The Big Question: Are Robots Contributing to Inequality?

(upbeat music)

Hal Weitzman: Are you in danger of being replaced by a robot? Nearly half of US jobs could be susceptible to automation over the next two decades, according to a recent study—and it’s not just blue-collar positions. Jobs in accounting, legal work, technical writing, and lab work are also considered vulnerable.

Robots—once the preserve of carmakers and other big manufacturers—are starting to become more affordable for small and medium-sized companies, while 3D printing, driverless cars, and delivery drones promise more turmoil ahead.

So what effect is all this having on labor markets? Is the rise of the robots contributing to inequality? And what, if anything, should governments do to help affected workers?

Welcome to The Big Question, the monthly video series from Capital Ideas at Chicago Booth. I’m Hal Weitzman, and with me to discuss the issue is an expert panel.

Brent Neiman is an associate professor of economics at Chicago Booth. A former staff economist on the White House Council of Economic Advisers, he’s a faculty research fellow at the National Bureau of Economic Research, and an associate editor of the Journal of International Economics. In February, he was named an Alfred P. Sloan Research Fellow.

Matthew Notowidigdo is the Neubauer Family Assistant Professor of Economics at Chicago Booth. A former associate at Lehman Brothers, he’s also a faculty research fellow at the National Bureau of Economic Research and an expert in labor economics, public finance, and health economics.

And Peter Marsh is the former manufacturing editor of the Financial Times and the author of The New Industrial Revolution: Consumers, Globalization, and the End of Mass Production.

Panel, welcome to The Big Question.

Brent Neiman, let me start with you because you’ve done research about how much national income goes to workers and you’ve shown that over the past few decades, that share of national income has been declining. So tell us a bit more about that.

Brent Neiman: Yes, so the work really does two key things, I think, like you said, labor share has been a statistic people have looked at across countries for a very long period of time, and it’s the the share of total income in an economy that’s paid in the form of wages to workers as opposed to the money that’s paid to capital owners. You can think of profits. You can think of rental payments, things like that.

And it’s been very difficult to measure across countries, for example, the labor share, because in some industries, like let’s say agriculture, it’s very difficult to separate the implicit payments for the work that the farmers are doing as laborers versus the the work that they’re doing as the owners of that land or capital. And so there’s been limited work in a cross-country context.

In the United States, one of the first things people are taught in early macro courses is that labor share has been very stable for a very long period of time. And so the first thing that we do is we show that the labor sure has been declining since roughly 1980 or so, and that this decline is actually quite pervasive across countries and industries. And the second thing we do is we look at where the decline was steepest. And the decline happens to have been steepest in the places where the cost of capital goods has declined by the most.

Hal Weitzman: OK, but before we get onto that. Just to be clear, so when we’re talking about labor share, that’s not necessarily jobs, right? So how do those two interact?

Brent Neiman: Right. So it’s not about the number of jobs or about the wage on its own. Total compensation to labor is the product of the two. It’s the total wages times the number of jobs, and since labor share is comparing that product to the amount of money paid to capital owners, you really have to always be thinking about both sides of the equation to make sense of it.

So for example, during recessions, the number of jobs typically goes down. Unemployment typically goes up. But the labor share, in fact, in recessions typically goes up because profits go down even more than labor.

Hal Weitzman: So the share going to capital is decreasing, not—

Brent Neiman: Correct. Correct. So you want to think of them as distinct things, and I should note that our work is really about a very long-term decline seen in many countries, some of which have had, you know, decreases over the long run in their unemployment rates. Some that have had increases. Some that have had no long-run trend in unemployment. So you really do want to keep the two concepts distinct in your mind.

Hal Weitzman: And so if there’s a difference between labor share and jobs, what’s the significance of the decline in labor share?

Brent Neiman: Yeah. So labor share is important for a variety of things. One thing people very frequently associate labor share with is inequality. And the reason they do is because capital typically is a more concentrated ownership in society, you know, than labor income. And as a result, as you shift the amount of the total country’s income from a less concentrated toward a more concentrated category, you know, people presume—and generally speaking correctly, at least—that effect increases inequality. There’s a lot more richness to it we can discuss later, but that’s one reason.

Another is that labor share happens to tell us a little bit about the way in which production combines labor and capital. So for labor share to have been constant, as we had seen before 1980 for such a long period of time, it led economists to believe that labor and capital substituted for each other with something called an elasticity of substitution equal to one. The basic idea is as one factor of production—whether it’s laborers or capital—got cheaper by 10 percent, you would use more of that factor by 10 percent. And these off-setting responses would perfectly offset to stabilize the share that each obtained. If we find that labor share moves one way or the other, it might tell us—and my work suggests it does tell us—about the way in which production has changed in its ability to substitute between labor and capital.

Hal Weitzman: So how do we explain what’s been happening? Some people think it’s, you know, the loss of jobs is at least in part due to low-cost competition from overseas. Others say it’s because of ever cheaper technology and automation—the rise of the robots. How do you explain it?

Brent Neiman: Well, if we’re talking about the decline in the labor share, you’re right: some people think, and it probably does reflect a multitude of factors.

What we show is that in the cross-section of experiences, whether comparing across countries or across industries, those that have had the bigger decline in the labor share seem to be those countries or industries with the bigger decline in the relative price of investment goods. And what I mean by that is, you can think, for example, like, the price of computers, the price of equipment—capital goods that are then combined with labor to produce. Because we see the labor share declining by more in the places where the cost of capital declined by most, it suggests that in fact compared to the early parts of the 20th century, economies are able to substitute capital and labor much more than before.

Hal Weitzman: OK. Matthew Notowidigdo, what’s your view on trade versus technology as an explanation for what’s been happening to jobs or labor share?

Matthew Notowidigdo: Well, I think the consensus that’s emerging among labor economists and to here I’m trying to speak for people like [MIT’s] David Autor, [University of Zurich’s] David Dorn, [Harvard’s] Gordon Hanson and Larry Katz, who’ve done a lot of work on on these broad trends and labor markets, I think they would probably characterize what’s going on as primarily related to technology in the past and potentially increasingly being driven by trade in the future.

So what I mean by that is that if you look in the past several decades, it’s hard to find clear evidence that trade has extremely big effects on employment. But going forward, I think it’s reasonable to think that it might have a bigger effect in the future.

Hal Weitzman: OK. Why is that?

Matthew Notowidigdo: It’s a good question. One interpretation I’ve heard is that we’ve seen so much automation and computerization that you might see less and less of that going forward. And we see rising imports from countries like China that may start to affect the labor market in ways that we haven’t seen in the past.

Hal Weitzman: OK, Peter Marsh, you spent many years covering debates around manufacturing and industry. Do you think there’s a kind of misconception out there that what’s happened to good old-fashioned middle-class manufacturing jobs has been because of jobs disappearing overseas rather than technology?

Peter Marsh: Well, I mean, I think your explanation showed, I mean, it is a mixture of things. And the idea of just saying, oh, it’s due to the Chinese is wrong, I think. Trade certainly does affect this whole equation because of course the Western world in particular is trading much more with the what we call emerging economies than it used to. And it’s bound to have an impact on the kind of incomes people have and the jobs people have.

But I think overriding all this is this march of technology, and I think your findings really kind of sum up or provide more evidence for the intuitive feeling that virtually everyone has got that labor as a share of operations, really—in certainly manufacturing but in other parts of the economy as well—has become less important. I mean, obviously the easy way to see that is just look at one element of all this, i.e., jobs.

You look at the United States or the European Union, and combined they’ve lost 12 million jobs in manufacturing since 2000. In other words, gone down by something like 10 percent, and those jobs are very unlikely to come back again—in fact, impossible to come back again for some of the reasons that we’ve been talking about.

But I think the really interesting point is, firstly, what is that doing to our societies? Is it . . . some people might think it’s good. Secondly, is this going to continue in the future? And thirdly, if you do think it’s not a very good effect, i.e., as you say, this leads to inequality, this leads to social problems, what can you do about it?

Hal Weitzman: OK, thank you for giving me the questions for the rest of the episode. (laughs)

Peter Marsh: Oh, that’s your job. (laughs)

Hal Weitzman: No, those are all good questions. But before we get to those, I just want to talk a little bit more . . . you talked about this context since 2000, Peter. If we go back decades, there’s been a long-term secular decline in, you know, the kind of middle-class jobs, particularly in industrialized economies.

Matthew Notowidigdo, you’ve done research on this secular decline. Tell us a bit more about that.

Hal Weitzman: Yeah, so as was stated, manufacturing as a share of total employment’s been going down in the US and most other developed economies for several decades. It did not start in the 2000s, although it potentially accelerated in the 2000s.

One thing I want to make clear is that people who lose work in manufacturing, many of them do find jobs in other occupations and other industries, although a lot of them also remain not employed for a very long time. So what I think of is going on,when you see manufacturing declining, it’s caused a shift in the composition of work or the kinds of jobs that people have.

I really like the way that [London School of Economics’] Alan Manning describes it as what’s been happening over the last several decades is we see job polarization into lovely jobs and lousy jobs. And that the reason why I like that terminology is you may think of manufacturing as being the jobs that would be in between, maybe not particularly lovely but also certainly not lousy. And so as the manufacturing jobs have gone away, this polarization of work into lovely and lousy jobs, I think, is something that’s going to continue for a very long period of time.

Hal Weitzman: So not only inequality, but a kind of big gap in the middle there.

Matthew Notowidigdo: Yeah, that’s right. I mean, and it’s not really just manufacturing. So the way that I think about this is that jobs can be categorized into the the kinds of tasks that are done on them. And those those tasks, I think, can be broadly thought of as in three main categories: abstract tasks, sort of thinking about your job being creative on your job; manual tasks, carrying heavy machines, heavy materials, driving a truck; and then routine jobs, and these routine jobs include manufacturing, but they also include other types of jobs.

And the reason we call them routine is that they’re susceptible to being automated or computerized. And I think those are the jobs that we see disappearing over the last several decades. It’s manufacturing jobs because they’re being routinized through automated production processes, but there’s also a lot of other relatively low-skill service jobs such as clerical jobs, administrative jobs, where we see employment in those types of jobs declining as well. And it’s a different kind of routinization. I think of it as a computer-based routinization that’s causing employment in those sectors to decline. But that doesn’t mean that there’s gonna be no jobs available. What I want to make clear is I think we see this as shifting the kinds of jobs that are going to be available in the economy.

Hal Weitzman: OK, and that throws up its own set of challenges. But let’s turn first to inequality. Brent Neiman, you touched a little bit on this. Tell us more about, you know, the effect that this falling labor share is having on society at large.

Brent Neiman: Well, I should . . . my work so far, which I should note has been joint with Lukas Karabarbounis, my coauthor here at Chicago Booth, has focused on what the change in the labor share means for a representative agent of the economy. This is sort of technical lingo to say: we pretend like all workers are the same and all capital is the same.

And one thing that I think is lost when thinking about the decline in the labor share is that, on average, in that framework, it becomes clear that if technology or if capital and labor are very substitutable, and labor share is declining purely because of cheaper capital goods, that there’s a positive impact at least that this exerts on average. Much like people talk about, you know, free trade potentially being very good, on average, for the economy, even though there certainly are some workers that can be dislocated, and it might not be good for them in particular.

So the first thing I think I should say is, if technology is getting cheaper and better, and it turns out that technology and labor are quite substitutable, this is in many ways a positive thing. Now the way in which it can influence inequality in particular and make it a bad thing is that capital holders will disproportionately or can disproportionately benefit. And capital holders are typically a concentrated part of society. That’s one way in which inequality can increase.

Another is that if the mechanism really we’ve all been describing is in fact at play, and what’s happening is capital is getting cheaper, well then one wants to consider whether certain kinds of workers are more complementary—are better able to use those computers—than others. And so for example, there’s, you know, very good work talking about how the skill premium, the difference between skilled and unskilled labor, might have in fact, you know, grown a lot in recent decades precisely because capital got cheaper.

And in future work together with Lukas as well as with a grad student named John Adams, we’re essentially considering the same effect in an international context, but where we have more than one kind of worker, to see how the kind that is more complementary to capital fares relative to the kind that is less complementary to capital.

Hal Weitzman: But if your explanation was primarily about technology—all of you—rather than trade, does that suggest that this inequality is sort of an inevitable byproduct of of technology, technological advances?

Brent Neiman: Well, within types of laborers, it’s not clear or obvious that all technological advances, you know, have to favor high-paid workers versus low-paid workers. And in fact, you know, one certainly could imagine a world in which what technology substitutes more for are people that are typically at the high end of the scale. So that, I think, is something that remains to be seen.

We don’t know that the flavor of technology progress over the next 10, 20 years will be the same as what it had been in the past, and there’s no law that suggests it has to widen within labor inequality. But again, the other part of the, you know, the equation is what is it doing to, within capital, inequality? Whatever the shock is, whatever the change is that is causing labor share to decline simultaneously is probably acting on the within-labor inequality, the differences in wage outcomes, and the difference in outcomes on your portfolios, the within-capital income inequality.

And so all these factors enter together in a pretty complex way, and that’s why I think, you know, there are more answers and a lot more work to be done before we have a very clear, simple relationship of labor share to inequality.

Hal Weitzman: OK, Matt Notowidigdo, what’s your view on how that secular shift that you talked about, the lovely and lousy—

Matthew Notowidigdo: Yeah.

Hal Weitzman: How does that feed into—

Matthew Notowidigdo: Well, I think it’s a major force in inequality trends in the US and other developed countries. I think that, and it’s very intuitive that if you take middle-skill, middle-income jobs that require average levels of education such as manufacturing and administrative and clerical work, and you cause those jobs to be declining by a lot as a share of the total number of jobs in the economy, and then you’re left with lovely and lousy jobs. Well, it kind of mechanically is going to cause inequality to go up, and so I think it’s been a major force in inequality trends in the US and the rest of the world. And if you’re going to see continued decline in manufacturing employment and employment in these other jobs that I think we say are susceptible to routinization or automation, I think a natural prediction will be it’s going to cause inequality to continue to rise.

And when I’m talking about inequality here, I want to think about it as inequality for the 99 percent or within the 99 percent. My impression is economists do not have a good handle on what’s going on in the top 1 percent, like what’s driving the inequality growth in that group. But setting that group aside, and you’re looking at the other 99 percent, I think this job polarization that I’m talking about I think is a major factor in inequality growth.

Hal Weitzman: So the Lousys are gonna be always, you know, poor, and and the Lovelys are always gonna be getting much more, you know, wealthy.

Matthew Notowidigdo: Well, we know historically that the kinds of work that we put into this job category that I called “manual work” tends to be relatively low-wage work and it tends to require relatively little education. And so I think if you’re seeing the share of people employed in those jobs going up, it’s a pretty natural prediction that that’s going to cause rising inequality.

Hal Weitzman: I mean it does seem to overturn a sort of traditional view from economics that, you know, progress, technological progress would make everyone better off, a view you described earlier. I guess we’re getting to a point where that’s not necessarily the case.

Brent Neiman: Well, I don’t think that anyone ever thought it had to make everyone better off. The buggy whip manufacturer being the classic example, obviously disrupted by automobiles, even though clearly I think, on average, the world is better off for having automobiles.

So I think that would be kind of too extreme to say that everyone’s better off, but I think many of the same issues that, again, free trade or innovation have classically brought up still remain, even though, as Matt described, there clearly are some workers that would be made worse off by automation. It’s not at all clear to me, there’s nothing I’ve seen in the data that says that this has to be a bad thing, or in fact has to, you know, persist in making inequality worse and worse moving forward. The nature of what kinds of jobs that can be mechanized, the nature of what technology does, and trade is very similar to technology in the sense that in a deep sense, it doesn’t matter whether a new way of doing thing emerges, you know, from heaven or emerges from a foreign country.

But I think it’s very hard to predict the effects that those will have. And so no, I think that the analysis that we’ve typically made about technological growth typically being good, although of course carrying with it some burdens and some harm for some people, certainly holds today as well.

Hal Weitzman: Peter Marsh, you’ve written a lot about the history of technological advances and say that we are on the cusp of another industrial revolution. So how does that relate to inequality?

Peter Marsh: Well, as everyone has said, which I agree with: this is part of a long-term trend. You can go back to the ’30s, really, and start to see the prices of capital goods falling with this experience curve as people do things, they make things, they learn from doing that, prices go down.

And that, of course, has been reinforced, that basic trend, by the cheaper cost of doing things, which is partly to do with technology and partly to do with trade, with China entering the global economy.

Now the key point is: Is this going to continue? And another key point, of course, is: If it is increasing what I consider to be social problems in the form of inequality, what do we do about it?

Well, I think that there is a case for saying that, actually, there is a bit of a point where we won’t continue to see these changes, certainly, going through at quite this pace that we’ve seen before. Because I think there are some elements to do with the way people make products and create things that perhaps lending a little bit more emphasis in how they do that to this kind of touch labor, as it’s called.

In other words, people who basically come from a manual craft–based background, who haven’t necessarily been to university, who aren’t necessarily part of the the big capital masses, the richest part of society, you can at least make a point or case for saying there are elements in the manufacturing economy where there’s a bit of a role for these sorts of people.

Now what am I talking about there? I’m talking about things like making crafts-intensive metal products, for example, where you certainly need a lot of automation technology. You need the Chinese-made components to some degree, but you’re also using this craft-based labor at the level of putting some of this stuff together, very often in a high-cost country close to where it’s being consumed: leather goods, high-end footwear, high-end clothing. You can start to see that happening.

So you can start to see some elements of technology actually helping people in those positions to not necessarily reverse all these trends, but at least arrest them a bit. 3D printing, you could argue, is going to help a bit. Some other technologies, really clever but quite cheap machine tools at the level of . . . that can be used by people without huge elements of training may also help. This is only kind of a glimmer at the moment, I think, but I think it’s something very interesting that people should study and look at for the future.

Hal Weitzman: And I know you write about manufacturing, but, I mean, the advances in technology threaten all kinds of jobs. You talked about, you know, administrative jobs. A lot of these, most people work in the service industry, and many of their jobs are very easily automated as well.

Peter Marsh: Well, yes, I mean, I think there’s this kind of interesting idea oddly that . . . go back 30 years ago, and it used to be the office jobs were considered the best jobs, and the manufacturing jobs were considered, as we’re kind of saying, absolutely awful. You know, working at a steelworks with molten metal pouring all over the place.

And now, to some degree, for those manufacturing jobs that are left, but compared to the masses of service jobs, it’s to some degree reversed. In other words, it’s the office jobs in many cases, people under the whip, as it might be called, while in the past, they could just get away with anything, really. And in the manufacturing jobs, for the people who are left in manufacturing, they’re given much more leeway to make decisions, do things, have control of what they’re doing, aided by technology.

Hal Weitzman: Actually you might find that some of those manufacturing jobs move into the Lovely category and out of the old-fashioned Lousy.

Peter Marsh: You could make that point. The perception about all this is still—this is what . . . I’ve been in Chicago now for a few days and talking to one of the schools which is trying to do something about this, talking to some of the companies here as well. They can’t get people, especially young people, to consider manufacturing as a good thing to do. But in reality, if you know about it, actually, it is quite the interesting thing to do if you can work for a good company who knows about all these things, it’s very good. A lot of these perceptions haven’t yet spilled out into the broader community.

Hal Weitzman: Well that brings up the point of what do we actually do about this. What do we do about the . . . there’s going to be a transition period, presumably, between where we are now. There’s a lot of concern about long-term unemployment and people not having the right skills, and how do we get to that, you know, the kind of better economy where there’s more of the lovely jobs.

Matt Notowidigdo, what’s your view on kind of the . . . what should the policy response be to this trend?

Matthew Notowidigdo: Well, I think it’s challenging. I think one of the first answers you hear when you talk to almost any labor economist is they bring up education because the United States used to be the leader of the entire world in terms of the share of the population going to college, graduating from college. And the United States is no longer the leader of the world on that, so I think one of the first answers that people always give is figuring out ways to increase college going and raise college completion rates.

And that opens up an entire set of questions about what’s the best way to do that. I think if you look at different types of new schools, like for-profit schools, I think, may play a role in this. Figuring out ways to make college more affordable could play a role in this. But I just want to be clear: I think it’s a real challenge. There’s been generations of Americans that have for whatever reason decided not to go to college, and figuring out how to get more people through the college system to try to give them a chance to get jobs that would require a college education, I think, is a really great challenge.

Hal Weitzman: Brent Neiman?

Brent Neiman: If I could answer the question saying one of the things that I think we should not do?

Hal Weitzman: Right.

Brent Neiman: A lot of people, I think, look from the perspective of the United States at the labor-share decline and have in mind a very simple notion of how international trade relates. They might think the US is shipping labor-intensive jobs abroad to, let’s say, China and you know the remaining amount is more capital intensive, which causes a decline in the labor share. Some people’s knee-jerk reaction to that kind of a view is that we should impede international trade flows, raise tariffs, things like that.

And I do think it’s important to make the point that the labor-share decline in those labor-intensive countries with which we do a lot of trade, like China, like India, like Mexico is even steeper than the labor-share decline that we see in the United States. And so I think sometimes it’s lost from the conversation. The assumption is that this is a rich world, developed country problem that is being fed, you know, on the backs of labor-intensive countries, and I just wanted to say that my research with Lukas does not bear that out. It says that this phenomenon really is global, and therefore I certainly would say there’s no argument I would make on the policy side suggesting we should treat international trade any differently given this emerging pattern.

Hal Weitzman: And actually, that does raise an interesting point, Peter Marsh. You’ve written a lot about the, you know, the possibility of rich countries like the United States being able to compete actually, they, as wages rise in China, become much more competitive. So actually could this, you know, some of the lousy jobs actually make the US more competitive?

Peter Marsh: Well, I think so. I mean, now, if you do look at the statistics of manufacturing output, look at shares of manufacturing output according to the latest United Nations figures, you can—and it’s all done currently in dollars, so it’s looking at the snapshot at economy in each year, global economy—the American share of global manufacturing has been falling for a number of years. Last year, for which we really got figures, it’s 2012, when it was 17–18 percent.

Interestingly, in that year, according to the United Nations figures, the percentage market share, if you like, of the whole manufacturing output of the world was done in the United States, went up for the first time for some years. By 0.3 of a percentage point, but still a little bit of a change.

Is this reshoring? Is it something else? I mean reshoring is only one component—

Hal Weitzman: Reshoring meaning bringing—

Peter Marsh: Bringing stuff back that was lost, which of course the emphasis on reshoring isn’t all that interesting. What we’re really interested in is the net effect of what’s going on in manufacturing. So in other words, some companies may never have put jobs or activities outside of the country in the first place. And they’re retooling, reinvesting. Other companies, of course, are continuing to go out of business. So we’re interested in the net effect, and of course, in these figures, the net effect has been mildly positive. And if you talk to the people who study these things, then the best guess is that for 2013, for which we don’t have accurate figures, but this market-share thing went up again by just a little bit, another 0.3 percentage points.

So it could be that some of the trends that may be re-equipping parts of American manufacturing and possibly parts of the other bits of the economy as well to compete in this new world may be kicking in and doing something.

And I’m glad you said also about China, because China is finding that it can’t compete in the way it’s been doing manufacturing in the past. They are finding they have to eliminate jobs in many parts of the manufacturing economy. They’ve got 100 million people, something about that, working in that field, a lot of them doing things unproductively. And gradually, even within these state-owned enterprises, you know, people have kind of been got rid of.

But China’s going to have massive social problems, and because it’s got an authoritarian government who doesn’t have the capability to cope with some of these dislocations. At least in these Western governments, they can kind of—or Western countries—they can kind of cope up to a point, you know. There’s a riot every now and again, but they can just about handle it. There’s not mass-scale killings in public squares.

Hal Weitzman: Right, and of course the US political system is perfectly functional, with no problems at all.

(Marsh laughs)

Anyway, well, on that note, our time is up.

My thanks to our panel, Brent Neiman, Matthew Notowidigdo, and Peter Marsh. For more research, analysis, and commentary, visit us online at chicagobooth.edu/capideas, and join us again next time for another The Big Question.

Goodbye.

(upbeat music)