Apr 17, 2009

Econometrics vs. knowledge

The undercover economist has an interesting review of Economic Gangsters on reason online where he wonders if parking tickets explain why poor countries are poor, or, in other words, if precise econometric studies can help us answer big questions about development. As he writes, "because a randomized trial offers compelling evidence on what works but not why it works, the lessons from such evaluations tend to be hard to transport to a different context".

He goes on saying that "famous “tectonic” results, such as Levitt’s findings on abortion and crime or Acemoglu’s study of institutions and settler mortality, are fiercely contested by other researchers. That is not surprising: They are inherently open to challenge because of the scope of the question and the need for sophisticated statistical techniques and careful interpretation of raw data. Nobody is arguing about the statistical analysis performed in the parking ticket paper."

Wait a second. He thinks these papers are fiercely contested because of the shaky econometrics? I seriously doubt so. It's always an ideological issue. If you are a conservative economist, you are against abortion as it is a matter of "life" and hence you are against Levitt's thesis and try to destroy it in the Review of Economics and Statistics. If you are more liberal, you don't bother, you like the idea.

About the Acemoglu story, everybody knew that Europeans had settled in Canada, the US, Australia and New Zealand, and even South Africa, because of the hospitable climate before their paper came out. Everyone knows these countries are rich because they are a continuation of European institutions combined with indigenous population apartheid or genocide. To say that income differences between this group and Congo is due to the different colonial institutions is obviously the academic and politically correct hypothesis. But to attribute the economic success to the institutions and not to the transplantation of a rich society and mass killing of a primitive one is not convincing. This paper gives more answers.

To show that institutions are the key to growth we need a natural experiment that is quite opposite to what Acemoglu et. al. exploit. We need colonies that acquired good institutions without European settlement. But anyway, as Tim says, economists can provide precise answers to small questions, but only vague answers to the big questions. But to determine how Africa will get rich we don't need to answer big questions. It will.