Tim Worstall riffs on William Baumol to fascinating effect:
One way of putting which is that increasing labour productivity in services is more difficult than improving it in manufacturing. Canonically, we cannot get a symphony orchestra to be more productive by playing at twice the speed. So, ally this with wages being determined by average productivity, we'll see the amount we need to spend on labour to get services to rise against the amount we need to spend on labour to get manufactures. Services will become more expensive relative to manufactures over time.
However, this is not certain. A tendency, yes, but not a certainty. For it is possible, through innovation, to turn a service into, if not a manufacture, at least an automated operation. Think replacing bank clerks with ATMs. Skilled typists with dictation software. We can record the symphony once and play it many times on a gramophone/Walkman/iPod.
Which leads us to another part of Baumol's work. What system, what structure, boosts innovation?
Note that "innovation" here is not taken to mean the invention of new stuff. Rather, the dispersion of such new inventions through society, enabling people to think up ways of using it in new and interesting ways: boosting labour productivity as they do so.
One point he makes is that the Soviets invented some pretty spiffy stuff but as anyone who was there either during those times or in the rubble following 1991 will know, almost none of it ever got used by anyone.
No, it's a market system that encourages the use, experimentation with and thus improvements in productivity, that new inventions allow. Planning doesn't, the State doesn't, markets do.
Which brings us to our conclusion.
Let us agree that middle incomes in the US have been stagnating (they haven't, just not growing very fast but….). Let us also agree that we'd rather like to get them rising again.
It's that nontradeable sector, healthcare especially, which we need to worry about, the one where innovation, which determines labour productivity and which in turn determines general wage levels, is more difficult. And the way we know how to increase innovation is through having a market based economic system.
Another way of putting the same sort of point is that inventions are only any good (for raising the standard of living) if they cause time-saving in the fulfilling of your needs. The more we do that for some needs, the bigger loom the needs we have not yet so improved. Fifty years ago, people spent a lot of their income on food. Now we have made food so cheap, time-saving in the production of food hardly touches people's lives. A new model of combine harvester results in a tiny improvement in the cost of a shopping basket. But because food is so cheap, people can afford to spend a lot of their income on health, a sector that is very good at life saving, but not much good at time-saving productivity improvements (especially because either tax funding or insurance funding gets in the way).
Presumably this is why catch-up growth in China is faster than pack-leading growth in America and Japan and Europe. The easy industries are being mechanised and efficientised first.