THE NOVEMBER 2019 JOBS REPORT: SEASONALLY ADJUSTED VS. NOT SEASONALLY ADJUSTED DATA AND EMPLOYER VS. HOUSEHOLD DATA

Yesterday (Friday, December 6, 2019), much attention was given to the estimated increase in U.S. nonfarm payroll employment.  Of that attention, people emphasized the increase in employment based on employer surveys, often called establishment surveys, and with seasonally adjusted data.  The announced figure was a gain of 266,000 jobs, again on a seasonally adjusted basis.  Readers can find the employment news release from the U.S. Bureau of Labor Statistics (BLS) at the link below.

https://www.bls.gov/news.release/empsit.nr0.htm

This estimate clearly surpassed most if not all forecasts.  Do other estimates echo the findings of the seasonally adjusted employer survey data?

For those not familiar, economic activity often if not usually or always is greater at some points in time rather than others.  For example, U.S. retail sales would be expected to be greater in November and December of a year compared with most if not all other months due to holiday shopping.  Seasonal adjustments can be made to reported data so that people can make comparisons that in at least some cases are much more meaningful in terms of the relative performance of the economy compared with data from the previous month or the previous quarter.  To give another example, if retail sales and other economic activities usually decrease in the first quarter of the new calendar year after holiday shopping, we do not necessarily want real GDP to decrease in that quarter compared with its level from the fourth quarter of the previous year if the decrease in retail sales and other economic activity in the first quarter is at a level consistent with economic expansion (for example, over the first quarter of the previous year).  Applying a seasonal adjustment to the data can help to make better judgments about the relative health of the economy.

Further, both employers and households complete employment surveys.  Household surveys are described as being more volatile.  Nevertheless, to the extent that the household surveys are valid, they paint a different employment picture for November 2019 in the U.S. 

Returning to the employer or the establishment survey, based on my calculations, data that are not seasonally adjusted indicate that employers added about 622,000 new jobs!  That figure is much larger than the seasonally adjusted estimate.  It probably should not be a surprise that employment increases near the holiday shopping season.  The U.S. BLS establishment data not seasonally adjusted (and seasonally adjusted) are available at the link below.

https://www.bls.gov/news.release/empsit.t17.htm .

However, on a seasonally adjusted basis, Table A-1 from the U.S. Bureau of Labor Statistics available at https://www.bls.gov/news.release/empsit.t01.htm shows that from the household survey, the number of people employed increased between October 2019 and November 2019, but only by about 83,000, based on my calculations.  That figure is rather far below the seasonally adjusted estimate of a gain of 266,000 jobs reported based on the employer surveys.  Moreover, I calculate from the data in Table A-1 that the number of people employed in November 2019 based on data not seasonally adjusted actually decreased by more than 120,000!  Whether seasonally adjusted or not, do these figures from the household survey better match the ADP estimate of 67,000 private sector jobs added in the U.S. in November 2019, as noted by The Atlanta Journal-Constitution (December 5, 2019, p. A9, no author listed, “November survey:  Just 67K jobs added”) and others?  Also, does this better reflect the slight decrease in the labor force participation rate reported in Table A-1 in November 2019 regardless of whether using seasonally adjusted data or data not seasonally adjusted.  Further, the release from the U.S. BLS points out that (1) from the household survey, the number of people not in the labor force increased by more than 130,000 on a seasonally adjusted basis and by more than 350,000 on a basis not seasonally adjusted in November 2019, and (2) the average monthly job gains in 2019 have been less than in the year 2018. 

(Note that data revisions that may follow could change the calculations in this blog post, and that most if not all figures are estimates.  Also, note that clicking the web links roughly one month or more from this posting could possibly result in finding a data release for December 2019 or a month in the year 2020 or later.)

Do these findings support the hypothesis that part of the surge in jobs reported from the employer surveys is due to people working multiple jobs, as I asked in my previous blog entry available at the link below?  

https://harrisonhartman.blogspot.com/2019/12/some-questions-about-job-growth-and.html

If more than 620,000 jobs were added using data not seasonally adjusted and more than 250,000 new jobs were added after seasonal adjustments in November 2019 based on employer surveys but the change in the number of those employed based on the household surveys was either an increase of approximately 83,000 using seasonally adjusted data or a decrease of more than 120,000 using data not seasonally adjusted, then does this mean that at least some of the people accepting at least one of those newly created jobs already had a job?  And if more people are working multiple jobs, then what does that suggest about the ability of the economy to generate jobs that pay enough to earn a living?

Whether the job news from yesterday is unexpectedly bad, unexpectedly good, or consistent with expectations, will future releases continue that relationship?  Further, will the observation of Steve Liesman on CNBC yesterday that he attributed to Mark Zandi be maintained about the number of jobs created and the number of people filling jobs?  Regarding both of those questions, we’ll need to wait and see.