Evan Soltas
Sep 1, 2012

Gordon on Growth

I read Robert J. Gordon's new paper, "Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six Headwinds," with great interest earlier this week. (An ungated version is available here.) Although I am not in truth a pessimist, I think there may be a piece of empirical evidence for Gordon's argument which went overlooked: population growth.

The annual rates of global and developed-world population growth look very much like Gordon's historical graphs of real per-capita output growth over the very long run. More specifically, the initial rates of growth, the acceleration's timing and manner, and the peak's timing are all shared -- as are the more-certain demographic forecasts with Gordon's expectations for a normalization of economic growth rates.

It is tempting to ask whether this is more than simple coincidence. And I think there is a plausible argument that the paths of population growth and real per-capita output growth are tied up in a single mechanism of causation.

By increasing carrying capacity, technology shocks may induce increases in population and output per capita over the very long run.

Suppose that Gordon's three technological shocks or "industrial revolutions" create discrete "stepwise" jumps in the carrying capacity of the planet. Mechanization of agriculture and the development of chemical fertilizers, for example, made the Earth able to support a much greater population and increases agricultural productivity, leading to an increase in output and living standards. To such a shock, population and output both respond with a lag, growing logistically from the old level of carrying capacity to the new carrying capacity with technology. The technology takes time to diffuse through society and generate downstream inventions; populations require time and pre-existing people to grow.

(An assumption one must make here is that the trade-off between increasing population and output per capita is relatively fixed -- that is, that society cannot easily "choose" to take an increase in productivity as mainly population growth or real per-capita output growth. Gordon's shocks transmit themselves in roughly constant fractions into population and real output per capita. They tend to result in a somewhat better-off and somewhat larger population, rather than only a larger or better-off one.)

Here's where looking at population growth as a guide to long-run output growth gets interesting. Population appears to follow a logistic growth model, as does Gordon's path for per-capita output. If population and per-capita output are both steered in the very long run by technological shocks to carrying capacity, then one may infer from the slowdown in the population growth rate that the technological shocks are dissipating; and from this, that the real per-capita output growth rate should slow down in the very long run.

This model has an interesting implication: economic growth without population growth in the very long run is not likely to happen, as any technology shock will diffuse itself into both. That may be grim, but it's the flip-side implication of what Julian Simon said in The Ultimate Resource, that population growth will be met by economic growth as the new population produce ideas and not just strains on resources.

Slowing population growth may offer some evidence for Gordon's contention that a winding-down of technology shocks is set to slow real per-capita output growth in the very long run.