Reproducibility package for What's In A Name? Implications Of Extensive Margin Measurement In International Trade

Recent years have seen a sharp increase in the availability of micro data at the firm and firm-to-firm level in international trade. Data platform providers often use proprietary algorithms that match reported firm names to assign identifiers to companies engaged in trade. We show that identifiers in one such platform suffer from substantial mismeasurement—for example, the same firm may be assigned different identifiers across transactions. We propose an algorithm to clean firm names and generate more accurate firm identifiers. Using these, we compare key exporter and importer indicators, as well as firm-to-firm trade indicators, with those based on the platform’s identifiers. The resulting biases are stark: using platform IDs shrinks the measured population of exporters and importers, inflates their average size, and overstates the concentration of trade among a few firms. It also artificially inflates firm entry into and exit from international markets. These distortions extend to firm-to-firm trade networks, which appear spuriously denser, less concentrated among top sellers or buyers, and far more volatile. Our findings caution against the growing reliance on readily available proprietary firm identifiers in studies of firms’ trade responses to global shocks, particularly through changes in their buyer–supplier networks.