Automation and History

History always has something doesn’t it?

Here is John Aziz writing at Pieria:

“So what does that mean today, as we get caught up in this great new tsunami of technological innovation and Schumpeterian creative destruction?

It means that we shoudn’t really fear the process we are going through. Yes, technology destroys some jobs. But the historical record suggests that automation will free us up to do more interesting ones. For sure, nobody precisely knows the future. And — just as there was during the Great Depression, when the U.S. economy transformed from a predominantly agricultural economy into a predominantly manufacturing economy — there may be great dislocation and major bumps in the road (one possibility is rising inequality if the robots are heavily centralized, although many argue not, and I am on the latter side of the argument).”

There is no doubt that the current wave of automation is changing the economy. That change is both having an accumulating impact. There are about 260,000 ‘robots’ working in America right now. That number will rise to an estimated 700,000 over the next decade.

To understand the economics behind the surge in automation think of this example: the cost of an automatic spot welder was in the region of $182,000 in 2005. Right now that price has dropped to $133,000. By 2025 experts estimate the price will be $100,000. Clearly it will become ever more attractive to replace labor with machines. Moreover one estimate I read had the human cost of auto industry welding as $24 per hour, whereas the same job done by machine costs $4 per hour. Those kinds of cost advantages are irresistible. Obviously the shift towards automatic production will gather pace. These estimates are from the International Federation of Robotics, so they are biased, but the effects will surely be similar to those the IFR predicts.

So is Aziz correct to say that this is not something to fear?

I think he has a deliberately optimistic reading of history.

It’s fascinating to see how economists and people who write about the economy rely on highly selective history. We don’t have many similar events in history to go back and study in order to unravel the probable impact on the economy.

Aziz rather glibly suggests that all will be well. After all, he points to a study, by the auditing and consulting firm Deloittes, that show that innovation has produced more jobs than it has destroyed over the past 180 years. I should hope so. But the studies examples point to one of the problems we are having. It focuses on the elimination of a great deal of drudgery that used to be done by hand. Laundry washers have, for example, became quite a rare breed, especially as a proportion of the population. The switch of employment out of agriculture is, of course, an iconic example too.

Personally I have no doubt that automation has had a great social benefit since the start of industrialization. It is, in no short manner, a great force behind the freeing of women from their older and more traditional role as providers of unpaid housekeeping services. It has reduced those hard laboring tasks to a far more scarce list, and even then it has added enormous leverage to those laboring tasks that remain performed by humans. So this part of the claim by the optimists I do not dispute.

What concerns me, and what I don’t think people like Aziz focus on enough, is that we have run out of sectors we can shift labor about in.

The huge rise in automation in agriculture that drove so many people off the land created waves of discontent and dislocation, but eventually — during the Great Depression here in the US — that surplus agricultural labor was absorbed into the then burgeoning industrial sector. So not only did the economy benefit from rising productivity on its farms, but it produced higher paying jobs that enabled the newer working class to become upwardly mobile and aspire to something we have became to call the new middle class. The new jobs paid well enough to compensate for the dislocation of the prior automation. The economy took a step upwards.

Mind you the entire process of absorption took a full one hundred years if we go back and start the clock running at the start of industrialization. Not only this: along the way there were enormous political and social changes that made the end result — a generally higher level of prosperity — possible. This brighter future was not, contrary to the sunny arguments of the libertarian economists, a result of the magic of the marketplace, but was, rather, the result of generations of activism and social protest that eventually put in place truly democratic institutions to mitigate that more dire consequences of capitalism. It is no accident that modern democracy is a much newer arrival on the scene than is modern capitalism.

As a side note: it was the success of the fight for democracy that forestalled the Marxist historical narrative. It is one of modern history’s great ironies that the only full implementation of Marxist thought took place not in the context of an industrial proletariat, which is where Marx taught it ought to take place, but in the context of a backward agricultural underclass, which is where marx considered it near impossible. But that’s a different discussion.

The advent of a new age of machines enabled the launching of what we know as modern capitalism. And its first notable consequence was to benefit a rise in returns on capital. It was this early history of industrialization that Kuznets so famously captured in his eponymous curve.

But since Kuznets we have observed an enormous shift backwards. Not only has inequality risen to near historic levels, but there has been a decided bias towards returns to capital once more.

And one force creating this unfortunate result must be automation. The very thing that Aziz so happily embraces.

What we now need to think about is where the new surplus of labor — those being made redundant by the current wave of automation — will end up. When we look at the American economy the answer is obvious: that labor has to be absorbed in the service sector which currently accounts for about 80% of all employment.

Here’s the issue: the service sector covers a very wide range of activities, from those remaining laundry washers to brain surgeons. Along the way it includes the hairdressers and bartenders that the Deloitte study highlights. It also includes the engineers, designers, and sundry bloggers that Aziz says he would prefer to be rather than those old fashioned laundry washers.

The problem is that many of theses jobs produce lower income than the manufacturing or industrial jobs being displaced by automation. So the new history is radically different from the older history.

Whereas the old displacement eventually created a more prosperous and plentiful middle class, this new displacement may not. Indeed if we take the studies of people like David Autor and his co-authors at MIT seriously, it certainly will not. At least any time soon. Worse as the pace of innovation accelerates automation seems to be working its way up the income scale. Not only are traditional blue collar jibs being automated, but many previously secure white collar jobs are going under also. The result is that we are experiencing a huge bifurcation in society unlike that of the last wave of automation. That last wave consolidated society around a fairly prosperous median. This wave is dividing society into two very different levels. The bulk of people are finding themselves compressed between stagnant wages and rising costs, with automation a major factor in the wage compression. Whilst, at the same time, a much smaller group benefit from the wages flowing to their skills.

At present I don’t see the older coalescence around a happy median recurring. This wave of automation is far less democratizing than the last.

As we fumble about trying to make sense of our contemporary economy, I recall David Hackett Fischer’s book “The Great Wave”. It’s a history of prices, and it has relevance to today.

Here’s a quote regarding what he calls the Medieval Price Revolution of 1180 – 1350:

“Men of wealth were able to profit by the price-revolution in many ways. Powerful Italian merchants, for example, obtained laws that allowed them to insist on being paid in gold florins or ducats which held their value, but permitted them to pay wages and taxes in silver coins which were much debased. As a consequence, rich merchants grew richer, and the poor sank deeper into misery and degradation.

This growing gap between returns to labor and capital was typical of price-revolutions in modern history. So also was its social result: a rapid growth of inequality that appeared in the later stages of every long inflation.”

Does this not apply to the past few decades?

In his conclusion Fischer lambastes free market thinkers for their myopia. They use too short a historical framework. And they attribute magic falsely to their revered market systems. Other forces grind more slowly and more deeply.

Are we at the end of another long inflationary cycle? Is history rhyming again?

The point of all this is that each of Fischer’s long price cycles ended in social and political disruption. Each cycle rose and fell for its own reasons — ours may well be the automation Aziz lauds — but each also had similarities. The tension between labor and capital being one.

Is the trend towards populist politics an early signal of that tension placing too great a strain on our modern world? I think so.

Indeed, I am certain of it.