The problems with mis-specifying proxies (part 2)

...continued.

The famous paper in proposing Resources Curses as being more or less a fact was Sachs & Warner (1995). In their paper, they regress, against 1970-1990 GDP growth, a bunch of variables including: primary produce exports to GNP, land per capita, number of coups, revolutions, assasinations etc., school enrolment rates, and the like.

This paper is super-influential: Google Scholar says it is cited by 1416 other papers. Unfortunately, though, the result it concludes---that natural resource `abundance' is correlated with low economic growth---is ill-gotten. It all comes down to the choice of proxy: in their case, using natural resource exports to GNP as a proxy for natural resource abundance. This was picked up by Stijns (2005), who replicated their regressions but using per-capita coal, gas, and oil production as the proxy for the same explanatory variable---not a perfect proxy either.

This next bit should come as a real lesson to economists and students everywhere about the choice of choosing a proxy variable. The picture below is lifted directly from Sachs Warner (1997)---an updated version of their paper. It shows Zambia as being the most `resource abundant' country in the sample. According to natural resource estimates for 2000 by the World Bank (2005), Zambia had a proven stock of natural resources of $17.5B USD. Venezuela, according to the same study, at the same time had an estimate of proven stock of natural resources of $658B USD---some 38 times the `abundance'.



An incredible problem with the choice of a proxy like this is that it not only that it messes up the magnitudes, but also the ordering. It allows researchers to say very little about what they are trying to explain.

As far as the Sachs Warner paper goes, although it shows fairly conclusively there is a negative relationship between natural resource exports and economic growth (though this is likely an endogeniety error; name a rich country whose exports are comprised almost exclusively of resources), it fails to say much about "Natural Resource Abundance and Economic Growth"---the title.