✨ 📈 The AI Paradox: When will artificial intelligence boost US economic growth?

Maybe we should take a lesson from Alan Greenspan and the 1990s boom

Back in October 1995, Federal Reserve Chairman Alan Greenspan gave a speech to the Economic Club of Chicago in which he publicly puzzled through an apparent productivity paradox. Why was the Information and Communications and Technology Revolution, such as “the semiconductor, the microprocessor, the computer, and the satellite,” not showing up in productivity statistics? Greenspan wasn’t the first to ask this question, referred to as the Solow Paradox. A decade earlier, MIT economist Robert Solow stated, “You can see the computer age everywhere but in the productivity statistics.”

Perhaps, Greenspan theorized in Chicago that day, all this apparent technological progress was mere “wheel spinning — changing production inputs without increasing output — rather than real advances in productivity.” As with electric power in the late 19th century and early 20th century, it might take significant time and investment for these technologies to fully translate into productivity gains. Maybe some "wheel spinning" might represent necessary transitional changes before industry could realize the full potential of these new technologies.

The other possibility: Greenspan suggested that with increased "conceptualization" of output (shift toward ideas and technology), official productivity and GDP measures — something devised when America had a wheat and steel economy — might understate true economic growth. Greenspan:

Our radios used to be activated by large vacuum tubes; today we have elegantly designed pocket-sized transistors to perform the same function—but with the higher quality of sound and greater reliability that consumers now expect. Thin fiber optic cable has replaced huge tonnages of copper wire. Advances in architecture and engineering, as well as the development of lighter but stronger materials, now give us the same working space but in buildings with significantly less concrete, glass, and steel tonnage than was required in an earlier era. These innovations are the extension of an established and likely irreversible trend. … Over the past century, our standard measure of output of goods and services, adjusted for price change, has increased by approximately three percent per year, but the actual physical tonnage of that output has gone up significantly less. The difference reflects the substitution of impalpable concepts for physical volume. The expanding conceptualization of output has also led to a cumulative build up of productive capital, which has meant less labor input per unit of output. This is a key to increasing productivity and, with it, our standard of living.

Greenspan’s speech reflected an important debate within the Fed. Wages were rising alongside stable prices and growing profits. Standard economics said this combination should be impossible without productivity gains, yet official data showed flat productivity. The stakes were immense: If productivity wasn't actually rising due to all these new technologies, the Fed needed to hike rates immediately to prevent inflation, as some Fed economists wanted. If Greenspan was right about mismeasured productivity gains, premature rate hikes could needlessly choke off a historic economic expansion. He ultimately concluded after considerable research that the productivity measurements were flawed, particularly in services, and successfully argued against rate hikes. The 1990s productivity boom was the real deal.

As Sebastian Mallaby writes in The Man Who Knew: The Life and Times of Alan Greenspan:

Years later, Greenspan’s productivity call was remembered as a high point of his Fed tenure. Drawing on his unrivaled feel for economic data, he saved the economy from a premature interest rate hike that even doves like Yellen had supported. Greenspan’s belief in the productivity acceleration “was intuited,” Larry Summers would marvel. “It must have come from some combination of good luck and a very deep acquaintance with a lot of data. And a very great feel for a lot of data. And I think the Fed staff didn’t see it. I don’t think most other people who could have been Fed chairman would have seen it.

Oh, and in 2000, Solow told The New York Times, ''You can now see computers in the productivity statistics.”

When will we see AI in our productivity stats?

Let’s take a step back: Productivity growth — a core theme of this newsletter — happens when workers produce more output per hour through better education and training, better and more equipment, and innovative technologies and business methods. More productive workers drive overall economic growth and higher living standards by enabling greater output from existing resources. And it’s my hope that recent advances in artificial intelligence will accelerate productivity growth here and around the world — maybe by a little, maybe by a lot.

But when? It's a good question.