Report Number



We examine why the gravity equation works and the implications for its use. There are three contributions. First, we demonstrate that the gravity equation as a statistical relationship (and not as a specific prediction about bilateral trade) can be generated from a model with incomplete specialization. Second, we analyze two facts or “puzzles” that are broadly inconsistent with the complete specialization models typically used to derive the gravity equation. In both cases, the data are consistent with the alternative hypothesis of incomplete specialization. “Home bias”, for example, is only puzzling when viewed through the lens of the complete specialization model. Third, we demonstrate that the explanation for why the gravity equation works has considerable relevance for how the gravity equation is interpreted and used, and how we view bilateral trade.

Date of this Version