The Data Quoted is Seasonally Adjusted. The data is seasonally adjusted to avoid the "panic" that would be created when the January Jobs Report reveals, on an annual basis, that the workforce contracts by roughly 2%. This column projects a range of Current Employment Statistics (CES) worker growth and Current Population Survey (CPS) job growth because there are no lists of published seasonal factors for the data. How does ADP know what the seasonal factors will be for the overall number, nonetheless the individual categories?


The Data Quoted closely resembles the Seasonally Adjusted (SA) Current Employment Statistics data. Reviewing the data provided on the ADP website reveals some differences in the data.

  • The February 2008 ADP data was  Lower than the SA CES data.
  • The February data for 2010 was higher than the SA CES data.
  • The February 2011 and 2014 ADP data was reported higher than the SA CES data
  • The February 2009, 2013, 2015, and 2016 data were reported lower than the SA CES data.
  • The May 2016 data in the current ADP report reveals a Worker creation of 53,000 when the June Jobs Report revealed a loss of 1,000 seasonally adjusted jobs.


The Sample sizes are different.The ADP data is obtained from 411,000 companies. The CES data is obtained from a survey of 147,000 companies.The CPS Data is obtained from a survey of 60,000 Individuals. The CES data is often referred to as the Establishment (ES - CES) data and the CPS data is often referred to as the Household (Person - CPS) data.  You would think that the ADP data would be more accurate due to the larger sample size. The question of accuracy or precision is in the seasonal factors. Accuracy is hitting the target, Precision is hitting the same spot consistently. The data, to this researcher, appears accurate, not precise.


The CPS data measures Jobs. The CES data measures workers. Workers can work multiple jobs. The number of workers in the workforce (SA CES) were 121.483 million during February 2016. The ADP number was reported lower. Close enough. The SA CPS data was reported at 151.059 million. We do not have 30 million workers working multiple jobs. The highest we have had is more in the neighborhood of 8 million workers.


We lose workers on a regular basis. The Red NSA CES histogram details why there is a seasonally adjusted component that is reported on a regular basis. The peak pre-recession job market was July 2007. If the NSA and SA CES data is aligned with the peak jobs market the histograms revealed that we lost 11.4 million SA CES workers between July 2007 and January 2010. The blue SA CES histogram reveals that we lost 8.6 million workers. Do we have 123.3 million SA CES workers or 121.1 million NSA CES workers. We have both. Where do these 2.2 million people live?


How will the January data be revised? If the January SA CES data is revised upward, as was the ADP data, then that reduces the bump from the advance January value. If we add 300,000 jobs the SA would be 123.578. If they say we added 50,000 more workers than originally reported, 123.328 then that 300,000 workers is reduced to 250,000.


Sometimes the ADP number comes in higher than the CES data. Sometimes the ADP data comes in lower than the SA CES number. The seasonal factors used to convert the NSA CES data to the SA CES data has been skewed for month, if not years. The ADP number is a best guess. The number promoted in this column is a best guess.  Neither one of us has the exact seasonal factor used by the government. The good news is that we should have grown the economy during February. How much we grew will be revealed this Friday.


It's the economy.