[pjc50]

The original formulation of Baumol's cost disease is that it's a relative effect caused by increasing efficiency in automatable fields making non-automatable fields look bad.

If a job has to be done by a human in the West, of course it's expensive compared to those that have either been outsourced to cheaper humans or turned over to machines.

The "exchange rate" between human-produced goods and machine-produced goods looks worse and worse over time. The classic example is whenever you hear someone describing "flatscreen TVs" as a lavish expense. They're not. All TVs are flatscreen and you can get perfectly adequate ones for under $100. Whereas ladies' haircuts can easily exceed that - after all, it's a job you can't export to the Far East. And a college education costs several hundred televisions.

[oli5679]

Baumol's model is interesting. It's definitely part of the story.

It seems that there are other factors at play as well though. Scott shows that wages haven't risen as fast as costs in many of the problem sectors, and that there is significant cost variation between countries with similar wealth levels.

[joe_the_user]

An extension to Baumol would be that not only have wages in non-productive industries had to keep up with wages in productive industries but capital spending rates in non-productive industries have kept up with capital spending rates in productive industries. At least in education and health care, you see buildings, equipment, administrators and so-forth taking more and more of the expenses relative to the salaries of service provider.

Admittedly that's automatic given cost increases without concomitant salary increases but I think one can this to standards of capitalization of productive industries bleeding over to non-productive industries. And there's a constant belief/hope/snake-oil that these increases in capital spending will make the industries productive. Indeed, the "captains" of these would never frame the industries as static, non-productive support industries.

[bluGill]

Except that that doesn't seem to be the case. A house in 1960 costs an inflation adjusted $100,000 today - The average house last year sold for about $300,000. So we can say infrastructure is 3 times as expensive - that doesn't seem to account for the 10x cost increase.

[joe_the_user]

Obviously, the kind of thing I speculate on above is an industry-by-industry effect.

However, as mentioned in the article, the average house that is built today is larger than a house built in 1960 and that can account for some of the increase also.

Moreover, a 3x increase in one cost factor is a good start on explaining a 10x increase in overall costs. You can't expect any economic process, from grocery bills upwards, to yield an exactly proportional result between two "back of the envelope" estimates, now can you? One inflation estimator might not be akin to another etc, etc.

[ikeboy]

You're switching between labor/machines and western labor/eastern labor without explicitly flagging that. Those are different things.

[smallnamespace]

The distinction is actually western labor vs. other labor/machines.

[rdlecler1]

This is a good point. These areas are not operating in a Moore's law domain. Even while computers get better and faster exponentially their costs are dropping more linearly. Whereas other domains rely on finite resources: real estate & talent which spurs demand side competition & inflation.

[crpatino]

I think that position is simplistic at best. Sure, it would explain an relative increase in human-intensive services compared to machine-empowered ones. What if fails to explain, though, is:

1. A sustained, inflation-adjusted increase in human-intensive services over time. Specially when combined with stagnation in the compensation of this (supposedly) expensive, and specialized Labor over the same period of time.

2. A perceived trend of decrease in quality of the service, in a futile attempt to keep #1 under control.

If I where to venture an explanation of that would take automation and technology into account, I'd try to approach the problem from a systems theory point of view and suggest that maybe we are observing an overzealous attempt at partial automation.

This industries, as you correctly pointed out, cannot be fully automated/outsourced. This however does not prevent upper management from trying to achieve at least a part of the goodness that is benefiting other, more malleable, industries. They will of course try to automate some non-critical part of their workflows, which will upset the balance between the different subsystems of the whole and place more burdens in the critial parts. This, paradoxically, will force them to hire more personnel to keep the operation afloat, which then seeds the way for future interventions when technology advances makes posible the automation of some of those extra positions.

Even in a very bread-n-butter manufacturing environment, that type of death-cycle is extremely damaging. Take a look at Eliyahu M. Goldrat's novel "The Goal" to see a fictional example unravel.

[hyperpape]

Baumol's cost disease predicts growth in prices above inflation. After all, inflation is overall growth in prices, and consists of things that suffer from cost disease as well as things that actually get cheaper (TVs, etc).