John Quiggin concludes a post with a “very wonkish note” on Purchasing Power Parities:
(Very wonkish note) Although PPP numbers are often treated as if they are are raw facts, they are index numbers which are fundamentally imprecise (even if the underlying data is perfectly accurate, which it isn’t). From work I did with Steve Dowrick in the 1990s, I estimate the difference between upper and lower bounds at around 10 per cent. It’s likely that any bias in PPP numbers favors the US. That’s because they are a generalized kind of Laspeyres index, and (as I understand it) the base data is derived largely from Europe.
I call for more wonky blog posts about PPP data. This kind of information is sort of the most dangerous kind—it’s easy to look up, it produces precise-looking ordinal lists, but few people (certainly not me) really understands how it works. Take a country like China, where the nominal GDP per capita is about $3,800 but the PPP GDP per capita is a much-higher $6,600. China is also a very large and diverse society with substantial class divisions. Presumably the typical “basket of goods” purchased by an urban professional is very different from the basket of goods of a peasant farmer. So whose purchasing power are we talking about here? I think relatively few of us who mention this data now and again really have a firm grasp on it.