Scott Sumner  

Automation and trade

PRINT
Trump's Picks versus Reagan's... Henderson on Scott Pruitt...

I've been arguing that the problems often attributed to trade are more likely due to automation. Indeed in a sense both trends are two sides of the same coin --- creative destruction. It might help to make this point with a simple example.

Suppose you have an air conditioner factory in Indiana, which employs 2000 workers. Now consider two possible corporate decisions:

1. Move 1000 of the 2000 jobs to Mexico, and import much of the resulting production.

2. Automate the factory so that the same output can be produced with 1000 workers instead of 2000.

In practice, both of those hypotheticals occur all the time, although the automation example is far more important---indeed it's the only one happening at the global level, as obviously Earth doesn't trade with other planets.

Now let's think of secondary effects, starting with the export of jobs and import of AC units. An increase in American imports tend to reduce the value of the dollar, which boosts exports. Will the rise in exports replace the jobs lost to imports? Probably not. Partly because trade may not be "balanced" (although if measured properly the US actually does have roughly balanced trade.) And partly because a given value of exports may require fewer workers than that same value of imports.

If 1000 jobs are shipped to Mexico, then we might end up with an extra 200 jobs in exporting high tech goods like Apple phones, Facebook ads and Boeing airplanes, and another 800 jobs in areas like home building, restaurants and health care. Critics of trade will point out that the 1000 new jobs might not go to the very same 1000 workers who lost their jobs at the AC plant, and that concern is legitimate. But it holds equally well for automation.

If 1000 jobs are automated, there will be some jobs created in building robots, and other means of automation. But almost certainly not 1000 jobs, otherwise there would be no cost saving for automation. What would be the point? More likely, you might get another 200 jobs building robots and other forms of automation, and another 800 jobs in areas like homebuilding, restaurants, and health care. The effects are essentially identical to trade.

You might argue that the extra 800 jobs in the non-traded sector assumes proper monetary policy. I think it's more sensible to think in terms of assuming non-perverse monetary policy, but in any case, the assumption applies equal well for trade and automation.

If you are opposed to free trade but not opposed to automation, you probably wrongly think trade results in a net loss of jobs and automation does not. It doesn't. And almost no one seems to oppose automation. I conclude that arguments against trade are bogus, based on ignorance. Trump and his apologists in the media have simply not thought through the issues.

Trade and automation are largely responsible in the rising living standards for average Americans over the past 50 years. (And yes they are rising; another myth is that they are not.) As I drive down the highway, 95% of the cars I pass are luxury cars, or more precisely cars that would have been viewed as luxury cars in 1966. Automation has made that possible. In another 50 years, lower middle class Americans will drive cars that are better than BMW 700 series cars, or perhaps I should say "riding in" cars that good--I have no idea if they'll be driving them.

Screen Shot 2016-12-11 at 1.41.31 PM.png




COMMENTS (33 to date)
E. Harding writes:

I don't think it's about jobs; I think it's about output. I think protectionists believe it's better to have stuff made here, and, thus, contribute to America's GDP, than made overseas, and, thus, contribute to foreign countries' GDP.

John writes:
Trump and his apologists in the media have simply not thought through the issues.

I'm sure Trump has economic advisors who have filled him in on this. But to win the election he knew he was dealing with people who live in the delusion that there's a major distinction there. He was happy to align with their worldview and, in so doing, collect their votes.

I would actually say that Trump's primary talent as a politician is his ability to create common ground with whatever support base he's courting without being encumbered by ideological baggage of his own.

Ak Mike writes:

I must be missing something. If the 200 new jobs in export result in 800 new support jobs in restaurants, etc. - then why don't the 1,000 lost air conditioner factory jobs result in an additional loss of 4,000 support jobs in restaurants, etc.?

Jeff G. writes:

E. Harding: if you simply replace jobs with output, Scott's points hold up equally well.

Pajser writes:

I think Harding answered that.

what happens if all American capitalists tomorrow move their means of production in Mexico, India etc? Poverty in USA, all Americans die and government cannot do anything. If it tries to tax capitalists, they move to Mexico too.

What happens if all capitalists tomorrow replace their workers with robots? Complete unemployment but GDP is the same or even greater; the government can redistribute the profit to prevent famine, and see what to do next.

I guess that the same effect will be dominant if the change happens on the margin.

MikeP writes:

I don't think it's about jobs; I think it's about output.

It is entirely about jobs, not GDP. The protectionists who elected Trump are not capable of more subtlety than that.

My favorite thought experiment on these issues is the magic factory in the middle of the Pacific. Anyone in the US who wants to can take a container ship to this factory and fill it up with whatever he wants to sell in the US. Optionally, that person can fill the containers on the way out with any goods he wants and dump them into the Pacific next to the factory.

Is the US economy better off or worse off because of the existence of the magic factory? Is the US economy better off or worse off sending the containers out empty rather than full?

Now call that magic factory "China". And while China doesn't want completely empty containers, it is okay with them being only half full. Is the US economy better off or worse off because of the existence of China?

jc writes:

Humans are not wired to feel the same degree of tribal rivalry with machines.

Jonathan K writes:

I agree that automation contributes to the continued increase in living standards because it allows us to purchase items at a relatively cheap price. I believe that in the future this may be reversed due to the new policies that the Government may try to implement. The campaign to bring jobs home may result in a price hike and less of an incentive to produce. With the rapid growth of automation even if jobs are returned home there is no guarantee they be enough.

What do you guys think??

Jim Glass writes:

And almost no one seems to oppose automation.

Oh, you are not paying attention. E,g. over at Reddit the econ and political pages are full of it and of worse, that Elon Musk's coming Artificial Intelligence singularity is going to destroy employment across all industries en masse.

Of course the alarmists never see the bulk of the benefits of increased productivity. They see more profits for the owners of AI, so Elon gets richer and 60% of the work force becomes unemployed. And with these masses of unemployed unable to buy anything the conventional economy implodes into some sort of high-tech feudalism with Elon and the lords of AI residing in the High Castle. QED.

To help them grasp that they might be missing something, a few times I've posited this example: Elon waves his magik AI wand to convert water into an environmentally friendly perfect substitute for fossil fuels (oil, coal, gas) available in any quantity and thus at negligible price. BLS tells us that there are 250,000 workers in the energy industry to lose their jobs as a result. At an average $50,000 pay per job that's a quarter million people losing $12.5 billion in wages to Elon's AI. The nightmare come true.

OTOH it also creates a real savings to the current consumers of fossil fuels (businesses and individuals) of a trillion dollars annually, every year going forward. Is a trillion dollars annually of freed up spendable funds going to result in any increase in new investment, innovation, consumption, growth in other industries? Free energy! A bad thing?

A crude example responding to a crude argument. Yet it's actually changed a couple minds -- and in Internet arguments, how often does one see that!?

(And yes, it is the same with trade, which is as has been said a magic technology for converting wheat into oil when wheat costs you less than drilling your own -- and some people at reddit have actually been able to see that too.)

Hey, Elon, bring it on!

Jim Glass writes:

For those who don't want to click through links given above to reach the source story for the automation discussion, from Forbes:
~~~~

93% of Investors Say AI Will Destroy Jobs, Governments Not Prepared

224 global investors with more than $100 billion of investable capital say that governments are not prepared for the coming of artificial intelligence and the massive job loss that, they say, will result.

Web Summit, the massive technology conference in Lisbon, Portugal, polled 224 investors face-to-face at Venture, one of the subconferences within the Web Summit framework. 93% of them said that governments are not ready.

A smaller percentage, only 53%, said that it is “inevitable that Artificial Intelligence will destroy millions of jobs."

Arguably, we're already seeing this happen. Self-driving cars and trucks are closer and closer every day. Earlier this year, iPhone manufacturer Foxconn replaced 60,000 factory workers with robots....
~~~~~~~~~

These ain't hicks from Rube Town who just got duped by Trump.

Todd Kreider writes:

"Trade and automation are largely responsible in the rising living standards for average Americans over the past 50 years."

I don't see how this can be correct, and I'd like to see how even a rough estimate was attained to support this claim. In 1958, Robert Solow estimated that 80% of the U.S.'s long term growth was due to increases in productivity, but automation accounts for only a small percent of that.

jonk k writes:

@Todd i want to try and understand your reasoning for dismissing the theory that Automation and trade has not driven this economy. In 1958 automation was nowhere near the level it is today even though i understand it is hard to compare years without the right statistics

Karl M writes:

People picketing mcdonalds over automated cashiers would tend to suggest that people are starting to feel the same level of enmity towards automation as they have towards trade in the past.

MikeP writes:

In 1958, Robert Solow estimated that 80% of the U.S.'s long term growth was due to increases in productivity, but automation accounts for only a small percent of that.

Productivity increases in 1958 may not have been due to automation per se, but they were certainly due to industrial leverage of another kind. From mechanical looms to steam shovels to robotics, the greatest productivity increases come when humans can utilize non-human tools to effect greater production than they could physically effect without them.

Todd Kreider writes:

Scott seems to be equating automation with productivity and this is incorrect. Take a simple hypothetical: cars improved a lot from 1965 to 2015, his 50 year period. There was more automation but that impact has been small relative to improvements in humans designing better cars in many areas as physics and engineering marched on over the past 50 years.


Maurizio writes:

@jimglass

"And with these masses of unemployed unable to buy anything "

But does this model hold? I mean, if they are unable to buy anything, who will the "AI lords" sell their stuff to?

pyroseed13 writes:

Sorry Scott but you continue to be an obscurantist on this point. The evidence overwhelmingly shows that most of the jobs lost in manufacturing from 2000-2008 were due to trade with China, not automation. And as you said yourself, the new jobs are not going to the same people. How can you call those people when ignorant if their jobs our outsourced and then they end up in new jobs at lower wages? I'm not endorsing Trump's proposed solutions to this problem, which are indeed bad, but to continue to carry on as if this is not an issue at all is not going to help economists improve their reputation with the public.

jonk k writes:

@Todd i want to try and understand your reasoning for dismissing the theory that Automation and trade has not driven this economy. In 1958 automation was nowhere near the level it is today even though i understand it is hard to compare years without the right statistics

Hazel Meade writes:

Both groups opposing automation and trade are making the same fundamental mistake - the idea that more work equals more prosperity. It is more obvious in the case of automation, since it is clearly better to produce good more efficiently, so as to enable lower price. But it is equally true in the case of trade - if it is cheaper to pay someone else to make something for you than to make it yourself than you should do so.

The problem is that it's not individual workers automating or outsourcing their own jobs, so they don't see the direct benefits of it. If you got paid the same amount, but spend 50% of your paycheck hiring a Mexican to do your job and then took a second job at the same wage - that would be a 50% pay raise for you. Workers don't own their jobs though so they can't contractually do that. They get laid off and then you have a class of losers that is directly harmed instead of benefitting. Everyone else benefits, but not the people whose labor is being replaced.

Speculatively, perhaps the solution would be to pay off the workers with some fraction of the gains from trade to compensate them for the loss of employment income. One could buy them off with shares of company stock, for example, from which they could receive dividends. If company profits go up as a result of automation or relocating a factory then the workers get a share of the gains made from the improvement in efficiency, as if they had automated or outsourced their own job.

Benjamin R Kennedy writes:

"And almost no one seems to oppose automation. I conclude that arguments against trade are bogus, based on ignorance."

There is a pretty big distinction between trade and technology, and it goes to how human beings process fairness. Humans accept the bargain of creative destruction with technology advances because it's man vs machine - there are no fairness implications. Sorry elevator operators! Humans even accept the bargain of comparative advantage - we should import oranges from Florida rather than Greenland because, duh. I think we're all happy we are not living in the 1600s.

Carrier-style job losses are different. It isn't a change due to technology, and there is nothing in the air of Mexico that makes it more conducive to assembling cars. It is regulatory and cultural arbitrage where for various reasons it is efficient to yank the job from one person and simply give it to another. It's a tenth of a step away from nepotism, another practice that we consider rather unfair. This is obvious when you see the wages of the workers. Foreign worker wages are set in a market consisting of other foreign workers, not set against the wages of people in Indiana.

The issue with trade is the efficiency gains are paid for by local workers. If we want to have fancy environmental regulations that Chinese steel manufactures want to disregard, why should local workers be the ones that pay for it by losing their jobs? It the same reason we oppose minimum wage - we're pricing our own industries out of the market. See "steel, Pennsylvania".

Hazel Meade writes:

@Jim Glass

Elon gets richer and 60% of the work force becomes unemployed. And with these masses of unemployed unable to buy anything the conventional economy implodes into some sort of high-tech feudalism

Yes. I always like to answer this sort of scenario by pointing out that if nobody can buy anything then Elon Musk won't be selling any products. The price HAS to fall to a point where people can afford it, otherwise the product won't sell. In other words, if magic free energy is too expensive for people to buy, then it's not going to take over the market and drive out oil and gas.

If you think about it, it almost tautologically HAS to be that case that the price drop is going to be larger than the loss of purchasing power otherwise you end up with logical absurdities like Elon Musk becoming king of an economy in which nobody can afford to buy anything from him.

Scott Sumner writes:

Harding, You said"

"I don't think it's about jobs; I think it's about output. I think protectionists believe it's better to have stuff made here, and, thus, contribute to America's GDP, than made overseas, and, thus, contribute to foreign countries' GDP."

Yes, it may be about that too, but of course that perception is incorrect. Trade doesn't reduce GDP, it increases it.

AK Mike, They are not "support jobs" in the direct sense, just jobs that occur as the natural result of creative destruction. If you assume AD is unchanged, there is no "multiplier effect" if that's what you are thinking about.

jc, That's a big part of it.

Todd, When I use the term "automation" I am referring to productivity.

pyroseed13, You said:

"How can you call those people when ignorant if their jobs our outsourced and then they end up in new jobs at lower wages?"

That's not my claim. I'd be frustrated too if I were in their shoes.

As far as your claim about job loss since 2000, it is not correct, AFAIK. The studies cited most often (Autor, et al) are cross sectional studies, which tell us nothing about aggregate job loss. Check out my post on the steel industry, similar posts could be done for autos, and many other industries.

Benjamin, You correctly describe many peoples' perceptions, but these perceptions are not correct. The issues in trade and automation are the same, and we need to educate the public of that fact. I predict that within 30 years my point will be widely understood. As manufacturing jobs disappear all over the world, including China, there will be no more scapegoats to blame.

Benjamin R Kennedy writes:

"Benjamin, You correctly describe many peoples' perceptions, but these perceptions are not correct."

Scott, you can't really go a person and say "your perception of fairness is incorrect". Part of my disillusionment with pure free market outcomes occurs when I take a step back and ask "is it fair?" And sometimes, the answer is "no". Granted I am not a moral realist, but I can still feel bad for my fellow citizens who are clearly suffering. Policies that have different impacts in people across nations involve very fundamental questions including "what is the purpose of countries". An analogy is firms - within firms, free market economics often don't apply - yet firms clearly exist to benefit the employees. I think there can be similar dynamics with countries.

MikeP writes:

Policies that have different impacts in people across nations involve very fundamental questions including "what is the purpose of countries".

The purpose of countries, as was once self-evident, is to secure unalienable rights -- including rights of association, contract, and trade. More pragmatically, the purpose of countries is to prevent the imposition of other countries' worse laws and enforcement on a population.

yet firms clearly exist to benefit the employees.

Citation needed.

Firms exist to benefit the owners by serving the consumers. Benefiting employees is a second-order consideration.

Floccina writes:

Many people have a view that most people that do not know, especially foreigners are out to get them. So they fear that trade with foreigners will lead to inability of the US to produce goods. They believe that once that happens the foreigners will then refuse to sell to sell to us driving us into poverty even at their own hurt. Such is anti-foreign bias.

pyroseed13 writes:

To Benjamin Kennedy's point, it is not clear why the economists' perception of what is a fair and moral outcome is necessarily the right one. Sure, I could argue that as a long as the "aggregate gains" from trade with China are positive then the outcome is just. But someone could just as easily argue that because those gains are not evenly distributed, and impose large costs on a certain segment of the population, that this policy is not just.

Scott, you argue in your China post that that employment in the steel industry and other manufacturing sectors is declining because of increases in productivity. The problem with this is that once you remove the computer industry from manufacturing, productivity has been basically flat since the late 90s for the U.S. See here:

https://www.washingtonpost.com/posteverything/wp/2016/10/18/dont-blame-the-robots-an-interview-on-manufacturing-automation-and-globalization-with-susan-houseman/

I'm confused by your "aggregate loss" point. Are there certain manufacturing sectors that have been particularly affected by automation but not trade?

Hazel Meade writes:

@pyroseed13:
I think what many people think is unfair is the counterintuitive result that the people whose labor is being saved by the labor saving efficiency improvements are not the ones reaping the benefits.
If everyone was a self-employed independent contractor, they could would celebrate trade - their jobs just got easier, they can make more money now. But because they are employees they get none of the direct monetary benefits of the improved efficiency. They get the benefits only very indirectly if at all.

Todd Kreider writes:

"Todd, When I use the term "automation" I am referring to productivity."

OK Scott, thanks for clearing that up, but I think you then defined automation in an unusual way since automation is normally considered a subset of productivity.

A common example is a company that reorganizes some aspect of how their employees interact that raises productivity without any computer/machine change. Another might be further training.

(I realize you know this but others reading might not.)

Peter Gerdes writes:

Scott,

While the two changes do accomplish exactly the same result (jobs disappear from the US while the goods appear in the US for cheaper than they did before) I think we can explain the difference in attitudes fairly simply.

Just postulate that (as seems psychologically and evolutionary plausible) rather than caring about absolute welfare/gdp people primarily focus on relative measures. In other words what people care about is whether we are doing better than china (are we the stronger tribe) or mexico not our overall utility.

Thus, while failing to automate is a pure loss of relative status for the US (our factories don't produce as much stuff as they would if we automated) exporting jobs may actually be a loss in our relative status if it sufficiently improves the economic welfare of the other country.

Morally this is horrific. We shouldn't be stamping others down but I think the issue is sufficiently complex people manage to think around that part of the analysis.

Scott Sumner writes:

Benjamin, I didn't mean their sense of fairness was wrong, but rather they were wrong about the facts of the case. They see trade as being different from automation. I know this because I interact with lots of people, and many are shocked by my claim. Many don't believe it.

pyroseed13, You asked:

"Are there certain manufacturing sectors that have been particularly affected by automation but not trade?"

Yes, look at my recent Econlog post on the steel industry, a classic example.

Jim Glass writes:

"While the two changes do accomplish exactly the same result (jobs disappear from the US while the goods appear in the US for cheaper..."

What makes anyone assume "jobs disappear" as a result of automation? In fact, how can anyone assume that when looking at the real world?

Western economic history since the start of the Industrial Revolution has been machines doing ever more things ever more and more better than people -- continuously ever more automation -- and as a result, in addition to gaining a previously unimaginable cornucopia of ever better and cheaper goods, we also have an unemployment rate of 5%.

Henry Ford automated the auto production line to reduce the labor involved per car by 90% -- and employment in car production exploded for generations ... ATM machines and online banking automated far more than any bank teller could do before them -- and now there are more bank tellers than ever ... airlines install auto-pilots with ever greater capability to replace crew members and fly planes by themselves, and the news is full of stories about pilot shortages.

Really, if you can't think right off the top of your head of a dozen other industries where increasing automation is going on with increasing employment you aren't even trying.

"Technology has created more jobs than it has destroyed, says 140 years of data."

And, yes, trade does "accomplish exactly the same result".

Jenn Landreth writes:

I believe the Luddite fallacy must be the central focus when analyzing automation and job loss, which proves that the advancement of technology in fact does not lead to an overall unemployment in the economy. Instead, it just changes the way jobs are distributed in the economy.

arqiduka writes:

One could agree that automation and trade both have an impact on jobs, and still go after trade alone because it is far easier to put up a tariff than to conduct a witch-hunt and burn robots. So, of course trade and automation are both instances of creative destruction, but the effects of one are far easier to counter, at least up to a point (although incentives for automation would increase under a protectionist regime).

In a sense, that we have protectionist sentiments spreading and not luddite sentiments is a testimony that people still care about leaving as free a market as possible, while also taking care of the employment issue.

Comments for this entry have been closed
Return to top