Do islands trade more or less? A meta-analysis

This study investigates the island-trade literature in order to examine two things. One is to examine the underlying effect of islands on trade. Two is to investigate to see if publication bias influences the direction of reported effect sizes in the literature. The study uses the methodology of meta-regression analysis as its empirical workhorse and examined 52 studies with 1,433 estimated effect sizes, and over a period of 10 years (2010-2019). Contrary to the study’s hypothesis, results estimated using a mixed-effects method (MEM) indicates the underlying effect to be positive. From the same approach, the study found the presence of a moderate publication bias that is negative, and from all included studies. However, robustness check on this result using outliers and controlling for an important study (ID 17) shows these results to be inconsistent. Hence, we cannot on the basis of this study conclude that the underlying effect of island on trade is positive; and that a moderate publication bias that is negative exists in the literature.