Quality sorting, Alchian–Allen effect, and distance

When identifying trade costs, the assumption of iceberg-type costs causes biases when specific (per-unit) costs exist. Although some studies measure the magnitude of specific costs, they suffer from self-selection bias. Furthermore, studies that account for selection and quality heterogeneity do not incorporate specific costs. Ignoring one of these three elements, that is, selection bias, specific trade costs, or quality heterogeneity, leads to a biased trade cost estimation. By incorporating all these elements, I identify trade costs and the condition under which quality sorting occurs. An empirical analysis shows that the magnitude of specific costs is large, with distance elasticity being three times larger than that for ad valorem costs. The parameter estimates are inconsistent with a model of quality sorting without specific costs, but consistent with one that includes specific costs, which confirms the significance of specific costs.