Growth slowdown (2): GDP is a measure of a mid-20th century industrial economy

(Back to the overview post)

The invention of modern national accounting, which includes the construction of GDP estimates, is credited to Kuznets and others, who did pioneering work on measuring output in the 1930s and 1940s. Ever since it was invented, thoughtful economists have cautioned against using GDP as a measure of welfare; all the while most other economists do just that.The standard criticisms of national output are outlined in clarity by the Stiglitz-Sen-Fitoussi commission, or more poetically by Robert Kennedy:

I would express these specific criticisms in a common way:

GDP measures the things that seemed important in 1940. #

National accounts take a picture of the world through a 1938 Kodak 35 rangefinder camera, and if, like a vintage filter on Instagram, they make every year look just a bit more like 1940 than it actually is, we shouldn’t be surprised. In the America of 1940, the economy was all about things you could could stub your toe on, and ultimately exchange money for, be that a bushel of wheat or a 1939 Studebaker Champion. ‘The economy’ did not include unpaid domestic work, because women worked in the home and always would, and so with a brief ceteris paribus we could ignore them. It did not include environmental degradation - Kennedy’s ‘air pollution’ - because Silent Spring hadn’t been written yet, and the environment had a boundless capacity to absorb our waste.

In short, the further you travel from 1940 - in either direction - the less relevant GDP becomes as a measure of anything very interesting.

While Angus Maddison heroically attempted to estimate GDP going back centuries or even millenia, these numbers are next-to-impossible to meaningfully interpret once we move too far from our own experience. Apparently, Haiti in 2008 had the same real GDP per capita, as Germany in 1500 (around $690 in 1990 prices). And yet Haiti has 4.2 million cellular phones (2011), whereas that kind of sorcery would get you burned at the stake in 16th century Germany. In Haiti, the life expectancy at birth is 63, miserable by world standards. But in 16th century Europe, it was an even-more-miserable 40. So what does GDP tell us? Was life in Germany in 1500 anything like life in Haiti in 2008?

Similar issues arise as we travel forwards in time. Consider, for example, that in 1940, Encyclopaedia Britannica was sold door-to-door, presumably recording substantial revenues, which fed into GDP estimates of publishing. Today, Wikipedia has entire displaced the traditional encyclopaedia (and many other reference books besides), and its revenues are negligible. The encyclopaedia-publishing sector has simply vanished from the national accounts, and yet we have better encyclopaedias than ever before. The same problem applies to open-source software, and free access to entertainment (music & videos on Youtube, for instance).

This is a well-known problem. GDP captures market exchange, but what we really care about is an imaginary and not-directly-measurable idea that economists call consumer surplus: the degree to which you value something more than the price you pay for it.

Some commentators think this is not a big problem today. Much of the economy is still involved in producing things that would be familiar to our 1940s researcher. But that’s gradually changing, and there’s no question it distorts GDP comparisons over time. The actual calculation of GDP by statistical agencies is a patchwork of minor repairs to deal with issues like this, and I’m sure they can go on repairing it for some time. But now, more than ever, we need to be careful about how we interpret GDP. And I would say one of the worst ways to interpret it, is to use GDP growth per capita as a measure of technological progress.