I see that my co-author Jared Bernstein has been pondering this question. While this sort of thinking can get you thrown out of the church of mainstream economics, I think that he is very much on the mark. Let me throw out a few reasons.
First, there is an issue about the money available to firms to invest. While larger and more established firms likely to have little problem financing investment in the current low interest rate environment, smaller and newer firms may find it difficult to get access to capital. For them a rapidly growing economy can be strong sales growth and higher profits, both of which are strongly linked to investment. This is a finding from an old paper by my friend Steve Fazzari and Glenn Hubbard (yes, that Glenn Hubbard.)
A second reason why productivity can be tied to growth is that firms will have more incentive to adopt labor saving equipment in a context of a rapidly growing economy. When they see additional demand for their products, they have to find a way to meet it. Of course they can hire more workers or have the existing workforce put in more hours, but another option is to find a way to produce more with the same amount of labor. Of course profit maximizing firms should always be trying to produce more with the same amount of labor, but they may not follow the economics textbook. Meeting increased demand can give them more incentive to do so.
A third reason is changes in the mix of output. At any point in time we have many high paying high productivity jobs and many low paying low productivity jobs. When we have a strong labor market, people go from the low paying, low productivity jobs to the higher paying high productivity jobs. This means that many people now working at fast food restaurants, the midnight shift at a convenience store, or as greeters at Walmart will instead find better paying jobs in a strong labor market leaving these low-productivity jobs unfilled.
The rapid growth of jobs in low-paying sectors in this recovery has been widely noted. Rather than reflecting an intrinsic feature of the economy, this could be the result of the failure of demand to create enough growth in the high-paying sectors. This is again a story where the causation goes from growth and low unemployment to high productivity.
These stories are hardly conclusive, but I have always found the causation from low unemployment to productivity growth is be quite plausible, glad to see Jared is on the case.
I also want to comment on Jared’s Greenspan story. Jared gives our former Fed chair credit for recognizing the upturn in productivity growth before others and using it as justification for allowing the unemployment rate to fall below the conventional measures of the NAIRU. I also applaud Greenspan for this move, which provided enormous benefits to tens of millions of people in the form of increased employment, longer hours, and higher wages. It also gave us the budget surpluses of the late 1990s, contrary to the claim of the Clintonites that it was their tax increases and spending cuts that did the trick. We would not have had a budget surplus, or even a balanced budget, if Greenspan had prevented the unemployment rate from falling below the 6.0 percent level that most economists accepted at the time.
While Greenspan deserves enormous credit for this break with the orthodoxy, it is worth pointing out that his reasoning did not make any sense. First, his belief that we were mis-measuring productivity growth stemmed in part from the fact that reported productivity growth in the non-financial sector of the economy was substantially higher than growth in the non-farm business sector as a whole. He argued that productivity growth in the rest of the business sector could not be negative; therefore we must be mis-measuring productivity growth for the non-farm business sector.
The big problem with this story was that productivity growth for the non-financial sector was based on an income-side measure of output. The income side of the GDP accounts was rising substantially more rapidly than the output side during this period. (One obvious explanation is that capital gains income was mistakenly being counted as ordinary income, inflating the income side measure.) So the issue was not a difference in the sectors being examined, but rather the measure of output. (Greenspan was also mistaken that productivity growth could not be negative. If a store decides that it has to hire a security guard to protect against robberies and theft it would have negative productivity growth.)
The other part of the Greenspan story that was wrong was his view that if productivity was being undercounted then the economy could grow faster than was conventionally believed. This made no sense because if productivity was being undercounted, then output was also being undercounted. (No one thought there was a problem in measuring hours growth.) This meant that the economy was already growing faster than was conventionally believed. The result was that Greenspan did the right thing in allowing the unemployment rate to fall below the accepted measures of the NAIRU, but he did it for the wrong reasons.
There is one last item worth mentioning about this episode. The debate over the level of the NAIRU was somehow translated into a debate over the economy’s growth rate. It is not clear how the level of the “non-accelerating inflation rate of unemployment” could somehow be transformed into a debate about the economy’s growth rate, but such is the state of economic debates in this country.
The switch from a debate on unemployment to a debate on growth was unfortunate for those of us who wanted to see the Fed let more people work. “Growth” is an abstraction. No one sees 3.0 percent GDP growth as opposed to 2.0 percent GDP growth. They do see the labor market. They know whether they or their friends and family members are able to easily get jobs. From a political perspective, we almost certainly would have been better off keeping the argument about unemployment and jobs, which was in fact the issue. Fortunately Alan Greenspan got this one right, so this particular unforced error turned out not to matter.