Cofounder and Chief Executive Officer, Business Talent Group
Director of Policy, Economic Policy Institute; Former Chief Economist, U.S. Department of Labor
Economic Scene Columnist, New York Times
Partner, McKinsey Global Institute
With the capabilities of artificial intelligence quickly expanding and middle class jobs under threat of being automated, will the government need to step in to pick up the slack of a job market hollowed out by machines? Experts explore the potential for mass job loss created by technological advances and, in turn, the possible need for a large welfare state to care for an increasingly underemployed population.
LUND: Hello, and welcome to today’s Council on Foreign Relations meeting titled “Only Robots Need Apply: Automation, Job Loss, and the Welfare State.”
I want to welcome you all. I’m Susan Lund. I’m a partner at the McKinsey Global Institute. And I am going to preside over today’s meeting and moderate a discussion with three terrific panelists.
We’re going to spend about the first 30 minutes with a discussion here onstage, and then I’ll open it up to questions from members. So start thinking already about what you want to ask.
So, before I bring in the panelists, I want to start with two different quotes to set the stage for our discussion today. The first I’m going to read and see if anyone can tell me who said this: “In our lifetimes, we may be able to perform all of the operations of manufacturing and mining with one-quarter of the human effort with which we have become accustomed. For the moment, the very rapidity of these changes is hurting us and bringing us difficult new problems to solve. We are being afflicted by a new disease, namely technological unemployment. This means unemployment due to our discovery of means to economize the use of labor outrunning the pace at which we can find new uses of labor.”
AUDIENCE MEMBER: John Maynard Keynes?
LUND: All right. I knew at CFR somebody would know that. (Laughter.) That was John Maynard Keynes in 1930, in an essay that he published called “Economic Possibilities For Our Grandchildren.” So this was the time of the development of the Model T assembly lines and the automation of manufacturing, and Keynes worried now, you know, 90 years ago about technological unemployment.
Now, he went on to predict that economies would become so productive that we would all need to work only 15 hours per week to have all of our material needs solved. At McKinsey we say he meant to say 15 hours per day. (Laughter.) But he also projected that incomes would rise between four and eight times over the next hundred years, and in fact, they’ve risen eight to nine times when you look at the U.S. and Western Europe.
So, of course, this technological unemployment did not come about. In fact, employment in the United States and in Western Europe grew because you had women starting—after 1950 starting to leave the home and do paid work in the workplace. So, in fact, the ratio of employment to the population grew rather than fell. So although, indeed, many jobs in mining and manufacturing were automated, other ones came up.
So now I come to my second quote, by perhaps a less-distinguished thinker, Woody Allen. (Laughter.) He said, just because they’re paranoid doesn’t mean they aren’t out to get you.
So what’s the answer here? Is this time different? Should we be paranoid about technology? And I have three terrific panelists today that are going to tell us about it.
We have Jody Greenstone Miller. She is the CEO and founder of the Business Technology (sic; Talent) Group.
Next to her we have Eduardo Porter from The New York Times, who is the economic scene columnist, that most of you probably read.
And then we have Heidi Shierholz, who is a senior economist and director of policy at the Economic Policy Institute. And Heidi was formerly a chief economist at the Department of Labor.
So I want to start with you, Heidi, to say: Is this time different? Are we at a point in history where, in fact, machines might take all our jobs? What do we see?
SHIERHOLZ: That is THE question. That’s an absolutely fine question. And I am not a futurist. I can only look at what has already happened. And I think a crucial thing to just—I mean, people in this room all know this, but just to get this out there, we actually have a really good measure of how automation impacts the labor market, and it’s productivity growth. So, you know, if I was a factory owner and I had a hundred workers and I was producing widgets, and then I automate it so there’s one person who is now just looking over a hundred robots, productivity in that factory has just gone up because now, in one human hour worked, same amount produced—it used to take a hundred human hours, now it takes one. We can actually measure that economywide. And so that’s sort of a very key measure of how these dynamics are impacting the labor market.
We have seen—and I really liked that—the sort of opening with that Keynes quote. We have seen from the beginning technological change, automation happening. Productivity growth on average has grown since we started measuring it around 2 percent per year.
There’s been some interesting recent dynamics. From ’95 to 2005, there was a surge in productivity. We saw a big sort of productivity boom during that period. It grew about 3 percent. The thing that’s going on recently is we’ve seen a big pullback, that we’re actually seeing a real drop in productivity growth as opposed to some acceleration right now. Over the last 10 years, productivity is growing far more slowly than it was in the last period. And if you just look at the last five years, it’s growing extremely slowly.
And if you just focus on manufacturing alone, it’s essentially flat. Productivity growth in manufacturing has been 0.2 percent per year over the last five years on average, compared to, you know, earlier in the—in the century—in the 1900s it was growing 4 ½ percent per year regularly.
So there has been a big deceleration in the sort of how automation is affecting the labor market. So it is—we’re not yet—we re definitely—I think this is what I can say. There is—in no uncertain terms, we are definitely not yet seeing the footprints of this. So anything that we have to—that we want to conclude has to be predictions about what we’re doing going forward, and that gets trickier. Like, things are more up in the air when we think about—mmm hmm?
LUND: Let me ask you, though, about manufacturing. So, over the last five years, I mean, we went through a period where we lost a lot of manufacturing jobs, and now we’re slowly adding them back. Does that explain why we’re not seeing productivity growth in manufacturing? Because on the other side of the coin you can look at, like, purchases of industrial robots, and we know that, in fact, that’s an industry that’s soaring. So somebody’s buying them. What do you think is going on in manufacturing?
SHIERHOLZ: So I don’t want to take us—should I launch into the connection with trade here? So it’s a key—a key piece of what’s going on with manufacturing is—there’s this—you know, this is all interacting with trade. So, from 1965 to 2000, the number of manufacturing jobs was insanely stable. It just fluctuated between 17.5 million and 19.5 million over those entire, you know, four decades, and the fluctuations were basically just if we were in a recession or an expansion. But it was really very, very stable.
That changed in 2000. We started to see a hemorrhaging of manufacturing jobs. We now have 5 million fewer manufacturing jobs than we did in the late 1990s. And the difference—what changed between the earlier period and the later period was not that productivity growth accelerated; it, in fact, decelerated from pre-2000 to post-2000. The big change there was a huge increase in the trade deficit. So I think that’s the driving force behind this decline in manufacturing jobs.
I know my panelists might not agree with me on that, and I’m totally happy to have people push back. But I think that that’s the—that’s what I see in the data.
LUND: OK. Eduardo, what’s your take on—are robots taking our jobs now or in the future?
PORTER: I mean, there I have to—I have to agree that we just don’t know yet.
But I’d like to make some comments about what you just said. And one is, perhaps, to situate the productivity story into a longer-term period, because the slowdown in productivity is not just a phenomenon of after the Great Recession.
If you look at, like, 100, 150 years, you do see a quite sharp slowdown in productivity from, like, the 1970s to today, with a little blip which is the dot-com—the dot-com moment, starting in 1995. So there’s a long historical series that looks very much in tension with the story of robots taking everybody’s jobs because you’d need to see accelerating productivity. And, you know, because this is starting in the ’70s, this is starting in a moment where, you know, you start seeing a lot of automation, you know, you get from IBM through to—or to, you know, the moment in AI. And so I think this tension, we have to understand what’s—why these things seem not to mesh.
And even though it’s true that in manufacturing productivity has been much higher than in the economy as a whole—way higher, even though it has slowed down recently—I mean, we should also think, when we’re thinking about the jobs of the future, we’re not thinking about manufacturing. Manufacturing—there’s 12 million jobs in manufacturing. The labor market is 150 million jobs. And so, as we think about what technology’s going to do to all of our jobs, I think we’ve got to start thinking about, you know, services.
And so I was looking at—the BLS has this little chart of projections of what are the industries that are going to generate more jobs over the next decade. Of the top—of the top 10, four are some variant of a nurse: personal-care aide, registered nurse, home-care aide, nursing assistant. So, if we’re thinking about—so we’ve got to—and these jobs, the problem is that the quality of many of these jobs is very, very low. And these are quite low—I mean, there’s not a lot of technology in a lot of these jobs.
And so, you know, a personal-care aide, according to the BLS, earns about 20 grand a year. Now, if I’m thinking, well, what’s the problem of our future labor market, it’s a labor market where a large share of the people employed are earning 20 grand a year. And so my—so my sense is we need more robots in the personal care—I mean, we need more productivity growth in the personal care market. We don’t have—we don’t worry that there’s too little. We want, actually—we need more. And, well, of course, we need ancillary things too. We need a skilled component to this that needs to be upgraded. But, you know, right now, if our economy—you know, home care rate is 21 grand. Nursing is 25 grand. I mean, these are all very low-paid jobs. And so that’s actually—to my mind, that’s the challenge.
LUND: That’s a great point. And it reminds me to tell the audience, we have, from the McKinsey Global Institute, a report on automation that’s outside the door, I think, that you can pick up. And one of the things we conclude is, indeed, especially in aging societies, we do need machines to start to do more of the work if we want to see incomes and standards of living rise. Now, there’s a different distribution probably, which you’re pointing out, which is that a lot of the jobs that aren’t done by robots in home care and face-to-face personal services are not terribly well-paid.
So I guess, where does this put you on the spectrum of being a techno-optimist or -pessimist?
PORTER: Can I sit on the fence on that one? (Laughter.)
LUND: Wait and see.
OK. Jody, I want to turn to you. Jody—first, tell the audience, what is the Business Talent Group, and—
MILLER: It’s a—it’s a platform for high-end business talent that wants to work on projects. So it’s part of the gig economy, but the high end. These are the people who are choosing to work this way, who have very different drivers than what you’re reading every day about Uber. And this is a part of the economy that actually I think has to be considered when we think about what’s going on with automation and job loss for a number of reasons.
Just quickly, because a lot of these folks are going to go into the gig economy, and that’s going to drive up those numbers. It’s already, according to Larry Katz and Alan Krueger, at the low end, 16 percent of the labor force today is in something called alternative work arrangements. There are figures that go as high as a third to 40 percent. The growth at the high end is actually incredibly rapid. Fifty percent growth since 2011 in people earning $100,000.
And the problems that this gig economy is facing are exactly the same as the problems that people who are facing job loss to automation are facing. They are outside of the social net, because we have a unique—as everyone in this room knows—situation where our health care and our retirement are tied to a traditional vision of employment. They are outside of any kind of skill program. And so thinking about this part of the economy, in addition to what’s happening with automation, I think go really hand-in-hand.
And you may have had an additional question, but I was just going to comment on what Eduardo was saying. You know, I think absolutely we have to have a path to middle-class wages for the service worker, because that is a segment that isn’t going to be turned over tomorrow—although, I understand in Japan they actually do have home health care worker robots. So what will really happen is we will make decisions as individuals whether we want robots taking care of us or our children or our parents, or whether we will just subcontract some portions of that and then, as you say, increase productivity for the actual worker. But this has been an issue that, you know, Andy Stern has been talking about. I mean, creating better jobs and better career paths and better wages for this class of worker is becoming increasingly important.
PORTER: If I could—if I could actually just comment on this. I think that this phenomena you just mentioned, which is the growth of these alternative working arrangements, is really very important. To my mind, it’s perhaps bigger than robots, and it is not robots. It has more to do with changes in business strategies and forms of organization of business, this idea that big businesses siphon off so much of the work to—you know, to less-well-paid—
MILLER: Or, the high end is actually, when you are a free agent in Hollywood, in baseball, in technology, you will make more money. And so what’s happening is it’s actually rigidity in the labor market that is benefitting traditional employers that is just unfair. And if you were to get health care apart from employment, social net—safety net apart from employment at the high end, you would benefit people. And that’s what I think is kind of the dirty little secret around the employer attachment to benefits, that people at the high end are really being punished, and people at the low end are being punished because they’re being siphoned off and they have no social benefits. So both sides really need a new vision of employment that affects everything from the legal framework, IRS collection, to benefits.
PORTER: Yeah. And if you’re the janitor that’s, you know, cleaning the bathrooms and you work for GE, so somehow you were sharing in the rents of GE. But if you’re sent off to work for ADM, which is a janitorial subcontractor, there you’re going to be paid basically your marginal product, and it’s going to be, you know, minimum wage, if you’re lucky.
MILLER: Yeah. Exactly right. Exactly right.
LUND: So let’s go into solutions. We are supposed to talk about, do we need a new social contract? So whether jobs are automated, the jobs that are left are low paid, we might be threatened by trade, let’s start with income support. Any of the panelists, what do you think about universal basic income, the idea that every member of society, or certainly ever adult, should get some sort of minimal annual payment, whatever you do. And this will free up people to do what they want and for those replaced by machines would give them a way of a livelihood. Is this a good idea? We’ll talk about the political reality later, but just on economic terms what do we think about this?
SHIERHOLZ: So this is a very good question. And I would like to step back and say despite everything I already said, I do—it is totally true, even if I don’t think that automation is going to affect the total number of jobs in the economy, it is absolutely true that automation will displace jobs and it is very important to think about what we do for the people who are in that situation. So, like, any conversation about how this isn’t going to cause mass unemployment—which I actually believe it won’t cause mass unemployment—it will still cause real economic hardship for individuals. And I think that’s a very important conversation to have, so I’m glad you brought this up.
The universal basic income is one place people go. I tend to immediate—because that’s hard to—you know, universal goes to everyone. Not very well targeted. Like, it’s sort of a tricky thing, I immediately go, what about universal jobs? Like, to me, if you were going to go universal, having government be the employer of last resort, so there’s a job guarantee, if you’re doing—if you’re thinking about a really big move, like universal basic income, my big move would instead be to have the government be an employer of last resort—a decent job with a living wage and benefits to help people who have been displaced, is sort of where I would go.
That’s—all of this is pie in the sky, right? Like, those aren’t happening, and thinking about what other things we could do are important. But I go—when I go universal, I go jobs rather than universal—
LUND: Public works. I love that idea. For those—
PORTER: We have the money.
LUND: For those of you who have been to Acadia National Park up in Maine, they have these lovely little stone arch bridges all over the park and these, like, cobblestone paths so that people could take their carriages through the forest, and it was all built for the conservation corps. It’s really spectacular.
But, Eduardo, you’ve written about this. What do you think about universal basic income?
PORTER: I got into trouble with Andy Stern for saying it was a bad idea. (Laughter.)
SHIERHOLZ: That’s why no one can tell Andy Stern that we’re having this conversation.
PORTER: Yeah. Or Charles Murray on the other end.
MILLER: That’s what’s interesting about it, right?
PORTER: Yeah, they both kind of like it. I mean, I share your skepticism about this idea. It’s not particularly well-targeted. It’s a lot of money. Think about it. You know, how much are—how much is this universal basic income going to be? If you’re talking of $10,000 per person you’re talking about 3 trillion-plus (dollars), which is a good chunk of GDP. This country has never shown any interest in raising that kind of money from general taxation. People that support UBI say you can tax back that money from the rich, but it strikes me as extremely inefficient to provide a lump sum and then tax it back at a marginal rate. And what’s more, I mean, these things will have disincentive effects on employment.
And I think it’s way too soon to, you know, give up on the labor market, frankly. I think that the argument, the pro-UBI thing, is riding on this sense fueled in Silicon Valley that there will be no jobs, that technology is going to, you know, kill all of them, or kill half of them, or kill two-thirds—a big, large share of them. And if that were to happen, well, then I think that thinking about something like UBI might make sense. But I see no evidence right now that this is happening. I mean, I think we’re actually near full employment right now. We have an issue of quality of employment.
LUND: Yeah, so how do you address the very low pay of the jobs that are out there.
PORTER: That’s right. That’s right. And you know, hey, providing—Tony Atkinson wrote this great—you know, he was on your side, you know, proposing this jobs at a minimum pay that was guaranteed by the government. And that would be a great idea. Kind of pie in the sky-ish. But bringing it down to our present, I mean, we could do stuff like skills—I mean, training—the U.S. doesn’t do any training. And we’re kind of unique in that way. If we think that part of the problem is that you need to increase, you know, the productivity of the workforce and you need to—so there’s a skill component, the U.S. spends 0.03 percent in GDP in training, according to OECD data. Denmark spends half a percent of GDP. So that’s, like, something like 17 times what we do.
And training here has a bad name, because we’ve always done it really badly and on the cheap. But you can do training well. I wrote a column once about a training assessment by MDRC in New York, where the outfits that were providing the training were also going out there to the labor market and talking to employers about what they needed and whatnot. And they found a 17 percent increase in the wages of people who were in the program versus the control group. So this was a really—this was $2,000 a pop. And it was expensive. It was, I think, 6,000 bucks to train each of these guys. But compare that to, you know, the cost of all the other social, you know, consequences of having such a low-paid workforce, and it starts looking pretty cheap.
So I think that that is one part. I think there’s other parts, about, you know, changing the nature of the jobs. So if everybody’s going to be nurses, maybe we can get Medicare to stick something in their contract saying, you know, we will only hire you if the nursing has these sorts of qualities and these sorts of pay. I mean, I think there’s things that could be done on that side.
LUND: Federal standards.
PORTER: Yeah, on the standards side. But I think that we can do stuff that definitely—to affect the idea of the quality of the work.
MILLER: So I would offer a couple thoughts. We don’t know what the pace of this dislocation is, but what we do know is if it happens there’s going to be enormous wealth created. And that wealth is likely to be incredibly concentrated, more so than we’ve ever seen. So Bill Gates, I guess last week, said: Well, why don’t we tax the robots? So I think starting to think now about how we sync up, making sure that the wealth creation that will come in conjunction with this job loss, if it happens, is thought through, so that we can share that equity more broadly. And we need to start thinking about that now. I think it’s a great idea for an XPRIZE challenge. I think we should think about how we get the monetary benefits broadly distributed, if in fact those monetary benefits are going to come with this job loss. So I think that’s something very concrete that right now minds in this room could probably contribute to thinking.
On skills, I should disclose I’m on the board of a company that is in the coding and education area, so you can take what I say through that prism. But I think we have not thought about skill training broadly enough. I think the notion that you go to school, or you go to vocational school, and then you’re done, and then you get a little, you know, sprinkling here and there just doesn’t work anymore. And I think there are a couple things we should be aware of. One, even if you have what is, quote, unquote, a “traditional” job, employers—average employment, and you can keep me honest—four years and change. You know, employers just don’t have the incentive to continue to training. They’re looking for you to walk in trained. And that’s why you’ve got these coding schools, General Assembly, these companies that are coming up with these new skills.
But even that’s not sufficient, because you’ve got three issues. One is, how do you get timely data about what the market needs are? The Labor Department obviously would be a great place, but you need it fast. Once you’ve got the data, you need the capacity—broad capacity to turn that data into actual training that works and that can be distributed broadly. And it doesn’t all have to go through the prism of an accredited institution. And so we have to think about everything, from, you know, the MOOCs, which have actually moved from generic course on philosophy to more specific skills, companies that are enabling peer-to-peer training, like U-2-Me, companies that are offering certifications, all the coding boot camps. So you’ve got—the first layer is data. The second layer is you’ve got to be able to translate that into something that is actionable.
And then the third issue is cost. Who’s going to pay for it? And, you know, there are a lot of different ideas about how that should be, you know, thought—you know, how that should be handled. But unless we start thinking much more, I think, globally and bigger about training, we’re not going to solve this on the margins.
LUND: I have to say, in the—in the transition from agriculture to manufacturing in the U.S. and in Europe, you saw a huge shift, over the course of 70 years, in employment, where at the turn of the century about half of Americans were employed in agriculture and today that’s 2 percent. And in basically the first 50 years of the 1900s there was this massive shift. But at the same time, there was the rise then of universal high school education. It was a time that the government decided to train people for the new manufacturing and industrial future. People needed more education than they were getting in the agrarian societies.
Is it time for—I mean, I’ve heard a lot about training. What role—who—I guess, Jody, you raise it—who pays for this training and how does this actually get provided? And do we need a different sort of educational policy at either the federal or state levels to train people for the 21st century workforce?
MILLER: Yeah, first I should acknowledge, people like Byron Auguste are doing great work in this area. He’s got Opportunity@Work. He started Hope Street Group, came out of the Obama White House. I mean, there were people thinking about this at a big—but these are big problems. And they’re working with LinkedIn and Reid Hoffman, trying to get the data, trying to start, you know, these processes. And I actually think it’s got to come to the individual to drive their own agenda. The payment’s different, but I think the individuals got to look at say: Here’s what I think I’m good at. Here’s where the market’s going. And here’s the skills I need. And then, you know, I think government should—personally, my point of view—pay for some of this. And it’s not once and done. It may be every few years. I understand college interns now are expected to come in for summer internships with the skills they need for that internship.
So I have a daughter who’s a sophomore. She’s saying: How do I get an internship? I’m going to have to go take a class to learn what I need to do to get an internship for the summer. I mean, it’s a very different world now in terms of what the employers are expecting. And that means we have to train people to understand, they have an obligation.
LUND: All right. I want to open this up for members to participate in the conversation. There are roving microphones. And I’d ask everyone to state your name and affiliation and ask just one question, so we can get lots of participation. We’ll start right up here, and then go over here.
Q: Hi. Rob Quartel with NTELX, a technology company.
In this discussion of robots, it’s like they’re all the same. But they’re not. There’s robotic decision analytics and systems. There are autonomous vehicles. There are industrial robots. And it seems to me, teach one of these has a different impact. So, for example, who is autonomous vehicle displacing in a non-commercial context? Zero. In a trucking context, it might be replacing a truck driver. So have any of you thought about this? And we know that as industries become more efficient and have higher productivity, there are also circles around them of increasing supporting activities, which were not anticipated as it happens. So, I mean, I’m actually—I’m optimistic, but I think we don’t think enough about how each of these works out. Have you all?
MILLER: I mean, I can talk about the law. I was a lawyer by training and consulting. And, you know, what I would say is McKinsey tells us 23 percent of legal work is going to be automated. We’re already seeing enormous amounts of automation in discovery. You know, they can have automatic review discover. And what’s happened is the lawyers at the top of the chain are doing just fine. Their rates are actually going up. Their productivity is going up. Clients are saying: I’m not going to pay, you know, a very expensive law firm to do work that can be done, you know, by a computer or by offshoring. And there are a lot of temporary law companies now, Axiom Law, some of you may know, who will take this work out of the firms.
So I think you’re right. There’s a certain segment of work that will just be integrated, will improve the economics for those at the top, and then you’ve got the problem we’ve all talked about which is the people who are actually—if they’re not robots—who are being outsourced. Their jobs are not going to be very satisfying and their path and their wages are not going to be satisfying. So you still end up with, what are we going to do about that?
PORTER: Kind of like more generally, you have the work of David Autor and others—David Autor at MIT—who has kind of, like, looked at how does automation affect different bits of the labor market. And his basic proposition is that you have the hollowing out of the routine tasks, everything that, you know, you can get a machine to do, a machine will do. But that will also increase—that will be a complement to many people that are higher up on the—on the, say, educational scale. And so you’ll have a polarizing effect. And interestingly, at the very bottom, it has—apparently technology has much less of an impact as well, because you need much more—there’s in-person services and whatnot in that part of the economy. So you basically have a hollowing out in the middle in this analysis.
LUND: I would also encourage you to take a look at our report. I mean, we looked at—oh, perfect. We go through 2,000 different specific work activities across 800 occupations and consider everything from autonomous vehicles to intelligent machines, to what we traditionally think of automation as sort of manufacturing.
So, all right, let’s go—we had a question over here, and then we’re going to go here. Why don’t we take three at once, and then give everyone a chance, because I see a lot of hands up in the room.
Q: Joe Onek, the Raben Group.
Everything we’ve suggested is going to cost government money. If you do minimum wages for nurses, Medicare and Medicaid pay. Obviously, an increase in earned income tax credit, very popular program, money. Job training, money. Where is that money going to come from? You talked about the creation of wealth. That creation has already happened. We now have the most unequal distribution of income and wealth since the end of the 19th century. So it seems to me we do have to be willing to, quote, “tax the rich,” but is that even possible in the current society?
LUND: Good question. Let’s take Paula.
Q: Thank you. I’m Paula Stern.
And actually, I’d like to ask this question based on my work with the National Center for Women and Information Technology, NCWET. And it’s about computer science for all, the initiative which ultimately resulted in President Obama really trying to place emphasis on the need in the K-12 setting to have computer science be considered a science, that you have credit for taking and you accredit teachers for teaching. I would like you to address that fundamental problem. You did a good service of mentioning historically how universal high school reflected a change in our economy a century or so ago. And I’d like you to kind of give us, in the list of all the ideas of recommendations you’ve come up with, what’s wrong with having computer science in the K-12 curriculum as a matter of fact in every state? Because it is definitely not there.
LUND: OK, great. Let’s take one more question. How about way in the back of the room, and then let’s answer those three.
Q: Hi. My name’s Alexander Sienaert and I work at the World Bank Group.
We’ve heard a lot in the sort of recent election, and then since from the president about the impact that trade has had on—or the impact that he claims trade has had on jobs in certain areas of the country. I was interested in hearing at the outset of this conversation that if you look at jobs, manufacturing jobs—I think you were saying this—up until 2000, that the number sort of stays relatively constant, and then it begins to decline rapidly. And that sort of coincides with the increase in robotics, I think was sort of what you said. So does this suggest that these discussions and claims about trade are invalid? Was—you know, even I could imagine a scenario where jobs in the 1990s in manufacturing, even as they were remaining relatively constant in terms of number, they were changing because of trade. And that made these jobs more vulnerable to robots. You know, what do you guys—what credit do you all give to that trade argument versus the robotics argument?
LUND: OK. So we have three questions, one on inequality and wealth creation that we’ve already seen, how are we going to pay for all this? Second, what role should computer science and computer programing play in K-12 education. And I can say with two daughters now in public schools they’re going to go through and know calculus and multivariant calculus, and neither of them are going to know how to code a computer. And then thirdly, what role does trade really play?
SHIERHOLZ: Want me to go for it? So I can quickly—so I’m totally in favor of computer science in schools. (Laughter.) As far as the trade goes, it’s actually the reverse. So what we saw is very stable employment in manufacturing through 2000, and then the drop off. And the thing that changed there wasn’t an increase in automation, it actually was an explosion of the trade deficit. So I actually think that drop off can be far—it’s far easier to pin it on trade than the explosion of—than automation.
And the question about inequality I think is really at the core of this. When I think of what do we—how do we—the solutions for what do we do for people who are displaced, we’ve touched on these. And just—the framework I have in my mind is that it’s a three-legged stool for what we do for displaced workers. It’s a safety net. Like, we need to boost unemployment insurance. It’s training, which the rest of the panelists are much more expert on that than I am. And then I think the third leg of that stool, which has gotten touched on, is a good job strategy.
Like, one of the reasons that displacement is so awful for workers is that there are—there’s a real lack of good middle class jobs with decent wages and decent benefits that are available. And that is because we’ve spent the last 40 years over and over and over again making policy choices that erode the quality of jobs available for low and middle-income workers. And so that’s a choice. We can reverse that. The political will is the question. So I don’t have an answer for the political will, but the economics—it’s a choice we could make if the political will is there.
We could do things like boost labor standards, like minimum wage and overtime. You know, increase the share of workers who are eligible for over time. We can help workers balance work and family by providing paid sick leave and paid medical leave and affordable child care. We can boost unionization. One of the reasons being displaced from your job sucks is because in many cases you’re using a union job but, due to low union density, it’s hard to find another one that has that same—that provides that same quality of job boosted—so boosting unionization.
A full employment policy is really important, not just for the job that provides, but also when you have full employment workers have more bargaining power, not in a union sense but in a personal bargaining power sense, because if your employer knows you have outside options they have to pay you more, they have to give you a better job to get and keep the workers that they need. So you know, that’s where the sort of jobs for all may come in, or other kinds of job programs. Monetary and fiscal policy, sort of all that play there.
But those are the kinds of—that third leg of the stool, the good jobs strategy, that could help solve some of the pain from the displacement, from the workers who are displaced. And the political will is not going to make any of that happen right now.
PORTER: If I could just jump in and talk about political will? Walt Scheidel, a historian at Stanford, just published a book where he posits that going all the way back to the Stone Age, through Medieval Europe, through the 19th, 20th century, to the present, the only moments in time where we have mobilized resources to reduce inequality sharply have been wars, plagues, pestilences, natural catastrophes. So, well, we can hope for one of those. (Laughter.)
LUND: Wow, I thought this was—
Q: (Off mic)—says in his book, right?
PORTER: Yeah, exactly.
Q: But he asks the question, we have all this inequality. Is that possible to continue in democratic societies? And I think we’re learning what the answer is: No.
PORTER: Yeah. Well, if you look at it just on the redistribution side, the safety net side, back in the 1960s, the U.S. has been—the U.S. government has been—at all levels—raises about 20 percent of GDP in taxes. And this has been a really, really stable number, going back to the 1960s, which is when the OECD starts this series. If you look at pretty much every other developed country, not all of them but most of them, including all the social democracies in Europe, they started, in the ’60s, around a fifth of GDP as well. But they’re right now at 40 percent of GDP, you know, 35 percent of GDP. So, like, that gives rise to this thesis—
Q: And without the defense expenditure.
PORTER: And without the defense expenditures. And so there is this sense that, you know, as these societies have developed, the notion that you need to provide social insurance on a larger scale because our—kind of like our—we want to raise the bottom as the society becomes richer. Somehow, in the U.S., did not happen. Now, why it did not happen, I mean, I have my pet theories. I mean, one can think of—I think that racial hostility is really very important to explain this. But could think of other things. But, you know, I don’t really—I don’t see the political—I don’t see anything out there in the objective reality mobilizing political will in a different direction than there is nowadays. And who mentioned that Bill Gates said taxing the robots? He didn’t say tax the owners of the robots.
MILLER: Well, I actually think that’s what he—that is what he meant.
LUND: That is what he meant. (Laughter.)
MILLER: You know, I would just throw in, I actually don’t agree that the wealth’s already been created. I think there will be enormous—there almost will have to be enormous new wealth created. And I think that, as I said, is an opportunity now to think through how we want to do it, and not only taxing the robots but I’ve seen people talk about licensing the owners of robots so that they have to be responsible, just like you are if you’re a driver of a car, for what those robots do. So there’s a lot of things that we can think about now, to get a little bit ahead of the curve.
And on political will, I do think this is an issue which is—we’ve talked about, whether it’s Charles Murray, if he’s not getting shouted down by a university, you know, group, or Martin Ford, you know, we’ve got a lot of people who you wouldn’t expect to express interest in—whether it’s universal basic income—because they recognize you’re not going to have a consumption-driven economy—a mass-consumption-driven economy if you don’t have a group of people who are able to participate.
So I think there is more hope that you will, in fact, be able to bring the political will to bring the two sides together around this issue if, in fact, we start getting the kind of catastrophic job loss. And we are seeing—you know, I think the last election is an early warning sign that there are a group of people in this country who are dramatically unhappy with what’s happening to them. And we need to address that. And so I’m a little more optimistic.
And in terms of Paula’s question, I would just say I totally agree. And by the way, I wouldn’t even argue there should be a science. I mean, we require kids to take languages. And, you know, I love speaking French, but I got to say if I have to choose between speaking French and learning to code, for my daughter, it’s going to be coding, right?
Q: Coding is not computer science. Just want to—
MILLER: Right, but I think—I agree. We should have that.
Q: (Off mic.)
MILLER: We should have it.
LUND: Let’s take another round of questions. But I cannot let the trade issue stand. I would say my reading of the literature, I think David Autor at MIT has the best paper that is out there, how much of U.S. manufacturing job loss is due to trade. And he concludes I think it’s about a quarter. So yes, it did coincide to China, but that is our biggest portion. It’s the vast majority of the manufacturing trade deficit, is in fact with China.
So I have to say, I don’t think—I think that, yes, these two things happened at once. Manufacturing employment declined at the same time trade with China kind of increased. But as you know very well, correlation does not prove causation. And I think that people who have tried to disentangle the effects have concluded that trade is not the main culprit. And at any rate, whatever happened in the past 15 years, the manufacturing jobs that left the United States were very largely labor-intensive textiles and apparel, some furniture production, that have gone to very low-wage countries. And those jobs are not coming back. So the new manufacturing renaissance and the increases in manufacturing employment are now in more advanced industries.
But let’s go to at least three more questions. We can start out here, here, and here. We’ve got one on this side, right over here.
LUND: OK, go ahead. Lindy?
Q: Oh, hi. Lindy Miller with Deloitte.
I’m very interested—I mean, obviously it’s a very complex issue—when we talk about political will and we talk about the skills—and I think you said, you know, the onus might be on the worker to go and figure out what skills they need. I think that’s a very big thing we’re asking of people. So I’m from Georgia. And I’m interested in understanding how you envision this all playing out in terms of the urban-rural divide in this country. And I see a lot of things happening—for example, in a state like Georgia, which has a big divide and also politically along those lines—around technical colleges and universities and their relationship, say, to Chambers of Commerces in their towns or in their cities.
And I’m interested in understanding, how can a worker really know what skills to get, and particularly when we talk about folks who might come from, say, lower socioeconomic levels, how do they actually get that insight without—I think it’s asking a lot of people. So it’s one thing for me to do that and know what skills I need. It’s something for—you know, the bulk of people we’re talking about to go out and do that. And particularly kind of reframing the way we think about our education system—I know that’s a lot in there—but thinking through how to get 15-year-olds and 16-year-olds thinking about that early, and the other options they have, other than the ideal we’ve set to go to university—a four-year university.
LUND: OK, thank you. We’ve got one back here.
Q: Michael Donovan, Inter-American Development Bank.
I appreciate the historical perspective on this discussion. I understand that the—some of the original theorists of automation predicted an increase in productivity, but that would also free up our leisure time, that we would be able to spend more time with our families and communities. In terms of the time use data it seems like the opposite has happened. We have more working hours, less leisure time. How do you explain that paradox?
LUND: Good question. Good question.
Q: Thanks. Barbara Slavin from the Atlantic Council.
I’m curious if you’ve looked at the potential role of the alternative energy sector in creating more jobs, and how that’s going to be affected by our current administration’s apparently allergy toward renewable energy or at least promotion of dirty energy? Thank you. (Laughter.)
LUND: OK. All right. The historical factor, I would jump in and say, look, I think what’s happened is our standards of living have gone up more than Keynes imagined. If we wanted to live at 1930s standards of living, we could probably do it on 15 hours a week. But the fact is, we now have, you know, mansions of houses and numbers of electronics, et cetera. So that may be one of the paradoxes. I don’t know. Or maybe—
Q: A higher standard of living?
LUND: We have a much higher standard—(laughter)—
PORTER: And there’s a distribution, by the way. It’s not uniform—this is not uniformly true. The United States really does have a lot—you know, a lot of working hours. But if you look at Europe, there’s been a very, very consistent decline in the number of hours worked for year. And so you could say that it’s a trade of leisure for money, which if what, in fact—I mean, there’s been several theories proposed about this. One is that Europe has—that high taxation in Europe discouraged work. Then another one was actually that labor law—tight labor laws in Europe discouraged work. So these are two very theses very popular in the United States.
Olivier Blanchard, the French economist, unsurprisingly, disagrees. And he says it’s actually a taste for leisure, a different sort of tradeoff that Europeans were willing to make. But there’s also a difference at the income level. Because as you—if what you say is true in the average in the United States, very large number of working hours—if you look at it, we’ve also seen a big decline in the labor supply of low-educated men and women. So there you’re having your—you’re freeing up hours. You could call it leisure or you could call it, you know, some horrible experience with—
PORTER: —oxycodone and fentanyl. But it’s—so I think at the top end of the income distribution, of the educational distribution, yes, you have more work, because it’s highly rewarded and you’re rewarded in many ways. But there’s an inequality, say, which is working in an interesting way, which is you have—at the bottom end you have less work hours than you have at the top.
And I just wanted to—I mean, I’m not really sure about what the policy prescriptions are, but I think the urban-rural—the locational, the geographic dimension of these changes is very, very important. And I think that our understanding of the geographic manifestation of these changes is quite imperfect. I’ve seen work that suggests there is a really big urban-rural divide in outcomes for—in labor market outcomes and so on. But then there’s other work that suggests that it’s not quite true.
So if you look at, say, Angus Deaton and Anne Case, which put out this, I would say, excellent work on white men and women without a college degree who are really killing themselves at very high rates, either with suicide, drugs, and alcohol, they find really no urban-rural divide. They see it happening in metropolitan area, in large metropolitan areas, small metropolitan areas, and in rural areas as well, which strikes me as a little bit in tension with this sense that the decline in manufacturing jobs really, really hurt these places outside of urban centers, where a lot of these jobs were. And, you know, jobs—things like coal and stuff like that. So I’m—just to say that I think the question is really interesting, and I don’t know that we quite understand it well enough.
LUND: What about—let’s—don’t forget the question too about energy, and dirty energy versus clean energy, and is that a job creator for the future? If we start to think about what will those jobs be—
PORTER: Well, there are 60,000 coal jobs. So Trump went out there and saved 60,000 jobs with repudiating the Clean Energy Act. There’s, I think, more than—
LUND: There’s 150 million.
PORTER: I think there’s more like 200,000 jobs in wind and sun right now. I mean, that’s just from memory, but roughly. So, and that sector is growing very fast. And I don’t think it’ll stop growing, frankly. I don’t really think that ending the Clean Power Plan will do a lot to stunt the growth, because this is going to become a global market. I don’t think this is just custom made for the United States. So I really—I do expect that to—
SHIERHOLZ: But when you look at those BLS projections that you look at, the fastest growing is, like, wind turbine engineers. I mean, there’s—they’re fast-growing because there’s not so many of them now, but it is a—right, you can see the projections. Go ahead, I want to—
MILLER: I was just going to—one counter argument to people working more is we see at the high end of the independent market, one of the reasons people who have really in-demand skills want to be independent is so they actually can make the tradeoff between income and hours. And so you do see—I often have believed, as this market become more ubiquitous, the rate for summer work will go up 25 percent, because a lot of people want to take the summer off, and the same thing for Christmas vacation. And that’s what it should be, by the way, when you are employed in a traditional job—which is why I think hours have gone up at the high end and why you’ll start to see them come down—you don’t have that flexibility. You can’t make that choice. And employers, I think, pay premium to own you. And so I think you get a more fluid, efficient labor market, you’re going to see real variations in what people choose in terms of their time.
And just one comment on your side. I think you’re right. It is a challenge. But I actually think we have so many tools to address it. I think the whole ability to do remote work now on platforms is ultimately a huge way to try to level the playing field between rural and urban. Now, people will have to be educated to take on those, but you’ve got—you’ve got people doing work around the globe on these platforms, like Upwork today—you know, remote—enabling remote work. And I think as we get better at those three things, which is allocating, you know, real information about where the gaps are, having training programs, there will be a way that someone should be able to, like, log in and say, you know: Here I am in rural Georgia. And I want a job. And you should be able to pull up what those jobs are, and where you get the training. And, you know, LinkedIn is working—this is LinkedIn’s vision for the future. So I do think it will actually work for urban-rural.
LUND: One more?
SHIERHOLZ: I just wanted to quickly mention the thing about the hours. Another thing, in addition to what’s already been said that’s just worth making sure we just get out there, is one of the reasons that families as a unit are contributing to the paid labor market is to keep up with the fact that their wages have not been growing. So it’s just—you know, families actually need to put more hours into the paid labor market in order to maintain or see rising living standards. So I think that it sort of loops into this erosion of job quality that we’ve seen, in that sense.
And then the other thing that is—I don’t—the urban-rural divide thing. It just points to your comments about can’t expect people to know exactly how to identify the in-demand jobs, to train for it. It points to this real need for this important, good infrastructure in these one-stop shops that deal well, that, like, having accessible places that people can go and get that information—like, ask for it and get it to help them know how to best put their energies in setting themselves up for their future, I think is a really key part of the solution.
LUND: OK. Let’s take one more round of questions. We have one, two, three, and maybe one in the back four. And these’ll have to be really quick, because we are unfortunately running out of time.
Q: Yeah. How does immigration, legal and illegal, affect the issues you’ve been talking about?
LUND: Great. We have a woman over here, this side. All right. We’ll get you both.
Q: Heeral Coleman with the Fair Labor Association.
I was wondering if any of you are seeing any specific examples of successful strategies governments have employed to address the impact of automation in low-wage and low-skilled workers.
LUND: And right behind you.
Q: I actually—Patrick Theros. I run the U.S.-Qatar Business Council.
I have a question on definitions, which I don’t understand anymore. How do you measure productivity anymore? Is Uber, which has double the number of people driving cars and cheaper and causing taxi drivers to lose money, is that an increase in productivity? Is it a decrease? What is it?
LUND: And then one in the back, yes, right there. Yes.
Q: Bruce Stokes at Pew Research Center.
I do think we are politically short-sighted if we disconnect technology and trade from each other. I think that if you talk to workers, they know that robots were put on the floor of their factories in automobile plants because of the rising competition from Japan, who could use robots to produce cars that were of a higher quality, because human beings can’t produce cars with those low defects. And when they blame trade, they blame globalization, they know why those robots were put—it’s not because General Motors loved automation. It was because of the growing competition posed by globalization.
And if we look forward to the services sector, there are roughly three-quarters of a million call center workers in the United States in just the top four states that have call centers. We all think of India and the Philippines as the centers of call centers. But there are well over a million people in this country work in call centers. If they lose their jobs to automated call center work, because the alternative was to ship it overseas, they’re going to blame globalization. And they’re going to know that they lost their job because the alternative was to send it overseas. And I think that the political ramifications of that for services workers, we cannot underestimate the possibility that service workers will have the same populist backlash to this as manufacturing workers have had to globalization, and they’re tied together.
LUND: OK. Thank you for that intervention.
So we have three questions. Any order that you guys want to take them would be great. You go.
SHIERHOLZ: I can start—I’m fine. So the productivity question, this is a really tricky question, about how to measure it. And I don’t—I am not a productivity expert. But conceptually, it’s just the GDP—total value of goods and services produced in the country—divided by the total hours worked by all people.
So what was your particular example?
Q: Uber and regular cabs.
Q: They’re doubling the number of people out there working and everybody’s income is going down. Is that an increase in productivity?
Q: The same number of people riding in cabs.
SHIERHOLZ: Right. So if they are producing—say it’s the same number of people. So the output is the same. Hours worked?
Q: Doubled, because people are out in the street—(inaudible)—looking for a taxi, or for a—
SHIERHOLZ: So, in that sense, with this little experiment we’re doing—I don’t know if this is exactly how it’s measured—productivity would have gone down, right, because you’re—
MILLER: But there’s more rides being taken.
LUND: I don’t know if that’s how it actually—
MILLER: Rides has actually gone up.
LUND: All right. What about governments who—governments who we could look to who are coping with this better than ours?
MILLER: I think you mentioned—Eduardo mentioned it earlier. I think Denmark. You know, I think they have a policy that includes labor-force flexibility with security, and you need both. I think they call it flex-security, and I think that is a good model. But Eduardo also pointed out they’re spending a lot more money than we are spending on everything—on training, on social network. To me, that would be a good model.
PORTER: I mean, there’s also—there’s bits that you can look at in different places. So Germany’s educational trajectory is perhaps a place to look. So there, you know, at a fairly young age students are moved into two tracks, whether you’re going to go into higher education to get a Bachelor’s degree or higher or whether you’re going to go into more of a, you know, technical training path. And that will—that’s kind of like, well, sort of like—and there’s a good connection between this training, this pipeline, and the demands of German business.
Now, I’m not entirely certain that I like this idea, because this sort of like tracking has implications for inequality. You look at who gets tracked on the—on the, you know, cheaper route, and it tends to be immigrants. And so it’s not perfect. But in terms of kind of like matching skills with demand in the industrial economy of Germany, it has done a fairly decent job.
Just to mention—your question was very short, but the answer could be really, really long. Immigration. So the latest that we know from the—an NAS report of last year is that the net effect on the economy is about—it’s not very large. About $50 billion of the—if you’re looking at the surge of immigration that the U.S. got—I think they start counting in 1980 till now—was worth about $50 billion a year to the native population. So, excluding the immigrants—because for the immigrants they got much, much more, for the immigrants; it’s a huge increase in the wage for people that arrive from Mexico. So if you just care about what’s the impact on the people who are already here, so there was a net impact of 50 billion (dollars) plus.
But this has all sorts of distributional consequences, right? And so you can say—and the data does find—that the people that most directly compete with immigrant labor, which would be folks that mostly don’t have a high school education certificate, there will be—they do see a decline in wages, but the decline is really not very large. It amounts to like a 3 percent decline in this analysis. But this is very controversial. You have folks like George Borjas at Harvard, who has for the last 50 years been arguing that the impact is larger, but other economists like David Card and Giovanni Peri on the other side who think that it actually might be smaller.
So I think, yeah, it stands to reason that there would be some people hurt. The question is, how big a deal is this compared to what the economy gets from this? And what are the best tools to address whatever those vocations are.
LUND: OK. Heidi, any concluding thoughts? And unfortunately, then, we’re over time, and we will have to bring this very interesting and complex conversation to an end.
SHIERHOLZ: No, I’m good.
LUND: Yeah. OK, thanks. (Laughter.)
SHIERHOLZ: I meant to say I’m good.
LUND: You’re good, all right. Well, I think we might have raised more questions than we answered, but it was fun. So thank you. (Applause.)