IntroductionThe authors attempt to empirically validate the interpretation of the quality of the Melitz (2003) model. For the purposes of the study, the authors examine the exports of champagne producers, comparing exporting companies to producer ratings from two wine guides. This method would be impossible for many products where middlemen handle exports rather than the producers themselves. They show that quality is correlated with exporting to more markets, prices and the quantity sold in each market. The most attractive markets are served by exporters who, on average, produce lower-rated champagne. Their quality sorting results follow the model's predictions, but under two conditions. First, they look at differences in quality rather than productivity. Second, idiosyncratic demand determines export performance. Compared to other papers examining industry-level data, Crozet et al. focus on the Champagne sector. Methodologically, this article makes several contributions to the existing literature on quality and trade. First, it identifies significant selection bias in the firm-level regression of export performance on the observed quality measure. They propose a method for estimating firm-level export regressions on capacity measures and use Monte Carlo simulations to show that it corrects for this. Second, they propose extensions to their models to check whether their results on quality are sufficiently robust compared to the proposed alternative mechanisms. In this model, quality can account for more than 27% of the firm-specific factors that determine export value. Industry choice Champagne is an excellent example of a differentiated product that is widely exported. Many characteristics of the industry also make it suitable for empirical studies. Champagne...... middle of paper ......r the probability of reverse causality. It is possible that Juhlin's ratings are based on a wine's popularity in markets outside of France. In this case, exports would determine quality rather than presumed causation. Another problem is that experts often seem proud to give very high marks to obscure producers. This would change the direction of causality. The authors attempt to address this concern by adopting a different scoring guide. The results are very similar to those obtained with the Juhlin guide. Although this result does not rule out endogeneity bias, it strengthens the underlying relationship between quality and export success. Extend the model to account for other sources of firm heterogeneity, include non-homothetic preferences for destination countries, and account for non-iceberg transportation costs. do not destabilize the model.977 words