The contribution of institutions, trade, and geography to the development process: How robust is the empirical evidence to variations in the sample?

This paper uses an influence spectrum to identify influential subsets in a stylized cross-country data set and finds that institutions, geography, and trade (policy), all appear to play a significant role in the development process for a relatively large sub-sample of countries. For example, equatorial distance, a proxy for geography, becomes positive and significant (originally negative and insignificant) after removing only eight countries or observations from the original sample of sixty-three, while controlling for institutions and trade. In fact, for this set of fifty-five countries all three variables have the correct sign and are statistically significant. As another example, the trade variable becomes positive and significant (originally insignificant) after removing only two countries from the original sample.