Taste Parameters as Model Residuals: Assessing the 'Fit' of an Armington Trade Model

Abstract

We note that calibration parameters in a multi-country Armington trade model play a role similar to that of econometric residuals: they allow the model to fit the data exactly. We use this premise to evaluate the "fit" of a standard multi-country computable general-equilibrium model. We find that the model relies heavily on these parameters to explain the pattern of trade. In 33 of the 46 commodity groups we assess, modeled economic behavior explains less than 20% of the variation in bilateral trade. In a calibration-as-estimation experiment, we estimate the commodity-specific elasticities of substitution consistent with a well-fitting model and find that they are substantially higher than widely accepted estimates.