The Great Divergence in living standards can be recovered from a number of proxies. Economists pay too much attention to per capita income. If we pay attention to that, the vast gulf between rich and poor nations is evident. But since per capita income grows exponentially in the modern growth regime, we get a poor sense of when the divergence began. In what follows, the income data is from the Maddison Project and the rest is from Clio-Infra. We arbitrarily define rich nations as those with per capita income at least half that of the United States in 2000.
The median of rich nations’ per capita income is close to $38,000; for the poor nations it’s closer to $7,300. We can track this better if we log transform the data. We can see the structural break in the mid-nineteenth century. But incomes don’t really take off until the second half of the twentieth century.
We can also pick up the scent from other, more kosher, measures of living standards. I’ve argued before that actuarial variables like life expectancy are a much more reliable measure of living standards than national income statistics or anthropometric measures like stature. What is clear from two graphs above and below is that the poor world shared in the uptick, but the rich world pulled further away. (Note that we don’t have enough data for 2010 — we shouldn’t read too much into the rise in life expectancy in the poor world during 2000-2010.)
Stature is confounded by Bergmann’s rule — bigger bodies are found in colder climes. But gain in stature is a good measure of the divergence in living standards. Stature began to diverge at the turn of the century. By 1980, people in the rich world were 7.8cm or 3 inches taller than people in the poor world.
Clio-Infra’s Well-Being Index displays a similar pattern.
Crucial to the story, although often overlooked, including by the present writer, is the demographic transition. Economists think that high fertility rates are characteristic of all premodern societies. Anthropologists know better. Hunter-gatherers have much lower fertility rates than sedentary people, largely on account of the fact that it is difficult to stay on the move with toddlers. Thus, the Malthusian regime itself has a history — population growth could wipe up gains in living standards only after the long Neolithic transition, c. 10-5ka, and even then it took millennia for packing thresholds to bind. Indeed, it is not until well into historic time (beginning at Uruk, c. 3200 BC) that Malthusian discipline began to affect most agrarian populations on the planet. Recall that the spread of the Neolithic way of life was very slow.
The Western world had slightly lower fertility rates than other agrarian societies. But fertility rates started falling rapidly after 1870. The Fertility transition occurred in the rich world nearly a century before the poor world.
Britain and India are characteristic. Fertility began falling in earnest in Britain c. 1880 and in India c. 1970. As we can see from the graph above, this century-long gap is typical.
Galor and others have traced the fertility transition to increasing rewards for human capital in the core of the world economy as a result of the Industrial Revolution. What is particularly interesting about Galor’s “Unified Growth Theory” is that the international division of labor that emerged in the late-19th century not only put downward pressure on fertility rates in the north. It also put upward pressure on fertility rates in the south, as the periphery came to specialize in primary production for the world’s factories up north. This is a much more compelling channel from the global division of labor to the great divergence than the one on offer in standard core-periphery schema. The picture of the rich growing richer and the poor growing poorer as a result of differences in profit rates or terms of trade (“development of underdevelopment”) is not persuasive — the rich world largely traded with itself. Much more compelling is the dynamic picture. As Braudel put it, it was better to specialize in wine than barley.
The level of per capita income is highly correlated with fertility rates. The standardized slope coefficient (with both the response and the predictor regularized) is -0.81, meaning that a one standard deviation higher fertility rate predicts a 0.81 standard deviations lower log per capita income.
But what is the proximate channel through which fertility rates affect income levels? Part of the answer is investment in human capital. Education is as correlated with income (beta=0.81) as fertility (beta=-0.81).
Fertility rates are very tightly correlated with education as well (beta=-0.92, although the quadratic is a much better fit since fertility rates eventually stabilize at a low level).
If fertility affects income levels only through education then the gradient ought to vanish once we control for the latter. This is not the case. The gradient is attenuated but remains robust, suggesting that there are other channels from fertility to income levels beyond merely mean years of schooling.
Nothing changes if we change our response from per capita income to life expectancy or Clio-Infra’s well-being index.
But level variables are confounded by all sorts of factors. It is almost always better to look at growth rates. Again we find that change in fertility rates are highly correlated with change in per capita income (beta=-0.76).
While robust in the simple regression, the fertility gradient vanishes once we control for national gains in average years of schooling. The implication is clear. Nations where the fertility decline is less marked invest less in educating their young (perhaps because you can’t educate all eight of your kids on a meager income) and this, in turn, predicts lower gains in per capita income as a result of slower human capital formation.
The interpretation of these patterns is simple but compelling. The demographic transition is a crucial part of the story. The rise of the modern world economy in the mid-19th century magnified slight differences in skill levels by triggering a virtuous feedback loop in the advanced zone whereby people began to have fewer children and invest more in them. It may also have arrested development on the periphery, by driving fertility rates even higher — although the data to test this hypothesis is sparse.
Galor is much more interesting and serious than I had realized. The non-existent hump-shaped relationship is at best a correction term in an otherwise interesting model of the hockey stick and the great divergence.