Abstract
We re-examine Lynn and Vanhanen's argument that gross domestic product (GDP) depends upon IQ. We argue that their analysis suffers from three types of biases, each of which would tend to erroneously favor their hypothesis. Despite this stacked deck, we find that their results are rather fragile. Rather, education has a stronger impact on GDP than does IQ, whose effect we find to be insignificant. In other words, it is a country's actual human capital, rather than its potential human capital, which determines its GDP. In short, we are unable to replicate their results.
Keywords:
Notes
1 Since L&V use interpolated data, their observations are not independent of each other, so standard inference becomes misleading. An example in a much simpler context is the following. If one were to flip a fair coin 10 times, the probability that it lands on heads ten times is 0.5010. But if one were to flip a coin once, and record its value 10 times, we would have a dataset of 10 observations, but we cannot make the same probabilistic claims as before, because the observations are not independent of each other.
2 For a deeper understanding of the political and institutional effects on growth read: Barro (Citation1991, Citation1997) on the size of government and growth; Mauro (Citation1995), Knack and Keefer (Citation1995) and Knack and Zak (Citation2001) on corruption and investment; De Hann and Sturm (1999), Gwartney et al. (2005) and Lim and Decker (Citation2007) on economic freedom and growth.
3 Given that we estimate regressions in cross-sections, the data are all averaged over time, so that linearly interpolating the missing economic freedom data, from when it was measured in 5-year intervals, would make no material difference to the averaged value and the resulting regression estimates.