Three centuries of climatic variation and the world income distribution

John C Bluedorn, Akos Valentinyi, Michael Vlassopoulos, 15 December 2009



Policymakers around the world are being urged to address the effects of climate change. Formulating a unified policy on climate change is one of the central goals of the Copenhagen Summit on Climate Change now underway. A key input into that decision-making process is an accurate appraisal of the possible economic consequences of climate change.

Studies of the empirical relationship between climatic variation and economic outcomes provide one means of appraisal. Nordhaus (2006) found a negative, cross-sectional relationship between current temperature and income using a global gridcell dataset. In a country-level panel stretching from 1950-2000, Dell, Jones, and Olken (2008) found a negative and persistent effect of current temperature upon economic growth for poor countries. These papers support the earlier evidence of a negative relationship between standards of living and temperature, presented by Myrdal (1968), Kamarck (1976), and Lewis (1978), among others.

Delving deeper into the past

Much of the previous research has focused upon temperature variation from the latter half of the 20th century. Recent paleoclimatic estimates of temperature have opened up the possibility of investigating the effects of climate from much earlier. The paleoclimatic estimates are based upon a diverse set of indicators (such as ice cores, tree rings, geological measures, and so on) that are related to climate. With these in hand, we can push further back than the instrumental record for temperature to consider the effects of climatic variations over the past three centuries upon current incomes. Such an effort yields striking results (Bluedorn, Valentinyi, and Vlassopoulos, 2009).

Persistent and surprising effects of historic temperature

We find that the negative temperature-income relationship appears to reflect the long-run effect of temperature variations in the 18th and 19th centuries – not the effect of current temperature alone.

Moreover, the effect of historic temperature upon current income depends upon the time of its occurrence.

  • 18th century temperature has a positive relationship with current incomes,
  • 19th century temperature has a negative relationship.
  • After controlling for temperatures in the 18th and 19th centuries, 20th century temperature has either a weakly positive or no effect upon current incomes; this finding survives a host of robustness checks.

Temperatures from up to two and a half centuries ago have a stronger influence on current income than does current temperature. Figure 1 shows the simple relationship between 19th century temperature and current incomes.

Figure 1 19th century temperature versus current incomes

How could effects be so persistent?

Dell et al. (2008) found a permanent, negative effect of 20th century temperatures on poor countries’ economic growth. With the power of compounding, even a small, permanent growth effect can translate into a large level effect given enough time. This is what we see in our results – historic temperature variations have a powerful effect upon current incomes. Unlike Dell et al. (2008), we see this pattern even within a sample of high income countries.

Why does 18th century temperature have a positive effect on current income, while 19th century temperature has an even larger negative effect? Our interpretation is that there are important interactions between temperature and historic events across centuries. For example, if there are complementarities between technologies and the climate of their originating countries, then the rise in technological innovation associated with the Industrial Revolution in the 19th century could lead to differences in technological diffusion across countries. Through the power of compounding, these would translate into large differences in incomes today. Additional evidence is required to fully assess the importance of such historic interactions.

Historic temperature and the world income distribution

Accounting for historic temperatures changes the implied long-run effect of movements within the temperature distribution upon income. As an example, if we only consider the simple relationship between current temperature and income, moving from the median to the 90th percentile of the temperature distribution would be associated with a 25% drop in income. If we consider the more complex relationship with historic temperature and a similar movement within the temperature distribution each century, income falls by twice as much – 50%.

More concretely, consider the cases of Canada and Sudan, which are near the extremes of the temperature distribution. Current Canadian income is 25 times the size of Sudanese income. If we only account for 20th century temperature differences between the two countries, we would predict that Canada's income was about 4 times larger than Sudan's, explaining only 15% of the income gap. If we also take account of 18th and 19th century temperature differences, then that increases to 7 times larger, explaining almost 30% of the income gap.

How does historic temperature affect current incomes?

Unlike current temperature, historic temperature cannot have a direct effect upon current incomes. Historic temperature must be affecting income through its effect upon some other driver in history, which in turn either persists or affects yet another driver.

Our reduced-form approach allows us to recover robust, general patterns in the relationship; it does not allow us to resolve the exact mechanism by which historic temperature affects current incomes. Looking at various candidate channels, the most promising appear to be human health and political institutions. Accounting for historic temperatures leads to marked increases in the explanatory power of temperature for the cross-country distribution of these channels.

The message for policymakers

The core message to take from this empirical work is that temperature variations can have effects upon standards-of-living that last for centuries. Moreover, these persistent, historic effects are more important than current temperature differences for current standards-of-living. To the extent that these historic correlations continue into the future, a corollary of these findings is that the full consequences of climate change for the income distribution will likely not be felt for many years. When they are, they will last a very long time.


Bluedorn, John C., Akos Valentinyi, and Michael Vlassopoulos (2009), “The Long-Lived Effects of Historic Climate on the Wealth of Nations”, CEPR Discussion Paper7572, November.
Dell, Melissa, Benjamin F. Jones, and Benjamin A. Olken (2008), “Climate Change and Economic Growth: Evidence from the Last Half Century”, NBER Working Paper14312, June.
Kamarck, Andrew M. (1976), The Tropics and Economic Development, Johns Hopkins University Press for the World Bank.
Lewis, W. Arthur (1978), The Evolution of the International Economic Order. Eliot Janeway Lectures on Historical Economics in honor of Joseph Schumpeter. Princeton University Press.
Myrdal, Gunnar (1968), Asian Drama: An Inquiry into the Poverty of Nations. Pantheon. 3 volumes, sponsored by the Twentieth Century Fund.
Nordhaus, William D. (2006), “Geography and macroeconomics: New data and new findings,” Proceedings of the National Academy of Science, 103(10): 3510–35177 March.

Topics: Environment
Tags: climate change, geography, history

John C Bluedorn

Economist, World Economic Studies Division, Research Department, IMF

Cardiff Business School, Institute of Economics, HAS, and CEPR Research Fellow

Lecturer in Economics, University of Southampton

Vox Talks