There’s a paper over at VoxEu* which postulates that US economic growth is coming to an end.
The paper is deliberately provocative and suggests not just that economic growth was a one-time thing centred on 1750-2050, but also that because there was no growth before 1750, there might conceivably be no growth after 2050 or 2100. The process of innovation may be battering its head against the wall of diminishing returns. Indeed, this is already evident in much of the innovation sector.
— Robert J. Gordon, 11 September 2012
I have some issues with this, but not because of a knee-jerk “that can’t happen to US” response. It obviously can and must happen, if only because of thermodynamics. Our ability to provide energy to our global economy is going to hit a wall sometime in the next 300-400 years. I’m not talking about peak oil, I’m talking about some combination of having to cover the earth in solar panels vs raising the surface temperature to the boiling point.
On the surface, Gordon makes an interesting case: that we’ve already cleared the technological low-hanging fruit, so that future productivity gains will be harder to come by; meanwhile, structural issues in US society add additional headwinds that will help drag our performance back down to colonial levels.
I guess my initial problems with this paper stem from three issues: he’s given up on the computer revolution too soon, he’s ignored some new technology that will have a major impact on productivity, and he’s identified a start point, but not an end point for the productivity drop.
First, the computers. No. First, the technology. Technology always takes longer to have an impact than its inventors realize. As I tell my students when we talk about bringing new technology into a firm, assume that it does exactly what the vendor says it does; what else has to work in order for it to be successful? My favorite example is frozen food. Clarence Birdseye invented flash freezing of food in 1922, but frozen food didn’t become an American staple for over thirty years. What happened? What else had to work? Refrigerated trains for distribution, glass topped freezer displays in the general store for sales, home refrigerators with freezer compartments in place of ice boxes. It always takes longer. Computers are a post-WWII phenomenon, and general purpose business computing really only started with the IBM 360 in the 1960’s. Fifty years later, we are just on the verge of ubiquitous computing — as Cory Doctorow says, a hearing aid is a computer you stick in your ear, and a car is a computer you sit in.
As for new technology, I see nanscale engineering and 3D printing as game changers that will have major impacts on how things are done. Nanoscale materials, so far, are giving us self-cleaning cloth, paint, and glass; medical drug delivery techniques, and new ways of embedding computers. 3D printing will let us have the cost advantages of mass production combined with the customization potential of the job shop — single unit mass production. Are these advancements equivalent to building the railroads across the West? Probably not, but we won’t know their true impact until, say, 2050.
Third, Gordon claims his ‘headwinds’, about which more anon, will bring US productivity back to colonial levels before that date, 2050. But he doesn’t say what might lie on the other side. The structure of US society isn’t going to freeze at that point, and some of the drags on productivity (such as old folks like me) will be starting to ease up (OK, die off) by that time.
Now, he has an interesting graph, possibly the best labeled and easiest to understand of the lot, showing the change in actual and predicted levels of GDP per capita in constant dollars, 1300-2100.
GDP/capita 1300-2100 Source: Gordon, CEPR Policy Insight No 63.
What makes the graph compelling is that it’s a classic S-curve, headed for a rolloff at about $90K/capita sometime shortly after 2100. What makes the graph less than compelling is that there’s no indication we are actually past the inflection point, and until one passes the inflection point there’s no sure way of predicting what that upper bound will be.
Next week, I’ll address some other issues. Like, does it matter? Is it good news?
Did I say next week? I meant “Next time I have a chance to work on this”.
* A web portal operated by the European Centre for Economic Policy Research