MethodLogical is now at

MethodLogical is now at

Due to some persistent technical issues we've been having with Blogger, we're now posting at Please update your RSS feeds, etc. For the time being, our new posts will automatically be mirrored here, but you'll have to visit the new site to comment.

Sunday, December 12, 2010

PPP Income Statistics as Worthwhile Lies

When we talk about poverty our first impulse often is to go to the numbers. Pictures can tell a story but it is much more compelling to back them up with data that shows that the typical person in Malawi earns one one-hundredth of the income of her American equivalent.

The numbers we use to make such claims are usually the annual average of purchasing-power-parity adjusted (PPP) gross domestic product per person (GDP per capita). That typically means we count up the value of all the goods and services consumed in a country each year and divide by the number of people. We then do something clever: anyone who has traveled outside the "developed" world knows that a dollar goes farther in some countries than in others, so we do a "PPP" adjustment to account for that.

Even with this correction it's easy to object to this measure. For one thing, income is heavily skewed: the typical person in the country may consume nowhere near the average income. Since we construct mean per-person income by adding up expenditures and dividing, we can't even talk about inequality if we want to - it's not measured.

But I can easily go the other way as well. Ben Elberger recently posted a link to a fascinating followup story on a Bangladeshi family that had been featured in "Portfolios of the Poor". It's a great article, I highly recommend it for its upbeat and frank take on the family's finances. What stuck out to me, however, was how low the family's income was - and this after substantial progress. At purchasing power parity their daily income is $4.28, which comes out to $128.40 per month.

Bearing in mind that this already is supposed to account for stuff being cheaper in Bangladesh, how in the world are Hamed and Khadeja able to afford rent? The cheapest unsubsidized rent I've ever heard of in the US was $200/month and that was for a single room. The answer is probably that they live in a house that is so poor in quality that no American lives in anything remotely equivalent. In part this may be due to regulation, but it's also because Americans simply demand higher-quality housing because they can afford it.

Economic data shows that there is no market for haunted houses in the U.S.

What I've highlighted here is an example of countries with incommensurable consumption patterns. This is a big problem for PPP adjustments. It turns out that the calculation of PPP exchange rates involves constructing a "ring" of countries around the world where each sequential pair in the ring consume somewhat similar goods. Price comparisons across countries end up involving a whole lot of extrapolation and it can get pretty ridiculous.

It's clear that we shouldn't be taking these figures at face value - so why maintain them at all? Because even though they're not true, they're still useful. Despite concerns about the distribution of income, for example, it turns out that the incomes of poor people in any given country track closely to the mean income. And a low mean income is the single best predictor of the odds that a civil war will break out in a country.

There are tons of other examples like these - mean PPP income turns out to predict all sorts of variables that actually are real and that we do believe in and care about, from life expectancy to the odds that kids go to school. We should certainly measure it as well as we possibly can, but never lose track of what PPP GDP per capita statistics really are: useful tools rather than ironclad facts. Institutions like the IMF and World Bank that publish them need to make this much more obvious.


  1. Hey Kerwin- this is less of a comment than a shameless attempt to get you to do my research for me. WHO defines cost-effectiveness in relation to GDP per capita: Highly cost-effective (less than GDP per capita); Cost-effective (between one and three times GDP per capita); and Not cost-effective (more than three times GDP per capita)

    What's not clear is if they use PPP or not. It looks like the figures are unadjusted, but hard to tell. I suppose if government's are buying goods from international sellers, PPP is less relevant, but it's all murky to me. Any insights?

  2. A little poking around on the WHO website reveals that their cost-effectiveness analyses are done in "international dollars", which is just a fancy term for PPP:

    This is probably the right way to do it - even if they're buying from international sellers, if those goods are available in the country in question then they'll be factored into the PPP exchange rate.


Note: Only a member of this blog may post a comment.