Joel Mokyr, Philippe Aghion, and Peter Howitt won the 2025 Nobel Prize in Economics for their work in the theory of sustained growth through creative destruction. Today, a friend and I went to see Philippe Aghion speak about this work at Ekonomikum in Uppsala. The lectures are organized every year by the university and are open to the public, with no pre-registration. My rough notes are below [L: As usual, my own comments are embedded like this.].
Intro to the speaker
Transition from everlasting poverty to sustained growth is one of humanity’s greatest achievements. How to preserve this is one of the main question.
Sustained growth comes from technological innovation. But where does it come from? How is it generated, sustained, diffused? What kinds of policies and institutions allow for it?
Philippe Aghion introduced a framework that places the entrepreneur-innovator at the center of sustained growth. By turning light to agents of innovation and the competitive environment in which they live, the framework allows us to study incentives and conditions that enable innovation.
The power of creative destruction by Philippe Aghion
Philippe Aghion did not expect to get the Nobel Prize. He was in Stockholm about two months ago for an honorary degree. A colleague asked him at the time: “Could you send me a paragraph telling us what you’re working on? I have no clue!”
When he was told on Oct 13 that he was one of the laureates, he thought he was speechless, but apparently they’d never heard such a talkative speechless person.
“Fail, fail, fail!”
Aghion was a Harvard student between 84 and 87. His dream as a Harvard student was to enter the Harvard Society of Fellows, which is a sort of highly coveted first-class postdoc.
Aghion took took an oral exam for the Harvard Society of Fellows, but at dinner later was told he didn’t really understand the question he was meant to be answering… He did not get into the Harvard Soviety of Fellows and so ended up at MIT instead, where his neighbor was Peter Howitt. Aghion told Howitt that he had an idea: why not write from scratch a model that will embody creative destruction?
If Aghion had succeeded in the exam to the Society of Fellows, he would not be here today. And so, failure is extremely important: “Fail, fail, fail!”
Creative destruction
A process whereby new innovations displace old technologies – Joseph Schumpeter
- Long-run growth is driven by a cumulative process of innovation. You stand on the shoulders of giants.
- Innovation results from entrepreneurial activities motivated by the prospect of returns – innovation does not come from heaven.
- Creative destruction: new innovations displace old technologies.
On one hand you need innovation grants to motivate innovation, but on the other hand innovators want to prevent this because they themselves don’t want to be victims of creative destruction.
In our systems, innovators are afraid of being out-innovated.
Distinctive predictions
- Growth is positively correlated with firm turnover.
- Positive correlation between firm age and firm size.
- In the US, if you are a talented innovator you get capital and are allowed to grow more, evicting worse innovators. If you survive you are also better. Selection effect on top of direct effect.
- In some other countries, when you run out of brothers-in-law you stop growing (mostly joking).
Competition and innovation
Let’s say in a classroom we are all equally good, but some are at the top and bottom. You introduce a very good element into the class. The top of the class will work harder to remain top of the class, but the bottom of the class is already discouraged and will get more discouraged. With frontier firms, when there is more competition they react by innovating more. For non-frontier firms it is a discouragement effect. The more developed you are, the more growth-enhancing it is to have more competition.
Secular stagnation: Rise and decline in TFP (Total Factor Productivity) growth in the US.
- Between 1995–2005 in the US there is a growth boost: IT revolution.
- Then a big decline 2006–2017 – why? This is particularly true in IT-using sectors.
- High-markup firms became pervasive. Some firms – Google, Microsoft, Amazon, Walmart – knew how to harness the power of IT better than other firms. They expanded much more than other firms. That’s why average markups went up. The problem is that entry of other firms went down. Discouraged new entry.
- Can we adapt competition policy to the IT revolution? If we could, we might have avoided the fall in entry. E.g.: enforce sharing information. That was not done in the US. If we reform competition policy we can maybe get out of this entry downfall. Interesting policy application: these large firms harnessed the power of IT but ended up discouraging new entrants.
Middle income trap
The Soviet Union was growing and then stopped; maybe they lost accumulation of capital, etc. We see something like a middle income trap.
- You can grow in two ways: catching up with the world frontier, or growing by innovating upon yourself. Imitation vs innovation-based growth.
- South Korea: grows very fast and then slows down. If you are below the frontier you grow much faster to catch up. During the catching-up phase, big conglomerates appeared in South Korea . They discouraged entry of non-conglomerate firms and put pressure on government to prevent competition. This increased the probability of a middle income trap.
- Japan: Between 1945 and the 1990s Japan grows very fast. Same thing. Conglomerates discourage entry of non-conglomerate firms and put pressure on government.
Europe
“It is good to lecture other people but we are not that much better.”
After WWII Europe goes down a lot. Between 1945 and the mid-80s we catch up with per capita GDP of the US, rebuilding capital stock thanks in big part to the Marshall Plan, and high education levels.
Over time China hugely increases proportion of high-technology patents. The drama is that in the EU we are good at research. If you look at original ideas of patents in disruptive technologies, they cite many European papers. Research takes place in Europe but not so much breakthrough innovation.
Over the past 20 years, patenting and R&D investments in the US moved from mid-tech toward high-tech – not in Europe.
In the US, top patenting firms between 2000–2023 changed. This is innovation. In Europe they are the same (Sweden is better). The US has higher turnover and more breakthrough innovation.
In the EU we do not have policies that favor successful breakthrough innovation. We miss a single market – each member state has its own regulation on top of EU regulation. Not enough funds for innovation and not enough competition. In the US and China there is a true innovative market.
To do breakthrough innovation you need to allow people to fail. If you don’t fail, you don’t try. In the US they do not punish failure [L: They don't? Seems riskier to fail in the US..]. Foundation of long-term research: you can fail in the short run, it’s not a big deal. In Europe you are writing applications instead of researching [L: Not sure how to reconcile with him saying EU is good at research before.]
- In the US you have lots of venture capital – no big deal if you fail in the short run. We don’t have enough VC in Europe.
- Who is good with pension funds and financial ecosystems? Sweden.
- In the 1950s in the US they were losing to the USSR (Gagarin in space). They had basic research and needed to translate it into putting a man in space. Money came from defense; team leaders and program managers elicited competing projects.
Innovation and inequality
In the US there is a U-shaped curve of top income earners. Share of income of top earners has gone up. What is the source and what do we do?
Top income inequality source: innovation. Innovation increases top income inequality, but does not increase broad inequality. Because it involves creative destruction, it is a form of social mobility.
More innovation intensity gives a higher share of income to the top 1%, but no effect on bottom 99% Gini.
So, where innovation is highest, social mobility is highest. By contrast, lobbying reduces social mobility and increases broad inequality. Lobbying is bad for social mobility as you remove creative destruction and hinders new entrants.
Example: Steve Jobs became rich by creating Apple; Carlos Slim by heading a poorly regulated monopoly.
Don’t totally handcuff innovators –> but tax them.
Rethinkning capitalism
The US does poorly on inequality and social protection. Europe does poorly on innovation.
Gini 0: we are all the same. Gini 1: one person earns everything. Gini is higher in the US than in France, Germany, and Scandinavian countries.
The US has a higher poverty rate, and mortality per 100,000 is increasing sharply. Aghion tells the story he heard from a driver in the US, who had to sell his house to afford medical treatment for his wife - “Shitty social model”. We are lucky in Sweden and France not to face this.
But the US is better on innovation. Ideally, we could combine the American innovation model with European protection and inclusion. It does not have to be an either-or, and can be done by focusing on three policies:
- Flexicurity (Denmark)
- Education (Finland)
- Competition (US)
AI
Technology and growth
- AI boosts productivity and the production of ideas.
- The cloud is dominated by Amazon, Google, Microsoft - by the same superstar firms. And the market force is dominated by GPU.
- So Aghion is very optimistic about the technology itself, but institutions have to adapt to encourage innovation and avoid a conglomerate situation. Need to account for future entry and boost competition.
- Will we have the courage to put in place policies that will enhance competition?
- Prosperity is a joint result of technology and institutions.
Employment
- Overall, firms that adopt AI increase employment compared to those who do not because they increase sales.
- Job creation: Firms that adopt AI become more productive and competitive. Therefore they increase employment - not uniformly everywhere but overall.
- Job destruction: AI adoption for administrative purposes on highly exposed jobs reduces jobs.
- AI additionally increases turnover.
- Enormous potential, but needs the right policies.
- Successful harnessing of AI for innovation and growth in economic system requires education and flexicurity systems. Otherwise, you die.
Education
In the US, there is too much for-profit in education. Needs to move to nonprofit schools.
Education needs to focus on learning how to learn.
What works well is to concentrate on reading: “Books, books, books!” No computer, no iPhone. Books! Children should read Grade them - not to punish them but to let them improve. Let children calculate! Children need to learn how to reason. How to prove a theorem. You learn to become adaptable, and then you have adaptable citizens and flexicurity works well.
Ability to speak freely
Aghion was at a conference in the US, focused on second-order policy improvements for innovation. He presented this example of what happened during the conference:
He had said we are bewildered in Europe because we know we should improve on innovation, but not copy the US social system. And he joked: Trump has read the Draghi report and decided to reduce the gap by shooting himself in the foot. They need openness and good students; instead, Trump decided to make their lives hell: cutting research funding, driving talent away, etc.
The non-Americans laughed; the Americans didn’t dare to smile.
Europe is a continent where you can still joke and smile freely. We have freedom, openness, democracy, and a social model the US does not have. This allows us to harness AI much better. Europe has huge power to attract the best people to work with us. We need to do the right things to return to frontier innovation.
But innovation must be inclusive. If it isn’t, you get populism. France faces the threat of the extreme right coming to power. Don’t despise those who vote for them: they do so because they feel abandoned and despised. They have tried everything they can think of, so they use the one thing they think they have left: voting for the extreme right. Inclusive growth is essential to prevent this.
To make growth sustainable, it must be inclusive.
Aghion wants young people to take over: “I chose creative destruction because the day I become obsolete, it validates my theory.”
Q&A
(All paraphrased - Aghion was so excited about the topic that sometimes the questions were not completed before the answers came, hence some truncated questions)
Q: You talked about creative destruction in the sense of old firms being replaced by new firms. Can you comment on creative destruction within firms?
A: Creative destruction is everywhere; it can involve activities within firms. It sometimes becomes difficult in large firms. Within firms, there can be barriers to new innovations. Sometimes innovation happens through spin-offs: someone spins off [L: Similar thing happened with Google and Character AI, I think]
Q: You are researching innovation and regions. How do we treat geographical regions that…?
A: I’m starting to work on geography. For example, it’s interesting that some innovators start outside cities, join a hub, then leave the hub because they’re afraid someone will imitate them. There is a geographical lifecycle of researchers. Should you have one university and one hub? But if everyone goes there, you have a parental problem: who will educate those not born in that region? How many Uppsalas do you need? This research is in the making.
Q: Industrial policy — we used to be very united on that…
A: We need good industrial policy. But better no industrial policy than poor industrial policy where you pick the wrong winners. Picking the wrong winners is terrible. You need good governments.
Q: You said lobbying is a bad thing, as rich people stay richer. Would you advocate a policy to eliminate lobbying?
A: It is very important to have firms that innovate and a state that implements competition policy. But there is a risk of states colluding with conglomerates. That is why civil society is so important - media and anything that can expose lobbying. You can make it illegal to lobby in political campaigns, etc., but exposure through media and civil society is crucial.
Firms innovate, government regulates, and civil society exposes and reacts. With separation of powers between the judiciary and the executive.
Q: You mentioned venture capital and how it is useful for anchoring new innovators. When some companies are funded by VC, they just grow and grow and corner the market without becoming profitable. For example, WeWork raised billions but didn’t make any profit. This seems to be a trend. What do you think?
A: Failure is a good thing. You need a rate of failure to get breakthrough innovation; it’s the price to pay. What’s great about VC with no collateral is that you share the upside in terms of control and revenues. As companies grow and expand, you reduce control and share of revenue. Some will fail, some will grow, some will IPO or get acquired. That’s the usual lifecycle of a firm. Many fail along the way, but it’s better than not having VC at all. The alternative is bank finance, which is not adapted to this kind of risk-taking. You need failure for breakthrough innovation.
When you are in the business of imitation, bank finance is enough. For frontier innovation, bank finance is not enough.
Q: What about the issue that those companies don’t need short-term profit and may corner the market? They may become monopolies without producing anything sustainable.
A: If you fail to innovate, why would you become a monopoly? You become a monopoly because you did something better than others. If you fail, you fail - so how can you become a monopoly unless you have government support that you shouldn’t have, or unless there is corruption? There is a risk of firms remaining monopolies for too long if they innovate successfully. Failures cannot become monopolies.
Everything can have perverse effects, which is why we need some regulation. The problem with too much regulation is that incumbent firms know how to cope with it, while new entrants do not. With the AI Act, I am not in favor of heavy regulation that becomes an entry barrier [L: How to balance this with new entrants being known for moving fast and breaking things - including in the areas of user privacy/safety/security?]
(Tangential: my notes on AI Act Conference 2024: Navigating the AI Act, Legislating Technology)
Q: In the US there is now what some woudl call “technofeudalism”…
A: Yes. There is a risk that the same companies will still dominate in five years. The danger is that giant tech companies remain monopolies.
Q: Some of these companies are in a new branch of AI and are only about two years old. Will this remain the same, or will they evolve or create new types of firms?
A: One example: an extremely smart guy left Meta and came to France to do something new. Europe can attract people like him, but they need proper financing and a good environment. There will at some point be a new generation in AI - some think it will be better than LLMs. Why not ensure it is done in Europe? The best response is to innovate more yourself.
Q: There was recently an EU initiative to attract researchers from the US after policy changes there…
A: Yes, anything like that we try to push. We should finance labs and long term research. Ten-year research programs. Some of this won’t be done at the EU level. It will be countries working together. You don’t need the entire EU to participate, just a coalition of the willing.
Q: Innovation in emerging large economies like China, which is rising and moving from catch-up to the technological frontier. What is the challenge in sustaining this growth?
A: There is an enormous amount of innovation in China. Some believed China would never take off, like Zambia. That turned out to be wrong. China is a vibrant innovator. How did they catch up? Through a combination of education and openness. When China joined the WTO, it could absorb technology from developed countries, which allowed it to move fast.
In France, we built high-speed trains and went to China for 1–2 years to teach them how to do it. After that, China said, “Thanks, but we got this - we don’t need you anymore.”
- China produces top research and encourages researchers to come back: reverse brain drain.
- China runs several funds at the same time and sees which works best. Additionally, there is competition among regional leaders. They have their own way of combining industrial policy and competition, and that works well.
- What might not work as well is total breakthrough innovation. China is currently at the top in research and frontier innovation, but will it produce another “AI”? AI foundation emerged first in Europe, then the US. For that, you need more disruption in the system. China may not encourage disruptive behavior efficiently.
My reflections
Philippe Aghion was an incredibly passionate and humorous speaker in the best ways. I knew nothing about creative destruction going into this, but his presentation made the topic fascinating even for a complete layperson.
In addition - how lucky am I and everyone who lives nearby to get to just drop in to a lecture by a Nobel laureate on a Saturday? I hope we never start taking these opportunities for granted.