Estimating exporter's quality: Do importers agree on rankings?

A large and growing literature focuses on product quality differentiation and its positive and normative importance for international trade. In this literature, there are a number of approaches that attempt to extract a quality ‘signal’ from noisy price or price–quantity data (i.e., unit value and demand residual approach). I propose a simple test for the robustness of these approaches to quality inference: when purchasing goods from a common set of exporters, do two importers agree on which goods are high quality? Results from Spearman's rank correlation tests show that as we use import demand residuals to measure quality, quality estimates become progressively less consistent and worsen the agreement among importers. Such inconsistency cannot be explained by trade cost effects on exporters' quality suggested by Melitz (2003) and Alchian–Allen hypothesis (Hummels and Skiba, 2004). It is also uncorrelated to differences in the compositions of imports. However, demand residual approach produces strong quality ‘signal’ relative to the noise only in highly differentiated products. Further analyses confirm the role of the differences in consumers' preferences and tastes across importers in the distortion of quality signals.