Why the U.S. Economy Can’t Grow Fast Anymore
Before 2021, the U.S. economy as measured by Gross Domestic Product (GDP) had not increased by at least 3% in a calendar year since 2005.¹ The only reasons annual growth exceeded 3% in 2021 were comparison with the lockdown-triggered economic contraction of 2020 and now-ended government payment programs covering the vast majority of the American population.
How poor is the U.S. economy’s 21st Century performance by historical standards?
Unfortunately, it seems like routine growth expectations from previous generations are now completely out of reach, but why might this be? What changed?
Historically, the way many economies grew was by low-productivity agricultural workers moving from the countryside into cities and obtaining higher-productivity manufacturing work. This type of migration drove double-digit annual GDP gains in China through the early 21st Century. Because China eventually ran out of countryside residents who were willing to migrate and could be much more productive in cities, annual GDP growth above 10% hasn’t occurred there since 2010.²
By 1960 in the United States, 70% of the population was already urbanized, so rural-to-urban migration hasn’t been a big economic driver for a long time, but what about other types of low-productivity to high-productivity migration?³
In the 1980’s, word processors and laser printers superseded typewriters for office document creation. Accounting with spreadsheet software represented a quantum leap over paper ledgers.
However, and this is where it gets interesting… what if there is a limit to how much economic growth automation can generate, just as there is a limit to the number of countryside residents willing and able to move to a city for a factory job?⁴
If replacing farm work with factory work can support 10% annual GDP growth, replacing paperwork with computer work can support 4% GDP growth, and holding a device with access to almost all the world’s knowledge in the palm of your hand can support 2.5% GDP growth, what comes next?
The point is not that the pace of human innovation might be slowing down. Even if it is speeding up, the economy has evolved to a high enough level where new ideas cannot have the economic impact they once did. If fusion energy became practical tomorrow and generated electricity at half the cost of natural gas, it would not change your personal or professional life to nearly the extent that the electrification of society did for Americans in the first part of the 20th Century.
But what about the recent advances in artificial intelligence (AI)? Couldn’t replacing word processors with text prompt input boxes be as important for economic growth as typewriters replacing pens or word processors replacing typewriters? What will the economic impact be once at-home viewers can order up an original action movie starring their favorite actors (living or not) with a few clicks on their remote and start streaming within a couple of minutes?
Despite its massive potential for disruption, AI is starting from the base of current human knowledge. For example, if I ask for a photorealistic portrait of George Washington, the result I get will be based on existing portraits and sculptures of George Washington created by humans and combined in a new way.
What happens, however, if another AI scrapes my result above and uses it the next time someone asks for a photorealistic George Washington portrait? Well, the basis is still human work, humans still need time available to view it, and there are still just 24 hours in a day.
Okay, in that case I’ll just ask an AI to write a Medium article that will make me an amount of money equal to a day’s wages. Then, I could skip my work shift and have more time available to view AI-generated images.
Such an approach might succeed at first but, once more authors try it, Medium’s payments to those authors at current rates would exceed its revenue from the finite number of people on the planet willing to pay for a subscription. This thought experiment highlights another human limit that’s increasingly difficult to circumvent as birthrates fall in many countries.
So, what can be done to push past current human limits? Following highly questionable outcomes from a variety of past attempts centered on pharmaceuticals (e.g. steroids to enhance athletic performance), some believe success will be found by engineering human beings themselves to be more productive and even mass producing future economic participants in “artificial womb facilities.” Others might view these types of concepts as immoral or insane.
Even the most ethical forms of progress, however, do not come for free. During the last Century, it became possible to recover from and even avoid previously incurable illnesses using technological means to augment the abilities of the human body’s natural immune system.
While these medical advancements certainly allowed for economic growth beyond what would have otherwise occurred, this was only because individuals were expected to use much of the extra time improved health afforded them to perform productive work. In other words, once barriers limiting individual potential are lowered, expectations of the individual are raised.
Footnotes
¹ GDP growth (annual %) — United States, World Bank national accounts data, and OECD National Accounts data files, retrieved 20 Dec 2022.
² GDP growth (annual %) — China, World Bank national accounts data, and OECD National Accounts data files, retrieved 20 Dec 2022.
³ Urban population (% of total population) — United States, United Nations Population Division. World Urbanization Prospects: 2018 Revision, retrieved 20 Dec 2022.
⁴ The point at which migration to cities no longer generates economic growth is called the “Lewis Turning Point” after Nobel Laureate economist W. Arthur Lewis. AI researchers Tshilidzi Marwala and Evan Hurwitz extended the Lewis Turning Point concept to automation.