Multinational firms, using their foreign affiliates as export platforms, are the largest players in international trade. The exporting behaviors of these multinationals differ systematically from those of local firms: Using the Chinese customs data, I find that the Chinese affiliates of foreign multinationals bias their exports towards the markets close to their headquarters. I incorporate this headquarters gravity into a general equilibrium model by, as in Head and Mayer (2019), allowing the export costs faced by multinational affiliates to depend on the proximity between headquarter and destination countries. To draw its aggregate implications, I calibrate my model to the Chinese customs data and perform counterfactual exercises, finding that (i) headquarters gravity accounts for about 20% of the Chinese exports in the early 2000s, and (ii) ignoring headquarters gravity would substantially bias our quantitative evaluation of trade shocks like the recent US-China trade war. I also consider the scenario in which the Chinese customs data on multinational sales is unavailable. I demonstrate the usefulness of my model in this scenario by constructing exact bounds on counterfactual results using only bilateral trade and multinational production (MP) data.
Read full abstract