Fables, Tales and Job Reports

Before too much rancor sets in among readers, I’ll admit that my content has strayed just a bit lately from health policy, etc. to politics and economics.  This too shall pass and rather quickly.  This post is for a friend and reader who e-mailed me earlier about the ADP job report and what it means for the current political debate regarding the economy.  The following is my brief answer.

For those who don’t typically follow economic data, the ADP report is a monthly barometer tied to private, non-farm, non-governmental payroll data.  ADP is the largest processor of payroll in the U.S.  Their report is the result of accumulating payroll data and arithmetically, modeling the data into employment changes (jobs added, jobs lost).  Today’s report indicates that 158,000 private, non-farm, non-governmental jobs were added in October.  On first blush, this is a plus as most economists were forecasting less than 100,000 new job adds in the month.

Politically spun, this a plus for Obama and while not a total downer for Romney, a shot across the bow.  The report rallied the stock market as expected.  Coming less than a week before election Tuesday, the report will either gain momentum based on tomorrow’s BLS report (federal job and employment data from the Bureau of Labor Statistics) or turn idle if the math doesn’t jive.  I suspect a high degree of alignment.

To the title of the post and the reply to my friend and reader: The accuracy of the ADP number and the BLS numbers is highly suspect.  While their respective releases make prominent news their corrections don’t.  Consistently and over time, the corrections are where the real story is.  Before anyone, including my friend, “jumps the shark” (a reference to Happy Days and Fonzie) and takes today’s report and tomorrow’s report as indicative of anything, let alone a sign to vote one direction of the other, consider the following.

• Even at 158,000 new jobs for October, the ADP report if accurate only indicates very slow growth.  Job losses for the month were still above 350,000.  Push and pull at the two with fifth grade math skills and a bit of common sense, this is not a sign of robust growth or even a foot-hold on longer term recovery.
• The ADP report does not cover the “core” of relative job data.  For example, we don’t know “what type of jobs” (part-time, full-time, permanent, etc) or at what rate of pay.  As is typical at this time of the year, seasonal retail is bulking-up and part-time, low to moderate wage jobs are added.  These are not permanent jobs with benefits or for that matter, “game changers” for recovery.  Similar to the last BLS report that dropped the unemployment rate, free-lance, part-time, ad hoc and so forth can be counted a variety of ways and reported as employment or jobs.
• ADP has recently changed its calculation methodology to “more accurately reflect” real time changes in employment.  Important to note is that ADP’s data is proprietary and only results are shared.  A quick glimpse difference in this report is a rather large shift to job growth among large businesses.  While I won’t state openly that this is troubling from a validity standard, it is outright curious as to this point and through recent periods, large business job growth has been “zip”.  Also somewhat curious to me is the strong results in construction job growth against a decrease in manufacturing.  I buy the manufacturing but I question a 23,000 jump in construction if for no other reason than I’d like to see the type of job, especially at this time of year.  True, new housing construction is up but commercial is flat and government spending for infrastructure is at low tide.
• Finally, ADP like the BLS data is consistently “wrong” and not just by a little.  Post period revisions are common and rarely, especially of late, are the revisions “up”.  For example, the BLS data and the ADP data are effectively the same in their raw state yet the difference between the two over recent periods (last three years) annualized to 400,000 jobs; ADP overstatement.  The ADP methodology revision I referred to is supposed to correct this gap but as it is new (first month), only time will tell.  I am skeptical at best.  Under the old estimating method, September’s report was 162,000 new jobs later revised to only 88,000.

Economic data like jobs reports, etc. point-in-time or snapshot reminds me of a phrase used by former British Prime Minister Benjamin Disraeli: “There are three types of lies – lies, damn lies and statistics”.  For any of this data to truly become meaningful from a complete economic perspective, it must be consistent over time.  Jobs are only a fraction of the issue with the greater weight of type of job, wage, benefits, sector, etc. all required additions.  Similarly, new jobs as a sole measure must balance out organic labor growth (new workers), existing unemployment levels at the U6 level (the total number of people unemployed and underemployed including those who have given up looking for a job which today remains precariously close to 15%), and rolling job losses.  At 158,000, if accurate, this is approximately a net “gain” of 58,000 jobs as by consensus measures, 100,000 new workers enter the economy monthly.  The truest measures are wages/income and percent of total population capable and willing to work, working/employed on a consistent basis.  Don’t look for this economic measurement to be truly positive and reflective of a go forward change in momentum prior to next Tuesday or for that matter, any time in the near future.