Import sourcing of Chinese cities: Order versus randomness

Capitalizing on the geographic detail of Chinese customs data, we show that buyer heterogeneity plays a major role in import sourcing. Hierarchy compliance, a core prediction of supply-focused models, is tested by measuring the frequency with which cities import a narrowly defined good from the country observed to be the preferred source in the province. Hierarchy violation is widespread: 92% of province goods have at least one non-compliant city. We show that introducing granular importers into a standard heterogeneous firm model leads to a prediction of 73% compliance, close to the observed average of 66%. Extending the model to allow buyers from a city to share an orientation towards specific source countries, we calibrate a heterogeneity parameter to match the average observed compliance rate. The results imply that the supply side explains on average 44% of the variance in city-level sourcing probabilities, leaving the majority of variation due to heterogeneity in buyers across cities.