Claims about electricity adoption and technological unemployment

This is from a recent working paper (pdf) by Miguel Morin:

When the adoption of a new labor-saving technology increases labor productivity, it is an open question whether the economy adjusts in the medium-term by decreasing employment or increasing output. This paper studies the effects of cheaper electricity on the labor market during the Great Depression. The first-stage of the identification strategy uses geography as an instrument for changes in the price of electricity and the second-stage uses labor market outcomes from the concrete industry—a non-traded industry whose location decisions are independent of the instrument. The paper finds that electricity was an important labor-saving technology and caused an increase in capital intensity and labor productivity, as well as a decrease in the labor share of income. The paper also finds that firms adjusted to higher labor productivity by decreasing employment instead of increasing output, which supports the theory of technological unemployment.

You will note of course that the short-, medium- and long-run effects here are quite different, and of course electricity is a major boon to mankind.  Still, technological unemployment is not just the fantasy of people who have failed to study Ricardo.

Here is a short summary of the paper, via Romesh Vaitilingam.


That technological unemployment is a component of the labor equation seems pretty obvious. New machines are produced more easily than will workers adapt. Realities of technological innovation and human behavior there. It has always struck me as odd when people like ‏@pmarca tweet as if it will all work out quickly and equitably.

How much of that technological unemployment had FDR's high-wage policies as a pre-requisite?

"The paper also finds that firms adjusted to higher labor productivity by decreasing employment instead of increasing output, which supports the theory of technological unemployment."

Why didn't labor saving technology result in increased production when labor incomes were reduced?

Perhaps because businesses are not idiots in producing 50% more and selling exactly the same amount because workers have exactly the same income to spend buying production???

Perhaps you can explain the economic theory of farm production that increases the output driving prices to create sales below the cost of production and requires borrowing money on land already burdened with debt to fund the sale of output.

Do you think farmers losing their land and banks foreclosing on land that can't be sold for the amount of the loan and then the savers in the community lose their deposits and thus can't stay in business is the best way to create jobs and wealth?

The end of the 20s was a time of greatly reduced demand for grain because Europe recovered from WWI and could now produce all the grain they needed. The last gasp of creating wealth by government land redistribution in the 20s failed because it was based on Europe never returning to farming and European governments sending gold to the US to pay US farmers and farm equipment manufacturing workers. By the end of the 20s, Europe was selling goods to the US to pay lots of European workers and instead of buying from US workers, they were taking back their gold.

GDP is limited by labor income in the real world.

Care to explain how GDP can exceed the labor income that would pay for buying GDP?

"GDP is limited by labor income in the real world." = Marxist dogma

If GDP is limited by income (in the real world) it is surely limited by TOTAL income, not just 'labor income'.
Unless, of course, profits are also considered 'labor income', which would invalidate your world-view entirely.

Um, none.

Why so, Jan?

The period studied was during the Great Depression, which would seem to limit application of the author's findings. I would like to see the findings for the period preceding the Great Depression, a period very similar to the period preceding the Great Recession (with respect to both inequality and speculation).

Me, too.

Who are you going to believe, ideology or your own lying eyes?

That there are dead towns, and "belts" of towns, dying demographics, is a given. Every one of them had some kind of shock which, if everyone was an idealized economic agent, would have meant a fast switch to a new and expanding career. Humans don't work like that though, as your eyes will tell you.

As an aside, a local sheriff's department paid for early retirees to be retrained as IT professionals (Windows networking specialists). Did anyone really think that introversion and nerdiness were taught that easily? Of course, per classical theory, workers will retrain for opportunities, disregarding any notion of their own "nature."

"As an aside, a local sheriff’s department paid for early retirees to be retrained as IT professionals (Windows networking specialists). Did anyone really think that introversion and nerdiness were taught that easily?"

I often see money for worker re-training stated as if it were some kind of mythical economic super engine. It strikes me as fairly unlikely. I'm particularly doubtful about highly targeted retraining of older workers, as per your example. At this point lower level IT positions are probably saturated and it's highly unlikely to be economically viable to retrain high level IT professionals. You are just as likely to get the same results by attempting to retrain workers to be lawyers or doctors. And yet, no one ever suggests that a laid off factory worker can become a practicing lawyer with just a couple of years of classes. Sure it's possible, and undoubtedly a few percent would succeed at it. But the money would be wasted on the other 98% that don't have the cognitive skills to do the job.

It occurs to me now that it was a little bit like the Monty Python lion-tamer sketch, only in reverse. Would carrying a cable-tester on a belt loop satisfy a man who has been carrying a Glock? (I asked my neighbor, the newly retired motorcycle cop if he actually wanted to work inside all day. He shook his head, no.)

So yeah, I agree that "retraining" as something you do to workers is highly suspect. Better to provide wide educational opportunities and then to let people self-select for compatible fields. Here's hoping they can find them.

Do you have any evidence that those positions are "probably saturated"?

Saturated might be too strong a word, but your own link shows that growth in IT positions has slowed way down this decade.

"Every one of them had some kind of shock which, if everyone was an idealized economic agent, would have meant a fast switch to a new and expanding career."

Whos to say they didnt? Maybe they just didnt do so in that particular town.

By "they" I don't suppose you are so optimistic as to mean "all," but this is what is required for "technological unemployment" not to be a thing.

Well i dont think technological unemployment is not a thing, just that dead towns and so forth could just as easily be a sign that the place in question no longer has the value it once did. The workers themselves could have switched careers, fast or slow and simply abandoned the town that no longer supported it.

Its not that i disagree with you, i just dont think that ghost towns and rust belts really prove much.

If they really were ghost towns that would be a good sign, but they tend to be "depressed communities" instead.

Depressed towns often have lots of older homeowners.

Nobody can sell their houses so nobody leaves.

Better to be poor in a home you own surrounded by your friends and family than to run off to the Big City somewhere else and gamble with your future.

The young and unattached drift away. The old stay behind. The town slowly withers.

Alexander Field argued that the Great Depression was actually a time of great technological advance (A Great Leap Forward: 1930s Depression and U.S. Economic Growth (Yale Series in Economic and Financial History)(2012)). Perhaps the latter is one of the reasons for the former's lasting so long.

Individual workers can lose their jobs due to technological innovations, but the overall unemployment rate is determined by government policies. There was lots of technological change in the 1920s, and yet unemployment was low and falling. In the 1930s a combination of tight money and artificial wage increases (NIRA, Wagner Act, minimum wage, etc.) led to much higher unemployment.

Sound macro policies lead to full employment, no matter how fast technology is changing.

I am not an economist, so I may have problems reading the assumptions between the lines. Should I assume that when you say "the overall unemployment rate is determined by government policies" you mean relative to some minimum? That might make it OK for me, if the minimum could be used to represent the limits to mental and behavioral agility in the workforce. Similarly "full employment" would not be drawing every able-bodied adult out of the house, but would represent something more subtle, perhaps "those who should work, and are sensible enough to understand it." That is, it's a normative recommendation for adult behavior combined with a market theory of how to achieve it. By its nature somewhat aspirational.

John: "I am not an economist"

....not a problem. Apparently nothing is settled in economics and everything is debatable. formal training in the discipline seems to sharply increase uncertainty, doubt, conjecture, and verbal output of same.

as you can see here, the issue of technological unemployment is still a mystery to professional economists.... after two centuries of study.

Try explaining this to all the textile and shoe workers in New England that lost their good jobs and never had a good job again.

Of course their sons and grandsons now have great jobs in finance or high tech, but that only demonstrates that there is a very long lag between old industries dying and now industries emerging to create new employment.

Looking back now an economist can explain the high unemployment in New England from 1930 to 1960 as just friction in the system,
but that still does not get the displaced workers good jobs.

But he's not saying that they would have found equally 'good jobs' -- just jobs. When you have low-skilled workers earning high wages and they lose those jobs, no, they're not likely to find other low-skilled/high-wage jobs (especially not right where they already happen to live). Those occupational conditions are always pretty rare and not sustainable in the long term (eventually competitors come in to the market who don't overpay for unskilled labor).

Actually, he says below "other jobs are created elsewhere."

Given that we know a contraction in shoemaking cannot cause an expansion in air travel, it must be a faith that for every such decline, there will be an expansion, very close to 1:1.

I certainly get, endorse, celebrate progress but it seems lumpier to me than a series of 1:1 pairings.

I would think the ratio would almost always have to be greater than 1, because there's always an efficiency gain in the overall system from any technological improvement. Even if there isn't a gain in jobs elsewhere there is a reduction in price for goods, which means that you don't need to earn as much to have the same living standards. Obviously there will be some individual who will lose, but a net gain is almost axiomatic.

A simple thought experiment: A mechanical man is invented which can do 80% of general human labor. It's cost is $10K/yr. How do we translate that productivity gain to a _human_ jobs boom?

@John: In such a world, labor would mostly be covered by the robots. We'd have to do the minimum income thing, and for most folks 'jobs' would be things like teaching, writing poetry, making crafts, etc. Labor would no longer be necessary for most.

Gee Scott, when I look at the data for the 1930s I see employment and wages both falling sharply from 1930 to 1933 before FDR$ and all those things you cite took effect.

After 1933 when those policies took effect both wages and employment rose. during the 1930 the correlation betweeen wage growth and employment growth -- using annual data was 0.8.

% ch % ch
ahe employment

1930 -2.473498233 -4.380288701
1931 -6.702898551 -6.513817532
1932 -13.39805825 -7.909454061
1933 -0.896860987 0.0368053
1934 20.36199095 5.933985073
1935 3.383458647 3.381294964
1936 1.090909091 5.557555252
1937 12.23021583 4.726181545
1938 0.480769231 -4.180776244
1939 0.956937799 3.615604186
1940 4.423380727 3.896103896

What is your source of data that shows just supports your conclusions?

Thanks, Scott. The absence of the macroeconomy from this assessment also struck me. How much a firm increases output (and thus, possibly, employment) depends on its demand curve, but also on the state of the economy...

Erm, that's partial equilibrium. Sure, employment in concrete went down. But presumably the price of concrete fell, making some types of capital investment cheaper, and raising employment in other industries.

John, Yes, there is some sort of natural unemployment rate, even with good government policies. Bad government policies can affect unemployment in one of two ways. It can raise the natural rate of unemployment, say with a very high minimum wage rate---$20/hour, or it can cause the actual unemployment rate to be higher than the natural rate, perhaps due to bad monetary policy. In my view technological progress does not significantly raise the natural rate of unemployment, and it does not cause unemployment to be higher than the natural rate.

Spencer, Yes, they may have lost their jobs for the reason you suggest. But that job loss did not significantly raise the overall unemployment rate, as other jobs were created elsewhere. (I say "significantly" as obviously no one can say it has precisely zero effect.) It's only when bad government policy either drive us into recession, or artificially try to raise wage rates, that the overall unemployment rate rises. Creative destruction has worked very effectively throughout American history, when it's been allowed to work.

Two just-so stories to counter the individual cases of misfortune we might identify?

Probably works better from a lectern, before undergraduates.

I worked with a guy who has to declare personal bankruptcy and lost his house in Fort McMurray Alberta. A slump in the market, his job disappeared and he couldn't sell. I knew another family who owned property in North dakota, couldn't sell it, was lucky to find someone who would live there and rent at the property tax rate.

Things go up and down. But both of those people found work and opportunity elsewhere. They moved, applied their skills and did fine.

Both those places became a magnet for young people where they could get a job based on their desire to work or just after finishing their education. They made great wages, the opportunities were remarkable.

I also lived in a place where government kept unproductive enterprises alive with tax revenues, had generous safety net programs. It was characterized by very high youth unemployment, high unemployment generally, a vigorous underground economy and the migration of young people to places with more opportunity.

If you are a robot/AI optimist, you are kind of led to the idea of a minimum guaranteed income. Nothing else works. Anything else produces a large underclass.

(I am not such a robot/AI optimist that I think a minimum guaranteed income is necessary in the near future, but certainly a safety net is necessary.)

John, consider ceasing your willing blindness to benefits created by change. Isn't it purely your own ideology that keeps you from seeing New Englanders as better off than it was at any time; at least any time more than 20 years ago.

Some New Englanders are probably still opposed to electrification, God bless them.

I have benefited greatly from technological change and creative destruction, its just that I can see others have not. Once we acknowledge that, we can move on to appropriate policy.

Perhaps some don't want to acknowledge technological unemployment because they'd rather stop at the discussion phase.

John -- You say that you can see that others have not benefited from technological change. Granted. But that is just the first-order effect. It's like VTProf says. If the lower electricity costs are passed on to consumers, then those consumers have more dollars with which to purchase other goods and services, creating jobs in those industries and thereby leaving overall unemployment pretty much the same. If consumers save those dollars instead of spending them, that will presumably lower the cost of capital/interest rates, with -- eventually -- similar effects.

Well, looking back now we do have little sympathy for a Luddite who hadn't eaten in days. In the long run they were all dead.

And no, no one has given a mechanism by which any arbitrary technological change is always balanced by a labor increase "elsewhere."

Technology is not a uniform substance any more than labor is a uniform substance.

John -- I don't think metaphysics is any help here. Perhaps you will share with us why it matters if labor or technology are "uniform substances."

Nor do I understand what you mean by "arbitrary" technological change ... as contrasted with what?

Our society has a safety net in place to address some of the painful consequences of technological unemployment. Granted, you may think it is inadequate. But it provides a munificent income compared to what even employed workers would be earning today if we had "arbitrarily" slammed on the brakes and blocked technological change 10, 20 or 50 years ago.

If you close your eyes you won't see the mechanism.

See the mechanical man thought experiment above.

BTW, "we have a safety net" is agreement that yes, we do need societal answers because no, the market does not always immediately and automatically line up work for displaced workers.

has anyone ever doubted the possibility of technological unemployment? I thought everybody has heard of Luddites.

They have, but since the Luddists are insane, their ideas and beliefs must be insane. Ergo, no possibility of technological unemployment.

What's that phrase in economy again? Ah right..."there's always such thing as a free lunch."

It seems utterly uncontroversial that technological change could reduce the before tax labor share of income. Whether that resulted in unemployment would depend on macroeconomic policy. And whether it reduced the after tax labor share of income would depend on tax and expenditure policies.

I found that George Clooney in the last scene of "O Brother Where Art Thou" made a strong case that the New Deal's plan to "hydroelectric up the whole durn state" meant that "Yes, sir, the South is gonna to change. Everything is going to be put on electricity and run on a paying basis."

Perhaps what is needed is an educational system that can provide individuals with a broad array of adaptable skills.
Technological unemployment seems to me to be somewhat a consequence of over-specialization.

We have two kinds of workers mostly: highly specialized college-educated workers, and unskilled laborers who learn a single skill on the job. We don't turn out a lot of jack-of-all-trades types who can adapt easily to changes in the market.

Jacks-of-all-trades need relatively high IQ's. Really, the problem of technological unemployment does stem from the fact that workers specialize early in their careers. This comports with human nature (brain is more flexible in youth) and is generally good because increased division of labor promotes productivity, but then technological shocks send some workers "back to the starting line" specialization-wise-- even if they can learn a new specialty they have their earnings reset at an inconvenient time, they lose the chance to ever reach the top of their field because (in the new field) they have "started too late," have to compete with more energetic youngsters, and will age out prematurely.

There's a good argument for social insurance as technological unemployment insurance. To the sufferer, technological unemployment feels a lot like any other kind of work-impairing disability. Suppose a pianist loses a hand-- maybe she can build a new career as an insurance adjuster, but maybe not, and it might never pay so well. We agree that the shock of physical injury to a career should be insurable; our pianist will get at least a "partial disability" check. We really should treat technological shocks analogously-- I would even be willing to pay premiums for private "technological disability insurance."

What works better as insurance though? Specializing early in one's career creates risk. Being a generalist reduces risk at the cost of potentially lower productivity gains from specializaton early on. So generalized knowledge acts as personal insurance against technological unemployment. Isn't some productivity later, after the technological change, better than zero productivity later and living on disability? Would you rather have people receiving a welfare check or being able to change jobs readily, even if it reduces the gains of specialization? I would think that the more rapidly technological change occurs, the fewer years that the highly-specialized will likely be working at peak productivity, and thus, the more advantageous it would be to generalize more so they can continue being somewhat productive for a longer period of time. Specialization does create gains, but if you can only achieve those gains for 5-10 years, maybe those gains aren't large enough to justify being rendered obsolete for the rest of one's life.

"Technological unemployment seems to me to be somewhat a consequence of over-specialization."

Because that's what the business community, particularly corporate big business, hires for. They want plug-and-play, specialized employees who can start doing the job day one with minimal to no training. Businesses don't want to invest in growing their talent. They justify this by blaming a mercenary workforce, not realizing that their preference not to train exacerbates the very mercenary mindset they decry.

On top of that, human resources recruits primarily by keyword matching against a resume database. This steers the process toward credentialism and the accordant, false assumption that certification in a particular skill--e.g. coding in a certain language, project management--indicates: a) actual competency in that skill as opposed to competency in the certification training course/materials provided by the certifying organization; and b) broader competency in other and/or more general skills--communication, critical thinking, social perception, adaptability, etc.

The whole hiring framework as currently constructed disincentivizes--and, I think, in some cases precludes--hiring generalist, jack-of-all-trades types. To be fair, capturing the sort of broad competencies that generalists possess in a resume, much less a searchable keyword cloud, can be difficult. Recognizing those competencies requires communication and reading between the lines of someone's resume...which requires actually reading someone's resume. Which HR doesn't do anymore.

Unfucking this situation is going to require some combination of employers once again investing greater time and energy in recruitment and training and the educational system developing better metrics for general skills and competencies.

I certainly don't have any sympathies for corporations that blame the workforce, or the government, for not providing them with enough plug-and-play trained employees.
It also seems to me they are only harming themselves by relying on incompetent hiring practices and lazy HR.
Workers who want to remain relevant in a rapidly changing environment would do well to develop general skills, and play the system to get hired if that's what it takes. Corporations can't blame employees for doing what's in their best long-term interest. (Example: I can write several different versions of my resume/CV tuned to particular fields, with the appropriate buzzwords dropped depending on which job I'm applying to.)

This is a great comment, 100% correct, and it's astounding how few people understand the fact.

Steve, in the part of Kentucky my family is from until the 1930s people lived in the valley (hollows) and used the slate bottom creeks for roads -- you could drive a wagon up and down the shallow creeks.

In the 1930s the government built paved roads and electric lines along the ridge lines. Soon almost everybody moved from the valleys to the top of the hills.

Comments for this post are closed