Slowing Down, Slowly?
Is change in sector composition reducing the US real output trend growth rate?
Today's post is more microeconomic than I'm used to, and I think that's a good thing. Macro is important, but it's easy to slip into an "aggregate dreamworld," and miss so many elements of the economy which are just as relevant as the business cycle or inflation. One of those issues is the sector composition of economies -- the allocation of labor, capital, and other resources to various ends.
These are not questions which, in my experience, come up as much as they should. An economy which produces mainly industrial goods meant for export to global markets -- South Korea or Germany, for example -- behaves very differently than economies with different sector compositions, perhaps service-based and domestic-consumption-oriented -- the United States, most obviously. Microeconomic differences such as sector balance generate different patterns of macroeconomic growth and recession, and I would imagine they also dictate strikingly different optimums for public policy.
For instance, a while back, The Economist discussed why Kurzarbeit, a work-sharing program sponsored by the German government, could be efficient in Germany but inefficient in other economies:
The path of economic recovery may partly depend on whether employment stayed firm because of companies’ own decisions rather than government subsidies. Short-time work schemes introduce distortions: they could impede growth and productivity once recovery begins by discouraging workers from moving from firms in declining industries to growing ones. Firms that hoard workers voluntarily do so because they expect to need them when business bounces back and, having invested heavily in their training, do not want the expense of replacing them. That suggests a more upbeat view of the future.When microeconomic differences generate macroeconomic differences, they require public policy differences. In Germany's case, the inefficiencies of the labor hoarding encouraged by Kurzarbeit are offset by the losses generated from unemployment -- skill atrophy and search frictions -- given the violent volatility of trade volumes. That would probably not be the case in the United States, where recessions tend to induce large-magnitude changes in sector composition -- that is, when a recession reduces employment sharply in a particular sector, that change in demand for workers tends to be permanent, and thus work-sharing would effectively delay the inevitable and be inefficient public policy.
In countries whose recessions were due largely to falling exports, such as Japan, Mexico, Germany and Korea, the OECD notes that companies were more likely to cut hours and productivity than jobs. The export drop “might plausibly have been viewed as being a largely transitory phenomenon”, reflecting global conditions rather than a structural change in the economy. Such firms are also more likely to use highly skilled labour, and the report reckons that these companies are far less prone to cut employment when sales fall.
Let's move onto the main question: does the changing sector composition of the American economy explain and/or predict slowing macroeconomic growth?
To quote "American City Suite" -- yes, I know the lyrics to a song which hasn't been popular since 1972 -- "[we] know that it's changin' / [we] see that it's changed." The American economy, at the microeconomic level of sector composition, has undergone dramatic change in just the last 10 years, and naturally in larger time increments too. We can see this by looking at a graph of the fraction of total nonfarm employment held by specific sectors:I'm using employment, rather than output data, because it's what FRED has available, but note that it's not markedly different in terms of what the findings are. We can see here the gradual contraction of manufacturing employment as a fraction of all jobs -- it's been happening steadily since the 1950s -- and the rise of service industries: education and health, leisure and hospitality, professional and business services.
It should be pretty obvious -- if it's not, see my post on Baumol's cost disease -- that productivity growth in such service industries tends to be significantly slower than in goods industries. It should be an easy logical leap from there to suggest that trend productivity growth, which is the ultimate driver of real output and incomes, has been on a secular slowing trend for decades.
I quickly -- and note: unscientifically -- used the sectoral data and my own "ballpark" estimates for annual productivity growth to generate aggregate productivity numbers. Indeed, I found that the changes sector composition should reduce trend productivity growth. I plugged in a variety of specifications for sectors' productivity growth rates, and this result is robust.Obviously, this doesn't reflect exactly what's been going on with productivity growth (see 1, 2, 3), much of which has been driven by productivity gains within the sectors -- the computerization of manufacturing, for instance. It does follow, though, the historical behavior of real growth.
I think it's a fair question, maybe to add to Tyler Cowen's "Great Stagnation" thesis, that sector composition changes could begin to cause economic growth in the US to follow a slower path.