Exporter Heterogeneity and Price Discrimination: A Quantitative View

Abstract

We quantify a class of commonly-employed general equilibrium models of international trade and pricing-to-market that feature firm-level heterogeneity and consumers with nonhomothetic preferences. We demonstrate theoretically that the models lack the flexibility to match salient features of US firm-level data. Consequently, we outline a theoretical framework that can reconcile the documented price dispersion across firms and markets, while maintaining consistency with cross-sectional observations on firm productivity and sales. We calibrate the model's parameters to match bilateral trade flows across 71 countries as well as the productivity and sales advantages of US exporters over non-exporters. We find that the calibrated model accounts for the majority of the dispersion in prices of tradables across countries of different income levels, while maintaining a tight quantitative fit to firm-level data. Given its additional flexibility, the model quantitatively outperforms the existing alternatives and yields welfare gains for the US that are 14-54% higher, but at the cost of loss of tractability.

Quantitative Trade Models: Developments and Challenges

“…Similarly, Simonovska (2015) examines data from an online retailer and provides evidence of nonhomothetic consumer preferences, implying that a country's trade elasticity varies with its income level. Along a similar line, Jung et al (2015) show that to properly match salient features of the data, we have to escape from the notion of having a constant trade elasticity. Our hope is that these insights on nonconstant elasticities will be helpful for figuring out ways to estimate trade elasticities that differ both by industry and by bilateral-country pair.…”

International Trade with Indirect Additivity

“…31 Jung et al (2015) demonstrate that the distributions of firm sales and productivity depend uniquely on the Pareto productivity shape parameter in existing models that feature consumers with directly additive preferences, including quadratic preferences (as in Melitz and Ottaviano, 2008, but without the outside good) and those of Behrens et al (2014) and Simonovska (2015). Therefore, these models cannot jointly reconcile moments from the two distributions observed in US data.…”

“…35 Simonovska and Waugh (2014,a,b) demonstrate this fact for models that rely on homothetic preferences, while Jung et al (2015) analyze models that belong to a class of directly additive preferences.…”

Quantitative Trade Models: Developments and Challenges

“…Similarly, Simonovska (2015) examines data from an online retailer and provides evidence of nonhomothetic consumer preferences, implying that a country's trade elasticity varies with its income level. Along a similar line, Jung et al (2015) show that to properly match salient features of the data, we have to escape from the notion of having a constant trade elasticity. Our hope is that these insights on nonconstant elasticities will be helpful for figuring out ways to estimate trade elasticities that differ both by industry and by bilateral-country pair.…”