Rethinking Industrial Policy with Nintil
Can government support for research actually slow down progress?
Yesterday on the podcast I published a, dare I say, scintillating interview with José Luis Ricón Fernández, an independent researcher who runs the blog nintil.com and one of my favorite thinkers on the internet. He took issue with last week’s ChinaTalk on US industrial policy with Rob Atkinson and the simplified narrative on government getting all the credit for innovation.
We discussed whether the state really invented the iPhone, lessons from the Soviet experience in the space race, why the way we fund science is broken, what mass conversions to Mormonism would mean for the American economy, and how to live forever.
Below, I featured a few excerpts most relevant to China and US industrial policy. Click here for a link to the full podcast. Be sure to read down to where José makes the case that under certain circumstances government R&D money can in fact retard scientific progress.
What Rob Atkinson Missed On How Industrial Policy Works
Jordan: So you take issue with Rob Atkinson, president of ITIF and guest on ChinaTalk a few weeks ago. What does he get wrong about the extent to which industrial policy has contributed historically to the West’s technological and economic preeminence?
José: He wrote this white paper that talks about the way you should react to Chinese industrial policy, and as part of that he briefly touches on the history of industrial policy in the U.S., arguing that the U.S. has been doing industrial policy since its origin, people typically point to Hamilton as evidence of this.
From there, you can either say that it contributed somewhat to the U.S.’ success, you can say this was key to the success of the U.S., or you can say it was detrimental, it cost more than it got back from those policies. The U.S. has indeed engaged in industrial policy (IP) since the beginning, tariffs ranging from purposes from raising revenue to restricting trade so that the U.S. could protect its infant industries in various sectors, like ship building, or at one point machine tools. There is no disagreement that the U.S. did that, but now the question becomes: was this effective for the long run success of the U.S. in those sectors? And then second, what was the counterfactual?
And here’s where we are running into the general problem of industrial policy.
The latest review of the whole literature around IP was by Nathan Lane in 2019. It’s called “The New Empirics of Industrial Policy.” And if there is a theme around this literature, it’s that it's extremely difficult to tease out the extent to which IP works.
And indeed Atkinson himself grants that economics cannot really judge industrial policy because it's kind of holistic in some sense, it can touch the entire economy, it can change the advantages of countries and other things.
I think that economics can be used to assess industrial policy and if it cannot, then in that case, we should probably be more agnostic about it actually. But I'm not implying that just because there are no randomized trials that we should just discard the whole thing. That would be committing the parachute fallacy, that just because there aren’t randomized trials whether or not parachutes work we should be agnostic with parachutes. Parachutes of course work.
My beef with IP discourse is that it's very one-sided. You have people saying, maybe more on the free market side, “Oh, it never works. There are the loose papers saying, ‘It didn't work in East Asia and it didn't work in the West and it will never work.’” On the other hand, you have, “No. Actually, it's great. It works for developing countries, for developed countries, and in East Asia, in the West and it's the best thing since sliced bread.” There are indeed studies that say that it worked and it didn't work, but it's important to consider the whole thing in the aggregate and then see what these studies actually say.
This latest literature review I was pointing to by Nathan Lane, let me just quote from the end of his paper. He says, “Nations have and will continue to shape their industrial destinies. Nevertheless, even with the recent papers reviewed in this study, literature in industrial policy is still thin, dwarfed by the attention these policies receive from policymakers. This review is a proposal to take these interventions more seriously as subjects of inquiry.”
I'll be very careful with being overly optimistic about IP. If there is one conclusion from the recent empirics of IP it’s that it's pretty much dependent on which industry, which country, in which period of development it is applied. I guess it's kind of like a death by nuance to some extent. I think it's true that it worked for East Asia. And I think that some of it can work for the current U.S., even though those things won't be the same things that worked for [East Asia]. I don't think the U.S. should just go all in on steel or heavy chemical industries. I think that the IP for the future will be very different from what it has looked like so far.
Did the US Government Invent the iPhone?
Jordan: Before we talk about future policies, let's do a little more historical investigation. One of the books that you've taken issue with is Mariana Mazzucato’s The Entrepreneurial State, a centerpiece of the literature arguing that the U.S. government and U.S. government research money was integral in creating the world we live in today. What issues do you have with that narrative?
José: There are different ways one could read the book. Suppose that you believe that innovation is just a matter of scrappy entrepreneurs in a garage and that's how you get everything. That is, of course, fake news. And the book is a good corrective to that view in that it shows that, indeed, governments and, in particular, the U.S. government have been funding lots of innovation.
Now that's the tricky thing. When we say that “the government did X or DARPA invented the internet,” what do we mean by that? And that's where things get complicated. One example is the iPhone.
Some people say the state invented the iPhone. Now there's this chart that has its title, “What Makes The iPhone So Smart?” and we can see things like lithium-ion batteries, liquid-crystal displays, microprocessors, multi-touch screens, and all those things feeding into the iPhone.
In those boxes, for example for multitask screens, we can see DoD, CIA, NSF. Whereas for, let's say, HTTP or HTML, which is basically the underpinnings of modern websites, it says CERN. So from this chart, we may think that it says DoE or DoD it means that there was some kind of program, that these departments of the U.S. government consciously designed or fostered the development of screens. Hence the state is not just a patron of basic research but in addition consciously guided the development of this technology.
What Mazzucato is saying is not just that the U.S. government has funded research or basic research, which is the kind of externalities or public goods problem that most economists would say the state should intervene in to fund more basic research, but that the state has done more than that. In the multi-touch screen's case, for example, if one digs into the actual story of what happened there its that there was this guy called Wayne Westerman and John Elias at the University of Delaware.
Westerman was a PhD student under Elias and they were studying something else. They were studying neuromorphic systems, whatever those are, and they were indeed being funded by the NSF and the CIA.
Now, this is not unlike the way most basic things are funded. You have agencies and they give money to researchers to fund things. But here they were not funding multi-touch screens. In the history of multi-touch screens, they stress that they were working on something else. And then in getting to their actual goal which they were paid to do, they developed on their own these multi-touch systems to aid them in their research.
But it doesn't end there. If you can keep tracing that innovation back, you get to this guy called Bob Boie at Bell Labs who had actually invented multi-touch screens before them. But you can still keep going and find this guy called Nimish Mehta at the University of Toronto in Canada, who developed a multi-touch system before even the Bell Labs guy. Although in his case, he wasn't able to touch screen.
In general, you can separate the extent to which it was basic research literally funded in a more blue sky ways or without a lot of intention behind it. You can also point to the fact that it happened in various places at once. So one specific program wasn't completely required for actual development.
Another example is that Maria Mazzucato will say, not only did the state somehow consciously bring all of this into existence rather than fund the underlying ecosystem that actually developed all this, but actually the state funded, directly, Apple: that Apple had a God called state venture capital right at its inception.
There is some truth to this, indeed. The naïve story was that the state invested half a million dollars in Apple in 1978 or something like that. Well, if one looks into that, actually it was not the state. It was the Chicago Investment Bank, a corporation, that did that. This bank was the Small Business Investment Company, which means that the loan they gave to Apple was kind of secured by the government, so the government was subsidizing them to do that. But again, it was the bank that decided to give the funding to Apple. It was not a conscious decision from any [government] agency.
Then on top of that, Apple was already pretty well funded anyway. So by the point where they got that money, they already had gathered...
Jordan: It wasn’t tipping point money.
José: Yeah, exactly. It was just like one more investor in Apple, it was not what either made or broke Apple. I think that that's the general theme with the book: the real facts are broadly correct but the interpretation you can draw from them seems somewhat tilted towards the point of view the book is making.
So what makes the iPhone the iPhone? It's not necessarily the touch screen or its square shape. There were already phones with multi-touch screens. It was the combination of design and software that basically made the iPhone.
Jordan: And there were plenty of enormous multi-billion dollar companies that could have put all these things together at that time.
José: Yeah and Apple was mocked. Steve Ballmer from Microsoft said that people will never buy a phone without a keyboard because people want to type with actual keys. So I think that the allocation of credit to a state and Apple has to be put in that context in that yes, indeed, there were those basic science investments were there, but the key point with further investments is that they were not purposeful. They were not done with the intention of, “We want to champion this technology and I have an overarching program to make iPhones.”
Basic vs. Applied Research
Jordan: To what extent do nations capture the benefits of the research that they fund, as opposed to it going into a global commons that makes all boats rise?
José: In the very long-term, scientific publishing is made available to everyone. The basic research that the U.S. does can be used by a firm in China. China has been investing a lot in applied research because the U.S. is funding a lot of more basic stuff. That makes sense for them, given that they are catching up to the frontier [of science]. But the U.S. cannot afford to say, “You’re piggybacking on basic research.” It would be difficult to stop that breadth of diffusion.
Now, that process is not equal. It doesn't work equally for all countries. You need the know-how and the background industry to actually leverage that research. You could go online and read as many semiconductor papers as you want, but you're not going to be making 5nm chips in your garage. You need to have fabs to do so. there's an introduction between having the capability to absorb the research (having scientists and engineers and industry that can actually take that research in) and actually having access to the research.
You will end up diffusing [research] but the extent to which it ends up being used will depend on a range of things, including preexisting advantages and some industrial policy as well. For example, if China didn't invest heavily in semiconductors, they wouldn't see a homegrown industry there and they wouldn't be leveraging U.S. research in that field.
But if you don’t have a good environment to have businesses working on that, and you don't have engineers and scientists and universities to do the research, it’s difficult to get anything off the ground, even if you naïvely read the papers and have the knowledge.
Jordan: What do you think of the argument that government funding in certain fields can, in fact, retard scientific progress. I'm thinking about the Where Is My Flying Car book and the arguments he makes about nanotechnology.
José: That's an interesting argument that rarely gets made. In most economic models, you throw R&D into the bucket, and then out comes more economic growth. Decreasing marginal returns more R&D means increased productivity but at a slower rate. But only a handful of people are thinking about whether it could be the case that funding more research could actually be detrimental.
There are some cases where this might be true. One case would be in the space industry, actually. Historically, the U.S. military has paid a lot of money to U.S. defense contractors to build rockets to carry military payloads into space. Likewise, the U.S. government financed the space shuttle.
When you have the defense sector involved, you usually end up in a situation where the defense sector doesn't tend to care much about costs. They care about performance and getting things done without those things necessarily being cheap, which leads to rockets being stagnant in an era where they work, they are robust, they are safe, but they don't come down in costs.
Then Space X comes in (with some help from NASA under their commercial apprenticeship program).
They say, we're going to make rockets super cheap and we're going to make them land and all that. This is something that doesn't make sense if you are a defense contractor. It's hard to imagine that the U.S. would not want lots of military satellites, but suppose we didn't have that. Then it could be that industry could have geared itself towards more cost efficient rockets that cater to the civilian world.
Another area would be the Manhattan Project and nuclear energy, which may sound surprising. Nuclear energy is clean, very efficient, and a generally great form of energy.
We owe nuclear energy, largely, to its origins in the Manhattan Project. Its objective was to build nuclear weapons. That it was funded in that way means two things: one, it was born “tarnished” by the association with the military; second, it was rushed into production. In a way, we got nuclear energy before quote “we should,” which means that we got, in the case of the U.S.S.R., Chernobyl, and in the U.S., Three Mile Island. These are isolated incidents. In general, nuclear energy is very safe. But in the public imagination… people got very scared about those things.
Now you could imagine an alternative world in which there is decreased funding for nuclear energy from the government, or in a more laissez-faire way with various labs working on it independently. At some point, Westinghouse picks up the development of nuclear reactors and it develops more slowly. It could be that we wouldn’t have had all these disasters and that people would be more optimistic about nuclear energy, leading to more nuclear energy installed and less coal and fossil fuels being used.
We will never know the extent to which this is true because it's past history and we will never have another case where we would reinvent nuclear energy.
Another case of this might be in fusion energy. There are a bunch of scientists that have written editorials to Nature that have complained that most of the funding for fusion has gone to one project, ITER. If you want to work in fusion as an academic, to some extent you have to work in this one paradigm because that's where the money is.
This has stifled the development of other approaches, which may or may not work. But I think having some diversity of funding is good. If you are just channeling money in one direction and your winner fails, then you're left with nothing. I think it really pays off to find diverse ways to fund, instead of trying to focus funding in one thing.
Jordan: Yeah, it's interesting because that's sort of the Chinese industrial policy model over the past 15 or 20 years. You give everyone a ridiculous amount of money and then you have 20 companies that are overfunded and eventually whittled down from there as opposed to the Soviet model where you just have one champion and if it works or it doesn't, you're stuck with the date you came to the party with.
José: During the space race, the Soviets followed a more decentralized approach in that they had lots of design bureaus, to some extent, competing with each other to build the different kinds of rockets and to propose different kinds of plans to go into space.
To some extent that might have been a success of Soviet space industry. They may have lost the space race for reasons that I think are due to their manufacturing capabilities, which were not as advanced as the U.S. making it difficult for them to manufacture reliable components. But the Soyuz rocket has historically been a relatively reliable and cheap way of sending things into space to an extent that the U.S. has not matched until recently.
It shows that, even in a Soviet context, if you have a diversity of people trying different approaches and then you pick a winner at the end, you’re probably better off than if you put all your money into an SLS or a space shuttle.
LCDs, T-shirts, and Semiconductors: The Geopolitical Case for IP
José: So LCD screens appear as NIH, NSF, and DoD in the chart. If I look at LCD screens in the U.S., the Army funded a lot of work in this space and they actually funded a consortium. People might have heard of SEMATECH, a public-private partnership to foster the development of semiconductors in the United States. The U.S. tried to do the exact same thing for an LCD screen. Back then, it was facing a lot of competition from Japan and they tried to both impose tariffs against Japanese LCD screens, as well as try to subsidize U.S. LCD manufacturers.
That didn't work in that case. You could say that either that it failed and it was a waste of resources, or you could say on the other end that actually the U.S. government should have doubled down on LCD screens and funded them even more… which is again circling back to the issue of assessing industrial policy.
Jordan: So let's broaden out a bit. We have a lot of complicated histories of a lot of really important products that may have had their timelines pushed back or forward a little bit by government intervention, but these aren't necessarily the slam dunk cases that the proponents of today's industrial policy really want to use as points when they're arguing for a hundred billion dollar push to support semiconductors, or what have you. Given that context, what do you think are the smart justifications for having a more aggressive industrial policy?
José: The most textbook, tame, empirically supported case for R&D funding is that the private sector will not invest sufficiently in R&D because it’s difficult for a given company to capture the downstream effects of that research. Now, this doesn’t mean that the private sector doesn’t do any research. It just means that it doesn’t do enough of it. So that’s one case as a justification for that, but one can go beyond this.
We could potentially just make one given country, let's say the U.S., into a semiconductors superpower. The free trade argument for why countries should not do that is something about competitive advantage: different countries have different competitive advantages and that they should specialize in what they are good at. But that doesn't quite work in general. The original case was about wine. David Ricardo was talking about wine production in Portugal and I’m sure probably Canada will never be a wine producing superpower. Sure. But you can change what your country is good at by targeted investments. So it's a completely different case. If you let things be, you can end up in a place where you didn't want to go.
At the same time, you could say that the price to make your country into some kind of superpower is, in a way, pushing against the “trends of history,” which seems to be pushing towards Taiwan in this case. If you succeed, then you end up with an advanced manufacturing capability in your country. If you fail you just throw a bunch of money to waste.
Jordan: It’s really interesting thinking about Taiwan, José, because I don't think you could sit there in 1980 and say the trends of history are leading Taiwan to have the most important semiconductor company in the world.
José: That's completely correct. My meaning is that, right now if the U.S. doesn't do anything, it's going to be Taiwan [leading the world in semiconductor production]. You could also say that maybe it will go to China because China will actually try harder to do it there.
Countries can indeed shape the future development of their industries, but the arguments around the extent to which this quote “works” (and works is a loaded word here is) is, is this good for the welfare of the population in the country. That is, the U.S., for example, benefits greatly from just buying chips from TSMC in Taiwan.
Now if TSMC were in the U.S., the U.S. could benefit in various ways. It could benefit from lower prices and it could benefit geopolitically, which I think is probably the strongest argument for industrial policy. Suppose there were some kind of confrontation between the U.S. and China, or even without being a confrontation, some kind of geopolitical contest of some kind. China could say, “Well, you are not doing this so we will not give you rare earths,” for example. China produces [80% of global rare earth supply], which are used in all sorts of things from electric motors to semiconductors. Then the U.S. would be in trouble.
Moreover, if the U.S. doesn't have manufacturing capability and it needs to build stuff, let's say in the worst case scenario, then it's difficult to build up advanced manufacturing capability if you don't have it. So there is a case to maintain it, even though it's not “in the short term, economically efficient.”
Geopolitical dominance is, in my view, the strongest and most clear-cut case for industrial policy. The other ones can also be made but I think this one is the strongest and I think it justifies more than other IP cases.
For example, the protectionist case for industrial policy, or, sort of the mercantilist case, will say, “the U.S. has a trade deficit. The U.S. is importing more than it is exporting, and we should fix that.” I think that's not necessarily a problem in and of itself. The deficit estimates that the U.S. is buying more than it sells, and that eventually the U.S. may have to export to pay for that. But the dollar is a reserve currency. Typically, when you import a lot, your currency depreciates, which then makes your exports more attractive, so your deficit fixes itself.
Now the U.S. can keep buying stuff without, quote “paying for them” because the dollar is a reserve currency and there's lots of demand for dollars outside of the U.S. It’s not a great argument.
Then we also have the argument that one should focus on industries that are export oriented. Famously, Studwell documented that East Asia industrial policy was all about fostering strong, advanced, export-oriented industries and trying to get those competitive in export markets.
For the U.S. it shouldn't necessarily be exports. It should be, of course, about advanced industry. If you need to set up a new t-shirt factory in the U.S. because you need garments that you cannot get any more from Malaysia because of some blockade, sure, you can do it. But if you need semiconductors and for some reason TSMC isn’t available to you anymore, you cannot just stand up a new fab in one year. You need that to be there to begin with.
But there is one more thing in the trade deficit case for IP: natural resources.
If you want to be geopolitically independent or have some power in that regard, you need to have resources available to you to just make thing. You need oil. Maybe not in the far future, but right now, you need oil. You need rare earths. You need steel and iron. So, in that regard, you can start by having a stockpile of these things. For example, when somebody like China is going into Africa to secure resources there, they just could pursue a similar policy to make sure that the raw materials that ultimately go into manufacturing are also available there.
Back in the fifties, when you had the first strong defense of IP by Raúl Prebisch in the context of Latin America. They were thinking about IP not in a country-way, but for the whole of Latin America. That is, Latin America as a whole could do IP in that let's say different countries could do different things and work together well. The U.S. for example, could say maybe you shouldn't care much if production is in Germany. Take ASML, one key company that supplies tooling for semiconductor manufacturing is a Dutch company. Europe, in general, is an ally of the United States, which is also geographically, not that far. It means that the U.S. shouldn’t necessarily try to get ASML into the U.S. but they should make sure that ASML doesn't just disappear from Europe and go to China.
Tweets of the Week
Thread. A masterclass in how to read the news.
Any readers out there can recommend me a good English language history of pre-partition Poland?