Economics, Pethokoukis, U.S. Economy

Will robots terminate the US middle class? A Q&A with Tyler Cowen, author of ‘Average is Over’

Image Credit: Shutterstock

Image Credit: Shutterstock

Flash fact: While most of the jobs lost during the Great Recession were mid-wage occupations, most of the jobs added in the recovery have been low-wage jobs. Many of those disappearing middle-level jobs are what economists call “routine, manual tasks” that can be easily automated. If economist Tyler Cowen is right, that trend is merely a taste of things to come.

In my Ricochet Money & Politics podcast, I chat with Cowen about his new book, “Average Is Over: Powering America beyond the Age of the Great Stagnation.” It’s a follow-up to his popular e-book, “The Great Stagnation: How America Ate All the Low-Hanging Fruit of Modern History, Got Sick, and Will (Eventually) Feel Better.”

Cowen is the Holbert L. Harris professor of economics at George Mason University and co-author of the popular blog Marginal Revolution. He also writes the “Economic Scene” column for The New York Times. In addition, he is cofounder of the online education venture Marginal Revolution University.

Here are the edited highlights of my chat with Cowen:

Your new book is “Average Is Over: Powering America beyond the Age of the Great Stagnation.”  It’s a follow-up to your popular e-book, “The Great Stagnation: How America Ate All the Low-Hanging Fruit of Modern History, Got Sick, and Will (Eventually) Feel Better.” 

In “The Great Stagnation” you made the case that U.S. was stuck on a growth plateau or on an innovation-productivity plateau after picking those low-hanging fruit that you mention in the title: free land, educating all the smart uneducated kids, huge technological innovations such as public sanitation, the internal combustion engine. 

But now, in the new book “Average Is Over,” innovation may be ready to re-accelerate, but with mixed results.  As you write:

The basic look of our lives in the surrounding physical environment has not been revolutionized all that much in 40 or 50 years.  That’s about to change.  One day soon, we’ll look back and see that we have produced two nations– a fantastically successful nation working in a technologically dynamic sector and everything else.

So what you do in the book is you sketch a sharply bifurcated American coming decades, one that looks like a mash up maybe between Downton Abbey and Elysium based on these two principles:  First, machine intelligence can replace human labor. And two, machine intelligence can augment the value of other human labor for many individuals, which is where we get your famous 15 percent will do great, 85 percent not so much. 

Is there evidence that this phenomenon is already happening. Do we see signs of it already in the current slow recovery? 

I think we see signs in the slow recovery that most sectors are still in the great stagnation. But we have already had one dynamic sector, incredibly dynamic, and that’s information technology. So we already see that happening. I just tried to play that out a bit. So imagine that we get driverless cars and get Watson doing medical diagnosis.  We get more smart software and we automate even a lot more manufacturing jobs. That to me is quite plausible for the next 20 years.  And that’s a scenario the book is trying to spell out.

Essentially what you’re talking about is a sort of thinning out of the middle class. 

Well, since 1999, median income is down about 9 percent, which is really quite a bit.  Even since the beginning of the recovery, median income is down over 5 percent.  I find those both very striking.  There’s also a lot of evidence from the work of David Autor and others that in areas where you have jobs being automated inequality is tending to rise to a greater degree.  So I think in the numbers already there’s evidence that we’ve seen these trends start.

But are you also saying that we’re going to get faster economic growth, better productivity statistics in the future than we have in the past? There was this big productivity slowdown in the early ’70s.  Productivity picked back up in the ’90s, strong for parts of the 2000s, then dropped again.  So what is going to happen with both productivity and innovation going forward and how those gains are distributed?  

Keep in mind there’re several different productivity statistics and they don’t always move in the same direction.  So if you look at per hour labor productivity, you can get that up just by firing people essentially.  We’ve also had some of our productivity statistics go up because we’ve outsourced more work to other countries.  That’s a kind of productivity gain, doing things more cheaply, but I distinguish it from major technological advances.

So I think since 1973 we’ve had a fair amount of productivity gains of the kind lowering costs, but we’ve done much less to make American workers more productive.  What I think the future will look like is a world where no single productivity statistic captures very well what’s happening.  You could put it this way; the great stagnation will end for some people, but not for everyone.  And just ask yourself the question: are you more productive working with the computer or is maybe the computer better off without you?

So those folks who are good at working with computers – and it’s broader than just science, tech, math, engineering, those kind of professions — that for those folks the economy will be better than it’s ever been, right? 

Much better.  And you already see this.  There are urban enclaves in the United States where wealth is just fantastically high, real estate prices are booming, social indicators are very good.  This is not a science fiction fantasy.  For part of this country this is already true.  It’s going to be all the more true in the future.

And we’re not just talking about the “1 percent.” In the book, and I realize this is somewhat of a ballpark number, you talk about the 15 percent. 

Much bigger than the 1 percent.  I think that’s an important point, not some sliver of billionaires.  There’ll still be billionaires of course, but you’ll really have large numbers of Americans with lives as good and as happy as those which current millionaires have.

The number, of course, is going to depend on your time horizon, but just start by asking what percent of people get a decent higher education.  And I think in the time horizon I’m looking at, which are, say, 20 years from now, those who are really high earners, they’re going to exceed that percentage.  You know, there’ll be some people who just leave high school and then they become tremendous yoga instructors.  And they’re yoga instructors for billionaires and they may earn a lot.  But I think when you look at just how well educated are we, that’s a pretty reasonable number.

If you look at Singapore today, Singapore measures itself as having about 17 percent millionaires already.  So in essence, you know, we’re running to catch up to Singapore.  We have about 4 percent.  So I think it’s a ways to go.  But when you look at how many workers are unskilled or didn’t finish high school, which is, you know, 30 percent right there, I think it’s really hard to think in the short run, it will be, you know, 60 percent of workers living like millionaires, maybe in a more utopian much longer run.  The book mentions that.  But in the short run, it’s still a minority.

So the top end sort of the income bracket will be people who can deal with machines.  And then, you’ll have machines — from robots to software taking all sort of those –  taking the middle-income group of jobs.  And then the lower skill jobs, that’s your other 85 percent of the carbon-based life form population.  But those jobs don’t sound really like the kind of jobs I’d want my kids to have.   

Well, keep in mind, those people – a lot of them will have very happy lives.  They’re going to get a lot of free things, for instance, will enjoy a lot of trickle down benefits of tech.  There’ll be much more free material on the Internet.  So you could think of that as say 85-80 percent of the population living like today’s poor.  A better picture of that is to think of high human capital bohemians who don’t necessarily have great jobs, but do interesting things.  They do take some financial risk.  I’m not saying this is going to work out well for everyone.

Think of some parts of Brooklyn, but it’s not as if we’re simply having most of the United States look like some kind of ghetto.

And our norms for status will change also.  I think that’s very important.  So what appears to be a menial service sector job, a lot of those jobs are quite fun.  They’re more fun than the old manufacturing jobs we used to have, which were also very dangerous and physically exhausting.  Those jobs will be very mixed in quality.  Again, I don’t want to present a Pollyannaish view, but the notion that somehow everyone is simply flipping burgers at McDonald’s is not what is going to be.  Some people will teach chess to the children of rich people.  Other people, you know, will be tutors.  Those can be very challenging, very interesting jobs.

So what will the labor market will look like 20 years from now? DO you see a higher level of unemployment or low unemployment, but less labor force participation, people having shorter work weeks?  Because there’s been a lot of talk about technological unemployment, you know, robots taking our jobs.  What would that labor market look like if I were to look at those economic statistic 20 years from now? 

The labor force participation rate will continue to fall.  I don’t think unemployment as we measure it necessarily will be higher.  It may in some transitional sense, but not in the long run.  People will find jobs again.  In part, it depends how we run our welfare state.  Whether people who have falling incomes, do we pay them to work or do we pay them not to work?  Right now, we do a mix of both.  In general, I prefer to pay people to work.  But that’s up to us, you know, how many of those people will just cash in and not really pursue jobs.

And in the 15 percent, people sort of instinctively understand that you might do pretty well if you are computer expert, software expert, somehow involved in technology.  But you also mention other kinds of jobs which people don’t think about as technology jobs – management jobs, marketing.

Well, management is very important.  So computers will make many business enterprises much more productive.  But you still need good people to manage others, right?  So the productivity is there.  Computers are very far from being able to be good managers: sit down, set expectations, measure performance.  You need humans to do that.  I’ve found most human beings are not good managers.  Those who are I think will learn much, much more.  It’s already the case.  Salaries for managers as a category have been rising.  I think managers, motivators, marketers– when you’ve higher inequality, whoever can do things to get the attention of rich people, that becomes much more valuable.  And that’s a very psychological skill.  It’s not really about being a great programmer.

I think the best managers and marketers will earn more than a lot of the best tech people because some of those tech jobs, of course, can be and are outsourced.

And an example you use of being able to work with machines — how people working with machines can do a better job than just people by themselves or machines by themselves  –  is that of  chess players who work with chess software. They’re actually able to beat either humans working by themselves, obviously, or the machines by themselves.  But that won’t always be the case, which makes me worry about how the scenario that you describe will eventually work out. Won’t machines be able to do more and more, and there’ll be fewer jobs left for people? 

You know, there’s a cycle in all of these areas.  So right now, we’re – you know – we’re maybe barely at the point of where man plus computer is better at chess than just computer.  But that’s expected to flip very, very soon if it hasn’t already, if you look at the performance of the very best computers on the very best hardware.  But that’s saving grace, so to speak, is that artificial intelligence comes to an economy slowly.  It comes to it unevenly at different paces in different sectors.  So we’re always going to have a bunch of sectors which graduate into machine-only, and then a bunch of other sectors which come online that machines can do something at all, and then they work together with humans.

Very, very far out in time, many, many decades what will this look like, I think we’ll have so many smart machines producing so much, either for free or very cheaply, that it will be just fine. I think the distant future vision is really quite utopian, even though I think the next 20 years or more will have some pretty tough transitions.

And when you outline the future, you point to the state of Texas …

Well, my main point there is simply that Texas has very cheap land.  And one thing smart machines cannot do is produce more lands for us in a way that we can use.  So as income inequality goes up, people who are earning less, they try to cut down on their living expenses.  One relatively easy way to do this is to move to a cheaper area.  So I think we’re going to see more geographic segregation.  States which have cheaper real estate will become more popular.  Cities like San Francisco and Manhattan, they will become more and more the province of the wealthy.  And ghettos will disappear.  We’ll look back on the 20th century as this strange era when so many poor people lived in these fantastic cities and not quite be able to understand that.

There will be great jobs in Manhattan, in San Francisco, service sector jobs, right, for people who do not have a lot of education.  A lot of those jobs will pay pretty well.  The living costs will be a big, big problem.  I mean, I’m talking about the present here too, not just the future.  So you will find people living in significantly smaller quarters or many more people to an apartment or very long commutes or some mix of those.  And I think when it comes to living space, that’s one area of life that will be very much squeezed.  That is ultimately a regulatory decision we are making.  I would say it’s a big mistake.  I would much prefer to deregulate building in our best cities.  But I don’t think that will happen because homeowners hate it.  And so our poor and lower middle class will find their living quarters more and more squeezed.

A concept which you’ve written about a lot on your great blog, which you mentioned sort of briefly in the book, is that of zero marginal product workers.  Who are those people? 

Those are individuals who find it hard to get a job at all because they are perceived as just not having that much to offer at any wage.  And I see this as a growing problem.  I suspect it’s about 1 percent of the workforce.  People you just wouldn’t hire no matter what.  So stimulus won’t help them.  Flexible wages won’t help them.  I do think it’s an exceptional case.  But I think the more complicated production becomes, the more coordination is involved.  The more people with bad morale can wreck value, and you just don’t want to hire them.  It’s not a question of them not being smart enough.  It’s a question of they take way more from an enterprise than they add.  And I think we’ve all known workers like this.

But do you think that given a society where the workplace is going to become more technologically sophisticated, that those ZMP workers, maybe they’re 1 percent now, but but that percent will increase over the next decades. 

Well, they may be ZMP in tech sectors, but they’re not in general ZMP in service sectors, right?  There’s always a service sector job a person can find.  Some people may not take those jobs.  Other people, you may not, you know, even want them flipping burgers.  They’re just destructive.  And my point is just the better we get at measuring value, which you will get with surveillance and software and just plain flat out right measurement, the more we will discriminate against those people.  And that is a social problem.

We’ve had very slow job growth coming out of the past three recessions.  And people wonder whether there is something deeply wrong with the US economy versus not have enough aggregate demand. And it sounds like what you’re saying is that really the bigger problem might be that technology is accelerating the trend of companies laying off workers and then turning to machines.

You know, employers don’t like to fire workers really under any circumstances.  It makes the other workers, the good ones, more worried.  But they run out of money, they have to.  And that accelerates a lot of labor force trends.  So I think what we’ve been seeing in the data is very consistent with the picture I paint in my book.  It’s both a cyclical problem and a structural problem.  But the further away we get from 2008, the more it’s a structural problem.  You know, if you look at nominal GDP today, that’s more than 10 percent over its pre-crash peak.  So for how long can we say it’s all aggregate demand?  I think it’s partially aggregate demand, but less and less over time.

And also technology is making it easier for employers to fully understand who are the productive workers, right? 

Exactly.Employers already check your credit score or they look at your Facebook page.  So we’re just being measured so much more.  I find this somewhat disquieting actually.  I don’t see any way to stop it.  It does and will have big benefits.  But I’m not altogether comfortable with it either.

Given your thesis, how you can be pro-immigration in the sense that you see the need for more low skill immigration given these trends?  How would you explain that?  I mean we all understand that it would be great to bring more highly education people, entrepreneurial people to America. But do we still need more low skill immigration? 

Well, outsourcing is another big recent trend, moving production to other countries because wages are lower.  If we don’t have more low skilled immigration, we’ll outsource more.  Ultimately, I would rather have that production in the United States that will be more complementary jobs for people who are already here.  I’m not suggesting immigration will solve our labor market problems.  It won’t.  But I think somewhat counter intuitively, the arguments of the book actually imply we should have some more people living in this country, including at the lower end skilled – lower end part of the skills’ ladder.

We live in a world of competing economic clusters.  And the main competition right now is some parts of Northeast Asia.  And if you just look at the numbers, numbers alone, North America isn’t quite there.  So geopolitically, economically, I would like North America to be the dominant cluster.  So I would like to see the U.S. have a much larger population.  It would address a lot of our fiscal issues.  And again, when the alternative is outsourcing, there’s a lot of evidence that having more immigrants, even low skilled immigrants, it doesn’t lower or much lowers U.S. wages at all.

I think in the longer run, it will create more complementary jobs by keeping more of the production in this country.  Again, I’m not saying it’s a fix to our problems, but it’s good for the immigrants and I think it’s fine for us.  And geopolitically, I think it’s actually very important.  So I absolutely want to go in that direction.  More immigration, high skill, absolutely, but low skill too.  Keep in mind, low skilled laborers free up high skill laborers to be more productive by doing personal services for them.  So you also get more creativity.

The general tone of the book is fairly  dispassionate: Here are the trends.  Here is the evidence.  Some parts of the book readers may like, others they may find scary. But are these trends so overwhelming that policy is useless?  Or are there policy things we could do to maybe create a more optimal situation?

I think we could do a lot to ease the transition.  And one thing I talk about quite a bit in the book is deregulating buildings.  I think in general government is spending too much money on the elderly and not enough on the young.  I think we should get rid of most occupational licensing.  I think insofar as we have welfare, we could rethink and improve it.  But we could do all those things.  I don’t think we’re going to prevent rising inequality, but I think we can make it much more bearable.

There’s no reason why should need a state license to be an interior designer or a barber.  I think that’s just crazy.  It’s just drawing value.  It’s hurting lower middle class workers.  It’s just a terrible idea.

Do you get any sense that policymakers in Washington really understand these trends?  Are they going to be blindsided by these trends?  Are we doing anything to deal with them?  Or is this just kind of going sneak up on the politicians? 

I think at various high levels, these issues are well understood.  Does that mean every or even most policymakers get it?  No, probably not.  But I think people in the world, they know like something’s changing, something’s going on.  It’s not just this country.  You look at the United Kingdom.  You look at Germany.  They’re seeing economic recoveries and growth.  But you know, they still have falling wages or very flat wages for most workers.  And they know something’s not quite right compared to what we used to expect.

On an earlier podcast, I talked with the Erik Brynjolfsson, a co-author of “Race Against the Machine,” another book looking at how technology is going to affect labor markets. How does your work sort of dovetail with that book as well as some other economists, like Robert Gordon, who have looked at whether productivity, innovation, and growth are stagnating?  

Well the Brynjolfsson and McAffee book, “Race Against the Machine,” that’s a great book.  It’s influenced my thinking.  I just read their second forthcoming book, “The Second Machine Age.”  They focus more on automation than I do and less on inequality and much less on social issues.  But I think of myself as thinking along the same lines as they do.  But they and I, we differ a lot about the past.  So they don’t think the past has been a great stagnation.  I agree with them a lot about the future, but disagree with them a lot about the past.

Gordon, I disagree with him about the future, but agree with him about the past.  So Gordon, like me, sees a great stagnation.  And he thinks it will never ever end.  I think that’s crazy.  Even if it were true, how would you know?  But I see a lot of areas, not only artificial intelligence, but medicine and genomics, where the advances are not on the table now.  But it’s hard to believe there’s not going to be a lot more coming.  Science is very healthy.  There’re new discoveries all the time. The lags are much longer than we’d like to think, but absolutely progress is not over, and we’re about to see a new wave of progress over the lifetimes of our children.

Follow James Pethokoukis on Twitter at @JimPethokoukis

8 thoughts on “Will robots terminate the US middle class? A Q&A with Tyler Cowen, author of ‘Average is Over’

  1. 1000 years ago – will the horse drawn plow decimate the middle serf? Or will it release folks from the chore of farming so they can do other things, like create crafts and form guilds, and small trading villages?

    Hmm?

    Will the printing press put hand scribes out of work forever, or will new quicker ways of printing cause the print market and books to explode?

    Hmm,

    Will the truck destroy the jobs of horsemen, or will replacing horse drawn carts with trucks greatly expand the way transport goods and create more trucking jobs, then ever existed with horses.

    Will gas lighting permanently destroy jobs of the whalers?
    Will Airplanes decimate international travel.
    Will computers cost everyone their job?

    I think in every case technological innovation has caused markets to expand as they made goods cheaper for the rest of us. And in the process created more jobs, that pay more money than ever before.

    Lets cut everything back if you are so worried, and go back to the hand drawn plow, or have everybody farm by digging holes in the ground with their hands. Or shoot farming would hurt hunter gatherers. Lets all go look around for Acorns, and dig up roots, like in North Korea.

  2. So, the middle class is wiped out by technology, the elite top 15% are the new masters of the universe, and they outsouce all they can while flooding the nation with millions of immigrants so they can continue to have cheap nannies and leaf blowers and profit from lower wages , while passing on all the social, economic and environmental costs and burdens to the communities they doubtless do not live in.
    This future world masters forget when they change A and B, C will not necessarily stay the same. C is waking up.
    Immigration is a racket, it is a transfer of money from labor to capital. Those who would experiment with society and not expect to experience any consequences of their actions have selective memories of human history.

  3. When I started in the workforce back in 1980 it was 3 draftsmen, one personal secretary, one typist and one runner per every engineer. Our had a staff of 5 engineers, 10 engineers in training, 20 draftsmen, 5 personal secretaries, 5 runners, 1 office manager and a receptionist for a total of 47 people. When advances in word processing became available in 1987 it became one office secretary per 3 engineers and no typists. When automated computer drafting became available it dropped to 1 CAD draftsmen per engineer. When email became available it went to one runner for the entire office. We are down to 5 engineers, 3 engineers in training, 2 CAD draftsman and one office manager – a staff decrease of 90%. This is great for productivity and the huge increase in salaries for the engineers, (Myself included), but I can’t help but think that in the long run this is just setting us up for some kind of eventual political upheaval should we get into an economic depression.

  4. There is no doubt that just as machines created masses of jobs at the beginning of The Industrial Revolution they will eliminate masses of employment opportunities across the economic spectrum (but, probably hitting the Middle Class hardest) as technology advances. There is no stopping that. It is the nature of humans to advance. Those who can keep up will. It is the number of those who can’t keep up who will become a potential problem. It won’t be because these people are stupid, lazy, etc. We simply are not all born with the innate abilities that will be demanded to control increasingly complex (and able-minded) machines. The abilities of these machines will allow industry to operate more and more efficiently… reaping greater and greater profits. There’s nothing wrong with that. Businesses are not charities. They need only create/sustain the minimum number of jobs necessary to operate. That, actually, is their obligation. I am not certain what the solution to this potential crisis is. In the past the loss of manual labor jobs to one or another circumstance was overlooked or brushed off by members of the Middle Class as being what happens when a person doesn’t try hard enough, get a good education, and so on. When people who have tried hard, who have gotten the proper education, etc. begin to discover that they are the ones who are becoming obsolete… then the potential uprising will not be by (assumed) marginally educated laborers. It will be by educated – and very disillusioned – people. Instead of shrugging, Atlas may find himself running for the hills.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>