Armchair Economics

In a recent New York Times editorial David Brooks does a fine job laying out the real effect of globalization on job growth and loss. His main point is not that jobs (especially manufacturing jobs) have been lost to globalization. If manufacturing industries have been leaving the US for countries with cheap labor, then one would expect US global manufacturing share to decline -- but this is not the case.

"...U.S. manufacturing output is up over recent decades. As Thomas Duesterberg of Manufacturers Alliance/MAPI, a research firm, has pointed out, the U.S.'s share of global manufacturing output has actually increased slightly since 1980."

"...manufacturing productivity has doubled over two decades..."

"The chief force reshaping manufacturing is technological change... Employers now require fewer but more highly skilled workers."

So US manufacturing market share is up slightly in about a generation, but productivity (i.e. output per worker) has more than doubled. You can't quite compare those two concepts (since share is a percentage of a number not given), but if global manufacturing output doubled in that time (a guess), then there would have been no job manufacturing job growth (and even a big perceived loss, as the population went up by about 75 million people during that time).

But I believe there's a more subtle problem that Brooks misses, or at least ran out of space to address: flexibility. In an automated factory, output can be increased only to a certain level -- then more capital equipment needs to be purchased, installed, configured, etc. and that is very expensive. In a completely manual factory, adding capacity means hiring and training new workers. In the US at least, when times get tough, it is pretty easy to shed workers. But the resources spent on capital equipment generally cannot be recovered.

To me this means that US manufacturing flexibility is a bit like French job creation: employers are reluctant to take on new expenses that cannot be easily undone if times get tough.