One of the most important points of consensus in modern macroeconomics is the importance -- indeed, the supremacy -- of productivity. That apparent certainty makes me afraid to even go near this topic, but I am inspired by Tim Duy's recent post on the distributional effects of monetary policy. I want to reexamine productivity growth in the United States, because I have the overwhelming sense that something is not working the way it used to, and, I would contend, the way it should.
I don't think it's controversial to say that economists think the long-run cost of productivity gains is zero. And I think one would have to be a little bit blind not to acknowledge that productivity gains do have costs in the short-run, and arguably there are scenarios in which these costs are quite extreme. (The deindustrialization, disemployment, and depopulation of Detroit and the state of Michigan, for instance.)
My focus for this post concerns the medium-run costs of productivity gains. In particular, my thesis is that the medium-run costs of productivity gains may be substantially higher in periods of low capacity utilization than in periods of high capacity utilization. Or equivalently, the "break-even point" in social welfare in terms of the number of years after a productivity shock will be substantially further off during the former than in the latter. All together, I think the evidence amounts to two different types of productivity gains, defined by their distinctly different the medium-run economic footprints.
As a side note, I see macroeconomics as sharply partitioned into the short run, in which most of the factors of production are fixed, and the long run, in which none are. There are virtues in this division, but so much of human activity really depends on the medium run, when we have the ability to change some factors of production but are not yet dead. In this respect, the short-run/long-run division can be problematic in that it tends to give short shrift to the medium-run. As far as I could tell, there is not to much work in economics on the medium-run, and what exists largely confirms that the topic is under-examined.
Olivier Blanchard wrote in 1997 that:
Macroeconomics is largely divided into two subfields. One focuses on the short run, on the study of business cycles. The other focuses on the long run, on growth and its determinants. The assumption implicit in this division is that the medium run is primarily a period of transition from business cycle fluctuations to growth. This simplification is clearly convenient, but it is misleading. Modern economies are characterized by medium-run evolutions that are quite distinct from either business cycle fluctuations or steady-state growth.
Robert Solow echoed Blanchard in a 2000 paper in the JEP:
[T]here must be a medium-run, five-to-ten-year time scale at which some sort of hybrid [i.e. Keynesian and neoclassical] transitional model is appropriate. Prices play a role, but not simple market-clearing, so income-driven processes may dominate events. Is that sort of approach respectable, or even possible?...I can easily imagine that there is a "true" macrodynamics, valid at every time scale. But it is fearfully complicated, and nobody has a very good grip on it. At short time scales, I think, something sort of "Keynesian" is a good approximation...At very long time scales, the interesting questions are best studied in a neoclassical framework...At the five-to-ten-year time scale, we have to piece things together as best we can, and look for a hybrid model that will do the job.
First, I worry about the costs of productivity gains in the medium run because it is timely. We are moving into the medium-run period of recovery from recession, and the path of productivity during this recession is visibly different from the past. Here are two graphs which show the level of productivity (measured in output per hour in the nonfinancial corporate sector) on the left axis and in blue, and the year-over-year percent growth in productivity on the right axis and in red.
You can see how dramatic and historically unusual the post-2008 behavior of productivity has been -- the sudden spike, followed by zero growth -- in the first chart, and then in the second chart the recent data is easier to see, although one doesn't have quite the same perspective. In January 2010, the nonfinancial corporate sector had 8.1 percent YoY productivity growth, the highest rate since 1954, but since then, productivity growth has been zero or negative, and this is one of the few productivity stall-out episodes on record not associated with a recession.
At that statistical level, the productivity gains we have seen recently, in this environment of low capacity utilization, look nothing like the gains we saw in times of high capacity utilization during the 60s, mid-to-late 80s, and mid-to-late 90s, which saw stable, high growth in the area of 2.5 percent annualized.
There is, in fact, far more than statistical behavior which seems to differentiate these two species of productivity growth. The very source of, as well as the motivation for, productivity gains tends to be quite different. (These explanations all refer to the graph immediately below.)
During periods of high capacity utilization, productivity gains come from increases of output which exceed increases in payroll employment (hours tend to be a negligible factor here). What prompts firms to seek increases in productivity appears to be the pursuit of higher profits by increasing revenues -- they want to be able to produce more output because there is demand for it.
During periods of low capacity utilization, productivity gains come from decreases in payroll employment in excess of decreases in output, or perhaps small increases in output. Hours tend to swing downwards, but again, the change is small. The motivation for firms into increase productivity during times of low resource utilization is not driven primarily by rising costs of resources, or by the quest to increase profits by increasing revenues. Primarily, productivity gains are about reducing costs, which, given downward nominal wage rigidity, boils down to reducing the use of labor.
These differences in the means of productivity growth, perhaps small, have very different consequences in terms of social costs. While both involve technology, in the former, workers are often retrained or repurposed to work alongside it; in the latter, the technology replaces labor inputs. The former has almost no social costs, given that no labor is disemployed by increased technological inputs -- and even if it were, the costs would be low, because labor demand in other firms is high. In short, when resource utilization is high, the medium-run consequences of productivity gains look a lot like the longer-run consequences. The latter has large social costs, given that labor is disemployed by the increase in technological inputs when labor demand is low -- because the labor is not put to use, there is a loss resulting from the expiry of unused resources, but also humanitarian costs, costs which will be borne by government, and costs related to skill atrophy during prolonged periods of unemployment.
In fact, the window of sharp gains in productivity during periods of low capacity utilization occurs as output is recovering but employment is still contracting due to its lag. Contrastingly, high productivity growth during periods of high capacity utilization come, almost exclusively, from increases in output, with no cuts to employment or hours.
(Incidentally, this data raises some questions which this examination of productivity cannot explain. How did output contract less than employment or hours during the 2008 recession? Do the statistics not capture off-the-books overtime, or people working harder to make sure they can hold onto their jobs? Or was it that employers fired zero-marginal-product workers? Both explanations fit the story of the burst in productivity and then the slowdown. In turn, neither explanation fits the story of the "job-full" or "GDP-less" recovery -- the ZMP story was originally developed to explain jobless recoveries -- unless employers want to hire unproductive labor.)
The next graph, which I've recreated from the Tim Duy post to which I linked earlier, shows the decoupling of productivity gains from compensation gains, to the extent where one could argue that an increase in productivity does not necessarily lead to an increase in real compensation, even in the medium-to-long runs.
What I find really interesting is that the social welfare gains from higher productivity accrue to very different groups of people depending on capacity utilization. Here too, capacity utilization is the determining factor of the medium-run consequences of productivity gains. When capacity utilization is high, the productivity gains tend to be shared broadly, i.e. they translate into gains in real compensation. Perhaps this has to do with higher bargaining power on the part of labor, and less effective monopsony power on the part of employers, in such an environment. When capacity utilization is low, productivity gains are not shared broadly; they tend to go to corporate profits and foster increases in income inequality. The graph below shows the ratio between the gains in compensation and the gains in productivity, averaged out over five and ten years to smooth out short-run volatility so that we can see the medium-to-long-run behavior.
The data leads me to identify two different species of productivity gains. Depending on capacity utilization, the source of, and motivation for, productivity gains differs, the social costs differ, and who seems to gain from productivity gains differs.