Monday, May 25, 2009

Mo' Money, Less Problems

. Monday, May 25, 2009

KPC just posted the above picture, taken from research by Justin Wolfers. The picture shows the correlation between the U.N.'s Human Development Index (HDI) and per capita income. The correlation is .95, meaning that the two things are almost perfectly correlated. In the social sciences, relationships this strong are exceedingly rare.

HDI is an attempt to compare development outcomes across countries in a more nuanced way than simply looking at GDP numbers. So it includes things like educational enrollment, life expectancy (at birth), and adult literacy as well as per capita GDP. The thing is, those things are all highly correlated with per capita GDP. So even though the intention is to give a more nuanced view of quality of life than simply comparing income levels across countries, what we end up seeing is that income levels are the primary determinant of quality of life.

This shouldn't be too surprising. After all, if you have more money you can buy better health care (and thus live longer) and afford to send your children to school (thus boosting literacy and educational enrollment). However, simply comparing incomes across countries is often thought of as being a crass way of measuring quality of life. After all, money can't buy love or happiness, right? And if we spend our lives selfishly chasing money than we are surely missing other things.

That may be true, but it doesn't show up in cross-national studies. Wolfers has also done other work on the relationship between income and happiness (discussed here briefly; the actual research is here). This work (with Betsy Stevenson) re-examines the "Easterlin Paradox" that has caused much debate in the social sciences for decades. The Easterlin Paradox states that citizens of countries with more economic growth are not happier than citizens of countries with less economic growth, as was commonly assumed by economists. The inference was often made that there is no relationship between income and happiness. But the Stevenson/Wolfers research indicates that this isn't true, and the Easterlin Paradox doesn't actually hold: higher levels of GDP are strongly related with higher levels of subjective happiness.

This new research all points in the same direction: higher economic development leads to better objective and subjective outcomes. People in richer countries live longer, more enriching, happier, more fulfilled lives. So if we want to improve peoples' quality of life, then we need to improve their economic opportunities. And if we don't do that, we will have a difficult time improving outcomes.


Emmanuel said...

Kindred--why should we not expect HDI--a third of whose weight is per capita GDP--to display such a high level of correlation with per capita GDP?

As Amartya Sen said, HDI is a vulgar measure of well-being. Still, it has its purposes, although I would generally shy away from making too much of the whole resembling the part as it doesn't tell us much of anything.
It's like saying the Obama administration mirrors many of Obama's ideas.

There are many initiatives to develop more sophisticated indicators. Wolfers not even bothering to use just indicators of health and education to deal with multicollinearity is, to be honest, a n00b mistake.

Perhaps "the hidden side of everything" in Freakonomics is often just ignorance--like it is here.

Kindred Winecoff said...

Emmanuel -

I am not surprised by this, but I think that many progressives or liberal egalitarians would be. Yes, it is true that since per capita GDP is one-third of the input it should it be highly correlated with HDI, but .95? That means that per capita GDP is almost completely driving the index, and those other measures are superfluous. This is the point I was making in the post.

Multicollinearity inflates standard errors, making coefficient estimates lose some of their "true" significance. It doesn't change the coefficient estimates themselves.

And Wolfers isn't saying that these things don't covary; in fact, I'm sure he'd argue that they do. But he would probably argue that it's difficult to for individuals (or nations) to obtain good education and health without money. I don't think that's an ignorant inference at all. There is only one plausible counter-argument: that the other inputs of HDI are the ones that cause per capita GDP to go up rather than the other way around. But I think that's a somewhat difficult case to make.

Mo' Money, Less Problems




Add to Technorati Favorites