Skip to Main Content
HBS Home
  • About
  • Academic Programs
  • Alumni
  • Faculty & Research
  • Baker Library
  • Giving
  • Harvard Business Review
  • Initiatives
  • News
  • Recruit
  • Map / Directions
Managing the Future of Work
  • Newsletter
  • Partners
  • About the Project
  • Research
  • Faculty & Researchers
  • Media Coverage
  • Podcast
  • …→
  • Harvard Business School→
  • Managing The Future of Work→
  • Podcast→

Podcast

Podcast

Harvard Business School Professors Bill Kerr and Joe Fuller talk to leaders grappling with the forces reshaping the nature of work.
SUBSCRIBE ON iTUNES
  • 30 Oct 2019
  • Managing the Future of Work

How global trade and AI are resetting the terms of white-collar work

International trade expert and former presidential advisor, Richard Baldwin, discusses his latest book, The Globotics Upheaval: Globalization, Robotics, and the Future of Work. He argues that the speed and sweep of economic and social changes resulting from global connectivity and AI could provoke widespread dissatisfaction. These factors are already influencing white-collar, middle-class employment. Work that can be automated or done remotely offers employers huge potential savings. Jobs that require onsite collaboration and interpersonal skills look less vulnerable.

Bill Kerr: In his new book, The Globotics Upheaval: Globalization, Robotics, and the Future of Work, economist Richard Baldwin discusses the disruption in advanced economies resulting from telemigration and artificial intelligence. He says the combination threatens bedrock, white-collar, and middle-class jobs.

Welcome to Managing the Future of Work podcast from Harvard Business School. I’m your host, Bill Kerr. Today I’m joined from Switzerland by Professor Baldwin, who teaches International Economics at the Graduate Institute of International and Development Studies. Richard is also the founder and editor-in-chief of voxeu.org, a prominent policy website of the Center for Economic Policy Research. Baldwin calls globotics the third great economic transformation—after the Industrial Revolution and the 20th-century computer revolution—only this time, the simultaneous rise of automation and globalization and the exponentially faster rate of potential displacement could trigger an even greater backlash. Baldwin is confident of the underlying technologies’ long-term benefits, but are we witnessing a rocky transition or a bleak new normal? Will there be enough human-centered jobs? And what are the implications for business strategy and public policy? Let’s find out. Welcome, Richard.

Richard Baldwin: Great to be here.

Kerr: Richard, you’re an expert on global trade and also a former policy adviser to presidential administrations. What led you to write this particular book?

Baldwin: What I found when doing reading about the future of work was systematic mis-thinkings about how globalization and automation in the service sector will affect the future of work. So, in some sense, I’m just going down to the village square and ringing the bell, “Look out, look out! This thing may be coming. It may be big, and we really have to prepare for it.”

Kerr: Yeah, and you spend the first 85 pages of the book kind of walking through the earlier steps of globalization. What was important about setting that stage?

Baldwin: This idea that technology really stirs things up, people push back, and then there’s a resolution, that’s nothing new.

Kerr: One of the ways you framed the sort of lessons from history is “Globalization is arbitrage.” Talk us through that particular phrasing of it and how this time is different as we move to the service sector.

Baldwin: When things cost different amounts in different countries, companies exploit those differences by buying or making them in one country and selling them in another. Now that is, in essence, a type of arbitrage. The thing about this arbitrage is that, for centuries, it was only really done in goods. In other words, price differences between goods led to trade arbitrage in goods, which we call trade—imports and exports. Now, come around 1990, with the Information and Communications Technology [ICT], a very different type of arbitrage became possible: arbitrage in know-how—in particular, manufacturing know-how. Factories in the rich countries—the G7, the United States, Europe, et cetera—were bundled together—microclustered, if you will—not to save trade costs, but to save communication and face-to-face costs. Now, when the ICT revolution came along, it became feasible, technically feasible, to take some stages of the factories in, say, Detroit and put them in central Mexico and still have the two stages work together in sync. So, in essence, a new arbitrage started, from about 1990 on, that was the arbitrage of know-how of firms in G7 countries to factories—offshored factories—in nearby developing countries. It has led to a lot of de-industrialization in the rich countries and industrialization of a few developing countries. Now, what I’m talking about as the next wave is arbitrage in labor services. So the idea is now, if you look around the world, the really, really big price differences—probably the biggest in the world—are the wages and salaries of skilled professionals around the world. You’ll frequently find a 20-to-1 difference. For example, I’m a Swiss professor, and if you look at a professor of International Economics sitting in Manila, maybe even with a US PhD, he would—or she would—be earning about one-twentieth of what I earn in Switzerland. Now, we haven’t been able to arbitrage those enormous wage differences because a lot of services require face to face, and that—my point here is that digital technology is changing that reality. It’s way beyond Information and Communications Technology. It’s making remote workers less remote and, thereby, enabling an arbitrage on wages and salaries of professional and white-collar workers. So that’s what I think will be very disruptive, and it’s why I like to think about globalization as arbitrage.

Kerr: You have a term that you’ve titled the book: “globotics.” Hopefully you’ve got a good copyright surrounding that term, because I’m sure others will pick up on it. We covered the first half, the globalization. Tell us about the robotics half of this —or the technology half—that you’re seeing.

Baldwin: Journalists, even academics—definitely governments—at some point use what’s been going on in industrial automation to think about what automation of the future in the service sector will look like. The ones in factories, I call those “steel-collar robots.” And the ones in offices I call “white-collar robots.” So basically, a steel-collar robot is just a machine. It’s a little bit smart machine, but it’s still a machine. White-collar robots are really just software. It’s a very clever piece of software, but it’s software, nonetheless. For one thing—take the United States—less than 10 percent of people work in factories now. So it’s really interesting that factory robots can do more than they could before, but it doesn’t really impact the job market that much. These white-collar robots, this software automation, is affecting the rest of the 80 percent or 90 percent of the workforce. So it’s really, really a big deal. The second thing is, the changes can come much faster. With an industrial robot, you have to build the thing, you have to install the thing. It has to be one place or another. With white-collar robots, it’s just software, so it’s instantly, costlessly, freely, and perfectly reproducible all around the world. The two really big ones are robotic process automation [RPA] ...

Kerr: You have a spectrum in your book that goes from RPA to “Amelia.” Can you tell us a little bit about the range of ways this could be impacting the white-collar work?

Baldwin: Sure. RPA is a little bit like a player piano. A player piano uses a real piano, it just passes through some code, and then the piano physically does exactly what the human player would have done. An RPA is just like that. Let’s suppose that when I did my book tour—and I launched this book in February of this year—I sent an email to my telecom provider here in Switzerland and asked them to add in local calls in America to my subscription. And somebody in Swisscom would have—a human—would have opened up that email, read what I wanted, closed my email, opened up a database in the subscription database, changed my subscription, closed that database, opened up a billing database, and changed my billing, and then moved on to the next. What RPA does is exactly the same thing, using exactly the same software tools—for instance, Outlook—to read emails, the same databases. The big difference is that now computers can read. The machine-learning-trained software robot can open up an email, read a random email, and decide what needs to be done. Then it does all those things exactly in the same way the human does but a hundred times faster. And since it’s all recorded, it’s actually less … it’s easier to pick up the errors.

Kerr: In the same process that was traditionally done.

Baldwin: Yeah. So that’s the low end. It’s basically people, a lot of people, are in the process of what I call “knowledge assembly jobs.” And those are in the process of being taken over very quickly by RPA. And RPA is getting smarter and smarter in the sense that it can read better, it can recognize things visually, it can produce output. And then, of course, there’s chatbots, which many of us have interacted with, where, basically, they listen to you, and they can even speak back, or they do it in writing. And at the very high end are these AI platforms. Maybe one of the more spectacular ones is called “Amelia” by IPsoft. And it has an avatar associated with it. So if you get it online, it looks like you’re talking to a woman. It’s a very presentational, healthy looking woman. That’s what the avatar is. You engage with that avatar in ways as if you were engaging with a real human. The more advanced versions have emotional recognition so they can see whether you’re upset. Now then, even beyond that, you have things like IBM Watson who can diagnose diseases or detect fraud in the trucking department, compare hundreds of thousands of property contracts that a company has signed in hundreds of different cities, for example. So it’s essentially, at the low end, you’re talking about really the low-end white-collar workers being threatened in some sense or certain of their task being taken over by automation. And the higher-end ones are a threat to professionals, including lawyers and doctors and architects and things like that.

Kerr: And for some people, this may be the first moment of anxiety they’ve had, is when they feel their own particular job, such as being a professor or being an accountant or others, are threatened by this technology.

Baldwin: A lot of service-sector workers just are not used to this. They’re not prepared. If you talk to factory workers anywhere in America, they’re all used to automation and globalization. They’ve been competing with robots at home and China abroad for decades now. But many service-sector workers are shielded from global competition, because services were considered to be non-traded, because the service provider and the service buyer had to be together in the same room at the same time, and that was just too expensive to do internationally. And they were also shielded from automation, many of the jobs, because computers couldn’t really think, at least in the ways that we’ve been discussing—reading, writing, speaking, that kind of thing. Computers couldn’t do that before 2016, and now they can. The people in the service sector, I’m afraid, are not ready for it and will react very badly when it does come.

Kerr: You spent some time in the book talking about Upwork and kind of it’s kindred platforms. How do you see that, those platforms, impacting the future of services?

Baldwin: Well, I think they already are impacting. Upwork went public last year, and I think it’s worth over $1 billion now, and its revenue is growing at, I think, 20 percent, 30 percent per year. So basically, I think that will be as impactful on services as eBay and Amazon were on retail services. Freelance platforms like Upwork do exactly the same thing for services. Just to give you an example, last week, I started looking for a personal assistant online, and I was discussing with Upwork what to put for the job description and how much I would have to pay. So then Upwork’s machine-learning-trained algorithm is going to find me a few matches. I will then interview them and hire them. So that’s the first really hard part of telemigration, is finding people abroad who are appropriate to you. And these freelance platforms help with that. Then I just put a credit card up, and there, when they work for me, my credit card gets charged. And that’s really good in freelancing, because a lot of freelancers complain that they don’t get paid. This way, they’re sure to get paid if they do the work. And then there’s ways to manage the workers. For instance, you can see screenshots of their screen while they say they’re working for you too to check that they really are are working for you. That is, in some sense, I like to think about it, the “containerships” of future globalization. These freelance platforms are how service exporters from countries around the world, mostly developing countries, will be selling their services into countries around the world, mostly rich countries, because the price arbitrage is enormous.

Kerr: And going back to your 20-to-1 wage differentials across countries, it suggests that, in future versions, you may even have a greater arbitrage potential for that. As you think about Upwork, it’s still relatively small compared to traditional staffing companies. And kind of building on your containerization metaphor, do you think the Manpowers of the world and these more traditional agencies will be going toward the Upwork-type model? Is that kind of the future of this whole a sector in business?

Baldwin: LinkedIn is using its massive network to try and get into this business of matchmaking of service providers. And if you just look at the basic economics, the basic commercial realities, of this, there’s an enormous arbitrage opportunity to be had. When companies can hire, can reduce their costs, their staff costs, by hiring really talented, low-cost foreigners sitting abroad, other firms will be forced to match that. In business, I think the consulting firms say it’s a part greed and a part fear, and it’s that snowball of greed and fear which drives big, big changes very, very quickly. So that’s what I’m a little worried about.

Kerr: And I want to kind of take you to a quote that you make where white-collar workers are paving their own demise through remote work due to the organizational changes that follow. Help us understand that part of the book’s thesis.

Baldwin: Once our companies arrange things that make it easy to slot in remote workers, they’re going to figure out that they could get some of those tasks done for one-twentieth of the price. So, in essence, when I give this talk, and I ask people to raise their hands, “How many of you telecommute from home?” and then I tell them probably your jobs will be the first ones replaced by telemigrants because you’ve already proven that part of your job, some of your tasks, can be done remotely, and those tasks could be done a lot cheaper than others.

Kerr: I bet that makes your talk very popular.

Baldwin: It makes people nervous. That’s when I say, with this book, I want to make people think harder about the future of globalization. It is not going to be like the last globalization because it’s coming to the service sector. And I like to raise people’s anxiety a little bit, partly because I think people ought to worry more it, but also excitement. If you really are a world-class service provider, you’re selling your services across the world as you couldn’t before.

Kerr: We often have the IBMs and the Yahoos of the world kind of pull back from remote work—they say everyone’s got to come into the office. How do you think about that ebb and flow of these programs to both get people out of the office, save on the building costs and so forth, but then somebody comes in and says, “Nope, we need everybody back in.”

Baldwin: Well, I would say there’s nothing new about that. We tried to move some stages of production to China, it turned out not to work, they moved them back. What you can think about this telemigration is people are essentially unbundling the service-value chain and the physical office and offshoring some jobs. And that’s a learning process, because we don’t really have a very good science of why it’s important for people to be in the right place. Having foreign, remote workers will never be as good as having domestic in-place workers, but it will be so much cheaper that people will arrange things to make it work. Maybe splitting up jobs. Think about your job. How much of your job actually has to be done by you in Boston? And how much could it be done by a telecommuter coming in, for example? If you wrote down a to-do list or a chore list for your job, probably some aspects of that could be done remotely, and I’m arguing that those will be done remotely. Very few entire occupations will be removed, only certain tasks. And when you talked about this ebb and flow—everybody has to come into the office—that’s not everybody. A lot of these tech companies have very large online freelance contractors, and the people who actually are running all these things have to be in the same building. And I think, actually, that is the future of work. I think if you hold on to your job, you will actually be in the same building with people or machinery, but you will be using artificial intelligence—software, robots—and you’ll be using freelancers to spin up your productivity.

Kerr: Richard, thank you for moving beyond that spot where you’re asking me to break down my job and see what we could offshore. Help us think through what you see as the attributes that would make a person more competitive in the future. What’s sort of the human superpower that “Amelia” or that other forms of automation can’t take away?

Baldwin: In a word, it’s soft skills, or the most human skills. First of all, machine learning is what this is all about, because we have incredible processing power and incredibly large data sets that weren’t large enough before 2016. Now, once you get a large enough data set, you can estimate one of these great big statistical models that does the guessing, and you can automate certain tasks. But in order to do that, you need a large structured data set. And what I mean by large structured data set is a data set where the question is clear and the outcome is clear. But many of the things in service sectors, it’s really hard to know what the question was. These are the most human things. Like, you go into a meeting. There may be an agenda, but frequently you don’t really know what the meeting’s about. And even though there may be minutes written down, you’re not really sure exactly what was decided. But everybody knows that meetings are essential process in getting consensus moving forward. Managing a meeting is where you cannot get a large structured data set because the questions aren’t really clear and the outcomes aren’t really clear. Those human things that require dealing with unknown situations—dealing to be innovative, to be curious, to be ethical, to deal with many people at the same time, to motivate many, many people at the same time, to interact in an interactive way in a creative process—those sorts of things—it’s really hard to write down a big structured data set. And, as a consequence, those types of tasks won’t be automated going forward. That’s one line of argument. No big data, no automation. Look for tasks where you can’t gather a large data set, or at least in the next five or 10 years. Who knows, after that. The second thing I think is useful is human’s evolutionary social power is social intelligence. Humans can easily cooperate in groups of a 150. What we’ve been able to do is cooperate. The key to that cooperation is social interaction. In essence, the way we can cooperate to do things is by having a society where we understand. There’s something called the “theory of the mind” in psychology, and what that is is your ability to imagine what other people are thinking. That’s the first level. There psychologists have done experiments where they can show there’s up to six levels of reflection, where I think what you are thinking about what I’m thinking about what you’re thinking I’m thinking, et cetera, et cetera. That kind of preempting and thinking through what other people are thinking is essential to social interactions and motivating people and building trust. Now that happens instantaneously with humans, and some humans are better than others. And robots now, AI—they call it “emphatic AI”—can do that one on one, or at least they’re getting better at it. But once you get several groups, you’re into the world of combinatorics. Once you say there’s 10 people, where I’m thinking what they’re thinking or their thinking and they’re thinking like that, the computational problem explodes, and it’s actually too large for computers that we have now. That’s why managing large groups of people, it’s a real talent. It’s a talent that’s cooked in, and it’s not one that’s explicit and logical that you can write down. That managing large groups of people, I think, is also structurally not possible to automate.

Kerr: In both cases, the technology is one that’s complementing the competitive workers. Do you see any future in which technical change can be supportive or helping those that are less competitive? Is there a future version of our technological progress that’s supporting the lower-skilled workforce in ways that we’re under-appreciating?

Baldwin: Machine-learning database pattern recognition is, in essence, taking wisdom or judgment that used to require tens or 20 years to acquire, turning it into a model in a way that people with very average educations can use it. I think it will create a whole set of semi-professional jobs, say between doctors and nurses. Let’s just drill down on that one, for example. If you take a really clever AI-trained diagnostic software that, say, diagnoses common children diseases—is it flu, or is it an earache, or what is it?—if you have a very good diagnostic tool and give it to a nurse, she’s way more productive than she was before. If you give the same diagnostic tool to a doctor, he’s a little more productive than before, or she’s a little more productive than before, but not huge amounts. In some sense, the fact that this is canned wisdom or canned judgment, and it can be used by people without advanced training, in many ways, I think we’ll see semi-professionals, it’ll take a while to fix it, but between, say, engineers and road chiefs, deciding how much concrete to use; between architects and draftsmen, you know, how to complete a particular floor; between lawyers and paralegals, how to do simple law things.

Kerr: Given your background in globalization, you, of course, have seen the barriers and the tariffs and other stuff that have arisen as people sought to protect various manufacturing industries and so forth. What do you think will be the likely policy scenarios that we’re going to encounter in the future? What’s going to be the way that, as globotics comes to rise, people that are affected by this are going to be reacting?

Baldwin: With the globalization side, funny enough, it is not a whole lot that they can do. Back in 1994, when the WTO was formed, countries all around the world signed the general agreement on trade and services, and technically what we’re talking about here is Mode 1 services—the service producer is in one country, and the buyer is in another country, and that was the basic rules of the road are already set. Most of the rich countries who never anticipated importing this stuff committed to not putting up barriers to it, so there are no barriers. And the countries have committed, it’s renewable, but have committed to not putting tariffs on digital transmissions. So, in essence, there’s no classic way of shutting off this trade. The globalization is coming, and it will be very, very difficult to stop. The automation may lead to lots of pushbacks like we saw on Uber or Airbnb. That is, the interests that are affected by it will use existing regulations to slow down the implementation of the technology. You saw that in the past as well with red-flag laws and featherbedding, where politically powerful groups of workers resisted the automation—not to say stop it, but just slow it down. And I think that will go on. But, personally, I think that may not be enough if it starts to come quickly. If it becomes an avalanche, it will be difficult to slow down with this type of existing regulation. Many office tasks just are completely unregulated.

Kerr: Retraining is often mentioned as something to help workers move from one task to another task or one job to another job in the face of displacement, either from globalization or automation.

Baldwin: Yeah, well some countries do it much better than others. US is not one of the better ones. Countries like Denmark are committed to retraining people, and that works better. Mostly Northern Europe, it works relatively well. This is just people who have one service job are going to have to find another service job. One more kind of optimistic point—that it’s different between services and manufacturing—is that service sectors typically have a skillset which is more flexible than manufacturing workers, and they tend to live in cities where there’s a lot more jobs. There’s lots and lots of work, lower incomes, and more precarious incomes. I think what the government really has to do is help people adjust. In some ways, it’s easier, but I think we just need to do a lot more of it, especially in countries like the US and the UK, who haven’t really helped workers adjust as they perhaps should.

Kerr: Possibly with the remote technologies, we might even be able to link them up to new jobs that are farther away and help them make those transitions.

Baldwin: Yeah, that’s actually happening inside the United States. But remember, that’s the first step to jobs going even further away.

Kerr: Richard, I want to take and think about just 2020. We’re in the midst of the beginning of the presidential election process. Do you think these themes are going to be picked up in the campaigns ahead? Obviously, Andrew Yang has put UBI, universal basic income, as one of his basic platforms. How do you see this story evolving over the next couple of years?

Baldwin: Andrew Yang is putting the focus firmly on technological automation, focusing on service-sector jobs, and he also talks about offshoring of service-sector jobs. I doubt he will get elected, and I doubt UBI, universal basic income, will get implemented. But I think Yang’s approach to blaming people’s dis-ease on technology, rather than China and immigrants, will be very seductive to the other Democratic candidates who are more likely to be elected. After all, you don’t want to deny that people in America are struggling. Lots of people are not feeling very good about themselves. And Trump, the Republican Party, or other parts, picked up blaming immigrants and China for that. Now, if you can put on technology and automation for a third cause of the dis-ease that’s not one that he has, I think that will be very attractive. On top of that, it spills in with the fact that people are feeling nervous about Big Tech’s control of our privacy, the fact that it’s Facebook deciding what is hate speech and what is political extremism. Why should they be deciding? I think if you roll those things all up together, there’s a fair chance that an anti-tech—or “tech-lash” as some of them are calling it—will become a theme in all the leading Democratic candidates.

Kerr: Richard, as a final question, what should I be doing for my six- and eight-year-old right now? As you look toward this future, their world’s going to be very different from my world, and the pace of change is only going to be accelerating. What advice do you give?

Baldwin: The first rule is don’t acquire lots and lots of skills in something that AI is going to automate very soon. Lots and lots of professions involve experience-based pattern recognitions—the classic being radiologists and many types of medical diagnosis or many types of legal reasoning where you have to read lots of paper and then digest it. Focus on the human skills—being able to work together with people, motivate people. The second, I think, is don’t focus on jobs that can be done remotely, because if you can do it remotely, somebody else can do it remotely for a whole lot less money. Focus on group skills, working together in teams. One example, I was at Davos this year, and the head of Cisco’s human resources said that they have stopped evaluating individual workers one by one, and they evaluate the team and the project as a whole. That’s to encourage teamwork. And I think the kind of education where the team as a whole is evaluated is something that we should be thinking about. In a nutshell, it’s soft skills. Now, of course, everybody will have to have a certain digital literacy in order to survive in this world, but that’s probably not the hardest problem for the digital natives that are already coming up. What I do like to push back against is this idea that STEM skills are the answer to everything—more, harder things—because lots of those things can be automated. It’s really the softer things that are going to be the successful jobs of the future.

Kerr: Richard, I fear this may give my eight-year-old cause for suggesting why he shouldn’t do his mathematics homework, as part of the automated future. Instead, recess, more recess! Richard, thank you very much for joining us. Richard’s new book is The Globotics Upheaval: Globalization, Robotics, and the Future of Work. Highly recommend that you get a copy. Thanks, Richard.

Baldwin: Thank you, Bill.

Kerr: We hope you enjoy the Managing the Future of Work podcast. If you haven't already, please subscribe and rate the show wherever you listen to podcasts. You can find out more about the Managing the Future of Work project on our website at hbs.edu/managing-the-future-of-work.

SUBSCRIBE ON iTUNES
ǁ
Campus Map
Managing the Future of Work
Manjari Raman
Program Director & Senior Researcher
Harvard Business School
Boston, MA 02163
Phone: 1.617.495.6288
Email: mraman+hbs.edu
→Map & Directions
→More Contact Information
  • Make a Gift
  • Site Map
  • Jobs
  • Harvard University
  • Trademarks
  • Policies
  • Accessibility
  • Digital Accessibility
Copyright © President & Fellows of Harvard College