The Return of the Luddite

If the Luddite fallacy were true we would all be out of work because productivity has been increasing for two centuries.

This quote from economist Alex Tabarrok is found all over the Internet as the definitive reductio ad absurdum dismissal of the Luddite fallacy.  It's on Wikipedia, and is almost certainly Dr. Tabarrok's most widely quoted quote.

Let me start deconstructing this by providing a counter example:

Productivity has been exponentially increasing for two centuries, and yet we are not all millionaires working one hour per month.

Back in the 1950s futurists projected that, with wage and productivity trends, there would be a major societal problem by the 1980s regarding what workers would do with all of their leisure time, when working four hours per week.  Needless to say that didn't happen.

There are two primary reasons for this: one, is population growth has kept the supply and price of labor in check, and two, workers have not seen most of these productivity gains reflected in their paychecks.

Economist William Easterly offers the following criticism:

[...]the "fallacy" of the Luddites lay in their assumption that employers would keep production constant by employing a smaller albeit more productive workforce instead of allowing production to grow while keeping workforce size constant.

Dr. Easterly never interviewed any Luddites I am sure, but there is no reason to believe that they expected employers to keep either production or workforce size constant.  Most employers are trying to continuously optimizing to produce more with fewer workers--the definition of productivity.  Automation allows them to do exactly that.  In other words, employers face no fundamental tradeoff between production and workforce size--they can and do have both greater production and smaller workforces as productivity improves.

Agriculture is a perfect example of this phenomenon.  We produce more agricultural products today than ever, with a fraction of the labor required in pre-industrial times, thanks to advancements like combines and fertilizers.  US manufacturing output is higher than ever, yet the manufacturing labor force is less than half the size it was back in the 1950s heyday.  The Luddites didn't make any false assumptions--they knew exactly what the plan was and they were right!

Now, one may point out that the agriculture story demonstrated how the loss of jobs and reduction in the cost of goods in one industry (agriculture) freed up labor and consumer spending to enable the manufacturing revolution, and a similar story can be told as manufacturing industry labor need gives way to that of the services economy.  Ok, but this leads to the key question:

When all goods and services production is automated, what's left for humans to do that someone will pay them to do?

The standard answer that new technologies will create new jobs and services is certainly partly true, but saying that there will be enough jobs created thusly to employ the growing population is an article of faith unsupported by any facts.

Immortalized in the tale of John Henry and the Steam Engine, people have recognized for a long time that technology advancement creates labor winners and losers.  Losers are humans replaced by automation.  Winners are those who fulfill new labor opportunities afforded by the new technology.  Economists usually make the following argument that sounds like a law but has no name as far as I can tell:

1. Technology improves productivity

2. Productivity lowers production costs

3. Lower costs means lower goods prices

4. Lower prices means higher demand

5. Higher goods demand results in more labor demand

There is no doubt that technology destroys some jobs while creating others.  I argue, though, that there is no promise that as technology advances, the process of "creative destruction" as von Mises put it will forever result in a positive job growth balance, which is necessary as long as human population is growing and we have a more or less free market system.

As this chart from Jared Bernstein's blog shows, something seems to have happened in the year 2000 that suggests that we may have reached a point where computers and robots have gained the upper hand:

US productivity per employee has accelerated, while employment growth has decelerated.  Certainly multiple factors are at work here.  One is overseas offshoring, which has contributed to US company topline revenue while suppressing labor growth.  Another is the rise of Internet businesses, which are capital intensive but require very small amounts of very skilled labor.  Google makes more than $1 million dollars in revenue per employee, an incredible productivity statistic.

We need only look outside the US to other countries that have high double-digit unemployment rates to see that there is no mystical rebalancing function in an economy that will create an equal number of similar-paying jobs when technology obsolesces a job function.  If there were, then horses would have found jobs by now after being displaced by motor power a hundred years ago.  And don't tell me horses can't learn new tricks.