dailynewsnblog
The Public Origins of American Innovation
Tim Phillips Talks to Paolo Surico
Tim Phillips
For the last 80 years, the US government has had a blueprint to use public money to fund scientific research. So today on VoxTalks Economics, the public origins of American innovation. My guest today is Paolo Surico of the London Business School and CEPR. He’s one of the authors of an ambitious new study of which innovations have contributed most to increasing productivity in the US and, importantly, the origin of these inventions. Paolo, welcome.
Paolo Surico
Thank you very much for having me.
Tim Phillips
Paolo, what is it about this research that leads you to call it the public origins of American innovation?
Paolo Surico
Because the most breakthrough innovations in the last century were seeded by publicly funded research. The mass production of penicillin, the computer, the internet, the GPS, the COVID vaccine. Yet, when we talk about innovation, people think about the private sector, the Silicon Valley, entrepreneurs, venture capital. Our point is that the government is able to shape the direction of innovation through publicly funded research. And this is particularly important when uncertainty is high, when fixed costs are high, when the returns are not easy to appropriate. So the idea of public origins is that really what matters for transformative ideas is where those transformative ideas take place. even though they are later scaled up and commercialized by the ingenuity of the private sector.
Tim Phillips
Our story today goes all the way back to 1945 and someone called Vannevar Bush Tell me who he was and what was this? He produced a blueprint for American science, didn’t he? What was this?
Paolo Surico
Absolutely. When World War II was coming to an end, President Roosevelt asked his chief scientific advisor, Vannevar Bush, to produce a report on what have we learned from the mobilization of research for the military sector and how this could be translated for the private sector and to develop an innovation ecosystem for America. And Vannevar Bush came out with his 1945 report, a fantastic bedtime reading, ‘Science: The Endless Frontier’, in which he really emphasized the importance of basic research and the role of research institutes at universities. In particular, he identified three pillars around which to found American innovation ecosystem: Government to really shape the direction of innovation by putting funds in a particular area of special interest; Research institute at a university, the second pillar, to be able to push the frontier of knowledge without having the constraint of commercial goals; And third pillar, the private sector, to transform the new knowledge into new products.
Tim Phillips
So what’s the legacy of this?
Paolo Surico
So what is important is that now we are seeing in our results that this combination of public funds through research institute and university, really have been able to make major breakthrough and then the private sector has been able to scale up and commercialize. So it’s really a joint effort. Often we tend to have a discussion on whether it should be the public sector to lead or be in the private sector. Here what we emphasize is that we have to be a joint venture. in which the direction of innovation is stress-emphasized from the public sector, and is the private sector that actually found a way of bringing that to mass production.
Tim Phillips
Let’s talk about this piece of research that you have done, which is all about trying to find out the contribution of publicly funded research to innovation. How can you separate in the data innovation that’s publicly and privately funded.
Paolo Surico
Let me say up front that when you come to innovation, this is a vast literature. However, there is a little bit of conundrum. We have a fantastic, well-identified, firm-level literature, and then we have all the data on aggregate R&D. Yet, what we miss is to figure out what is the macroeconomic impact of the different transmission channel. So, what we do in this paper, we exploit the granularity and richness of patent data in a bottom-up approach in which essentially we get time series of different forms of innovation. By doing this, we can explore the type of questions such as what is the macroeconomic impact on aggregate productivity, on standard of living, on whether innovation is funded by the government or the private sector, whether it’s funded by health and education agency or whether it’s funded by the Ministry of Defence, whether it’s funded by university research institute or for-profit organization, whether it’s developed by startup or by incumbent. We can do this because in particular in the United States, it is a legal requirement for any patent assignee to declare whether there was any government interest in developing that patent. And therefore we can categorize innovation whether we’re funded by the government and owned by the government. fully private, which means owned and funded by the private sector, or a mixture partnership, in which the government funds and the private sector is actually the one that own and develop. You can think of the Covid vaccine exactly along this third dimension.
Tim Phillips
But in this, you have the patent, which then is adopted, which then becomes an innovation, which then changes productivity, which then grows the economy. How can you make that link between the patent and the economic growth? There must be so many confounding factors here.
Paolo Surico
Absolutely. So what we exploit here is the fact that we take advantage of patents at the time of application. So essentially, you can think of this, we are comparing two quarters in time, two different years in time, one in which has a particularly elevating number of patent applications funded by the government, and another quarter in which there was a surprisingly low amount of patent applications funded by the government. Then, by controlling for condition in the stock of knowledge, in GDP, in inflation, in productivity, we can trace down what are the effects of what’s happening in those particular years, two years down the road, five years down the road, ten years down the road. And we found that those quarters in which there was a disproportional number of government-funded patent applications are followed by a period of very high productivity, three, four, six years down the road, relative to quarters that experienced a much lower number of government-funded patent applications.
Intermission
AI is the innovation we’re all talking about, but can AI also transform the process of innovation itself? Listen to Anton Koronek describe what could happen if AI becomes self-improving. in our episode, ‘Will AI Transform Economic Growth?’, from March 2026.
Tim Phillips
Your headline result is quite remarkable that government-funded patents are 2% of the total, but they account for 20% of productivity growth. Why?
Paolo Surico
We were surprised too. But then when you start to analyze what kind of innovations are those funded by the government, you immediately realize that those are more fundamental, more science-based, with larger spillover. The government funds most of basic research in the economy. A research institute and university are not only those that are more likely to conduct basic research, but also are the organizational type that are more likely to crowd in private investment. So, although they are small in share, they are very important in nature and content because they are really those that shape the direction of innovation. They allow research institute and university to push the frontier of knowledge, crowding in private investment that brings that breakthrough into mass production. And this would generate this large externality that, although small in share, accounts for 20% of medium-term fluctuation in productivity and GDP.
Tim Phillips
Is this result perhaps driven by a few really big wins in technology because we all know about the US government’s involvement in the research around that?
Paolo Surico
This is a fantastic question. I’m so worried that a few examples, for good or for bad, are cherry-picked simply to fit a narrative, sometimes an ideology. And in the paper, we come up with famous examples, whether it’s the internet or the GPS, And then we also come with terrible example in which government actually waste a lot of resources: the Solyndra award, or the failure of DARPA in trying to insert a microchip into some bags before the drone technology becomes much, much cheaper. What we try to do in this paper is to do a much more systematic approach. You can think of this paper as trying to identify regularity. I dare to use the word macrofacts. Is the case that, on average, when the government intervenes, it makes more good than bad? And what we discover is that, quite a few times, government investment… doesn’t really generate much of a return. However, there are a few other instances in which government investment innovation really delivers a big home run. And on average, the benefits of those home runs are significantly larger than the zero return associated with the failures. So this is not a story that the government cherry-pick the winner. This is not a story that the government always gets it right. This is a story in which the government goes for the high risk / high reward. The government takes that uncertainty of fixed costs that the private sector has not shoulder large enough to take. and therefore is able to generate breakthrough that otherwise will not happen. But to generate that, the principle is not value for money. The principle is high risk / high reward, which therefore forces the government to accept a large amount of failure, to be able to find out and identify the success story that is actually the one behind the significant gain in productivity and standard of living that we have seen over the post-war period.
Tim Phillips
And in this story, when we say “the government”, what we actually mean is a group of federal agencies that have managed to produce this research. Which ones have been particularly valuable, good at doing this?
Paolo Surico
This is such an important question, especially at the time in which we are hearing, especially in Europe, the increased budget for defence. So the two agents in the United States that seems to be driving those effects are the NIH, which is the correspondence of our NHS. And the NSF, which is Education, National Science Foundation, that is a little bit the correspondence of the UKRI, if you are based in the UK. Now, what is interesting is that we found that when innovations are funded by health and education agencies, they are able to generate productivity gains that are even larger than military innovation. So here I think there is a little myth that I would like to debunk. Education, health, military innovation, there’s nothing special in each of those. If the emergency is war, then military innovation is special because it generates a nuclear breakthrough. If going to the moon is the emergency, is the need, then space innovation is what becomes special. If the emergency is COVID, then is innovation, health innovation that becomes special. What all those examples have in common is an emphasis on basic research, fundamental research. And in the post-war period in the United States, NIH and NSF happen to be the agencies that do most basic research, that fund most basic research. And that’s why they are associated with innovation that generates the largest increase in productivity gain, even larger than defense innovation, which still is very significant, but quantitatively smaller than the innovation funded by health and education agencies.
Tim Phillips
Ask most people what has driven innovation, particularly in the US in recent years, and they would say it’s startups, probably backed by private venture capital. You have found out that startups outperform established firms, but only when they have this public backing behind them. Why is that?
Paolo Surico
So this is super interesting. I want to make first a divide between the profit and non-profit sector, and then I want to zoom in [on] each of them. So our first funding is that whenever government funds goes to non-profit organization, the innovation that they bring to the table has a much larger impact on productivity and standard of living than the impact of innovations which are developed by the private sector, by profit organization. And then, as you pointed out, when the government funds startups, these generate innovation, which are associated with a much larger macroeconomic impact than when the government funds incumbent. So two points here are worth making: a) the leading role of non-profit organization through university and research institute is because a), they focus on basic research and 2), they are not constrained by commercial goals. When you go to the for-profit-organization world, here is the second principle that comes in, namely that most of the innovation comes actually from startups rather than incumbent. As Aghion very clearly explained in his Nobel Prize lecture, innovation requires competition. Incumbents have a vested interest to defend their market share. Innovation is disruptive. The startup of today becomes the incumbent of tomorrow and that’s why you need a constant flow of startup. When the government funds a startup, it is given a direction, it is given a mandate, often mission-oriented, on what the startup should investigate and because the startup has a commercial incentive to gain market share from the incumbent it is able to deliver exactly the direction of travel, the direction of innovation, that the government has asked. The incumbent cannot deliver that, because it doesn’t have the capacity to innovate. And the start-ups funded by the private sector also cannot deliver that large macroeconomic benefits and improvement in standard of living, because private capital is interested in profitability rather than macroeconomic benefits and standard of living. This is why startups funded by the government manage to achieve the goal of the government, which is maximize macroeconomic benefits and standard living. And when the startups are funded by the private sector, they are able to achieve the goal of the private sector that is more about profitability.
Tim Phillips
Everyone’s thinking, ‘They’re talking about innovation. Why haven’t they mentioned AI yet?’ So I’m going to mention it now. I’m going to ask you to speculate: Are we going to see the same story when it comes to innovation and productivity improvements around AI?
Paolo Surico
I think I want to resist the temptation to indulge in macroeconomic forecasts. First, because I’m not sure how successful macroeconomists are. And second, because everybody out there seems to be a futurist. So I’m going to try to be the perhaps boring bookish academic that looks back at the past and tries to see what seems to be different now. In all of our research, what we have discovered is that really those large macroeconomic impacts come when the government shapes the direction of innovation and the crowding in private investment in that direction. AI might represent a break from this historical pattern. And this is something that already we see now in the data. First, most of AI investment comes from the private sector. Second, there is so much money in the private sector for AI, that now more and more firms are able to attract those talents that in previous industrial evolution would have gone to research institutes and universites. This matters because in our story, in our results, are research institutes and universities that push the frontier of knowledge through basic research unconstrained by commercial goals. But if now, with AI, all the resources are driven by the private sector and are driven out of research institutes or universities because firms offer more competitive packages, there is a significant risk that the direction of innovation is not going to be shaped by microeconomic benefits [such] as standard of living, but is going to be shaped by profitability because this is the goal of the private sector. What we’ve seen in the past to be so important for American leadership, for American innovation, has been exactly the public sector to be able to shape the direction of travel. If AI is going to be driven by private investment, it’s reasonable to expect that the direction of innovation is also going to be driven by the private sector and therefore by profitability. So, one possible scenario… is that we are going to see many high profitable firms, high evaluation among firms, but not necessarily the benefits of aggregate productivity and aggregate standard of living that previous industrial evolutions have had because of the leading role of the public sector.
Tim Phillips
A lot of the research I’ve seen on this, a lot of the research that has been produced has been very, very anecdotal. And it tends to take sides because of that. So people who want to talk about the value of government basic research would point to what Google had been able to do with it. People who think that it’s a bad thing. Well, Solyndra was one that you’ve already mentioned: a high profile failure. What gets lost if we focus on anecdotes like this?
Paolo Surico
So let me pick up on Solyndra. In that exact round of loans in which Solyndra got funded and then failed, also Tesla got funded and it became a huge success. So it’s very easy to cherry pick examples to kind of fit a narrative. We wanted to move beyond that and really go to a systematic analysis, try to understand what is the average effect. After all, that’s what matters for the government. are the successes more than compensating for the loss of the failures. And when you do that, not only you’re able to get to this average effect – and we do find that the benefits of the success significantly offset the cost of the failure – But also we can tell what are the channels exploiting this granularity of micro-patent data that have made it possible for those successes to thrive and for those failures to sunk.
Tim Phillips
A lot of the people who are having these arguments are having them because they have particular point of view on whether public R&D crowds out or crowds in private R&D? What do your results tell you?
Paolo Surico
This is very, very important. What is really special of public R&D, especially when it goes to a research institute or university, is exactly the idea of crowding-in private investment. When we look at private innovation, we also do find that they have a significant impact on subsequent productivity, on subsequent standard of living. But the effect is much smaller. One of our main results is that the return of every dollar of public R&D is actually more than double of every dollar spent in private R&D. And the key mechanism is exactly because for every dollar of public R&D, there is another dollar of private investment coming from the private sector. And the reason is what we described before. Many of those startups tend to be talent that used to be in a research institute or university that went on to commercialize a breakout idea. But it’s exactly the government what wants: the best talent to push the frontier and then either for them or someone else to move into the private sector to scale it up and commercialize so that the mass can benefit about that. So this is really critical for our result. When the government drives the direction of innovation in basic research, pushing the frontier of knowledge without constraint of the commercial goal and letting the private sector to commercialize, to benefit commercially from that scaling up, this is where we have the best of both worlds. When the public and the private sector work together, rather than in substitution, as a complement, with the public leading and the private sector scaling up.
Tim Phillips
So two policy questions for you, Paolo. The US first. 2025 saw huge cuts to federal science research funding. What does your research tell you about the likely impact of that?
Paolo Surico
Taking our result at face value, the effect could be quite dramatic. cutting funding for science or research more generally is really cutting at the heart the fund for basic research. While this might not be felt for a few years the long-run consequences of this could be dramatic, not just in term productivity but also in standards of living. We are talking of very significant cuts, we are talking of the possibility that America may lose its technological leadership, which are the foundation also of economic leadership. In fact, if anything, now we are seeing that most of R&D in the world is no longer in the Western world or in the United States, but is moving toward Asia, in particular toward China. When the UK was the economic leader of the world, it’s because they were also the technological leader of the world. Then, this shift towards the United States, when the United States becomes, because of their innovation ecosystem, the technological leader of the world, and with that, they become the economic leader. These cuts significantly undermine the ability of the USA to remain a leader in technology. And this may have far-reaching economic and social consequences.
Tim Phillips
And on the other side of the Atlantic, there is a region that would really like to be a leader in technology and research and innovation, and that hasn’t quite worked for Europe. Where to find that innovation in the future was a focus of the Draghi report and many of his recommendations.
Paolo Surico
Absolutely.
Tim Phillips
What do we learn from what your research tells us that might be applicable in Europe?
Paolo Surico
I think that there are two main lessons. People in Europe now are obsessed about spending more, whether this is defence or whether this is any form of public spending. The first thing we need to realise and recognise, that the major problem of Europe is not the level of spending, it’s the composition of spending. What we do is so much in procurement, especially on defence, and so little on public R&D and basic research. The good news is that Europe is not short of talent. Europe is not short of top research institutes and universities. Europe is not short of early stage startups. Where really there is a major difference between the US and Europe is the support to scale up later stage startups and to support innovation cycles exactly when it’s needed most. Particularly when you go to science to research, the innovation cycle is longer. So what often happens in Europe is that we have great universities, great research institutes, great start-ups that come up with great ideas and then are forced to move to the United States for later stage innovation. If Europe could simply tilt the composition of its spending, starting from defence, towards public R&D, that act as a venture capital and crowding private investment to support later stage start-up, perhaps the Draghi report could be that Vannevar Bush moment that Europe has missed after World War II, but for the US, was the major driver of the American innovation ecosystem.
Tim Phillips
A Vannevar Bush moment. We’ve been waiting 80 years for it. Let’s hope it happens. Paolo, thank you very much.
Paolo Surico
Thank you. It’s been a pleasure.
Tim Phillips
The Discussion Paper is called ‘The Public Origins of American Innovation’, it is Discussion Paper 20788 at CEPR, if you want to look it up that way. Authors Andrea Gazzani, Joseba Martinez, Filippo Natoli, and Paolo Surico.
Outro
VoxTalks Economics is a TalkNormal production. The assistant producer is Megan Bieber and our editor is Andrei Zagarian. Next week on VoxTalks Economics, Al Roth on what economics can contribute to the debate on assisted dying.
