Why Economic Growth Has Slowed Even as Technology Keeps Improving

It can feel strange to talk about slower economic growth while watching technology improve so quickly. A phone can perform tasks that once required an office full of equipment. Software can produce an answer in seconds that previously took hours of work. Yet the existence of impressive inventions does not tell us how quickly the whole economy can expand. The American growth slowdown that developed over recent decades reflects changes in the workforce, investment and productivity across ordinary businesses, including plenty of places where progress has been much harder to achieve.

The basic arithmetic is fairly simple. An economy can produce more by using more hours of work or by producing more in each hour. For a long time, the United States benefited from an expanding workforce. Population growth and rising participation brought more people into paid employment. An aging population makes that source of growth harder to sustain, while immigration can significantly change the outlook.

This is partly a question of economic size. Slower growth in the number of workers does not automatically mean that each person’s standard of living must stop improving.

That makes productivity especially important. Workers can become more productive through better equipment, improved skills and more effective ways of organizing production. A BLS analysis estimated annual productivity growth in the private nonfarm business sector at 3.3 percent from 1997 to 2005, compared with 1.4 percent from 2005 to 2018. It identified weaker gains in both capital intensity and the efficiency with which inputs were used. Those historical figures describe a particular slowdown, rather than an unbroken decline continuing indefinitely. More recent productivity gains are a reason to be cautious about treating weak growth as permanent.

One difficulty is that inventing something useful and putting it to work throughout the economy are separate achievements. A business may buy new software without changing the procedures that make it slow. Employees need training, existing systems need to connect and managers need to decide which tasks should disappear altogether. Research on the productivity J curve describes how these complementary investments can delay the measurable benefits of a major technology. AI could eventually produce substantial gains while requiring an expensive and uneven period of adjustment first. The size of those gains remains uncertain.

The composition of spending matters as well. Some things are much easier to produce efficiently than others. A factory can make more goods with fewer hours of labor, while caring for an older person still requires considerable human attention. As spending and employment shift toward activities where productivity is harder to raise, rapid progress in a few industries may do less for the overall growth rate. This is one mechanism associated with Baumol’s work on uneven productivity growth. It helps explain why cheaper electronics can coexist with persistent pressure on the cost of essential services.

Small changes in growth eventually make a large difference. If output per person grew by 2 percent a year, it would rise by roughly 64 percent over twenty five years. At 3 percent, it would more than double. That is an illustration of compounding, not a forecast, and it says nothing about how evenly the gains would be shared. It does explain why improving productivity deserves attention beyond the latest technology announcement. Lasting gains require useful ideas to become better equipment, better services and better working practices across the economy. An impressive demonstration is evidence of possibility. The harder part is making the improvement ordinary.