The great productivity delusion (corrected)

1 view
Skip to first unread message


Apr 12, 2000, 3:00:00 AM4/12/00

To see the graphs, you should view the PDF file:

The great productivity delusion
' S Vol. 18, No. 06r MARCH 31, 2000 30 Wall Street, New York, New
York 10005 0 www. grantspub. com

A crisis of faith for the acolytes of the New Economy may be just
around the corner. On May 4, at 8: 30 a. m. on the dot, the Labor
Department will report on the growth in nonfarm productivity for the three
months ended March 31. The number stands to be hugely disappointing-
just as disappointing, in fact, as last year's third and fourth quarter
data were uplifting. March 7, as you may remember,
brought tidings of a surge in output per worker at an annualized rate of
6.4% for the period ended December 31. It was the fastest quarterly
improvement since 7.4% late in 1992, when the economy was emerging
from a brief recession. The immediately preceding productivity report,
a 5% gain for the third quarter, was almost as brilliant.

Sharp gains in output per hour of labor are standard fare at the start of a

business expansion. They are rare in the shank of one- so rare, indeed,
that the 6.4% and 5% readings stand either as testaments to the mounting
efficiencies of the Internet age or as sheer statistical illusion. New
provided to Grant's by the crack economists James Medoff and
Andrew Harless, strongly points to the latter interpretation. If the
Medoff-Harless hypothesis is the right one (as we are persuaded it is),
the New Economy may soon lose some of its youthful bloom. We have
all heard it said that the step-function increase in output per hour of
observed late last year promises a new age of health, wealth and price
stability. Not so fast.

Last June, Robert J. Gordon, professor
of economics at Northwestern University, disclosed a major finding
The so-called productivity miracle in the U. S. economy during the
five years was entirely centered in one industry, he wrote in a paper
to the Congressional Budget Office. You will not be surprised to
learn which industry: "There has been no productivity growth acceler-ation
in the 99% of the economy located outside the sector which man-ufactures
computer hardware. . . ," wrote Gordon. "Indeed, far from
exhibiting a productivity accelera-tion, the productivity slowdown in
manufacturing has gotten worse: when computers are stripped out of
the durable manufacturing sector, there has been a further productivity
slowdown in durable manufacturing in 1995-99 as compared to 1972-95,
and no acceleration at all in non-durable manufacturing."

However, markets make opinions, as the sages say, and the Gordon find-ing
has had no appreciable effect on the economic and financial discus-sion.
Indeed, only last month, the Federal Open Market Committee
ruminated on monetary policy as if the existence of a wide-ranging
advance in the ratio of output to hours worked were a revealed truth:
"[ Officials asserted that] equity mar-kets had shown a remarkable
resilience to higher interest rates as earning prospects continued to be
marked up in association with the acceleration in productivity," said the
minutes of the February meeting.

Coming next is a deceleration in
first-quarter productivity growth, we hereby predict, caused by some of the

very factors that were responsible for the third-and fourth-quarter pickup.

The reported boom in manufacturing efficiency in the last half of 1999 was
the result of not one, but two misapprehensions, relate Medoff and
Harless. Not only did the government statistics exaggerate the output
of the computer industry over those six Y2K-obsessed months (for reasons
having to do with the way in which computer prices are calculated).
They also grossly underestimated the number of hours worked to achieve
those results. It was this twin error that, in no small part, enlarged the
legend of the productivity boom and uncorked the newest bottle of
moonshine. Rarely has a botched calculation delighted and
enriched so many guileless people.

No doubt, observe Medoff and
Harless, respectively professor of economics at Harvard and untenured
economist in Needham, Mass., manufacturers are indeed making better
computers all the time. "But how much better, exactly?" they ask. "Is a
Pentium III three times better than a Pentium I? More to the point, how
much more valuable is a PIII/ 700/ 128M/ 10G/ 56K/ DVD today
than a P90/ 8M/ 512M/ 14.4K/ noCD of five years ago? That's an awfully hard

question. (Think about it: five years ago, you would not have had to
Office 2000; and once upon a time, there were financial analysts who felt
they could accomplish quite a lot using 1-2-3 for DOS on a mere 386.)

"The Bureau of Labor Statistics solves for the answer by using a
'hedonic price index' produced by the Bureau of Economic Analysis
(BEA)," they go on. "The index is supposed to measure the cost of the
specific features of different computers in a way that allows newer ones
and older ones to be compared. One
could think of it as a measure of the cost of the abstract 'computing
power' embodied in whatever com-puters are being sold. According to
the BLS's measure (and the conven-tional wisdom), that cost is falling
very quickly. Between 1992 and 1997, the BEA's hedonic price index
for computers fell by almost a factor of four. To figure out 'how much' is
produced by the computer industry, the BLS divides the dollar value of its
output by the hedonic price index."

The potency of this adjustment is
clear in the accompanying graph, which presents three different
measurements. The top line, bearing a close resemblance to a
high-tech stock chart, divides the computer industry's output (using
the hedonic calculation) by hours worked; it conveys the familiar,
almost miraculous, productivity story. The middle line also measures
in the computer industry but omits the hedonic calculation;
instead, the overall Producer Price Index is used to arrive at the real
value of computer industry production. Productivity growth in this case
is positive, but something less than inspirational. The bottom line
describes productivity growth in every industry except computers. As
Gordon discovered, it goes absolutely nowhere.

It bothers Medoff and Harless that this is so. "One might say," they
write, "' so what if it's all in one sector? Productivity is still growing
rapidly, and maybe that one sector can save the rest of the economy. '
That point would be valid if computers were a consumer good that could
easily be substituted for other important consumer goods. With computers
getting faster and faster, we could start wearing them and driving them
and shaving with them and eating them, and productivity growth in
computer manufacturing would
transform the face of American enterprise. But aside from the obvious
problems with substitution (fast computers don't taste any better than
slow ones), computers aren't primarily a consumer good."
But there is a more specific and eye-opening problem, as Medoff and
Harless discovered. "Hedonic price indices can be distorted by the effect
of obsolescence. That is, if a particular item becomes obsolete, it will
appear that its price has come down. In fact, the price of an absolutely
superseded item may go to zero. But this doesn't reflect a true decline in
its price, or in prices overall. It certainly does not imply an increase in

productivity. "Last year, some older, non-Y2K-compliant
computers did become obsolete. People bought new computers
not just because they were faster, but because they promised to
keep working after the end of the year. To an observer unaware of the
Y2K issue, it might appear that the new computers must have possessed
an extremely large amount of computing power, otherwise, why would
people be so eager to buy them? And since the BEA's hedonic price index
doesn't factor in the Y2K issue, it would make a similar misattribution,
concluding that an extreme of computing power was now available at a
reasonable price. As a result, the sta-
tistics would show 'productivity' increasing particularly quickly. And
so they did."

Not only, as noted, did the government
statistics exaggerate the quantity produced, but they also minimized
the hours worked to achieve it. In calculating "output per hour" in
the computer industry (and the rest of the durable goods manufacturing
economy), the BLS assumed that white-collar, or "non-production,"
employees spent an average of 37.6 hours at their posts. Medoff and
Harless scoff at this, as, indeed, will anyone who observed the hectic
preparations for the supposed millennium crisis. In the months before the
new year, not a few info-tech employ-ees slept on the floors of their
in their clothes. Whatever the actual number of hours logged by America's
cyber-workforce in the third and fourth quarters of 1999, Medoff and
Harless propose, it was very likely more than the stipulated 37.6.

In the run-up to Y2K, the Fed cre-ated tens of billions of dollars of extra

Reserve Bank credit, and the world replaced uncounted numbers of non-Y2K
compliant computers. The Fed has now removed the last extra increment
of that monetary stimulus. And with the coming disclosure of the
first-quarter nonfarm productivity data, the market may start to remove
the misplaced increment of faith it had lodged in the
properties of high technol-ogy. Mark your calendar: May 4 is or
ought to be judgment day.

Reprinted with permission of Grant's Interest Rate Observer, copyright
2000 for subscription information call: 212-809-7994

Do You Yahoo!?
Send online invitations with Yahoo! Invites.

--------- End Forwarded Message ---------

--== Sent via ==--
Share what you know. Learn what you don't.

Reply all
Reply to author
0 new messages