Reconciling Armington and Melitz Trade Responses: A Parameterisation Study of U.S. Motor Vehicle Tariffs

This paper examines the consistency between Armington and Melitz trade specifications in computable general equilibrium (CGE) models, focusing on the parameterisation of Melitz elasticities for the motor vehicle sector. Using the framework developed by Dixon, Jerie and Rimmer (2018), we first restrict the Melitz parameter space by imposing bounds derived from theoretical constraints and potentially observable industry characteristics. We then assess whether any admissible Melitz parameterisation can reproduce the import response generated by a standard Armington specification in GTAP-FIN for a 25 per cent U.S. tariff on motor vehicle imports. We find that no such parameter combination exists. This incompatibility raises a parameterisation dilemma: when Melitz parameters are constrained to generate empirically plausible industry characteristics, the implied import responses substantially exceed those produced under conventional Armington elasticities. We therefore reverse the calibration exercise and identify the range of Armington elasticities required to match the import responses generated by the restricted Melitz model. The implied Armington elasticities are considerably larger than values identified in the literature. One interpretation is that conventional Armington parameterisations understate trade responsiveness in sectors characterised by firm-level heterogeneity. Another is that alternative representations of heterogeneity, such as small group monopolistic competition, may warrant further investigation.