Technology versus the Distribution of Workers in Aggregate Productivity

Dietrich Vollrath:

Technology versus the Distribution of Workers in Aggregate Productivity: There was a recent post by an engineer rebutting Robert Gordon’s (and others) thesis that technological change was slowing down. The evidence cited is a series of plots and figures showing how specific technologies (battery storage, energy efficiency, computer speed, etc..) are advancing just as fast as they have for decades, if not faster. And there were a number of responses along the lines of “See, Gordon is wrong!”.

The mistake here is that this doesn’t constitute evidence that Gordon is wrong. But the mistake is partly forgivable because Gordon himself indulges in these kinds of anecdotal arguments to advance his thesis, and so it seems as if you could refute his conclusions by offereing alternative anecdotes.

But the important part of Gordon’s argument is not that specific technologies are or are not advancing. It is that aggregate productivity growth is slowing down. And aggregate productivity growth depends not only on individual technologies, but crucially on the distribution of workers using those technologies. Arguing about only those individual technologies is like using an increase in the price of milk to argue that inflation must be high.

Aggregate productivity growth depends on what we can call “within-sector” growth, which is going to be tied closely to those individual technologies. But it also depends on “across-sector” growth, which is tied to the movement of workers from one sector (or job) to another. If workers are shifting from high- to low-productivity sectors or jobs, then aggregate productivity growth may fall even though nothing happened to actual technological change.

I’ll make this more clear below, but here’s a quick summary of what I’ll try to establish. Gordon’s critics could well be right about their individual technologies, and yet wrong about this having anything to do with aggregate growth, because those sectors may not employ many people. And Gordon can be right about aggregate growth, but wrong about individual technologies stagnating, because the movement of workers to low-productivity sectors may be dragging down growth. In short, you cannot talk about aggregate productivity growth without talking about both technology and the distribution of workers across sectors.

Gruesome Detail...