Teaching long-run growth in Principles of Macro can be a little difficult. After all, in the macro data, most growth in developed countries seems to come from “technological progress” and technological progress is measured as a residual. That is, we look at how much GDP grew, estimate how much was due to more capital, how much was due to more workers (or hours), and then see what’s left. That’s technological progress, or productivity growth.
So I agree with Nick Bunker that we don’t necessarily know where productivity growth comes from and why it appears to have slowed down recently. On the other hand, as Noah Smith points out, we know exactly where productivity growth comes from. As I tell my students, there are really only two sources of productivity growth: new products and services, and producing old products and services with fewer inputs.
When we’re chugging along, producing the same products and services with the same production technology, there is no productivity growth. There may be economic growth as we produce more, adding workers and capital in order to do so, but in order to have productivity growth, or technological progress, we need something new. One possibility is that we come up with new products: smart phones! electric cars! smart watches! This growth is actually fairly hard to measure as we need to compare something new that didn’t exist before to the old products. If both old and new survive, usually because the new one offers a different value proposition, this isn’t too hard. But if the new product or service kills off the old because it is better and cheaper (and therefore offers a superior value proposition), the growth could actually look negative at first. This is the argument that a lot of internet services have improved growth but simply aren’t showing up in the numbers. Not only are Google and Wikipedia free, but Encyclopedia Britannica is basically dead.
The other source of productivity growth is more straightforward to measure and involves producing the same products or services but with fewer (labor) inputs. This is what Noah focuses on in his post. The short-term fear has always been that as firms substitute new capital (and new technology) for labor that there is a disruption in the labor market and people lose their jobs. There has been an increasing amount of worry in this area as artificial intelligence could make a large number of jobs obsolete. But as Noah points out, so far technological progress has had a positive long-run effect for workers by making them more productive and pushing up their real income. It also often requires an increased investment in human capital which also pushes up wages. I think this is likely to continue in the future even if robots are doing most of the jobs that humans are currently doing. Humans will then find other things to do that either robots can’t do or that people don’t want robots to do. What are those things? I have no idea But 200 years ago most economists couldn’t have predicted all the 21st century jobs that didn’t yet exist either.
So from the micro side we know exactly where productivity growth comes from while on the macro side it remains a bit of an enigma and basically impossible to measure directly. The question that Noah asks in his piece of whether or not higher labor costs will lead to more productivity growth is an interesting one. A related question is whether higher aggregate demand will lead to more productivity growth. If so, that’s a possible argument for both a higher minimum wage and for reducing income inequality as there is evidence that the poor and middle class have a higher marginal propensity to consume that the rich. I have a paper that makes that argument, but it is not yet ready for prime time.