Saturday, May 23, 2009

The Mismeasure of Well-Being

What kind of metrics should we use to gauge the success of different country's economic policies?

Per capita GDP is the obvious choice, but as many have pointed out, it is imperfect. What about inequality? What about education? Health care? Prevalence of slow drivers in the fast lane? Etc.

Enter the Human Development Index. The HDI aspires to be a more complete measure of the human condition than standard GDP measures. Does it succeed? Bryan Caplan doesn't think so:

So what are the main problems with the HDI?

1. I can see giving equal weights to GDP per capita and life expectancy. But education? As a professor and a snob, I understand the appeal (though a measure of opera consumption would be even better). But in terms of the actual if not professed values of normal human beings, televisions and cars are a lot more important than books.

2. When you take a closer look at the HDI's education measure, it's especially bogus. 2/3rds of the weight comes from the literacy rate. At least that's not ridiculous. But the other 1/3 comes from the Gross Enrollment Index - the fraction of the population enrolled in primary, secondary, or tertiary education. OK, I feel a reductio ad absurdum coming on. To max out your education score, you have to turn 100% of your population into students!

3. The HDI purportedly gives equal weights to three different outcomes, but bounding the results between 0 and 1 builds in a massive bias against GDP. GDP per capita has grown fantastically during the last two centuries, and will continue to do so. In reality, there's plenty of room left for further improvement even in rich countries. But the HDI doesn't allow this. Since rich countries are already close to the upper bound, the HDI effectively defines their future progress on this dimension out of existence.

To a lesser extent, the same goes for life expectancy: While it's roughly doubled over the last two centuries, dying at 85 is not, contrary to the HDI, approximately equal in value to immortality.

The clear winners from this weighting scheme, of course, are the literacy and enrollment measures, both of which have upper bounds that are imposed by logic rather than fiat.

4. The ultimate problem with the HDI, though, is lack of ambition. It effectively proclaims an "end of history" where Scandinavia is the pinnacle of human achievement. Admittedly, I've never visited Scandinavia. But when I see it for the first time this August, I'm pretty sure I won't say to myself, "Wow, it can't get any better than this!"

I'm inclined to agree. Per capita GDP is imperfect, but is it worse than the HDI? I'm all for the idea of better measures of the human condition than GDP. The problem with current measures (see also: The HPI) is that they suffer from a consistent bias in favor of statist, intervention-heavy public policy.

Question for the day: Imagine a team of Hayekian libertarians get together to create their own development index. Which factors would they emphasize to create an index biased in a pro-market, pro-liberty direction?