Rarebird: Inflation;Wage pressures>Job growth. Part I
OCTOBER 11, 1999
Counter Intuitive
Friday's labor numbers aren't what they seem
By Gene Epstein
Has the labor market gotten so tight that there's no one left to hire? Or is
there actually a lot of hiring going on that the Bureau of Labor Statistics isn't
catching in its net?
As BLS statisticians are always ready to admit, that net has a lot of potential
holes in it. So when the agency reported on Friday that nonfarm payrolls had
fallen by 8,000 jobs in September -- an estimate that runs counter to
everything else we know about the economy's current strength -- then it
becomes appropriate to poke a little fun at the way the monthly payroll figures
are cobbled together.
But first let's look at what the BLS actually said. To begin with, it attributed
the reported decline in payrolls to Hurricane Floyd, which was on the
rampage during the September survey week. With millions of people
evacuated from their homes, and with stores and restaurants shut, jobs that
would normally have existed were temporarily out of commission. So if Floyd
hadn't been a factor, said the BLS, September payrolls would have expanded
by about 50,000.
The unemployment rate, which remained steady at 4.2%, wasn't affected by
Floyd. That's because, for purposes of calculating this figure, people are
counted as employed if they're temporarily absent from their jobs because of
bad weather. Also, average hourly earnings jumped 0.5%, which makes
sense when joblessness is so low.
But even that 50,000 adjusted estimate for September is nothing to write
home about, especially since it comes on the heels of an August advance of
only 103,000. And while two months don't make a trend, there are times
when they indicate turning points. So has hiring suddenly slowed to a crawl?
Probably not.
For starters, one major anomaly in the estimates is the
persistent contraction in manufacturing employment, which flies in the face of
all the available evidence that this sector has been on the rebound. In
particular, the NAPM index of manufacturing employment has been
telegraphing job growth for the past few months. As Jason Benderly of Vail,
Colorado-based Benderly Economics points out, the last time the NAPM
employment index was this strong was in 1994, when factory jobs grew by an
average of 50,000 per month.
Then, too, the low level of new unemployment insurance claims and the
Conference Board's Help-Wanted Index are indicating that overall payroll
gains should still be running about 250,000 per month. The rate of GDP
growth, which also correlates with employment gains, is saying the same thing.
So why the 75,000 average over the past two months?
True, the 4.2% jobless rate hasn't been this low on a sustained basis since the
late 1960s. So maybe those who believe there's no one left to hire have a
point. But if history is any guide, then at no time has low employment ever
prevented payrolls from growing along with the economy. In the late 'Sixties,
the unemployment rate continued to fall, hitting 3.5% by '69, and payroll
growth actually accelerated.
This tendency for job growth to quicken when labor is scarce shouldn't be
surprising. To begin with, firms become more eager to scoop up extra
workers in order to avoid a potential shortage. And for the same reason, they
become more reluctant to shed workers -- a process known as labor
hoarding. While the point is often forgotten, payroll gains reflect net hiring,
after layoffs.
Of course, this time may be different, which is another common refrain.
Maybe the mismatch between geography and skills is worse than it used to
be, and that's why there's nobody left to hire. But for all the talk about the
needs of high tech, non-tech employment growth is still dominant. Over the
past year, the retail trade alone added nearly half a million positions. And if
anything, labor is more mobile than it used to be, with better access to
information about where the jobs are.
Which brings us back to the paltry August/September payroll gains. First,
these numbers are only preliminary and are subject to revision. Maybe next
month they'll be revised upward.
Also , October's estimate could compensate if it comes in very strong, which is
a pattern that often occurs. For example, July's gain stood at a revised
373,000, and with the expected bounceback from the hurricane effect,
October might print at better than 400,000.
But if none of these things happen, we may simply have to wait until next year
when these figures are benchmarked against the unemployment insurance
figures, which provide a near-universal count. Then we'll have some idea
about whether the BLS's sample was inadequate, or whether the bias
adjustment that it uses to boost these figures was insufficient -- or whether any
of the other things that could go wrong with these estimates did go wrong.
And then we'll know whether the no-one-left-to-hire argument holds any
water. |
