That is a new suggestion made by Alwyn Young in the latest American Economic Review.
The cost disease argument suggests that some services do not augment their productivity very readily (e.g., the barber), and furthermore the demand for many services is income elastic and price inelastic, so with economic growth services take up a rising percentage of gdp over time. The relative cost of producing service output rises. And since services are more sluggish in productivity, and being weighted more heavily in output, we can expect economy-wide productivity rates to decline.
Young’s argument is to draw out the implications of the heterogeneity of workers. Let’s say that a worker’s skill in one sector is only loosely correlated with his skill in some other sector. That will mean when individual workers have comparative advantage in a sector, they also are likely to have absolute advantage in that same sector. The typist really is a better typist than the lawyer.
Now let’s go back to the services sector. It expands over time and sucks in labor by offering higher relative wages. That draws in more individuals with low comparative and also low absolute advantage in that sector. More concretely, the system ends up pulling in a lot of losers into law and medicine, while we are left with only the very best factory workers still on the job. Or think of all those mediocrities who flooded into punk rock in the early 1980s. It’s a bit like a Peter Principle.
The services sector will appear less efficient, but what is called “underlying true levels of productivity growth” — taking into account the average efficacies of the workers present in the two sectors — might not be changing much at all. In other words, a quite different mechanism can generate the same observation as Baumol’s cost disease.
The paper presents plenty of industry-level evidence that this declining efficacy of service workers is indeed the case.