Back in 2020, when the world was navigating the hellscape of the COVID-19 pandemic, economist John Haltiwanger discovered something really strange happening in the U.S. economy: Americans were creating new businesses at a record rate.

Haltiwanger is one of the top experts on new business creation in the United States. He even helped the U.S. Census Bureau set up official statistics that track it. A surge in new businesses, Haltiwanger says, tends to be a great sign for job creation, innovation and productivity growth in the economy. But, up until 2020, the data painted a sort of gloomy picture. Haltiwanger was writing research papers with titles like “Top Ten Signs of Declining Business Dynamism and Entrepreneurship in the U.S.“

And so, in the summer of 2020, when Haltiwanger first saw the data suggesting there was a startup boom, he and his colleagues were gobsmacked. “At first we thought, ‘There’s got to be something wrong,'” Haltiwanger says.

By November 2020, when I first spoke to Haltiwanger for the Planet Money newsletter, it was clear that the startup boom was real. But it was unclear whether the surge in new businesses was just a pandemic-related blip that would quickly fizzle out. Much of the boom could be seen in the online retail sector.

But then the startup boom kept … well, booming. We saw a second surge of new businesses beginning in 2021. So I spoke to Haltiwanger again in June of that year. By then, he was more optimistic that the surge of new businesses was a harbinger of great things to come for the economy. He was particularly struck by the revolutionary changes to business life created by widespread remote work.

But, to be honest, I was still a bit cynical about the boom. Lots of small businesses were devastated by the pandemic and closed up shop. Many people lost their jobs. Maybe a lot of them were creating new businesses in a desperate attempt to pay the bills or to get more federal relief money, or just out of boredom.

Most plausible to me: Maybe this surge in new businesses was just a reflection of the wild swing that we had been on to a weird pandemic-related economy and then back to normal again. The first swing created one-off business opportunities, like selling hand sanitizer or masks, or delivering cheeseburgers and fitness equipment to our front doors, or whatever. Then, as we swung back to normal and consumers began spending money outside again, that created new opportunities for the revival of face-to-face businesses, like bars, gyms and restaurants. These were the same types of businesses that had been devastated by the pandemic. Maybe, I figured, this was all just one long, arduous journey that ultimately brought us back to where we began, with limited long-term benefits for the economy.

But we’re now well past the pandemic crisis and even the pandemic recovery. It has been almost exactly four years since the startup boom began — and there’s still a bonanza of new business creation in the United States. It’s harder to be dismissive about it.

“I’d say we’re on a new plateau that started in 2021,” Haltiwanger says. Comparing the three years before the onset of the pandemic with the three years after it, the data suggests there are now, on average, almost 60% more new businesses being created each year.

The boom is real. It’s sustained. It’s visible in both businesses that are just a single person striking out on their own and, importantly, businesses that are likely to grow and employ people. It’s seen in traditionally underrepresented minority communities. And, Haltiwanger says, this boom could be a sign that something fantastic is about to happen to the American economy: a long-awaited boost in productivity growth, which is the magic sauce for making society richer.

But what’s driving this boom? And how optimistic should we be that it will end up having real benefits for the country?

What’s driving the boom

Haltiwanger says that, basically, two big buckets of new businesses are being created these days.

New businesses in the first bucket are capitalizing on a huge post-pandemic population shift. Many office workers are now either fully remote or hybrid. “People are not spending five days a week at the office in major downtown areas,” Haltiwanger says. Where people spend their time, they spend their money. Bad news for businesses in downtown areas. Good news for businesses where office workers live.

That’s why one of the big areas for new business growth is in food and accommodations, particularly in the outskirts of cities. Haltiwanger, together with Ryan Decker, calls this “the donut effect.” There’s now a hole lacking vibrant economic activity in many major business districts, and a delicious fried dough of new business opportunities in the suburbs surrounding them. Office workers need their doughnuts, coffee and sandwiches near their office, which is now more often at home.

However, if the story of the new business boom were limited to just delis, gyms and doughnut shops in suburban areas, the upside would be somewhat limited. Sure, remote and hybrid work is a revolutionary change for a large fraction of the workforce, but the business boom it has fed could be seen as mostly just a geographic reshuffling of economic activity. Fewer coffee shops in Manhattan. More in New Jersey or Brooklyn. That would likely have only a limited upside for the economy.

That’s why Haltiwanger is much more excited about the other big bucket of new businesses he has identified in the data: tech startups. This boom is proving, he says, to be the most persistent. These tech startups come in many stripes, but one subcategory has really caught his and other economists’ attention: startups working in artificial intelligence.

“I think we’re in a new tech wave,” Haltiwanger says. “I think AI is the poster child of this.”

What the boom could mean for the economy

The last time the United States saw a significant uptick in productivity growth was in the 1990s during the dot-com boom. Productivity growth means we can make more stuff in less time, which causes products and services to be more abundant and cheaper. It’s like fairy dust being sprinkled on the economy, lifting society’s standard of living.

“The first thing that happened back in the early 1990s was not increases in productivity and the like,” Haltiwanger says. “That came later.” What came first, he says, was a surge in new startups. He says startup booms are a “leading indicator” of this virtuous cycle for the economy.

Startup booms, Haltiwanger says, are both a reflection of technological innovation and a huge driver of innovation. Startups figure out how to use new technologies. They experiment and develop new products with those technologies. They force competitors to adapt and innovate.

I should mention, however, that when I spoke to Haltiwanger back in 2021, he was also bullish that the startup boom would result in a long-awaited boost in productivity growth. But three years later, he admits, we still aren’t seeing it. “The productivity statistics are pretty anemic,” he says.

There’s a cliché used in articles when discussing whether technological innovations will ultimately lead to productivity increases: references to the late economist Robert Solow. In the late 1980s, when the personal computer was spreading like wildfire through the U.S., Solow famously wrote, “You can see the computer age everywhere but in the productivity statistics.”

Haltiwanger stresses that Solow wrote this back in 1987. And, he says, “The productivity surge didn’t really show up until the mid-1990s.” It takes time.

The economist Erik Brynjolfsson and others have done research suggesting that the effect of new technologies on productivity growth may even follow a “J-curve.” That is, productivity may actually dip before it shoots up, because businesses need to invest in the new technology and spend time figuring out how to use it. Only then will they and the broader economy see tangible results.

However, it’s possible we could see AI in the productivity statistics more quickly than we did with the spread of the personal computer. With personal computers, it took a long time for people to buy the physical hardware, figure out how to use it and then restructure business processes around it. Most consumers of AI don’t require new hardware. AI is also smarter than old-school computer programs and may require less user input to improve workflows. Nonetheless, Haltiwanger says, it still could take a while for businesses to figure out how to exploit AI and enhance their productivity.

There’s currently a big debate about whether AI will revolutionize business processes and juice productivity in the economy. Economists like Daron Acemoglu argue that AI may have only a limited upside for productivity growth in the foreseeable future. Others like Brynjolfsson are more bullish.

Haltiwanger, focusing on the startup boom, continues to have big questions. Is it just a coincidence that we saw a distinct wave of new tech and AI startups around the same time we saw waves of new businesses created in reaction to the pandemic and the population shift precipitated by remote work? How much of this is just people wanting to be their own boss or achieve a better balance between their personal lives and work? Will the boom finally lead to a long-awaited boost in productivity growth? Haltiwanger says he’s also keeping his eye on whether we’ll see “gazelle firms” that run away from the pack and race to transform our economy, as the Googles and Amazons of the world did back in the 1990s.

Whatever this startup boom ultimately means for the economy, we look forward to talking to Haltiwanger again — for years to come — as the data keeps rolling in.