Sunday, May 17, 2009

History versus geography

I spoke previously about the debate between those that argue that institutions matter, versus those that believe geography is the most important determinant of current income.

As the evidence seems to point squarely to the geography camp right now, the response from the institutions group has been to ... completely ignore that there is a debate. The core of the institutions camp, Daron Acemoglu, Davide Cantoni, Simon Johnson, James A. Robinson, have a new paper on the importance of Napoleon's spreading institutional changes when invading Europe. While an interesting idea, I am sad to see they don't have any controls for the geography of the parts of Prussia invaded.

Nathan Nunn also has a nice summary of what the institutions argument is about. Geography though comes in only at the end, and looks rather weak when he claims it has been controled for in many studies.

The evidence is not weak: geography, especially malaria ecology, trumps when correctly controled for.

Sachs is correct though when he argues that the variable of interest, GDP today, is incomplete. Why is GDP growth never looked at?

These are the things going through my mind for the last few days as I have been finanlizing my dissertation, which partially looks at this issue. Pouring through the data, I can't find any evidence for the claims of the importance of colonial origins for current GDP growth rates, or even any out of sample evidence of current GDP.

I don't want to write a paper about this. I really just want the work using colonial origins as an instrument for institutions to end. The evidence is clear that the data is flawed, and the results do not hold up to different sample specifications.

No comments: