economics
technology

Is the Great Stagnation finally coming to an end?

Stanford University
Northwestern University
Genesis
Response
Penultimate
Finale

Erik Brynjolfsson

Stanford University

October 31st, 2021
Bob Gordon and I have been friends since I was a graduate student, with a tradition of regular annual dinners. He’s a giant in the field of economics and his book, The Rise and Fall of American Growth is a tour de force. However, we’ve long disagreed about the potential of digital technologies and prospects for growth going forward.
When I read the evidence, I conclude that we are the cusp of a productivity boom. There are four factors that are driving this:
1. A wave of powerful technologies, particularly those involving AI and machine learning, that are just beginning to be implemented. Machines can now recognize images, whether faces of friends or pathologies in medical images. For the first time, we are beginning to talk to our machines, and they talk back, though admittedly not in the most scintillating conversation. Machines can now make better decisions than humans for placing ads, recommending products, granting loans, hiring employees, or even deciding who should get parole. And it’s not just AI: mRNA vaccines have been developed at warp speed, the cost of solar power is dropping at rates that exceed all expectations, and we’re even beginning to see the long-awaited flying cars.
2. Impressive as the above technologies are, we shouldn’t expect them to instantly translate into productivity growth. The economics of powerful technologies typically follows a “J-curve”, where productivity growth is initially unexpectedly slow or even negative while firms struggle to reinvent themselves, and then takes off as the benefits of the new processes and products are harvested. Instead, a period of painful restructuring is required, and that’s what we’ve seen over the past decade.
3. That said, the pandemic accelerated 20 years’ worth of digitization into 20 weeks, compressing the J-curve. Individuals, companies and industries are rapidly learning how to work digitally, whether that means teaching a class from home, as I have been doing, or shifting a whole company to remote work. Many of the needed technologies were already place when the pandemic hit, making the transition far less painful. Now that we know how to do these things, we’re not going to go all the way back.
4. The aggressive fiscal and monetary policy that started with the last administration and continues with this one, driving down labor slack, boosting wages, and creating new incentives for productivity gains. For most of the past decade, the government ran the economy with a lot unused labor capacity and inflation often below the Fed’s target level. Congress has begun pushing the accelerator with a series of large tax cuts and stimulus packages, while the Fed kept rates low. As the economy runs hotter, it becomes harder to find workers and more attractive to harvest some of the productivity potential of new technologies.
In light of these factors, I recently challenged Bob to a bet: I predicted that productivity growth over the coming decade would substantially exceed the CBO’s official forecast of 1.4%. Bob graciously agreed and we memorialized our bet on the “long bets” website.
In fairness, I’ve long been optimistic about IT and productivity and not always been correct. In 1987, Bob Solow, the Nobel Laureate, remarked that “we see the computer age everywhere but in the productivity statistics”. My first academic paper took that “productivity paradox” as a challenge and concluded that a part of the problem lay in our flawed measurement tools and part was due to the lags associated with implementing IT systems and the restructuring. I felt vindicated by the technology-driving boom of the 1990s and early 2000s, but have been disappointed by the even longer period of slow productivity growth since then. If Bob and I had made our bet in at the start of the last decade, he would have won handily.
Looking at today’s situation, I see more similarities with the early 1990s than the early 2000s. In particular, AI and machine learning promise to be as transformative as IT was back then. In fact, creating intelligent machines is arguably a more important invention than just about any other invention we’ve ever made. Leading firms have been investing in business process redesign and business model reinvention creating an unprecedented magnitude of intangible assets. This digital capital is the harbinger of productivity growth.
0 Comments