Tuesday, August 11, 2009

A staggering term of productivity growth

Usually productivity growth tumbles during a recession. Firms incorporate economic conditions at a lag, cut marginal costs (i.e., jobs), and the unemployment rises. Not this time.

The chart illustrates annual productivity growth per quarter since 1950. As you can see, productivity growth dropped below zero in six of the last ten recession (including this one). Even more remarkable is: that this cycle ranks number two - behind the 2001 recession (2.0%), which didn't even see negative annual GDP growth after recent revisions - as the recession with the highest trough in annual productivity growth, 1.9% in Q2 2009.

Why is this important? Well, it means that firms have fired at a rate inconsistent with previous cycles - much faster, in fact. Brad DeLong suggests that firms are not "hoarding labor" and predicts a jobless recovery due to the fact that the historical relationship between output and job loss is diverging from that seen in previous recessions. It seems to me, though, that firms fired at record rates; and therefore, they may hire at lightening speed as well. Robert Waldmann at Angry Bear suggests the same.

According to the Wall Street Journal, the drop in service sector employment implies a speedy labor recovery. Here is an excerpt from yesterday's article:
The rapid pace at which businesses shed jobs during the recession comes with a flip side: Workers will need to be hired back quickly as the economy improves.

So deep have companies cut jobs that Friday's employment report, which showed that the U.S. economy lost a quarter-million jobs in July, was seen as a relief. Since the recession began in December 2007, U.S. payrolls have fallen by 6.7 million, according to the Labor Department. That's a 4.8% decline, a level not seen since the late 1940s.

"Firms were unusually aggressive in cutting costs and cutting employment," said James O'Sullivan, an economist with UBS. "The flip side of that remains to be seen, but it could mean that companies will be quicker to bring back people because they were more aggressive about getting rid of them."
These productivity numbers suggest that there is a significant probability that a resumption in demand for end-production will drive a sharp recovery in the labor market.

Rebecca Wilder


  1. The construction sector might be growing jobs aggressively, especially in the Phoenix area. New housing is hot - people who were locked out of the foreclosure homes like the look of a new house and are buying. A number of firms who built are gone but those that remain and pricing and building right. Ergo, jobs.

  2. As your chart shows productivity is highly cyclical.

    But the cyclical pattern has changed over the last few cycles with positive productivity growth in the most recent recessions.

    But productivity growth was still strongest in the recovery period of the last two cycles.

    So why shouldn't that pattern continue this time why strong productivity growth keeping employment growth weak?

  3. Let me put this another way.

    What has happened is the long term trend growth of productivity has shifted higher.

    But why shouldn't we see the same cyclical pattern around the new higher trend?

  4. Rebecca: Isn't it true that the Great Depression saw one of the greatest decades for productivity growth in US history? I think Alex Field wrote a paper about that.

  5. Hi Spencer,

    If indeed the long-run trend growth of productivity has shifted higher, then yes, cyclicalities around this number would be evident. A jobless recovery could certainly be expected (as in the last cycle). However, if I toss productivity into an HP Filter, I see that the trend series has slowed down markedly, around 1.6% annual growth or so since the last cycle. That suggests to me that the productivity swings are purely profit-driven; and that firms will be forced to hire if demand resumes.

    Hi Paul,

    I think that you are right - haven't read the paper or book, though. But I seem to remember David Wheelock (St. Louis Fed) referring to productivity swings in the 1930's. Will look for that reference.

    Hi Aunt Jane!

    Construction has not been associated with strong productivity growth. However, a resumption in homebuilding will get some of those workers back into the force (or employed).

    Thank you for your comments!


  6. I do not remember the source, but yes productivity was very strong in the depression.

    The generally accepted break points for productivity growth are:

    high -- prior to 1974
    low --- 1974 to 1995
    high----after 1995.

  7. Another very interesting post.

    Rebecca, I would suggest that the answer cannot be found in the aggregate. Different industries are affected by currently available technologies in different ways.

    For example, the trend toward light stocking in retail dates back years before this recession. One of the reasons is the growth of internet shopping and branch stores. For example, Walmart stocks stores with standard items, and floor workers can reorder, but it also has a pretty successful website, and customers can order there and pick up at their local store. This has the potential to really change retail profit margins per item sold, and therefore cut staffing and overhead. Auto manufacturers are increasingly turning to similar model, which has the potential to decrease the need for many dealerships, and thus cut staff levels associated with sales.

    Another example is the growth of automated self checkouts in retail. Robotics have had a huge impact on manufacturing, and I believe that they will have increasingly more impact upon warehousing and distribution.

    My impression is that the depth of this recession will accelerate some of these trends, and so I think that only talking to people in the major industries can provide much insight on this question.

    Aggregate productivity measures may have been greatly affected by trends in particular industries, such as the massive overgrowth of retail.


Note: Only a member of this blog may post a comment.