Income per person revised and expanded

We now have data for 219 countries and territories for 1800-2007, although the data for the 19th century are largely based on rough assumptions. For a couple of countries the revisions mean substantial changes. You can still find the old version of the indicator, under “For advanced users”. 

See the new indicator here. Note that the revisions also apply to all the “gaps within countries” graphs.

One major change is that the data is now based on the purchasing power parities (PPPs) of the 2005 round of the International comparison program (the ICP). In the earlier version of the indicator, which we will call “old version of income per person”, we used PPPs for 1990 based on Angus Maddison.

For some countries these revisions have meant quite substantial changes in their relative positions. Here we plot the new income per person against the old income per person to illustrate the differences caused by the revision. You find the old version of the indicator in the indicator list, under the heading “for advanced users”.

One widely discussed change in the new data is that the new income level of China is significantly lower. However, this does not, in itself, mean that the growth rate of China has been lower than older data implied. Here is one article discussing this. 

We will upload a detailed documentation of the indicator soon. In the meantime we give a short description of how the dataset was compiled:

  • We used the GDP levels based on the ICP for 2005 (hence the unit of the new indicator is 2005 international dollar, rather than 1990 international dollar). 
  • This was extrapolated to more countries using various other data sources, together with an estimated model. 
  • We use the national real growth rates to extrapolate to other years from the icp data for 2005. The major source for these national growth rates was the database of Angus Maddison
  • For countries where Maddison had no data for the 19th century we used the average regional levels of GDP of Maddison (e.g. his estimated average for Sub-Saharan Africa). We adjusted the data to make them comparable to the other data (which were expressed in 2005 fixed international $). 
  • The countries that use the same regional average will end up on the same position in the graph, which will make most of them invisible. Hence, we decided to “spread” the observations around the average, using various assumptions (e.g. reflecting the relative position the country had at a later date). 

The data for the 19th century are accordingly to a large extent based on regional averages, which in themselves are only more or less qualified guesses. Hence, one should not make too much of the relative position in the graph of most of the individual countries in the 19th century. 

So does the data for the 19th century contain any information at all? First, we do have some information for a couple of individual countries. It is quite uncontroversial that these included the richest countries at that time, countries like the UK where the industrial revolution had started to take-off. The exact income levels of these countries are highly uncertain, of course, but the level of magnitude of their income are not that controversial. Hence, this gives us an approximate upper bound for the average incomes in the countries of the world.

Second, it is hard to assume that any country have had a lower income level than something like 100-200 $ for any longer period. This would represent incomes well below the absolute poverty line, and would hence imply a continuously catastrophic-situation for the vast majority of the people in the country. This argument has been made by Lant Pritchett in “Divergence, big time”, who sets his lower bound to 250$ in 1985 years prices. All this taken together gives us an approximate range of likely values.