Posted - Filed under Payroll Survey, Release Observations.

Lower than the trend and expectations.  Also, a very large revision to last month’s number for July growth. More on those revisions below.

Seasonal factors were strained by a couple of highly seasonal transitions: Local government education saw a seasonally adjusted gain of 21,000 (enough to increase total government jobs in August) of which more than 10,000 were probably an adjustment error.  On the other side, the opposite seems to have happened in the movie industry, where 22,200 jobs (almost 6% of the total in that industry) were calculated on a seasonally adjusted basis to have been lost.  Those will come back next month.

The Trend


Click to enlarge

Much is made of each month’s growth number, but there is typically a lot of random fluctuation from month to month due to reporting and seasonal adjustment issues – and of course, real underlying changes in the jobs growth.  (The impact of those first two on the recent  are pointed to below.)

In the chart here we plot the rolling 3-month change in Private jobs,  in attempt to filter out some of the month-to-month noise.  Private job creation does seem to have been slowing down over the past few months.   Government jobs (not shown here) have continued their steady decline – notwithstanding this month’s upward blip, which is entirely due to that seasonal adjustment blip in teachers.


Revisions to Past Months

The change in total nonfarm payroll employment for June was revised
from +188,000 to +172,000, and the change for July was revised from
+162,000 to +104,000. With these revisions, employment gains in June
and July combined were 74,000 less than previously reported.  - BLS Release


Click to enlarge

The chart to the right shows not only the published BLS revisions, but their unpublished ones. Note that growth in June was not really revised by as much as the published BLS number, though April and May have also been revised downward. (The revised seasonal factors for even earlier months will include some upward revisions to balance these out over the year more or less.)

Of the hefty 58,000 downward revision to July, 24,000 was in Government (Fed, State & Local all moving down).

The remaining drop of 34,000 in Private jobs came mostly from a near 20,000 downward revision to growth in Motor Vehicles & Parts,  and a near 12,000 revision to Transit & Ground Passenger Transportation.  Both have bounced back somewhat in August, and both the Government and Transit revisions seem to have been enlarged by a seasonal factor effect, similar to those described, below, for August.


Seasonal Adjustment “Glitches”

Some employment sectors have such dramatic seasonal patterns, that a slight shift in their timing can throw off the seasonal adjustment process.


Click to enlarge

In the Motion Picture & Sound Recording industry, there has historically been a not-seasonally adjusted (i.e. real) drop in jobs reported for September of each year.  This year it looks like the drop came early enough to be counted in August.  The mathematical modeling used in the seasonal adjustment process cannot consider this, and thus has seen this as an unseasonal drop.  However, if we are correct and the regular not-adjusted September fall does not materialize (having happened in August) the chances are that the seasonal adjustment process will revise to reduce this month’s fall and show an unseasonable rise in September, in some combination such as to get the overall job level back on track.



Click to enlarge

For teachers, a slightly different glitch here, in which the seasonal adjustment process is just finding it hard to adjust for the massive swings in fall hiring.

From the plot of past adjustments, we might see a further calculated rise in teacher levels in September, before crashing back in October.