IMF Staff Papers, Volume 56, No. 3

Studies of the impact of trade openness on growth are based either on crosscountry analysis-which lacks transparency-or case studies-which lack statistical rigor. This paper applies a transparent econometric method drawn from the treatment evaluation literature (matching estimators) to make the comparison between treated (that is, open) and control (that is, closed) countries explicit while remaining within a statistical framework. Matching estimators highlight that common cross-country evidence is based on rather far-fetched country comparisons, which stem from the lack of common support of treated and control countries in the covariate space. The paper therefore advocates paying more attention to appropriate sample restriction in crosscountry macro research.