Dietz Vollrath at the Growth Economics blog:
Does Culture Matter for Economic Growth?: There’s been an increasing number of papers concerned with culture and its relationship to economic growth. I happened to just see this working paper by Di Tella and MacCulloch (2014), but the idea of culture being an important determinant of economic development levels has been hanging out there in the literature for a long time. Weber’s theory of the Protestant work ethic is probably the starting point for any discussion of this topic. More recent work tends to try and be more empirical than Weber, often using World Values Surveys as a means of measuring cultural elements. This is what Di Tella and MacCulloch do in their working paper. [If you'd like a good introduction to the culture literature, check out James Fenske's course materials, in particular his "Foundations of Development" course].
I think this is pretty interesting reading, but I’m starting to get a little antsy about the use of the cross-country empirical work. Not in a standard “Identification!!” way, although that’s an issue, but in a slightly deeper way. In particular, why bother regressing GDP per capita (or growth, or any measure of economic activity) on cultural variables at all? ...
So here’s the issue... If culture leads to different utility functions, which in turn lead to different measurable economic outcomes, then why should we bother with measuring economic outcomes? Let me take this from the opposite angle. If everyone has identical utility functions, then measurable economic outcomes (GDP, average wages) have some information about relative welfare across countries. But if everyone has a different utility function, then measurable economic outcomes don’t necessarily provide any information about relative welfare. If one culture derives utility from having massive families with lots of kids, and doesn’t really care about consumption goods, then what does their low GDP per capita tell me? Nothing. It doesn’t tell me they have lower welfare than a high GDP per capita culture.
If you tell me that culture is important for economic outcomes, then you’re telling me that utility functions vary across cultures. But if utility functions vary across cultures, then cross-culture comparisons of economic outcomes don’t imply anything about welfare. So aren’t the regressions with culture as an explanatory variable self-defeating, even if they are econometrically sound?
I could well be over-thinking this, and I’d be happy to hear a good argument for what the culture/growth or culture/income regressions are supposed to be telling me.