More on US Manufacturing and Productivity

Continuing the discussion from here and here (see comments) of US productivity, manufacturing capacity/activity, and what that means for both the domestic and global economy, David Steinberg pointed out that while total US manufacturing output has gone up, and US share of global manufacturing output has stayed surprisingly high, there's something else to consider:

I agree that the US's share of global manufactured production is the relevant measure regarding American control over global production. But, from what I can tell, the main reason why most people worry about the decline of manufacturing is the belief that the manufacturing sector has higher productivity than services (e.g. Rodrik's stuff). And if that is what worries you, then the manufacturing sector's share of GDP is the "better" measure. Maybe that's just my bias though.

If the worry is about productivity, then the data we're looking at should encourage us. After all, it shows that we're producing a lot more with a lot fewer workers... that's the definition of a productivity increase. But, as Steinberg says, we could be beneath our potential if we aren't using our workers in the most productive mixes. So take a look at this chart:



From the BLS via the Fed, which has a nice interactive tool at the link allowing you to compare sectors. I'm guessing the straight line for manufacturing pre-1986 means that these type of data weren't collected before then by the BLS, but in case the trend is pretty clear. Per-worker output has been going up for both manufacturing workers and those in services. Manufacturing productivity has risen fastest, which is I think part of what Steinberg is referring to (via Rodrik) but it's not clear that productivity in those industries is much higher than in non-manufacturing industries. Productivity growth has been faster there, but remember that we've been shedding workers during this period; those are likely the least productive workers. It's not immediately obvious that we could add a bunch more workers to that sector and keep productivity levels the same. There's some evidence and argument that quite a lot of these potential manufacturing workers generate very low marginal product. Meanwhile the marginal cost of hiring them remains relatively high. Automating a lot of simple tasks has made some workers's skills surplus to requirements, and offshoring to lower marginal cost workers for low-productivity tasks has further hurt those with few skills.

One (possible) way to think of these dynamics is as follows: the US economy is left with a high-skills, high-productivity economy in tradable industries. Those with lower skills get pushed into non-tradable sectors that are not easily automated, where their lack of productivity doesn't make them uncompetitive. Maybe that's okay if there are lots of jobs in construction or services, but it tamps down middle class wage growth while boosting high-skill wage growth, thus leading to more inequality.

If this is close to true, then we ought to see some political organization across sectoral lines, and some across factoral lines. Across sectors, we should see support for trade openness in nontradables and high-productivity sectors. Across factors, we ought to see a lot of opposition to immigration of low-skill workers from workers in nontradable sectors, but support for immigration for capital in those sectors. We should see more of the economy shift to high-skill services. We might expect a stronger social safety net too. To greater or lesser extent all of those have been present in the politics of the US over the past 10-15 years.

Has there been any research done on the politics of productivity?