Trade frictions and welfare in the gravity model: how much of the iceberg melts?

A key element missing from the structural gravity literature is an examination of the implied general equilibrium. By design the gravity equation is adept at predicting bilateral trade flows. To make inferences beyond trade flows, however, the theoretic models should be consistent with other observables. Structural econometric estimates from Anderson and van Wincoop (2003) allow us to evaluate their proposed general equilibrium along several dimensions. We Find that their gravity model predicts too large a difference between consumer and producer prices; excessive variation in the geographic distribution of consumer price indices; and an exceptionally large portion of output devoted to overcoming trade frictions. Under plausible parameterizations of the model at least 50% of output 'melts' in transit.