Stop me if you’ve ever heard some version of the following argument: government mandates and funding gave us the __________ [insert: space shuttle, internet, Silicon Valley], so the government should spend $ __ billions/trillions on stimulating innovation and shaping industrial policy today.
But that argument is built upon a simplistic and sometimes flawed understanding of what actually happened in the past. The true story is much more complex. In this episode, Paul talks with historian Christophe Lecuyer about the military’s role in the creation of Silicon Valley. And he then sits down with Cato’s Peter Van Doren to discuss the track record for government industrial policy in the decades since.
What is silicon logic, user logic and competitive logic? Does the military use new technologies first? What are the advantages of Pentagon contracts compared to private contracts? What is the difference between research and development contracts and manufacturing contracts?
Making Silicon Valley: Innovation and the Growth of High Tech, written by Christophe Lécuyer
The Technology Pork Barrel, written by Linda R. Cohen and Roger G. Noll; With Jeffrey S. Banks, Susan A. Edelman, and William M. Pegram
Universities Built Silicon Valley (with Margaret O’Mara), Building Tomorrow Podcast
Military-Industrial Complex, written by Christopher A. Preble
Regulations Gone Wrong, Free Thoughts Podcast
How the Powerful Captured the Economy, Free Thoughts Podcast
00:03 Paul Matzko: Welcome to Building Tomorrow, a show about tech, innovation and the future. I’m your host, Paul Matzko, and I wanted to return to a topic today that’s been niggling at the back of my mind since we did our Silicon Valley trilogy of episodes.
00:17 Paul Matzko: It’s the question of the role of the military industrial complex in the creation of our tech hubs and in propelling innovation. You see, this is, or really ought to be, a discomforting idea to folks like us who generally believe in free markets and limited government, because on the surface massive government spending on R&D, as happened during the Cold War, appears to have worked wonders. It’s part of the birth of Silicon Valley. And if that was true then it’s natural to wonder if it could be true today, perhaps we do need some kind of industrial policy where the government uses tax money to encourage technological innovation.
00:57 Paul Matzko: It’s a pretty basic challenge to the libertarian world view and you’ve probably heard variations on it, like when the pundits cite government funding of the Internet or NASA to support their argument for more government spending. So, I decided to seek out a couple of experts on the subject of industrial policy and other government interventions in the economy, and I think you’ll soon realize that the reality is much more complicated than the simple version you may already be familiar with.
01:26 Paul Matzko: You’ll hear from two interviewees today. Later on, I’ll sit down with Cato’s own favorite economist, the notorious PVD, Peter Van Doren. But first I wanted to talk with a scholar that could give us some historical perspective on the military’s role in the birth of the semiconductor industry, which really is the core of Silicon Valley. Dr. Christophe Lécuyer is a professor of the History of Science and Technology at the Sorbonne, and the author of Making Silicon Valley: Innovation and the Growth of High Tech, published with MIT Press. He has written extensively about the history of the semiconductor industry. Welcome to the show, Professor.
02:03 Christophe Lécuyer: Thank you.
02:04 Paul Matzko: Now, in one of your earlier books, Makers of the Microchip, you describe the challenges that microchip researchers faced as falling into three categories, which you called silicon logic, user logic and competitive logic. What are those in layman’s terms?
02:21 Christophe Lécuyer: Yeah, so silicon logic is really the opportunities and constraints offered by silicon and is where many great opportunities and many constraints. User logic is really the needs and demands of the users. So in that case it will be… In the case of microchips, it would be really the military at first and then commercial users. And then competitive logic is basically the competitive tensions between firms but also between countries in certain industries or technologies.
02:53 Paul Matzko: Let’s flesh that out a bit. So first user logic. That’s what, you know, different users have different interests and in the case of early Silicon Valley, it’s the Cold War and the military has an interest in semiconductor research. How did making chips for the military change the direction of the research and would that logic have been different, how might it have been different if the primary users had not been the military?
03:18 Christophe Lécuyer: Yeah, it’s quite likely, because the military was interested a great deal in reliability and performance. And this led the industry to move to silicon, and before it was working on germanium. And then it really started working on speed, and performance, so that I think it was essential for shaping the early days of semiconductor technology.
03:38 Paul Matzko: I didn’t realize there was another substrate they could have used. Germanium, you said, is that still… What’s the difference between a germanium-based substrate and a silicon-based one?
03:49 Christophe Lécuyer: So they are both semiconductor materials, so they operate in the same way. But the big difference is temperature, so silicon can operate at very high temperatures, but germanium doesn’t. So as a result, the military was interested in silicon because it can operate at much higher temperatures.
04:06 Paul Matzko: Are there things that with a different substrate that would have been possible? What are the advantages of the other approach is maybe what I’m asking?
04:14 Christophe Lécuyer: Of germanium?
04:15 Paul Matzko: Mm-hmm.
04:16 Christophe Lécuyer: I think the advantage at that time was that germanium was much easier to use and actually silicon was really a very hard material to work with in the ’50s and ’60s.
04:33 Paul Matzko: Those who have an interest in history of Silicon Valley and semiconductor research are going to have heard of the Shockley-Fairchild split and that infamous falling out among semiconductor researchers. I had never heard the story in this detail until I read your book, but there was a disagreement over whether they should pursue a research avenue over what’s called the PNPN diode or whether they should go with diffused transistors. And apparently, Fairchild thought the military would prefer transistors even though the PNPN, in your telling sounds like a superior technology or at least the more complex technology, but with less immediate military application. Now, I just threw a lot of jargon out there. What is the difference in basic layman’s terms between those two technologies and tell us more about what the role that the military demand had in that debate?
05:24 Christophe Lécuyer: Okay, there was an essential difference between the PNPN diode and a transistor. So a transistor is a discrete device, right? PNPN diode was what was called at the time a functional device, so it was basically a device that operates like a circuit but without having discrete devices. That was one approach to miniaturization of electronics. So the PNPN diode was something that was invented by William Shockley. He pushed it at the Bell Labs, and then he pushed it in his own company, Shockley Semiconductor Lab, but the problem with this is that it was really hard to make, it was very difficult to control the device and its manufacturing, and the group that left Shockley and moved to Fairchild felt that it was not a good product because it was not stable enough, and they felt that the demand was for transistors.
06:21 Christophe Lécuyer: When they moved to Fairchild, they started working on transistors. But what is interesting is as far I know the debate at Shockley was not on the military versus commercial markets, it was really on PNPN diode versus transistors. When the group moved to Fairchild, to start Fairchild, they went out and went to talk to customers, this is in Southern California, and they discovered that at the time that the market was mainly for military transistors. So as far as I know I know there was no debate about the market itself at Shockley. It’s only later that the group at Fairchild really realized that silicon transistors were for the military.
07:07 Paul Matzko: So it’s clear that military demand accelerated research into the promising new silicon substrate. And that led to very concrete breakthroughs that might have otherwise been delayed. But the military’s particular interest for these devices also funneled researchers into specific research directions. It’s not always clear that those directions were necessarily better than the alternatives from a pure innovation perspective, as with the PNPN versus transistor debate. But even if government money played an important role in silicon development, and there we can chalk up one for the industrial policy advocates, it’s vital that we recognize that the US adopted a much more hands-off R&D approach than other countries. Which is to say that while it might have been government money, it was private researchers competing in the marketplace that actually did the innovation work. That’s a very different model from other countries. Back to Christophe.
08:05 Paul Matzko: On competitive logic, we actually interviewed Margaret O’Mara for this show. And she has talked about… She contrasted the US decision to fund private research against how other countries that were attempting to research at the time with government-run research labs. How important was the competition between individual researchers, or I should say research centers, and different semiconductor corporations in promoting the pace of semiconductor innovation?
08:40 Christophe Lécuyer: Yeah, so the competition was intense, clearly, among firms that were making devices mostly for the military. And there was very strong competition tension between Fairchild Semiconductor and Texas Instruments, that were the two leading firms in microchips in the early days. And it’s clear that this competition between TI and Fairchild accelerated the development of the technology.
09:06 Paul Matzko: O’Mara makes that tacit comparison to how the US set up its government funding of research of private corporations and private research labs. How is, for example, was France conducting R&D differently at the same time period or another country that you’re familiar with?
09:25 Christophe Lécuyer: Yeah, I think she has a point. Namely that the decision was made at the beginning of World War II in the US to contract out research to universities and companies. And that was something new, that no other country in Europe did during World War II as far as I know. And after that the Americans stayed in that mode, if you will. And the Europeans stayed in the old mode, which was the arsenal mode, right? Things that don’t incite within the government.
10:00 Paul Matzko: The US avoided making the mistake of trying to centralize industrial policy in government-run research labs. It’s been reaping the rewards ever since. But there are other downsides to even the more limited role of government funding in the US model. Working on a government contract could be a frustrating bureaucratic nightmare at times. It also meant tying up a limited supply of researchers on projects that interested the military. That created a bottleneck that discouraged innovation in consumer-facing products. After all, close to 100% of integrated circuits went to the military up until the early 1960s, leaving consumer electronics dependent on vacuum tube and germanium-based tech later than might have otherwise been true. Back to Professor Lécuyer.
10:46 Paul Matzko: I’m also curious on the supply of genius side. Was there a bottleneck among engineers and researchers with expertise in semiconductors? Were there more projects available than there were people to staff them?
11:02 Christophe Lécuyer: Yeah, so there was clearly a very large demand for people who could do anything with semiconductors in the late ’50s and 1960s. And it’s quite likely that companies were looking for engineers for jobs and they couldn’t find them. So what is interesting to note is that the more dynamic companies such as Fairchild in Silicon Valley, were basically organizing raids and they were basically raiding the companies based in Boston or Texas and bringing their engineers back to Silicon Valley.
11:36 Paul Matzko: That’s great. So it’s a good time to be a semiconductor researcher, ‘cause I’m sure salaries are going up and yeah. Now that we’re kind of on the human side of the equation, what are the very basic advantages of going with Pentagon contracts for semiconductors as opposed to selling to non-military buyers? What, well, let me do first, what are the advantages?
12:01 Christophe Lécuyer: So you have to think also about, think of the timing. So the answer is going to be different in different periods. So in the early days of Fairchild, so late ’50s, early ’60s, the military was basically the only market. It’s either the military itself or military contractors. They were basically the only market for silicon-based transistors, diodes and microchips. And then things changed starting in ‘62, ‘63, around that time, maybe earlier. Why? Because the military started to change its procurement policies and it made selling devices to the military much less interesting.
12:39 Paul Matzko: What does that mean? How did they change their procurement system?
12:43 Christophe Lécuyer: So they changed it in two ways. First of all, they make major cuts in procurement. So they were buying less from… Fewer microchips and transistors, but mostly transistors and initially in microchips. And then they also changed the rules. They essentially ask all the firms to give them access to their accounting books. So the military wanted to know essentially how much profits the firms were making on military contracts in order to negotiate the prices down. And this was something that was really not well-seen in Silicon Valley, for good reasons, right?
13:18 Paul Matzko: Yeah.
13:18 Christophe Lécuyer: And many firms decided that the military was really too risky to work with and then they tried to move into commercial markets.
13:24 Paul Matzko: Yeah, there’s often a revolving door between the industry and the procurement process in the government. And so if you open your books to the Pentagon, all your competitors are going to see your books as well, potentially, leaks and all that kind of stuff.
13:36 Christophe Lécuyer: Absolutely. It’s really a bad thing to do.
13:39 Paul Matzko: Now, I also… In your book, you also mention that there were some intellectual property problems as well as the bookkeeping. What is that?
13:48 Christophe Lécuyer: Yeah. So at the same time, the military wanted to get access to intellectual property. And in some cases they would actually take the devices from companies and ask other companies to produce these devices for them.
14:04 Paul Matzko: So, one company’s done all this research into a new way of constructing a transistor and the military is now going to take it and let someone else do the very profitable production line work. Is that…
14:18 Christophe Lécuyer: Yeah, absolutely. So they’re basically moving the thing away from the creator of it and give it to somebody else to produce it. So that this happened a lot for microwave tubes and a bit for transistors as well.
14:29 Paul Matzko: So as military purchase declines in the ’60s, you now have all these companies that have excess capacity for producing and, well, I suppose researching as well, in the area. What industries pick up that slack? Where are we seeing these semiconductors go?
14:46 Christophe Lécuyer: Yeah. So actually what happened is that the companies such as Fairchild essentially created new markets for their products and they targeted two industries: Consumer electronics and the automotive industry, in the early to mid-1960s. They essentially created the market and they did so by lowering their prices on their products, which meant lowering the cost of manufacturing. But also, they were actually very creative and they designed devices like TVs, for instance, around their transistors and microchips. And then they would go to see the manufacturers in the TV industry and show them their designs. And this enabled them essentially to create markets, because the designs that were engineered at Fairchild were adopted then by these TV companies.
15:40 Paul Matzko: And these new monitors, what would the ordinary consumer have noticed, the difference between the old and the new? What difference does that make to quality of a TV or even a car?
15:51 Christophe Lécuyer: In the case of cars, I think it enabled the introduction of more electronics, if you will, inside the cars. Because the cars need… A car drives at very high temperatures, right? So you need silicon. So you can do a lot of electronics with silicon in cars that you wouldn’t be able to do with germanium. And I think in the case of TV monitors, it helped with the military addition of the electronics part of it.
16:21 Paul Matzko: Robert Noyce was one of the founding fathers of the semiconductor industry and he once said, “Government funding of R&D has a deadening effect upon the incentives of the people. They know that their work is for the government and that it is supported by government dollars, that there is a lot of waste. This is not the way to get creative, innovative work done. The best way to get something done is to have enough confidence in yourself and your men to do it yourselves.” He also complained about what he called “bullshit, make waste, make-work, and lack of incentive in government contracting.” I asked Professor Lécuyer if Noyce’s complaint was common at the time.
17:01 Christophe Lécuyer: That’s an interesting question, and my sense is that it was fairly rare at that time. But what is interesting to note is that Noyce was clearly not interested in getting, researching weapon contracts from the military. Actually, Fairchild almost got no contracts from them with a few exceptions. But he was interested in getting manufacturing contracts from the military. So that’s a big difference. So when you get manufacturing contracts, basically you design the device and send it to the military, military contractors, and you rely on government money, essentially. But if you get a researching government contract, you’re essentially tying your expertise and your competency to the perceived needs of the military. Fairchild was very much interested in military contracts and military markets. But it really defined the needs, if you will, of the military, and then selling its devices to the military according to what Fairchild expected the needs of the military to be. This is very different from trying to get the contracts from military to do research, where the military tells you what research to do. And in that case, the military is really controlling your engineering and your technical competency and directs it toward its own perceived needs, if you will.
18:15 Paul Matzko: So there’s a lot more freedom on the manufacturing end rather than on the research end, where it’s much more directed, “We want you to do this very specifically.” And then, for a researcher like Noyce, that’s frustrating, that’s limiting.
18:30 Christophe Lécuyer: Yes, that’s limiting. But also, I think Noyce was a very competent man, and he understood that the real needs of the military may not be the needs that the military perceived about itself.
18:43 Paul Matzko: Oh, okay. Interesting.
18:45 Christophe Lécuyer: The military at that time was giving lots of contracts for radio frequency transistors or high-powered transistors. And it turned out that there was no big markets for these types of products.
18:57 Paul Matzko: I mean, if you’re a functionary in the Pentagon, you don’t have the expertise of the folks actually on the research and manufacturing end. So you think you know what you want when the experts are saying; “No, you really want… This is a better approach, this will be better for you in the long run.” But there’s a disconnect there, I suppose.
19:02 Christophe Lécuyer: So the way I would see this is, the military identify its needs… There were clearly needs for high-powered silicon transistors and radio frequency transistors, right? But they also identified needs for logic devices. But the thing is that Fairchild focused on where the volume was. And the volume was in logic devices, not in RF or high power devices.
19:45 Paul Matzko: Very interesting. Now, in the intro to Making Silicon Valley, you argue that the relationship of Silicon Valley to the military industrial complex is not a simple story. And there are, I think, several different ways folks approach that story in the simplistic fashion. What are some of the different ways that you see kind of commonly being spread today?
20:09 Christophe Lécuyer: To me, there’s often a confusion between R&D contracts and manufacturing contracts, right? I think it’s important to make that distinction that the military had basically an impact in two different ways. One was giving R&D contracts to companies to develop technologies that they were interested in. And then the other ways in which the military was impacting the industry was with manufacturing contracts. And some companies decided that, as we talked about for Fairchild, decided not to go for military contract, R&D contracts, but for manufacturing contracts. I think it’s a big difference.
20:47 Paul Matzko: This distinction that Professor Lécuyer makes between what it was like to work for the government as a researcher versus as a manufacturer gets to a more general observation about the nature of industrial policy. In short, the more strings attached to a government grant or the more specific requests from a Pentagon contract, the less likely it was to have beneficial outcomes. Researchers often felt constrained by government requirements and the roving eyes of politicians always looking for a new ribbon to cut or constituency to please. And this appears to hold true, even outside of semiconductors in the formal tech industry. Economists have tried to assess when industrial policy succeeds and when it fails and why. It really is a mixed bag. So I’ve asked Peter Van Doren, the editor of Regulation Magazine here at Cato, to explain what the relevant literature says.
21:40 Peter Van Doren: It’s just a word for throwing money at stuff.
21:42 Paul Matzko: There your go, that’s your answer.
21:46 Peter Van Doren: Industrial policy refers to either appropriations or tax breaks in the tax code, either credits or deductions, that are directed not at a general thing, like R&D or love or the weather or some… But rather this particular industry defined as the government. The Department of Commerce has what are called SIC codes, Standard Industrial Codes, so every segment of the US economy for census purposes has a six-digit number. And so, in the law it says if you are an industry that does, fits in these codes, you’ll get so much from the appropriations bill or you’ll get this break on your corporate income taxes, if you spend it on… And then they usually fill in the blank by, you know, capital and equipment, and this and that, and they define, like the depreciation of this… Or you know, they’ll… So basically… We talked off the air, it’s… Congress throws money through the tax code or through appropriations at various particular sectors or industries in the economy hoping that good stuff happens as a result of throwing money at this particular sector.
23:03 Paul Matzko: So, it’s tweaking those formulas, right. There’s a form, essentially, fill in this blank with whatever, you know, economic sector ID number. And when they spend money on fill in blank category, they get fill in blank repay the money. Yeah.
23:16 Peter Van Doren: Then the government pays for it.
23:18 Paul Matzko: Credit or whatever.
23:19 Peter Van Doren: Exactly. It’s like buying a dishwasher and getting the visa card back from Bosch or whatever.
23:25 Paul Matzko: Yeah, yeah. When you put it that way, it’s less romantic than calling it industrial policy.
23:30 Peter Van Doren: It is, that’s right. It’s money. We’re throwing money at…
23:32 Paul Matzko: Throwing money at things, yeah.
23:33 Peter Van Doren: The behavior of particular sectors buying or employing particular inputs.
23:41 Paul Matzko: This has a long history. We’ve done a lot of industrial policy during the Cold War, as we’ve heard from our recent interview with Dr. Lécuyer. But it’s older than that, right?
23:51 Peter Van Doren: In my reading in my preparation for this interview, I came across an anecdote that was new to me. And I’m always… I mean, that’s why I keep telling people I’ve read a lot, but there’s always more to learn” And then it was the following: In 1836, Congress appropriated $30,000. And I haven’t done the inflation calculation yet.
24:12 Paul Matzko: That’s big money.
24:13 Peter Van Doren: That’s big money in today’s dollars, I suspect, to subsidize the first telegraph line from Washington DC to Baltimore. And so, industrial policy, I.e., throwing money at a particular thing because it’s an invention and it looks like it’s going to change society, goes back a long way in US political and economic history. And thus, we shouldn’t think of industrial policy as something named in the ’70s and ’80s to worry about, “the rise of Japan and the fall of the United States.” It goes back a long, long, long…
24:49 Paul Matzko: It is kind of the context when the term becomes popular, though, right? Yeah.
24:52 Peter Van Doren: Yes. Yes.
24:53 Paul Matzko: Decline of the US manufacturing sector or… I shouldn’t say the decline of the sector, ‘cause the sector actually, in terms of production is doing just fine. The number of people employed…
25:02 Peter Van Doren: Yeah, output. Because of productivity gains, it takes many, many fewer people to make widgets and that’s called the problem. Even though from our point of view, it’s actually a success.
25:14 Paul Matzko: It’s funny that we still use the term industrial policy, ‘cause it is that mindset. It’s like a ’70s… The steel mills are offshoring. Textile mills are leaving. It’s an industrial factory manufacturing mindset, when the term is much more broad than that. I mean, it can be any sector of the economy, right?
25:23 Peter Van Doren: Technically. But we don’t… No one that I know of. There aren’t bills in Congress to throw money at shopping malls or to prevent their decline. I mean, even though three quarters of US GDP is service sector, we don’t tend to think of that as…
25:53 Paul Matzko: Industrial.
25:55 Christophe Lécuyer: And we don’t… There doesn’t seem to be a whole… Although, I mean, I suppose, with locality… State and local governments are throwing money at Amazon to locate… Remember its second headquarters, there was an arms race which Northern Virginia won to… So there is state and local competition over parts of this so-called service economy. But I don’t… Nothing comes to mind about a national effort to make restaurants more efficient, even though restaurants… I mean, employment growth is huge in the service… Anyway…
26:33 Paul Matzko: Yeah. But it’s an odd… We use it in a very specific sense…
26:37 Peter Van Doren: Which is throwing money at factories where people deal with noise and heat and things, and things go boom and products come out.
26:48 Paul Matzko: Yeah. There’s a romanticism and nostalgia behind it too which is once upon a time, from the 1930s grandpa or dad worked in a factory. And they could work 40 hours in a union job and make a middle class living.
27:05 Peter Van Doren: Yes, yes.
27:06 Paul Matzko: And it’s very much centered around the vision of Detroit auto factories. And that’s lingered. That’s what we think of when we think of industrial policy often.
27:14 Peter Van Doren: Mines, auto, steel. The odd thing is, it just if… We romanticize the past about, my goodness. Injuries and deaths, mortality. People died in those. Yes, the wages were high, but that can be because of our competitors had been bombed in World War II, and we were it. And then unions restricted entry and shared the rents with workers. So it was… But it was smoky, dirty, and productivity was low. I mean, 500,000 people were employed in coal mining in 1950-something. And now we have 30,000. And wow, do we really wanna go back to have 470,000 more people in underground mines being killed in coal mines? But there’s misdoubt, yes, there’s cultural misdoubt among certain folks, the geographic regions in which coal and steel were foremost. If you go to those towns, if you go to Youngstown, Ohio, certainly not very pretty, and things aren’t happening there. I mean, that’s true.
28:28 Paul Matzko: Or Lordstown, Ohio, where the plant closed and…
28:32 Peter Van Doren: The GM plant.
28:32 Paul Matzko: You have whole town, communities built around factories.
28:37 Peter Van Doren: They were not economically diversified.
28:39 Paul Matzko: Yeah, yeah. They’ve struggled since. So industrial policy, very old, and complaints about industrial policy are very old. This is even before there were libertarians. People were complaining about government spending on internal improvements in the 19th century. But what…
28:55 Peter Van Doren: This goes way back. Yeah. I mean, federalists/anti-federalists fussing over whether something called the markets and private entrepreneurs and ingenuity can do everything or whether the government needs to, ought to, can, should, whatever verb you want, play a role in this process. And so we’re going through the latest iteration of this struggle.
29:19 Paul Matzko: And like the anti-federalists back in the day, there are many today, including most libertarians, who are critical of industrial policy. I asked Peter what the most common objections are.
29:30 Peter Van Doren: Well, first are the philosophical coming from libertarians and market-oriented folks, which is they’ll cite Hayek and they’ll say that life is so complicated that we don’t have enough knowledge to actually pick out winners ex-ante from losers, things like that. So we have general complaints, but those are constant. They’re not about any specific program. They just say no one has enough knowledge to figure out what to throw money at in particular. Instead, we need prophets plus market entrepreneurs making their own guesses with the right incentives, I.e., it’s their money at stake as opposed to other people’s money. That those incentives weed out really stupid ideas.
30:15 Peter Van Doren: Whereas if it’s tax payer-funded, ooh, it’s OPM. It’s other people’s money. And it’s something that constituents don’t pay for. Particularly instead, we all pay for it in general, and then it’s located in a particular place. So there’s often a geographic coalition associated with so-called applied R&D spending and thus it’s… The book. There’s a book I brought which we’ll tell our audience about. A Brookings book from the early ’90s. It’s one of the last and only discussions of all of this, the title of the book is The Technology Pork Barrel. It’s just like dams and roads. And so send money to boost this sector and the sector isn’t located all over the US. It’s located in particular places at particular times, and that creates the coalition to spend to save it or boost it.
31:13 Paul Matzko: Besides having to deal with the limited attention spans of politicians and voters, industrial policy also has to contend with politicians’ bias towards discrete projects with short-term benefits, the kind that they can point to at the next election. Industrial policy, in other words, favors specific earmarks with concrete measurable outcomes over broad general research grants. However, specific earmarks have a particularly terrible track record.
31:36 Peter Van Doren: So, this Brookings book has… It’s by two economists, and they discuss focused government technology efforts. And then they talk about in general, and then they have some case studies about failure.
31:36 Paul Matzko: And so this is spending on specific programs, not a like a general tax credit for R&D.
31:36 Peter Van Doren: Correct. The four programs that they talk about are the supersonic transport, this was an effort to, the French had the Concord and we needed our own, by golly, so we had a supersonic transport program, the development of the space shuttle, right? A cheap, which turned out not to be cheap, a reusable launch vehicle to take stuff back and forth from up there in the space station then back down to earth. Three is the B#breeder reactor.
32:36 Paul Matzko: Like nuclear reactor? Yeah.
32:37 Peter Van Doren: This was the Clinch River Breeder Reactor, and it was an R&D effort focused on a type of nuclear reactor, which would produce more fuel than it used, and this is when we were going to go all nuke and we were going to solve the oil crisis. And we weren’t going to be dependent, we were very worried about not having enough uranium in the US, and we were against trade and all…
33:06 Paul Matzko: Because a lot of the uranium is in Africa, Central Africa and other places? Yeah.
33:10 Peter Van Doren: It’s other places, and so then this would make as a result of nuclear reactions in civilian nuclear reactors instead of having, well, you’d have ways, but you would also make plutonium, and then the plutonium would be used in future of electricity production in a reactor. And so anyway, it looked right, kind of take a loaf of bread and make five loaves of bread, I mean, it’s magic. So there was that research effort. And then finally again, from the energy crisis, the synfuel program to convert coal into gas and liquids to make gasoline and natural gas out of coal. Under high pressure and temperature and engineers, you can do all these things. And it turns out, that we dropped, we spent a lot of money on that and it never went anywhere.
34:04 Paul Matzko: So all four of these, well, we spent large sums of money on, so how does that look? That’s individual companies are competing for grants, from…
34:13 Peter Van Doren: Well, they were federal R&D efforts, here they really weren’t existing companies building these things, instead it was companies that might build these things were then given money to research and develop prototypes and etcetera, etcetera, which didn’t happen.
34:35 Paul Matzko: Well, we did end up with a space shuttle. So that one, it seems like the most successful of the four.
34:40 Peter Van Doren: But it was so expensive we turned out not to be able to really, it wasn’t cost effective. I mean it did succeed, but the tiles, that whole effort to keep it from boiling up as it reentered the atmosphere that was, that didn’t… That was very costly. So, ironically, now we’re with Elon Musk, we’re doing the SpaceX effort, which is a private attempt to create a reusable rocket to launch and then recover, and it looks like it may work out in a way that the space shuttle did not.
35:19 Paul Matzko: So with these four attempts at specific targeted government industrial policy spending on these projects that were fantastically expensive and didn’t lead, didn’t produce the results that were hoped for, why does that fail to function? What’s the mechanism by which it doesn’t work?
35:44 Peter Van Doren: The conclusion of the book was that it’s, that the political coalition for particular industrial policy efforts is very, it’s a political coalition and therefore some problem catches the nation’s concern and it’s the decline of the rustbelt or something like that, or we’ve gotta catch up with the Russians, or some cause catches political imagination and then becomes in politician’s minds central to re-election efforts. For whatever reason we’ve gotta solve this problem, and turns out, R&D takes a much longer time to work things through.
36:26 Peter Van Doren: And the book basically says it’s not that the federal government is incapable of leading an R&D effort. You can hire scientists and you can have them work, I mean, the incentive problem I talked about earlier, is missing. It is other people’s money, but you can hire the same smart engineers and whatnot, and fund them and pay them and they’ll work on things. It’s not that they’re shirking, it’s just that congressional interest wanes and then the funding goes away, because they become troubled by what’s, ‘cause there are hiccups. These things always, don’t always work well, there’s failure. That happens in the private sector too, but those happen on the private side and don’t get much congressional attention and newspaper attention.
37:12 Peter Van Doren: But when it’s political and you’re sending money to something and there are hiccups, then, oh, why are we spending all this money? And, ooh, the problem we were worried about when we started this program isn’t today’s problem. That was yesterday’s problem. Let’s go onto something else. And so the message of the book is… A long run version of a famous political sign is Brookings Anthony Downs called the issue attention cycle, which was if you look at newspapers, whatever the problem was in the New York Times eight months ago isn’t what we’re talking about now. And if you’re gutsy enough you can ride out these political issue attention cycles.
37:52 Peter Van Doren: For instance, Governor Northam in Virginia, right? I mean, he was, he appeared in a medical school yearbook with blackface, and it looked like, oh my, this guy can’t possibly last. And while he just kinda hunkered down and waited and the Anthony Downs issue attention cycle is now focused on something else, and he’s still there. Alright, a long run version of that is applied R&D, where an effort will go for five or six or seven years, but it needs 30 years maybe. And the interest declines and…
38:03 Paul Matzko: That makes sense, yeah. Well…
38:32 Peter Van Doren: And the funding goes away.
38:32 Paul Matzko: The political coalition change… You might not have the critical mass of, I don’t know, say, there’s a combination of local congressmen who their district would benefit a lot from spending on having one of these research centers or…
38:44 Peter Van Doren: Right, but you need some other parts of the…
38:48 Paul Matzko: The other parts and the coalition… Something happens…
38:49 Peter Van Doren: Right.
38:50 Paul Matzko: And it intercepts. I can also imagine an issue where, you know, research isn’t always predictable and private efforts, you know, startup culture is well-known for the pivot. You’re doing research along one avenue, and it’s not leading to the results you expected, and so you… But you produce some results that point in another direction, right? And so you pivot the direction of the company and change the product, sometimes radically.
39:16 Peter Van Doren: But a congressional appropriations process…
39:18 Paul Matzko: They don’t pivot.
39:19 Peter Van Doren: They don’t pivot. It’s like, you know, we basically give you 2% more than we gave you last year and you write out forms, and you tell us how everything’s wonderful, and you did what you did with the money. And then, oh, what do you mean this is a total dead end and we need to do this. Well, no, that would be a different committee. And you see…
39:38 Paul Matzko: Right, right, right. They’re just not capable, they’re not flexible in that same way. We have a pretty good track record from Cohen and Noll and their contributors that particular spending on specific projects falls afoul of… The attention cycle falls a foul of, it runs into problems, tends to end up wasting a lot of money on average. And that matters, I mean, it would matter less, I suppose, if that money only supplemented private research, but isn’t there a literature on whether it actually whether complements or supplements or whether it substitutes for.
40:15 Peter Van Doren: Substitutes, yes.
40:15 Paul Matzko: And explain that for us, why is that a particular problem?
40:18 Peter Van Doren: Well, just for the reasons you said, which is economists have tried to figure out whether because of the public good nature of knowledge and the inability of the patent system to totally privatize the spillover benefits of R&D, is it the case that in effect public sector spending is a necessary complement to private efforts, because if you don’t have public spending, then the private effort will be insufficient because you can’t restrict the benefits of R&D to those who pay for it. There’s spillover benefits, people know that, and who wants to put all their money into something, only to have, in effect, other people use the knowledge and make money off of it. Again, the patent system is an attempt to privatize the benefits of R&D so that for a while you have monopoly rights to the research and the result, but it’s an imperfect effort.
41:25 Peter Van Doren: And so in the economics literature, there is this question which is, if the feds throw money at something, does private sector activity go down or go up and how does that relationship work? So, I’ll quote here from in the Journal of Economic Perspectives, the fall 2019 issue, very recent, had a very nice article on what economists know about R&D, both in general and this… Not so much on the applied side, which is the older Brookings book, but on the, on a more general if we throw money at scientists, what happens? So the stylized fact I should have known, but did not was that in this article from 1953 to 2015 R&D in the United States spending went up from 1.3% of GDP to 2.7, but during that time period, the federal role fell dramatically and the private role rose.
42:28 Peter Van Doren: So we drifted from a massive post-World War II, MIT Vannevar Bush military industrial complex kind of R&D federal effort to much lower defense spending as a percent of GDP, and defense R&D as a percent of GDP, and then the private sector rose quite a bit. And so it it seems like there at least federal R&D spending is not a necessary. I mean, it’s not that R&D fell off the cliff, because the federal dropped dramatically, the private sector picked up their role, but the composition is different. I mean, again, we’re going away from a lot of nuclear weapons and armament-type R&D to pharma R&D and health and and electronics and the Silicon Valley kind of stuff.
43:25 Paul Matzko: Yeah, which, I mean, I suppose it depends on one’s political and ideological priors. I tend to prefer the ones that aren’t means to kill other people.
43:32 Peter Van Doren: Yeah, yeah, exactly.
43:35 Paul Matzko: You know.
43:36 Peter Van Doren: The irony is how liberal states that are anti-war have a high defense R&D spending. California, Massachusetts and Connecticut have a lot of defense R&D and yet Joe Lieberman was senator from Connecticut for and it wasn’t accidental. He protected Electric Boat and Sikorsky helicopter and the United Technologies, and the technology pork barrel that went their way.
44:04 Paul Matzko: So the literature is pretty clear that general research is preferable to specific earmarks, but even with general R&D spending the results are still mixed.
44:13 Peter Van Doren: I have results both for and against.
44:16 Paul Matzko: Oh, okay, fun. Take your pick.
44:18 Peter Van Doren: Economists are into research designs that attempt to mimic what we have in science, I.e., experiments, and we look for unexpected what they call discontinuities in life so that we can say plausibly that if you’re on one side of this discontinuity or the other, it’s like random assignment in a clinical trial and thus, we can look at outcomes from different discontinuity results and say, oh, this is sort of like randomly assigning R&D money this way versus that way and seeing what happens. So one of the literature articles I’ll talk about is take all the finalists for NIH grants, the National Institute of Health, and then among the finalists some win and some don’t. Then follow their research career subsequent to that decision, how much of a difference does the NIH grant make? And the answer is one additional… For those who win, they produce one additional research literature paper over the next five years and every $10 million of additional NIH money produces about two to three patents over the next five years. And so this article didn’t do a cost-benefit analysis about whether that was a good, but at least there was some evidence that the public sector NIH grants did make a difference for outcomes.
45:51 Peter Van Doren: But I’ve got another NIH study that comes to the opposite conclusion, and this involved the decision of President Bush administration to stop stem cell research funding from the federal government over right to life concerns.
46:13 Paul Matzko: So we have another… Yeah, there’s a…
46:17 Peter Van Doren: Well, an exogenous, a plausibly exogenous shock. So no more federal money for stem cell research. Well, if our listeners remember, that was a big deal, it was like, “Oh, my goodness. We’re not going to discover cure of X, Y, Z and Q, which would save a gazillion lives and cure these diseases.” The scientific community was apoplectic. Okay, so the literature sort of studied, well, what happened to stem cell research? The answer was there was no change in research output and the scholars left publicly-funded academia and they went to industry. So here we find instead of the NIH result that I talked about earlier, which is, it looks like the public sector funding was a complement not a substitute, here it looks like the private sector could pick up the ball. There were enough wealth… Well, very, very rich philanthropists, because this was such a public thing, lots of private philanthropic money picked up the ball. So it’s not clear you can generalize this to more under the radar federal versus private R&D. But again, this was the world didn’t end, stem cell research didn’t end. The private sector picked up the ball. And you could argue that we don’t need federally funded medical research in this area because it doesn’t appear to be necessary.
47:43 Paul Matzko: So if there are some big principles I think in all this conversation, all things being equal, if you’re going to have industrial policy, if the government’s going to throw money at things, it’s better that they throw it at general scientific research, kind of a broad un earmarked spending.
48:03 Peter Van Doren: Not industrial policy, just…
48:05 Paul Matzko: Yeah, just broad tax credits for R&D.
48:07 Peter Van Doren: Yeah, something like that. And that, again, in the short run, may increase salaries, may not increase [48:14] ____, but if it’s not short-term, there’s a problem if you have a tax credit for anything or an appropriation for anything where there’s great uncertainty about whether it’s going to continue. So you need a kind of 30-year… We’re going to throw money at this for a long time. If it’s really basic R&D, it’s going to take that long.
48:36 Peter Van Doren: Take fusion research, for example, alright, thermal… Creating a limitless supply of energy that mimics what happens on the sun or any star, fusion. There’s a project at Princeton, there’s a project in Europe and they’ve gone through gluts and busts when it comes to funding, and we’re still not… It’s going to take a long time for them… And so even on the basic side, the ability of the public sector to commit to just throwing money at this for a very long period of time and seeing what happens, that’s very difficult, but if it’s going to do something, it should do that, not a particular publicized effort to bring out some particular technology in five years or six years or seven to solve some current crisis.
49:30 Peter Van Doren: And the one part of the federal government it seems that Congress loves is NIH, and they just, medical, throwing money at disease, basic science that has maybe a disease cure pay-off, the administration often wants to zero that out. Trump wants to lower the NIH budget. Congress loves NIH. And then, like I said, I’ve got evidence that it actually is useful and then some evidence that maybe it’s not.
49:58 Paul Matzko: I wonder why that is. Why health in particular, why that? It’s interesting.
50:02 Peter Van Doren: The politics of health research just don’t fall… The negative Brookings, the case studies of… But throwing money at NIH, Congress tends to…
50:13 Paul Matzko: Well, they’re not going to cut it now that we have the coronavirus scare.
50:16 Peter Van Doren: Probably not.
50:16 Paul Matzko: It’s probably safe for the next budget segment.
50:19 Peter Van Doren: Although, trying to create a vaccine under deadline, I’d worry about, although I hear Gillian has… Apparently there’s something in the pipeline and maybe they’re going to rush the trials and then… Okay.
50:34 Paul Matzko: I’m sure the response of government health agencies like the CDC to the coronavirus epidemic will be the subject of future papers like those Peter has referenced today, but I hope that today’s episode gave you a sense of just how complicated the story of government financing of innovation is. Next time you hear a simplistic version of the industrial policy pitch, you’ll know that you’re being sold a bill of goods. In reality, industrial policy is a mixed bag, though there are some broad applications we can make.
51:04 Paul Matzko: For example, if the government does do industrial policy, it should fight its natural instincts and go with general grants over specific earmarks. We should also expect that process to corrupt over time, leading to waste and distortions. We should question whether the innovation produced truly serves our interest as consumers, or if it’s about the interest of the military or other government agencies. Until next time, be well.
51:40 Paul Matzko: Thanks for listening. If you enjoy Building Tomorrow, please subscribe to us on iTunes, or wherever you get your podcasts. If you’d like to learn more about libertarianism, find us on the web at www.libertarianism.org.