What Rules in the ‘Deep’ Determinants of Comparative Development?

During the previous two decades or so empirical literature on comparative development of na- tions has turned to investigation of ’deep’ determinants of differences in income levels, such as institutions, trade, geography and human capital. In this paper I revisit this debate and make the following contributions. First, when run in one common framework, the previous results are generally not robust to the choice of measures of institutional quality as well as their respective instruments, and might be misspecified. Institutional measures of objective constraints should be strictly preferred. Second, a careful model selection across all the instruments for all the deep determinants reveals a clear pattern of instrumentation. Most notably this indicates that settler mortality proposed by Acemoglu et al. (2001) is not a dominant instrument for institu- tional quality for which legal or linguistical instruments should be preferred. Consequently I provide evidence that the theory of colonial origins is not institutional in its nature and rather supports human capital prevalence hypothesis. Third, human capital and geography come out as ’winners’ exerting separate direct effects on income levels with openness having indirect complementary effects either via institutions or natural resource exports.