Interview With Michael Spence

The 6th Bund Summit is themed Navigating a Changing World, which to some extent echoes your coauthored and very well-received book, Permacrisis: A Plan to Fix a Fractured World, the Chinese version of which has come out not long ago. Our two topics have arguably depicted the common challenges facing today’s world.

Q: In your view, how has the new changes in global political and economic landscape since Covid, including deglobalization and global supply chain restructuring, reshaped the way we cope with existing challenges such as climate change, demographic evolution, social inequity and slowing productivity growth?

Prof Spence:

It's a good question. These things are, in the short run at least, setbacks. They take time and energy to resolve.

The pandemic itself used up fiscal resources. Sovereign debt-to-GDP ratios are higher than they would have been otherwise, because governments correctly decided to protect the household sector and parts of the corporate sector from excessive damage, loss of businesses, and so on.

It's a little hard to add it up, and one can find silver linings in the cloud, but the supply side is now constructed in a much more complicated fashion, and the global economy as well.

These shocks and geopolitical tensions are producing a whole variety of changes, including diversification and some degree of home-shoring in pursuit of national security, all of which are not positive with respect to economic efficiency and growth, and they are contributing factors to the inflationary pressures we face in the west.

In the longer term, it's hard to say. In the book, we say that these powerful technologies available could advance major goals for the whole global economy, growth for the lower-income countries, sustainability for everybody, and productivity growth for places suffering from secular headwinds. But these will take time to kick in.

When you add it all up, with shocks, geopolitical tensions and secular headwinds on one hand, but powerful technologies on the other hand, I'd say the first two will dominate in the short run, hopefully not in a devastating way.

But then in the longer term—I'm talking not about decades, but at least 5 years by the end of the decade—it seems possible that we'll have a more positive-looking picture in terms of trends in the global economy. I believe by then, China will have rebalanced its economy and dealt with some of the issues that are associated with real estate and so on.

It won't be a return to the world that we lived in for a long period of time, where supply chains were basically constructed on economic criteria: efficiency, comparative advantage, trade, technology, and relatively free capital flow. Unfortunately, the world is going to be a more complicated place, and we'll have to adapt to it. Business is pretty good at adapting to it. I think that's where we're going.

Whether or not these trends have a positive or negative effect on the challenges with respect to inequality is very hard to say at this point. You can construct arguments that go in either direction. It's really a question of weights and which forces are the most powerful ones.

Q: With the climate crisis exacerbating, you have warned that delay in our action could result in a nonlinear rise in the economic cost, and we seem to be witnessing this process unfolding at the moment. Do we need a quicker fix to improve the efficiency of our climate actions? How can we find such fix or build more effective mechanisms?

Prof Spence:

We're late in getting on with this. I don't think there's anything we can do now that fully compensates for having been late in getting started. But there are some things we can do.

China has been a major innovator in green technologies, including electric vehicles, solar, battery technology and so on. What's happening now is that concerns about jobs and other factors are causing those products and technologies to be blocked with respect to the rest of the world. This is a huge negative for the sustainability agenda.

I can understand the concerns, but we need to find solutions to this. One direction to find a solution is (for China) to find a balance between exporting products on the one hand, and making capital investments and producing them in a wide range of countries around the world on the other hand.

But the simple approach, which is that western countries are not going to use the Chinese technologies; they will use theirs, which are more expensive. At a time when effective action is urgent on the sustainability front, this doesn't seem to be the right direction.

I'm hoping that whatever residual capacity we have for economic cooperation will help us find a path forward that is satisfactory from the point of view of all parties and advances the sustainability agenda. But we're now living with very, very frequent and fairly severe shocks from the climate side. They become a significant macroeconomic factor in terms of economic performance and so on.

So, the short answer is there are things we can do to accelerate the energy transition. It remains hard, because it requires international cooperation, and we've found that difficult at times. But maybe the urgency will increase the willingness to work hard at it. At least that's my hope.

Q: What suggestions do you have for governments and international organizations (IOs) to step up efforts to cope with climate change?

Prof Spence:

It's very hard for 170 or 180 countries to agree on things. But if you look at carbon emissions, the top six countries or regions accounting for 75% or maybe even a higher percentage of current CO2 emissions are North America, Europe, China, Japan, Russia, and India. There are also a couple of other smaller places like Australia and New Zealand in the higher income categories. A relatively small club of people could make a major difference, and they’re either high-middle-income or high-income countries, so you don't get this enormously wide spectrum of interests and capacities in that group.

So, one way to proceed is to do it in two tracks. We don’t exclude anybody, but have an accelerated track in which the main focus is the main participants and main sources of carbon emissions; then, we try to make sure that the two tracks are consistent with each other, so advances in the smaller club (to put it mildly or put it one way) then create benefits in terms of technology and so on that's available to the other sets of countries.

There is one other thing that's needed. The climate shocks are devastating for the low-income countries now, because the pandemic used up their fiscal capacity. They don't have the resources to buffer the shock. This is giving rise to desperate patterns of migration and etc. We should try to work harder to provide the resources.

It's hard enough. For example, when China was getting started, it wasn’t easy to accelerate growth and keep it up. China is the most successful case, but not the only one. But if you add the pandemic shocks and climate shocks on to the normal challenges of increasing the investment rates and so on, it looks very difficult.

So, another part of the agenda has to be taking that seriously. There have been commitments of around $100 billion per year to lower income countries, but not much of it has been delivered, so that's another item that should be put on the fairly urgent list, or we're just going to leave them behind.

To the point of the “convergence” that we used to talk about, I'm pretty sure it's still going to happen for a wide range of emerging economies in the lower-middle to higher-middle income category, but it may be too hard for the low-income countries. So, it's not that that isn't on the agenda, but it's a slow-moving agenda item, and it needs to move faster.

Q: Many countries are now formulating and implementing their own industrial policies. What have led to this phenomenon?

Prof Spence:

There are several factors.

One principal source of the industrial policies is the “geopolitical tensions”, or whatever you want to call it, and the lack of trust, which have caused national security considerations to be increasingly influential in the formulation of economic policy. What that really means is that you are overriding normal economic incentives and choices in pursuit of some other objectives. And that can be expensive.

A second source of the resurgence in industrial policy is we're trying to change the economic model in pursuit of sustainability. That doesn't happen just by paying attention to market forces. You have to have something that looks like what you might want to call industrial policies, including pricing carbon if you can rig up a way to do it, that changes the trajectory and incentives in the private market part of the economy.

If you don't have prices on carbon, you have to do it another way. The United States is a good example: it doesn’t have a nationwide carbon pricing system, and so the normal direction of the economy tends to ignore the negative effects of greenhouse gas emissions, and the U.S. has to counter that in another way, with a set of regulations and subsidies.

Third, when you get big technological opportunities of the type that we have now, say, in the digital area, you tend to introduce industrial policies to make sure that you're taking advantage of them, that you're at the forefront, that you're investing in both the development and the deployment of the technologies.

James Manyika and I wrote a paper on the productive potential of artificial intelligence (AI), The Coming AI Economic Revolution. But there's a very real chance, absent some kind of industrial policy that's focused on accessibility and diffusion, that you'll see a pattern of divergence, where the advanced sectors like tech or finance are early adopters and beneficiaries of this, while other sectors or small- and medium-sized businesses tend to get left behind.

So, policies that counter that are designed to ensure widespread accessibility and diffusion of the technology and an ultimate impact across the economy, because markets don't do that automatically on their own.

So, there's a variety of reasons why, at this point in history, we're seeing a kind of resurgence of what people call “industrial policy”. Some of them are protectionist, some of them are just designed to stay at the forefront, and some of them are designed to override market forces in pursuit of some version of national security.

You can see this in the Chips and Science Act in the U.S. There's a part of it that just flats out investment in staying at the forefront in the technology. It's a kind of beneficial version of strategic competition with China.

But then there's another batch that has to do with home-shoring at least some advanced semiconductor manufacturing capability. If you look at the current situation in the U.S., the semiconductor design sector is pretty advanced, but the fabrication or the manufacturing of semiconductors is not. In fact, the most advanced semiconductors in the world, as far as I know, are all being made either in the South Korea or the Taiwanese economy.

That's going to change. For China, the U.S., or anybody else, that's not a stable situation. That's the second component of the Chips and Science Act.

The third component has to do with China and denying China access to products and technology that would enable China to advance more quickly in the semiconductor area.

They're all in the same package. There are different objectives, but it's a useful example, because it helps you see that industrial policy isn't just one thing. It's a multi-objective set of initiatives. But there is no question that it's spreading.

Q: What are the good and bad things about industrial policy? What would you define as good industrial policy?

Prof Spence:

Good industrial policy is something that advances the state of the art, an industrial policy where you're investing in or incentivizing basic science or basic technology.

There is an interesting set of statistics from the National Science Foundation in the U.S. In a range of digital technologies, the funding for basic research—not proprietary research, but where people in academia publish openly available papers—unlike most funding of basic research in science and technology, in the digital area, only 1/3 of it comes from the government; 1/3 comes from the major tech mega platforms, on the American side that would be Microsoft, Google, Amazon, etc; then 1/3 comes from philanthropy. It's a very unusual configuration. But anyway, all of that is good.

You can be better or worse in terms of actual design, but anything that industrial policy designs that increases the likelihood of enabling the transition to a sustainable pattern of growth in the energy and related sectors is good.

But the way the three of us thought about the protectionist ones is—here you can have arguments about it—you're going to get some protectionist activity associated with national security, and that's almost inevitable. It's very difficult to see going back to a world in which most technology flows and most sensitive technology is open again.

Actually, it's always been true. Every country has a set of provisions that restrict the flow of militarily sensitive technology. What makes this period interesting is when a lot of the digital technologies fit into that category now, and unlike the old days when a lot of these technologies were developed for military purposes and then had non-military applications, a lot of these technologies were developed in the non-military world, and turned out to be relevant for the military.

So, from a military and defense point of view, it feels like there's a much bigger set of sensitive technologies than we had in the past, and that's probably true. That's why it feels like it's a different world, whereas it's really just an expansion of something that's been there all along.

But the challenge for the major players is to avoid having an excessive pattern of protectionism with respect to goods, services, capital flows, and technology.

On the American side, now that people have gotten the hang of the fact that there's a national security issue, you can use national security as an excuse for protectionism, even when it has relatively little to do with national security.

So, the challenge in terms of international cooperation is to limit the restrictions. We probably have to accept some of them, but try to keep the rest of the system open, not just for the major players, but for everybody in the global economy.

That's clearly consistent with the position that the vast majority of emerging economies have taken. They basically don't want to participate in a bilateral strategic competition with the U.S. and China. They’d rather be an economy as open as possible and pursue their own interests, whether they're with China or the U.S. or somebody else. They said, don't expect us to line up in the game you're playing. I think they're right.

The issue is excessive fragmentation and complexity in the global economy. We have enough shocks now that the pattern of diversification—which can be a pretty expensive one as compared with just going to the lowest cost place—is inevitable. I don't think we're going to reverse that. Even if you took away the geopolitical tensions, you'd probably have incentives to do that just from climate shocks, pandemics and wars that produce blockages all on their own that aren't in the normal immediate sense made by humans.

I don't think there’s a big stretch to that trend, though. Senior policymakers in the U.S. and China have had real conversations about which kinds of restrictions and competition we really have to have, and where can we draw the line and maintain productive cooperative relations, especially in the sustainability area.

Q: What advice do you have, both for a single economy and for international organizations (IOs), to better balance the use of industrial policy and maintaining global economic and trade ties?

Prof Spence:

In the book, there's a fair amount of concern expressed about that. I know Gordon Brown, who has spent the lion's share of his life after being Prime Minister and Chancellor in Britain—he was the longest serving Chancellor of the Exchequer in Britain—involving in international institutions and cooperation.

There are two issues. We absolutely need these organizations as a concrete way of both discussing and eventually implementing international cooperative activity. One of the concerns is that this trend toward nationalism and protectionism (“we’ll go our own way”) has at least created the threat of marginalizing a number of international institutions that we really need. But we don't want to do that, and it’s in nobody's interest.

In order to reverse that trend and limit that risk, first, you need reform of these institutions. The international financial institutions, like the International Monetary Fund (IMF) and the World Bank, need to have governance structures that reflect the economic influence and stature of its various members, and they’re way out of date on that.

Absent that, the newly-arrived big players including China, India and Indonesia, etc, are going to view them as having flawed governance structures and not be able to participate fully.

For example, the Head of the World Banks seems to always be appointed by the U.S. more or less. The IMF, it's the Europeans. That probably should change, too.

Once you get reform, then there's a good argument for capitalizing these entities at a very high level, much higher than the current, so that the World Bank is able to play a major role in global public goods like sustainability, and the IMF can have a bigger role or capacity to stabilize volatility in the global financial system.

But it goes beyond those, too. The World Trade Organization (WTO) has to find a way and needs to be supported in finding a way to reconfigure the rules for a more complicated world. But it also is an important enforcement mechanism, and if the major countries decide that WTO rulings are optional or suggestions, then reform doesn't help very much either.

So, the general point I think that Gordon would make if he were here with us today, is we need these institutions. Yes, they have to be reformed, but we don't want to drift in the direction of ignoring them or pretending that they're not part of the landscape that we need going forward. They're the practical means by which we find ways to cooperate, even in a world with greater complexity.

Q: What implications does increasing industrial policies have on global growth and inflation?

Prof Spence:

I don't think there's a simple answer. That's mixed. Some industrial policies can be easily imagined as pro-growth, for example, industrial policies that advance the state of the art, and technology’s diffusion and accessibility. If it produced a productivity surge in whatever economy it's happening in, that would definitely be pro-growth.

My belief is, we're operating in the west at least in mostly a supply-constraint pattern of growth. That's one of the reasons we had inflationary pressures. The demand side was running out ahead of the capacity elasticity of the supply side. A boost in productivity would be a major positive change in the direction of breaking the supply constraints.

There's definitional problems here. In the U.S., four main funding agencies fund basic research in science and technology: the National Science Foundation, the National Institutes of Health, the Department of Energy, and the Department of Defense. They provide around 70 or 80 billion dollars a year of investment in basic research going to universities, scientists, engineers, doctors, and biomedical science researchers. It doesn't have an immediate effect, but creates a foundation on which the rest of the advances occur as the handoff occurs to the private sector, the entrepreneurs and the venture capitalists and so on.

Some people call that industrial policy, and some people call it just whatever it is, depending on who's talking. But when the intervention in the market system is rather more specific, there are critics of that. They say governments have got no business doing that. For example, “promoting the semiconductor industry” is much more specific than “promoting basic research”, even though the U.S. is doing it and it's a pretty big investment program.

There's no consensus on the details for identifying industrial policy, but where it becomes controversial is usually when the government is doing something like that, and it becomes even more controversial if they're picking companies that they think should be the “winners”.

I guess the implementation of industrial policy does require attention both to the potential benefits, but also to the risks and potential inefficiencies that can arise. There's a long history of industrial policy. Some of them seem to work better than others. For a long time in the post-Thatcher and post-Reagan years, industrial policy had a bad name. But that’s a reasonable summary of where we are.

Imagine a spectrum, where you have a centrally-planned economy on the left, and an economy with almost no government on the right, then the U.S. has drifted to the left a little bit, though we never got there completely. China gave up centrally-planned economies, but it's a bit closer to the left.

My view is that there's a range in the middle that works. There's not one right answer, but you can drift in the wrong direction both ways. You can have the government doing things that it doesn't do very well as compared with the private sector; and you can have governments missing in action where they're really the only ones who can get something reasonably done.

There's no consensus on this, but I think a mature way of thinking about it is, there's a range of things that work. There isn't a particular point on that spectrum that tells you what's the optimal relationship between the state and the markets.

But that's where the industrial policy discussion comes in. Some people believe that the industrial policy pushes you in the direction of having the government doing things that they shouldn't. I don't share that view. I think there are good and bad industrial policies and good reasons for doing it depending on the circumstances, and those circumstances change. Maybe you need more industrial policy if you're in a world that's battling with sustainability than you had before.

Q: How do you foresee global economic growth, including its trend and paradigm, to evolve in the next 2-3 years and longer, 5-10 years?

Prof Spence:

In the next 2 to 3 years or maybe until the end of the decade, we’re going to see the pattern of slowing or slow growth. I don't see a way out of that.

There'll be differences across various parts of the world. If China is growing at 3%, that's considered a disaster; if the U.S. is growing at 3% in real terms, that's considered great. What you have to adjust for is potential. China's potential growth is still higher than in the U.S. and Europe.

The bottom line is, although we now have growth below potential because of imbalances and all the shocks we’ve had, at least there is still potential for significant improvement if you go further out in time.

If we really make progress on sustainability and a bunch of other environmental issues, if we do limit protectionist activity to things that we really have to live with in terms of national security, and if we have more powerful digital technologies, I can see a scenario once we get out past 3, 4 or 5 years where these things start to have a real impact, that looks pretty optimistic.

I don't think we're at the end of these powerful digital technologies coming. It seems more like the start, and it's hard to know exactly where that'll take us, but they do look pretty powerful in all kinds of ways: economically, in producing improved performance in terms of productivity and potential growth; in biomedical science, it looks like there is already a revolution going on there driven by digital technology, which could even enhance the speed of new discoveries.

Q: Technological advance is key to increasing productivity, but as you mentioned in Permacrisis, technology should serve as an empowering, rather than substituting tool, and what we are in urgent need of is augmenting AI rather than labor-saving AI. But the problem is, how can we do that? What does that require of governments and businesses?

Prof Spence:

Let me start by saying, when you look around the economy of the potential uses of the generative AI (gen AI), it looks an awful lot like the complements version rather than the substitutes version that writes the “first draft” of everything. Most people, whether they're doctors or software engineers, probably aren't going to end in the first draft and consider it done. Just coming to understand that you can use the machines in collaboration with people either to increase productivity, or enhance customer experience, or even create new things is step one.

But there is a kind of bias. You can hear senior executives who say, “Jesus, this is great. I can use AI and get rid of all my customer service agents.” That's probably a bad idea, and if they did it, they might regret it later on, but you do hear them say that.

In addition, AIs are benchmarked against human performance. That's how we measure their progress. That's probably problematic in two dimensions. One, it leads to the conclusion that “once the AI is past the average human performance, why don't we get rid of the humans?” That gives you the automation bias.

Also, there are applications of AIs that don't exceed human benchmarks that are pretty useful in the inclusiveness dimension. You can use image recognition software to detect skin cancer, eye diseases and so on. You can also use them in conjunction with X-rays or MRI technology to detect other problems.

They're probably not going to be better than the most advanced doctors in that field. But there are a lot of people out there that don't have access to those services, so even having them as a second-best choice or the first line of defense in primary preventive care looks pretty interesting.

There's a lot of examples of that. If you ask somebody who lives in rural China, would you like to be in Shanghai and have six banks right near you and all the services, they’d probably say yes, but they don't live in Shanghai or have the services next to them. Probably there are aspects of the digital economy and digital financial system that give them access to services that aren't a perfect substitute, but a lot better than having nothing.

So, people get consumed by the employment issue and don't have a balanced view of all the ways in which this technology can be used, where not only profitably but for the benefit of people in various categories.

I don't see a massive aggregate employment problem from AI deployment. Of course, we could be wrong. There is a centuries-long pattern of declining work weeks: if you go back far enough to the pre-industrial revolution era, most people were working 80 hours a week just to stay alive. But we don't have to do that anymore, and maybe we'll get more of that. So, there's questions out there about AI causing massive job losses.

Now, suppose we get the productivity surge. But we won't necessarily use that to just produce more stuff. There are a lot of people in the world, including in high-middle-income and high-income countries, that say we've got enough stuff, let’s not keep buying. I'm kind of sympathetic with that. So, we might do other things, such as arts or leisure.

If you don't think there's a big aggregate unemployment problem, then the question really is, at a more microeconomic level, what's going to happen? In what parts of the economy are there going to be job losses as a result of a big productivity increase?

The theoretical answer is in sectors that have low elasticity of demand. If you have a big productivity increase in a product or service that doesn't have a very elastic demand, you'll need fewer people doing it, so those people lose their jobs, not because of full automation, but because of the productivity increase. And they'll have to find something else to do.

Probably media writing or media copy will require fewer people. People often say software engineers, but we're building our whole economies on software, so it's not obvious, even if we have a big productivity increase and we're going to need fewer software engineers.

So, it's nuanced. If you look at the economy as a whole, it'll be disruptive, and people will be anxious. That's always true with any big transformation. Focusing on helping people in various categories of jobs and sectors get through this with new skills, new jobs and so on is the most constructive way of thinking about it. First, to admit that it’s disruptive and not easy, and then focus on making it as easy as possible.

Technologists often say, on your mobile phone, you've already got a whole lot of AI, and it's been there for a decade. It's just we're going to have more of it, and it'll be more powerful. There are lots of interesting things, for example, you can put the gen AI large language models in a robot, and that makes the robot capable of interacting with humans. Whereas in previous versions of robotics, they're either fully autonomous or they don't work. That seems like a big change.

I would imagine that over time, we might see a significant expansion of the footprint of robotics as assistance, even in fairly unstructured environments. Human beings have what Feifei Li calls the spatial intelligence. She and her and fellow researchers now try to develop the machine equivalent of that kind of spatial intelligence, but that's a big challenge that's on the horizon rather than a done deal.

Q: Is there any other issue you deem important to add a few words on?

Prof Spence:

I think we've covered most of the ground. Probably the biggest impacts in the short run of AI technologies is going to be in the sciences, including biomedical science, material science, probably also in the energy transition technology, mainly because the scientists can pick it up and run with it. They don't have the complexity of building new business models or retraining people, which tends to produce longer time horizons with respect to big change in the way the whole economy functions.

But those are pretty exciting opportunities. Alphafold developed by Deepmind now has three versions capable of predicting the three-dimensional structure of proteins. I talked to a leading biomedical scientist, a Nobel laureate, who said there are 2 million biologists using it every day. So that's a place to look both for progress and for investment opportunities as well.