Markets may be smarter than any of us, but they still can take a long time to learn when a new technology rewrites the rulebook.
Markets repeatedly were surprised by the gains from computing in the '90s. They could be surprised again by gains from AI in this decade.
I n 1994, internet pioneer Netscape introduced the first version of its “Navigator,” which made the internet accessible to even novice users. What followed was an explosion of new computer applications, and the concomitant diffusion of computing into virtually every corner of our lives. It all seems so obvious now, but one of the most interesting patterns in the financial data of the time was that markets repeatedly were surprised by the gains from computing.
In 1995, the stock market (as measured by the S&P 500) jumped 34 percent because of enthusiasm regarding computers. From 1996 to 1999, it rose by between 20 and 31 percent each year. One dollar invested at the start of 1995 turned into more than three dollars in that very short time.
If the market was perfectly efficient and understood the impact of the internet, then the market should have tripled right away in 1995. Of course, stocks began giving back some of those gains in 2000, but those who ran away from stocks because of the anxiety driven by the “bubble” chorus surely regret the subsequent financial boom they failed to participate in.
The lesson is that markets may be smarter than any of us, but they still can take a long time to learn when a new technology rewrites the rulebook. This matters today because the AI revolution is beginning to look like it will change the global economy more than the internet and do so at a much faster rate. Which raises the question: If the internet can make stocks triple in five years, what will AI do?
While I have been optimistic about AI, a recent interview in the Financial Times by Erik Brynjolfsson, my colleague at Stanford, suggested to me that markets may well be underestimating the economic changes that are afoot. Brynjolfsson, who ought to be a short lister for the Nobel prize in economics, built his reputation over the past two decades by designing ways to tease out hard empirical evidence on the economic impact of the internet. He now runs the Digital Economy Laboratory at Stanford and has turned his attention to AI.
In the interview, Brynjolfsson cites a recent paper he published at the National Bureau of Economic Research that found rapid and massive productive gains for call centers when they incorporated AI assistance. He goes on to say there have been a number of other studies that have all found “double-digit productivity gains” that occur rapidly, and these gains are showing up across a wide array of industries.
Extrapolating, it is easy for me to imagine an upside scenario for the U.S. economy where double-digit productivity gains spread like wildfire, and overall productivity grows at a 1950s-type rate above 2 percent per year. During the 2010s, productivity growth was typically below 1 percent.
While there are many reasons to be concerned about the downside risks associated with AI, which Brynjolfsson and his colleagues are also studying, consider the upside of productivity growth that high for a moment. It would take economic growth over the next five to ten years perhaps as high as 4 percent per year. Instead of having 20 percent more GDP over the next decade, we would have almost 50 percent more. And for equity prices? They are approximately inversely proportional to the difference between the risk-adjusted interest rate and the growth rate of earnings per share. As the growth rate closes in on the interest rate, the possibility for exponential growth gets higher and higher.
Back in the 1990s, markets saw the internet coming, but also worried that it might not be a big deal after all. In 1998, Paul Krugman famously wrote, “By 2005 or so, it will become clear that the Internet’s impact on the economy has been no greater than the fax machine’s.” It was not just him. That view was shared by enough people in markets to slow the equity boom down, but not stop it. Equities waited to see the data and responded as it became obvious that the gains were large.
That does not mean that jumping into AI-related stocks right now is a riskless bet or even advisable. Far from it. (And this article is not investment advice.) But bookmark the web page of the Stanford Digital Economy Lab and other sites like it. Or you could just bookmark Capital Matters, as we will be on the case.
If AI is really bigger than the internet in its effect, and it shows up repeatedly in the data, then there is a chance that one will be able to recognize the coming of a raging bull market before the markets do. If a dollar once again turns into three dollars in a short amount of time, then it will be because the future will turn out to be far better than any of us currently dare to imagine.