“You can see the computer age everywhere but in the productivity statistics.” — Robert Solow, 1987
The economy has been growing steadily since we started to recover from the 2020 pandemic and shutdown; the unemployment rate is and has been near the bottom of the range normally considered to be full employment; and, wages adjusted for inflation are the highest they’ve ever been. Nevertheless, the productivity slowdown observed by Robert Solow forty years ago seems to have returned, and surveys of consumer confidence are at or near their all-time lows. This essay explores possible reasons for the productivity slowdown and widespread economic uncertainty in the midst of our AI boom.
Measurement Problems
Among the possible explanations of the productivity slowdown or what merely appears to be a productivity slowdown is our limited ability to measure productivity in an economy increasingly dominated by services.
It wasn’t so long ago that productivity was measured in the sectors of the economy concerned with physical output such as agriculture and manufacturing. The productivity of office workers was either ignored or simply subsumed into the productivity of goods-producing companies.
There were problems of measuring quality differences in goods, but these could in theory be surmounted. For example, by comparisons of the prices of old and new vintages of similar products. And, the changing composition of production could be represented by slowly-changing weights in indexes of production.
As much effort as was put into measuring production, systematic discrepancies persist in indexes of the prices of services versus the prices of commodities. The CPI, for example, typically registers a 1 or 2 percentage point higher rate of inflation than the PPI. With the digital economy, now augmented by AI, a serious effort must be made to measure the production of information, entertainment and services. For example, an hour of broadcast time prior to the development of the internet might have cost something like $1 million (in today’s dollars). What about the several hundred thousand hours of video uploaded to YouTube nowadays?
Prototypes (as Opposed to Practical Inventions)
In the past, it took a while for breakthroughs such as the steam engine to have a discernible impact on production. It was more or less a century from prototype steam engines to when steam engines were first put to a practical use (in the refreshing of air in coal mines). Then, several more decades until steam engines were used to run industrial machinery. Then, yet several more decades until steam-powered locomotives were used in transportation.
Even when the steam-powered locomotive was developed, horse-drawn trains were more reliable. The story is told of a race in 1830 between a steam-powered locomotive and a horse-drawn train along the Baltimore & Ohio Railroad from Baltimore to Endicott City. The horse won the race since the locomotive broke down. Nevertheless, over the next century, horses lost their jobs in transportation.
Ditto, the steam-powered hammer. According to folklore, John Henry who—through superhuman effort—beat the steam-powered hammer, preserving jobs for the men of railroad work gangs. Nevertheless, over the next century, men lost their jobs to industrial machinery. Warehousemen who relied on their strength to manhandle barrels of pork, flour, and other produce were replaced by forklifts. Dockworkers lost their jobs to electric-powered derricks.
It is easy, looking back, to see the impact of the Industrial Revolution on productivity, on wages, and on standards of living. But, in real time, it took time for these benefits to accrue. In the meanwhile, workers had to deal with disruptive change.
More recently, in 1996, a super-computer named Deep Blue challenged Gary Kasparov, the world chess champion of the humans. Kasparov won the match 3 games to 1 game to 2 draws. However, the victory was fleeting. A year later, Deep Blue won a rematch; and, nowadays, a decent laptop computer can easily defeat a grandmaster. Today, in the current phase of the Industrial Revolution, computers, robots, and AI are coming after our jobs, not steam-, gasoline- and electric-powered machines.
Creative Destruction
Tremendous investment in digital technology (a positive) has been somewhat offset by the obsolescence of human capital (a negative). Many experienced professionals in information have lost or are at the risk of losing their jobs to the digital revolution. This includes experienced professionals in print and broadcast industries, college professors, and code-writers. This process is called the creative-destruction of the market process.
The speed of change is daunting. In the past, older workers might be able to hold onto their jobs until retirement (perhaps, by accepting limited opportunities within their occupation). Today, the speed of change is such that older workers may be forced into early retirement or else into lower-paying jobs. In either case, there is the loss to them and to society of a portion of their human capital.
Younger people don’t have as much to lose as older people. They should more easily embrace new technologies. Furthermore, the operation of new technologies is often simplified, even playful. The cell phone is not only a marvelous technological breakthrough, once learned, it’s easy to operate.
The speed of change moving forward may be less than feared. The success of the digital revolution is making resources formerly considered cheap or even free, become scarce. For some examples, water, electricity, and rare earth metals. As long as these resources are priced, the very success of AI will slow down its further growth. At some point, continuing change will have to be human scale (that is, something we can handle).
Changes in the Standard of Living
During the Industrial Revolution, average hours of work per week fell from 80 to 40 hours, that is, people used a portion of their higher income for additional leisure time. They didn’t use all of their higher income for increased material goods. “Leisure” is shorthand for all kinds of non-labor activities including entertainment, travel, and community activities. With a reduction of the average hours of work per week, the number of workers can be maintained even as productivity is greatly increased.
It might also be expected that lifetimes will change from a period of formal education followed by a period of work and then a period of retirement. Lifetimes might, instead, involve a mixed sequence of these things. It might be common to drop out of the workforce and then return. This opens opportunities for formal education, family formation, sabbaticals, and citizen soldiers.
Yet another sexual revolution may be upon us. The first sexual revolution involved the entry of women into the workforce, with consequent pressure on childbearing. With life featuring a mixed sequence of education, work, and leisure, it may be more possible for women to have children while still of child-bearing age, instead of postponing child-bearing until child-bearing becomes problematic.
Security versus Opportunity
For some, the opportunity provided by AI is exciting. Such as young people with no vested interest in the old ways of doing things. But, for older workers, whose human capital is being made obsolete, there is likely to be resistance to change. For them, AI is more like a threat than like an opportunity.
Looking back, fear dominated discussions of the Industrial Revolution. The picture was one of children at orphanages begging “More gruel sir,” as though hunger was unknown prior to the Industrial Revolution. The truth is starvation was common prior to the Industrial Revolution. The masses of people were trapped in poverty. Social classes, serfdom, and slavery were the rule.
The Industrial Revolution began sustained economic growth and upended the world. It improved the standard of living of the masses of people several times over, with consequent changes in human freedom and equality; and, with AI, we will continue to see such revolutionary changes, albeit, with continuing creative destruction.
