Peter Howitt

“I was fascinated by this,” Howitt recalls. “I asked my boss, why are these prices going up and down all the time? And he gave me a little lecture on supply and demand. And he said, if you want to really understand this, go study economics.”

Howitt says he was not a particularly strong student at the time but his boss took him under his wing and he respected his advice. He went on to study economics and political science at university, where his interest moved from the microeconomics of supply and demand to macroeconomics.

New things make old things obsolete. Some people win, some people lose. It's obvious, but economists were willing to assume that that didn't really matter.

How did Howitt and Aghion meet?

In 1987, during a sabbatical from Western University, Howitt joined MIT as a visiting professor. There he met Philippe Aghion, a young assistant professor who had just finished his PhD at Harvard.

Aghion told him that his Harvard teacher, the industrial organization economist Richard Caves, taught the intriguing idea of Schumpeter's called creative destruction.

“As Schumpeter put it, each wave of innovation leaves in its path a lot of destruction, a lot of bankruptcies, foreclosures, lost jobs, obsolete skills,” says Howitt. “That wasn't part of economic growth theory that we learned, that we taught, that was actually used at the time. And Philippe said, ‘Why not? You're the macroeconomist.’ And I thought, well, that's a pretty good question.”

Timing, as it often does, played an important part. Aghion had hoped to stay at Harvard and wasn’t planning to remain at MIT. “If he'd got what he wanted at the time, we probably wouldn't have met,” Howitt says. “Who knows what would have happened, but things would have been very different.”

How do you model creative destruction?

Building a mathematical model of creative destruction required a new approach.

In most markets, current leaders were once the ones producing the fresh, leading-edge technology, while new breakthroughs often come from new entrants. Any realistic model therefore needed to capture competition between incumbent firms and potential challengers, including attempts by established players to defend their position.

“What's involved is a strategic interaction between various incumbent producers and potential rivals,” says Howitt. “It's a kind of strategy that really requires the use of game theory to model properly.”

At the time, in 1987, game theory was transforming microeconomics, but it had yet to make a big impact on macroeconomics. Most macroeconomic models assumed markets were competitive and that firms simply accepted prices as given.

Economists have models that they really understand and work really well. We tend to ignore things that don't fit well into our models.

MIT also happened to be home to Jean Tirole, another future Nobel Laureate. Tirole had just finished a landmark textbook that summed up the state of game theory and included research on innovation races.

Building on this work, Aghion and Howitt embedded competition to innovate inside a broader macroeconomic model covering wages, savings and investment decisions, and monetary and fiscal policy.

“Lo and behold, we got a mathematical model that did predict what the growth rate would be and how it would depend upon the incentive to innovate,” says Howitt.

The model suggested that growth would be influenced by factors such as international trade, education and the environment. It also incorporated an insight from Howitt's work in monetary economics: because innovations arrive unpredictably, economic output follows a path in which major shocks can permanently alter the economy’s growth path. The first draft of their 1992 paper was titled A Model of Growth and Cycles through Creative Destruction.

Why had economists overlooked something so intuitive? Howitt likens it to an old joke about someone looking for their keys under a lamppost simply because that is where the light is. “Economists are kind of like that," he says. “They have models that they really understand and work really well. So, we tend to ignore a lot of things that don't fit well into our models.”

Creative destruction was one of those neglected ideas.

Why does it matter which technologies become obsolete?

Most new technologies both complement and replace what came before. Artificial intelligence has increased demand for computer chips, for example, rather than making them obsolete. Howitt argues that the technologies being displaced matter most, because they shape how people behave.

“The people whose livelihood depends upon the continual production and sale and use of those old technologies, they have a strong interest in trying to either stay ahead of the new technologies or, if they can't, to try to block them,” he says. Earlier growth theories often relied on what he calls a “rising tide lifts all boats” narrative. In his view, that approach missed an important part of the story.

Faster growth also brings greater labor-market churn. Drawing on research by labor economists, Howitt notes that in a dynamic economy like the United States, jobs created and destroyed in a single year can add up to roughly a quarter of all jobs. “It doesn't necessarily result in higher unemployment rates," he says. "It's just a lot more churning.”

Is competition good or bad for innovation?

Schumpeter believed aggressive antitrust policy could weaken growth because innovators need the prospect of monopoly profits to justify risk-taking. Howitt agrees that this effect is real. Without the opportunity to earn returns from innovation, companies would invest less in research and development.

But competition also has a second effect.

 “The more competition you have, the less you're able to do that and the more you realize, it's innovate or die,” he says. Aghion and Howitt call this the “escape competition effect.” Working with researchers at Oxford University, they developed a theory showing that these two forces pull in opposite directions. The result is an inverted-U relationship between competition and innovation.

When competition is weak, increasing competition stimulates productivity growth and patent creation. Past a certain point, more competition starts to stifle innovation. “What we discovered also was that in that sample, 90 percent of the industries were on the upward sloping part of that relationship, where more competition would actually result in more growth,” Howitt says.

Studies in other countries have found either the same inverted-U relationship or one that keeps rising.

Is AI becoming a monopoly?

Asked where the AI industry sits on the inverted-U curve, Howitt places it on the side where more competition would still help. Many of the current participants, he believes, are driven by the belief that the best model will take the whole market. He doubts that will happen.

“A natural monopoly is some industry where fixed costs are gigantic and marginal costs are almost zero,” he says. The cost of each AI query is tiny, but billions of calculations and heavy energy use add up. “It just seems to me that it's not a natural monopoly.”

Will AI destroy jobs?

When Howitt gave his Nobel lecture, he read a paragraph about the job-destroying potential of generative AI before revealing that ChatGPT had written it within seconds. “I hope you like that paragraph because ChatGPT wrote it,” he told the audience in Stockholm. He wanted listeners to experience the surprise that many people feel when they first use the technology.

Howitt expects plenty of creative destruction, but fewer job losses than many fear. He is skeptical of companies that blame AI for slower hiring. “It sounds a lot better to say, hey, because of AI I don't need new workers than to say, whoops, I made a mistake and over hired in the past,” he says.

Unlike earlier computer and telecom technologies, Howitt believes generative AI may not widen the gap between highly educated and less educated workers, because it automates routine cognitive tasks such as looking things up, recording and verification.

What makes a research partnership last for decades?

Howitt credits his nearly four-decade partnership with Aghion to their complementary personalities. “He's very energetic. He loves to race from one thing to another. I tend to be a little more methodical.” Over hundreds of hours spent working at blackboards, Aghion was usually the first to suggest abandoning a path that seemed like dead end, while Howitt preferred to persist. “Sometimes he was right, sometimes I was right. But I think it worked better than if either of us had been doing it on our own.”

Aghion also kept Howitt focused. “My tendency to get interested in something different and then just abandon what I'd done in the past never went away,” he says. “But he was very creative in getting me redirected and back to work again. And I'm glad that he did.”