Comparing Average Income to GDP per Capita

This article has no grand purpose. It just documents some technical stuff that I’ve learned about global economic data.

In the previous article I included a chart comparing a bunch of the world’s richer countries by GDP per capita. But I also described Max Roser’s estimate of the minimum economic growth needed to reduce world poverty to the same level as in Denmark, and that estimate is based on average personal income, not on GDP/capita.

Does it matter whether we compare countries by income or by GDP/capita? If so, which is better?

Fortunately, Our World in Data provides a handy bubble chart that plots average income (or expenditure) per day vs. GDP per capita, country by country. Unfortunately, this chart follows the usual convention of presenting income on a daily basis but GDP on an annual basis, so to understand it you have to keep multiplying and dividing by factors of 365. I’m terrible at doing that in my head, so I had to draw some lines on the chart to show where GDP/capita equals average income, where it’s twice the average income, and where it’s four times the average income:

Isn’t this interesting? In nearly all countries, GDP/capita is larger than average income. In the richest countries the ratio is pretty consistent: a factor of about 2 or 2.5. Elsewhere GDP/capita can exceed average income by a factor of more than 5 (Kazakhstan, Egypt, Nigeria), or, in a few of the poorest countries, not exceed average income at all (Haiti, Comoros, Liberia). For the world as a whole, GDP/capita exceeds average income by a factor of 2.66.

So what’s going on here?

Well, for one thing, there must be uncertainties in the data. My understanding is that the mean income numbers come from statistical surveys of households, which inevitably involve sampling errors and could also have systematic reporting errors. Governments compile GDP data by measuring production and trade and such, but these numbers won’t be exact either. How inaccurate should we expect any particular point on this plot to be? I have no idea, and my impression is that it would take weeks of digging into the data sources to get much of an answer.

Complicating things further, not all the “income” values are actually income. Some of the surveys ask instead about household expenditures.

Yet another complication is that a number of countries (Saudi Arabia, Oman, North Korea, Cambodia, Singapore, Argentina, Venezuela, Cuba, New Zealand, and many smaller ones) are missing from this chart, presumably because there are no good income surveys for those countries.

But the basic fact that GDP per capita is greater than average income (or expenditure) is no mystery. Roser explains it qualitatively:

GDP includes many items that are typically not measured in household income surveys, such as an imputed rental value of owner-occupied housing, the retained earnings of firms and taxes on production such as VAT. The gap is even larger when GDP is compared to surveys of household consumption – the latter concept excluding both investment expenditure and government expenditure on public services such as education and health.

Unfortunately, I don’t have a good feel for which of these factors contribute the most to the GDP/income ratios shown in the chart above.

What I’d really love to know is how these differences translate into people’s actual standards of living. Kazakhstan and Nicaragua apparently have the same average income, even though Kazakhstan’s GDP/capita is 4.5 times Nicaragua’s. Is life in Nicaragua more similar to life in Kazakhstan than it is to life in Nigeria, which has the same GDP/capita as Nicaragua but 4.3 times less average income? How does life in Nigeria compare to life in Mozambique, with the same (low) average income but a GDP/capita that’s less by a factor of 4.7?

Now let’s return to Roser’s estimate of the economic growth needed to reduce world poverty to the same level as in Denmark today. Denmark isn’t labeled on the chart above, but it’s one of the smaller dots just under the A in “Austria”, with a GDP/capita of $46,683 per year and an average income of $55.63 per day. Those numbers are 3.02 times higher and 3.49 times higher, respectively, than for the world as a whole (GDP/capita $15,469 and average income $15.94). Roser effectively uses the Denmark/world income ratio of 3.49, which he multiplies by a world population growth ratio of about 1.46 (roughly the projected world population in the year 2100 divided by the population in 2017 or so) to obtain a needed world economic growth ratio of 5.1.

If we prefer GDP/capita, rather than average income, as our measure of the size of the economy, then the only change we need to make to this calculation is to substitute the Denmark/world GDP/capita ratio, 3.02, for the income ratio of 3.49. Then instead of a needed world economic growth ratio of 5.1, we get 5.1 times 3.02/3.49, or 4.4.

The basic point, of course, is unchanged: The world economy needs to grow substantially if we’re going to lift nearly everyone out of poverty.