Gains from Trade under Quality Uncertainty
We add quality uncertainty to a two-country trade model with CES preference and mo- nopolistic competition. There are two kinds of firms - low quality and high quality. Quality is perfectly observable in the domestic market but not in the foreign market. Exporters use price to signal their quality. It is now well-established that in such a model with full information, the welfare gains from trade (GFT) can be captured by a sufficient statistic that depends on domestic trade share and the elasticity of substitution. In contrast, in a model with incomplete information, we show that within the class of separating equilib- ria, the sufficient statistic always under-estimates GFT, while within the class of pooling equilibria, the sufficient statistic could over-estimate GFT. Nevertheless, GFT are always positive. For an equilibrium refinement, we analyze the determinants of GFT. We show that the actual GFT under asymmetric information could be almost 2.5 times higher than that measured using the sufficient statistic approach.
Gains from Trade under Quality Uncertainty
We add quality uncertainty to a two-country trade model with CES preference and mo- nopolistic competition. There are two kinds of firms - low quality and high quality. Quality is perfectly observable in the domestic market but not in the foreign market. Exporters use price to signal their quality. It is now well-established that in such a model with full information, the welfare gains from trade (GFT) can be captured by a sufficient statistic that depends on domestic trade share and the elasticity of substitution. In contrast, in a model with incomplete information, we show that within the class of separating equilib- ria, the sufficient statistic always under-estimates GFT, while within the class of pooling equilibria, the sufficient statistic could over-estimate GFT. Nevertheless, GFT are always positive. For an equilibrium refinement, we analyze the determinants of GFT. We show that the actual GFT under asymmetric information could be almost 2.5 times higher than that measured using the sufficient statistic approach.