Reference Point Dependence for Specification Bias from Quality Upgrading

This paper argues that whether estimates of the welfare cost of natural or artificial trade barriers that do not discriminate by quality are subject to positive or negative specification bias when using models which do not explicitly recognize quality variation depends on the reference point used in counterfactual equilibrium analysis. We use numerical general equilibrium techniques to generate counter examples to the widely held view that (in the competitive case) incorporating quality upgrading will tend to reduce the welfare costs of quality invariant trade barriers. To do this, we use a trade-distorted equilibrium as the reference point, rather than free trade.