Abstract

Firms play a pivotal role in international trade, shaping the comparative advantage of the countries. We propose a ‘granular’ multi-sector model of trade, which combines together fundamental
Ricardian comparative advantage across sectors with granular comparative advantage due to outstanding productivity draws of individual firms. We develop a SMM-based estimation procedure,
which takes full account of the general equilibrium of the model, and jointly estimate the fundamental and the granular forces using French micro-level data with information on firm domestic and
export sales across manufacturing industries. The estimated granular model captures the salient features of micro-level heterogeneity across firms and industries, without relying on variation in
model parameters across sectors. The estimated model implies that one third of trade flows is explained by granular forces, and that sectors with the extreme export shares are more likely to be of
‘granular’ origin than sectors with average export shares. Failure of a single large firm in a granular sector has dramatic effects on the relative export standing of the sector. We further show that
empirically measurable proxies of granularity have a substantial predictive ability for trade flows in the estimated model, even after controlling for fundamental comparative advantage of the sectors.
Lastly, extending the model to allow for rm-level productivity dynamics explains the majority of mean reversion in country’s comparative advantage over time.