A set of more or less arbitrary lines continues to do very strange things to discussions about development. Most strange and arbitrary: country income classifications, with their division into low (<$1,036), middle ($1,036-$12,625), and high (>$12,625). Almost as strange and still pretty arbitrary: $1.25 and $2 poverty lines. We worry about aid allocations, who should give aid and contribute to global public goods, and poverty and middle-income traps all based on measures that only matter because we say they do: they're the placebo poison of development. To focus on the country-income classifications, here are two reasons they really can't be justified:
1. There is no natural grouping of economies based on GNI per capita.
Figure 1 lines 178 economies up according to their current Atlas GNI and then plots their GNI on a log scale alongside a trend line. If there were natural groupings of countries based on clusters of economies with similar GNIs separated by income bands with very few countries in them, the GNI line would have flat stretches followed nearly vertical stretches. But in fact, the GNI line follows the trend line very closely, showing that countries are spread fairly evenly in the space between richest and poorest. There is no strong sign of natural country groupings here. The dotted lines mark (approximately) the current income cutoffs. They appear to be poorly placed to capture what clustering does occur.
Source: World Bank
[view whole blog post ]