ADP has announced revised methodology to "enhance" its monthly job reports, no doubt because its prior numbers simply were grossly inaccurate.
Indeed, I stopped commenting on ADP numbers because I thought they were absurd.
Let's take a look at their revised methodology. Here is the ADP Jobs Report for September using the revised methodology.
Please note that last sentence. Compare to what ADP actually reported last month:"Private sector employment increased by 88,200 jobs from August to September, according to the September ADP National Employment Report®. The report, which is derived from ADP's actual payroll data, measures the change in total U.S. nonfarm private employment each month on a seasonally adjusted basis. Last month's employment estimate was revised down from 80,000 to 76,400 jobs."
"Employment in the U.S. nonfarm private business sector increased by 162,000 from August to September, on a seasonally adjusted basis. The estimated gains in previous months were revised lower: The July increase was reduced by 17,000 to an increase of 156,000, while the August increase was reduced by 12,000 to an increase of 189,000."
Got that? September private business sector increased by 162,000 but now we see September was revised down to 76,400 from 80,000 (not 162,000 as actually reported).
Revision Matching
Check out this statement from ADP FAQs.
Using this methodology developed for ADP by Moody's Analytics, our adjusted historical ADP National Employment Report data dating back to 2001 has a 96 percent correlation with the revised BLS numbers.
ADP revisions match BLS revisions over time. Lovely.
Spotlight on Revisions
ZeroHedge totaled up the ADP revisions for 2012 and concluded ADP "Cancels" 365,000 Private Jobs Created In 2012.
I conclude the same thing. However, things are even worse than Zerohedge states. Here is a chart that I put together of ADP revisions.
Note: numbers in charts and tables in thousands.
ADP Original Vs. ADP Revised Monthly Job Gains or Losses
ADP was way underestimating job gains in 2011 and way over-estimated gains in 2012.
The net effect was 136,000 jobs over 20 months (about 6,800 per month). However that is a very misleading way of looking at things as the following table shows.
Time Period Analysis
| Time Period | Cumulative Miss |
| 2011 Miss | 229.0 |
| 2012 Miss | -365.1 |
| Last 12 Months | -419.1 |
| Since Feb 2011 | -136.1 |
As you can see, I match ZeroHedge for 2012. However, for the last year, ADP was off by an even higher 419,000 jobs, nearly 35,000 jobs a month for an entire year!
For 2011, ADP was off in the other direction by 229,000 jobs, roughly 19,000 per month. Here is the complete table that I worked from.
ADP Data Points Table
| Date | ADP Revised | ADP Original | Revision | Cumulative Error |
| 2011-02 | 155.3 | 197 | -41.7 | -41.7 |
| 2011-03 | 215.9 | 199 | 16.9 | -24.8 |
| 2011-04 | 174.5 | 162 | 12.5 | -12.3 |
| 2011-05 | 156.4 | 47 | 109.4 | 97.1 |
| 2011-06 | 118.3 | 136 | -17.7 | 79.5 |
| 2011-07 | 180.3 | 125 | 55.3 | 134.8 |
| 2011-08 | 122.7 | 67 | 55.7 | 190.5 |
| 2011-09 | 197.5 | 105 | 92.5 | 283.0 |
| 2011-10 | 114.3 | 142 | -27.7 | 255.3 |
| 2011-11 | 173.2 | 226 | -52.8 | 202.5 |
| 2011-12 | 293.4 | 267 | 26.4 | 229.0 |
| 2012-01 | 218.9 | 182 | 36.9 | 265.8 |
| 2012-02 | 226.6 | 228 | -1.4 | 264.4 |
| 2012-03 | 89.7 | 204 | -114.3 | 150.1 |
| 2012-04 | 130.1 | 112 | 18.1 | 168.2 |
| 2012-05 | 81.2 | 131 | -49.8 | 118.4 |
| 2012-06 | 115.4 | 173 | -57.6 | 60.8 |
| 2012-07 | 145.5 | 156 | -10.5 | 50.4 |
| 2012-08 | 76.4 | 189 | -112.6 | -62.3 |
| 2012-09 | 88.2 | 162 | -73.8 | -136.1 |
By the way, these ADP revisions suggest the stated unemployment rate is blatantly preposterous, something I say in every jobs report.
Read more at http://globaleconomicanalysis.blogspot.ca/2012/10/adp-grossly-overstates-job-growth-for.html#dbQ45juVK7x07rix.99
