The World Economy Against the Backdrop of Increased Trade Tensions

Date May 28, 2018
Speaker Richard BALDWIN (Professor, Graduate Institute, Geneva / President, CEPR)
Commentator NAKAJIMA Atsushi (Chairman, RIETI)
Moderator KOMETANI Kazumochi (Consulting Fellow, RIETI / General Counsel for International Legal Affairs, Trade Policy Bureau, METI)
Language(s) Japanese/English (with simultaneous interpretation)

The world seems headed for a trade skirmish – if not a full-blown trade war – between the two largest economies. But given how integrated the world of trade is now thanks to the GVC revolution, the impact of higher US-China tariffs may have unintended consequences. This talk starts by discussion how 20th century trade tools – like tariffs and quotas – can have unexpected effects in the 21st century. It then goes on to discuss some of the political economy of trade war in the 21st century and the like reactions.



Richard BALDWIN's Photo


I think globalization has changed and will continue to change. This is important because much of the public discussion—including the rise is trade tensions—tends to assume that that globalization will proceed as it did before, but it will not. It has changed quite dramatically in the last 20 years, and it will change going forward. My basic message in the introductory part of my talk is that the all of the public discussion surrounding the rise of artificial intelligence (AI), software robots, and machine learning have missed the fact that these revolutionary technologies will also change globalization.

The new digital technologies are driving automation and globalization at the same pace, and they are affecting the same jobs, which are service sector jobs. In the past, both globalization and automation were mostly about manufactured, mining, and agriculture goods. Digitally-enabled trade in services will move globalization in a new direction, namely, into the service sector. Since the workers in these sectors have been largely shielded from both automation and globalization until recently, the rapid advance of globalization and automation in services will be especially disruptive since the service sector workers are not ready for either.

Dynamic duo of transformation

Automation and globalization have been the forceful drivers of dramatic structural change in the global economy. A technological impulse triggered the steam revolution in 1720. For almost a century, the transformation was due only to automation; modern globalization didn't really start until 1820. The introduction of mechanization made it easier for people who used their hands to do things in the fields and in the factories. The technical impulse back then—basically steam power and all of the mechanization it fostered—had very little implication for people who worked with their heads. The big productivity gains came in the primary and secondary sectors, not the tertiary (service) sector.

Another important aspect as to how this "dynamic duo of transformation" (automation and globalization) happened during the Great Transformation is the great lags between the two. First, automation came around 1720 (with the first commercial use of steam power) and then globalization around 1820. After 1820, the dynamic duo of automation and globalization fed into each other. Automation advanced more rapidly in the G7 nations since markets were expanding, and the heightened competitiveness that came with automation fostered greater trade.

A second massive transformation started around 1973—I call it the Services Transformation to distinguish it from the industrial nature of the Great Transformation (1720-1973). I choose 1973 as the year that marked the end of the Great Transformation and the beginning of the Services Transformation since that was the year that the "computer on a chip" was patented by the U.S. company HP.

Before computers-on-a-chip, we had computers which were very large. They were useful for some things but difficult to use in factories with machinery, like robot arms. Industrial automation advanced rapidly for many reasons, but miniaturization of computers was a key one. But importantly, the computer advances not only affected manufacturing. The technology provided better substitutes for people who worked with their hands, but also better tools for people who worked with their heads, i.e. service workers. This, in my view, is why this second transformation involved a shift of workers from goods to services. As far as automation goes, the technology was a substitute for workers in industry, but a complement for workers in services.

1973 is when the automation aspect really caught on. The "robotization" of industries from 1973 to 1990 was very disruptive; resulting in big gains as well as big losses. Then, technology favored, as I said, those who worked with their heads and substituted people who worked with their hands. This, in turn, drove disruption inequality, and people moving into service jobs. The inequality arose since the job-displacing effect fell on workers whose incomes where already below average (factory workers), while the productivity enhancing aspects benefited workers whose incomes were above average (professionals).

And then the technology impulse, the information and communications technology (ICT) revolution, passed from automation to globalization. From 1990, the ICT revolution allowed the dispersion of stages of production from rich countries to poor countries; therefore, it started changing globalization. In many various writings, I call this the second unbundling (the steam revolution drove the first unbundling).

Looping back to today's trade tension, the main point of this background is that the Donald Trump administration in the United States and the Theresa May administration in the United Kingdom are conducting policy as if the second unbundling never happened. They seem to be thinking about globalization as it was before 1990—as if there were no global value chains (GVCs). That is why I think they were misthinking things and causing difficulties for themselves and everyone else. Tariffs work in one way before GVCs; they work in a very different way after GVCs.

I want to shift the focus to what will happen to globalization and automation going forward. I think this will happen quickly. We will switch to thinking about future globalization, which is in the service sector. That will come to dominate a lot of discussions in the coming years.

Today's transformation

Forbes called 2016 the year of AI. The digital technology revolution provides a new technological impulse which drives a new type of automation and a new type of globalization. The automation from 2016 onwards was driven by machine learning and big data. Machine learning has a long history; it was not effective because our data were not big enough. From 2016, they started doing amazing things with machine learning.

Why do I distinguish between ICT and digital technology? Before machine learning, computers could think but only in a narrow way. Psychologists think of human mental processes in two ways: system 1 and system 2. They are called thinking fast and thinking slow. Thinking fast is instinctive and subconscious, like when you catch your balance after you stumble. Most of our mental processes are instantaneous. Thinking slow is when you are trying to make a calculation in your head or analyzing a decision.

Before machine learning, the only thing that machines could do was think slow because we had to teach them how to think fast with computer programs. Computer programs are step by step instructions about how to do something. Machine learning is allowing computers to think fast. It is teaching computers to imitate human thinking capacities in a way that they could never do before, because we didn't know how we did it; therefore, we could not teach computers how to do it. Machine learning, in essence, writes its own program through a statistical guessing process.

The other big change between ICT and digital technology is that of the physics involved. The physics of machine learning is the physics of electronics, not the physics of machines. Until now, automation had been limited to industrial or agricultural mechanization with robots. However, this new type of automation relies on software which is instantly and perfectly reproducible. The constraints applicable on the speed of automation in industries will not apply in the service sector because the physics involved is different here.


Machine learning is the jet engine; big data is the jet fuel. Digitally-enabled services trade will be a big thing in the next few years. This involves some things called "remote intelligence" or "tele-migration." I think it will advance rapidly. It will bring globalization to a section of the economy where it has not been before. Tele-migration is international freelancing. It will go to many other sectors. Four factors related to digital technology will advance tele-migration:

1. Domestic telecommuting
We and our companies are making it easier for telecommuting. It could become global; now, it is mostly domestic. The whole world is going towards remote work but mostly domestic. Japanese culture is very much an office culture, with regard to work. However, if Japan could see the efficiency and benefits of telecommuting, and adopt more of it, then it could change its work style and work culture in a positive way.

2. Online matchmaking platforms
These are like eBay but for services. There are already large successful companies where freelancers can register and find work around the world. One can hire, manage, pay, and fire someone using those websites. This is, in essence, the containership for this new business. These containerships are making it easier for companies anywhere in the world to find cheap and fast employees to get started on projects immediately and complete them as soon as possible.

3. Machine translation
Machine translation has become good and quite fast in both spoken and written language. Barriers to trade in services are going to fall. Machine translation is going to change the world, and it is getting good and fast. It will have a huge impact on the world of trade in services. Languages in which there is not much hand translated material online are insulated from this. There are not yet large data sets for these languages. Japanese is a good example. Since there is not much hand translated material available for the Japanese language, the machine translation services are not strong yet.

4. Advanced telecommunications
Telepresence rooms and robots make it seem as though people are in the same room. These are important for someone who could help you with computer systems or phone systems, for example. People could use robots to navigate around rooms and look at things. It is almost as if the person is in the room. This could have a tremendous impact on many industries, such as the medial field. A doctor on one continent could operate through remote-controlled robots on a person on a completely different continent.

Where is this going?

What jobs will be left? We cannot know because they do not exist yet. That happened in the great transformation. In 1973, when one-third of Americans relied heavily on industry work, we did not know what jobs their children would have because they did not exist yet. In the same way, we do not know what the jobs of the future will be because they do not exist yet. We can think about what tasks they will involve; they will do what the white-collar robots and tele-migrants cannot. If it is subject to machine learning, those tasks will be done by the machines. If it is something that can be done remotely, those will be done by foreigners. What is left is a more local, more human, cleaner, and richer economy. We cannot be sure we will get there without an upheaval or backlash.

The disruption that happened in manufacturing and agriculture will happen in the service sector, and that will be very disruptive for the rich countries, especially the United States. I think we will see a backlash against that. So far, the backlash has only been on automation. In Britain, there is an anti-Uber or anti-Airbnb movement. The U.S. Congress is passing a law to allow self-driving cars to be tested on U.S. highways without a permit, but not trucks because a teamster union got Congress to take the trucks out; there are too many jobs there. I think you will see that kind of backlash happening. Soon, it will come to online freelancing because I think that online freelancing will strike many people as completely unfair competition.


Q1: How would this whole situation of globalization related to ICT affect technology transfer, cyberespionage, information control, data, and AI? What would the effect be?

The first way this would be affected is technology transfer. In the 1980s, there was technology transfer between Japan and Korea. There were individuals from Japan going to Korea to help with the transfer. You could imagine that this kind of foreign freelancing would make that easier. If someone wants technological expertise that exists in the United States, but you are in Mexico, this is going to make it much easier to get an expert to provide that remotely. I suspect it will facilitate technology transfer. If you think about the past experiences of technology transfer, it involved people going to places and telling people things. A lot of technology transfer is people teaching people how to do things. This is making it easier for people working abroad.

Regarding cyberespionage, I do not have much to say about that. If things get easier to move digitally across borders, then cyberespionage becomes easier to do, especially if machine translation makes it so that everyone can understand languages. The protection of data is a critical aspect. Now, it is the Wild West. Some of these freelancing platforms have so much private information of the people who register. The General Data Protection Regulation (GDPR) is kind of a mess, but at least it started the conversation. I do not have anything particular to say about the GDPR, but it is a good start and it is a long way from perfect. It is good that we started having that conversation about privacy and openness.

Something more closely related is that the e-commerce initiatives are addressing these kinds of data issues. I know Japan is leading a group of countries to try and establish multilateral rules on e-commerce. The e-commerce that most people are talking about is goods. In the future, we can expect more rules to govern e-commerce and regulate the digital economy.

Q2: What is your view on the way U.S. President Donald Trump is approaching the world in terms of negotiations? Is it the right approach?

The approach you have seen with Brexit and Trump's trade policy is based on a view of the world that was true before 1990. I think a lot of problems that it is causing is based on the misthinking of globalization. The people involved are incapable of viewing the world in terms of being this new global value chain world. You need to import intermediates from China to produce manufactured goods at a reasonable price. That is the reality of today; every country who is a big player in the global economy does it.

A classic misunderstanding is if the United States makes itself the only place in the world where components are expensive, U.S. companies can produce things elsewhere. One of the things that is interesting is the lack of strategic prioritization in the U.S. trade policy. They seem to have picked a fight with Canada, Mexico, China, the European Union, and Japan at the same time. If Trump were strategic, he would have chosen one fight at a time. Instead, he went to all at once. That is not something you would have done if you had strategic thinking about national priorities.

It is clear that it is not a well thought out plan. A better interpretation is that nobody thought it through. Most countries playing against the United States realize that the midterm elections are important and coming soon. So, they are aware that the Trump administration is vulnerable to troubles. That is another sign that it was not thought through strategically.

My view of the U.S. trade policy is that it is driven by very stable forces; I view it as driven by vested interests. Those vested interests have not changed. Trump has very particular interests in trade and is being held back by the vested interests. The only thing he really did was pull out of the Trans-Pacific Partnership (TPP) agreement. He has backed off on some trade and tax policies because he has realized this is not the way to bring jobs back to the United States.

There is a history of accidents and that is what we have to worry about. If nationalism gets rolled up into this trade policy, I am worried that Chinese nationalists will get wrapped up in the retaliation. I do not think this will come to a trade war. If it does, it will be bilateral; the United States and China. What we used to call trade creation would now become trade diversion between the countries and everybody else would benefit through trade creation. If the United States and China exchange these big tariffs, it would be disruptive because of the supply chains.

Q3: Do you think that AI could take over the role that entrepreneurs play now?

I am not an AI expert, but I have read a lot about it recently. Entrepreneurship and management are some of the least automatable ones. The real problem is that there is not a big data set. To train a machine learning algorithm, the question has to be clear, the outcome has to be clear, and you have to have millions or billions of observations on this. Entrepreneurship is almost by definition encountering unknown situations. So, there is no data set, and I don't think there ever could be one. In order for AI to really work well, there must be millions or billions of units of data that AI could use to study and learn. That type of data is not available for entrepreneurship and management. What could help is using AI as a tool to check through things. It could be a good tool for entrepreneurship and management, but not replace it.

Q4: In the past view, globalization brought about divides. The technology revolution brought divides. Globalization will expand the divide. How do you see this? Do you think anti-globalization will expand further than today?

The steam revolution helped people who work with their hands not heads. The second revolution provided substitutes for people who work with their hands and new tools for people who work with their heads. This new revolution provides very good substitutes for certain types of mental processes. Where the question is clear, the outcome is clear, and there is a lot of data, automation could help.

Many lawyers are being replaced by "robo-lawyers" which can read through so much information and summarize it at a much lower cost. It is allowing less skilled workers to act as though they had decades of experience and are very smart. AI is providing substitutes for some of the highest income people in the world. It is proceeding in law, medicine, journalism, and finance. Bond traders are being replaced. Financial analysts are being replaced. A lot of low educated people may be less affected by this.

What would be upsetting is that many people will have to change jobs. Robotic process automation (RPA) is changing the world. There are many office jobs that are essentially information assembly line work and RPA replaces that. It is growing extremely fast, and it will replace lots of office jobs. The social instability is that the office workers will join the factory workers.

I am not sure that the backlash will focus on anti-globalization. Considering automation versus globalization in the old days, automation replaced a whole set of workers, but it provided more tools for the higher educated people. When the global value chain came along, the parts of the global value chain that were moved abroad were the ones that were intensive among the low educated workers. All of these things lined up; the automation, the old globalization, and the new globalization, more or less, hurt the same segment of society in advanced countries. That correlation will break down.

Q5: Regarding the trade policy issue, what is your possible prescription to make the trade policy tension go in the right direction?

First of all, we are talking about the future and making predictions on political responses. There is no right or wrong answer because it is not easy to know when we are right or wrong. Moving more offshore for U.S. production will probably be how U.S. manufacturing will respond.

The U.S.-China friction is the really dangerous issue because it is possible that they get nationalism involved and do not back off. The tariffs they announced are bilateral. If you lower the tariffs on the inputs, you raise the protection on the downstream goods. If you hold the downstream tariffs the same, you lower the tariffs on imported intermediates; that increases the protection of the local value added. This is called effective rate of protection.

As the world became fragmented, we found strong political economic tensions to lower tariffs on intermediate inputs. As you raise the tariffs on the imported intermediates, you lower the tariffs on the final goods, which is what the Trump administration is finding out now. I am confident that it will not spill out into global warfare because we are too connected. In essence, we have all exchanged hostages and we all have hostages, so there is no use getting in a war.

Q6: I wonder if there is a possibility to have telepresence similar to the images you showed in your slides. To realize this situation or to avoid this situation, we may need to speed up the whole debate. What do you think?

There are no answers, but we can discuss it together. Automation and globalization will make us richer, happier, greener, more human, and more local. The question is how do we get from here to there? In the meantime, a large number of people will have to change jobs. I think much of the public discussion about being upset about globalization will shift to the service sector.

The primary problem is domestic complementary policies. To a large extent, the proper reaction of the government is not trade policy but domestic policy. Policies should be made to help people adjust going forward. I think the services disruption will be more disaggregated. Helping people retrain, find new jobs, move to where there are jobs, and support income are exactly the same as the past. We just need more of it than we did before.

*This summary was compiled by RIETI Editorial staff.