The paper proposes a methodology for identifying the non-EU counterparts of Italian companies within the dataset provided by the Customs and Monopolies Agency. We identify the counterparts by leveraging Orbis Historical, a database containing financial statements for companies worldwide. The linkage between the two sources relies on the names and addresses of the counterparts and is achieved through cleaning procedures, name harmonization routines, and machine learning techniques, which proved useful for managing the complexities of the two large databases.
The method identifies foreign trade counterparts with a high level of accuracy, above 90 per cent for most of Italy's key trading partners. These counterparts account for a high proportion of Italian firms' foreign trade (85 per cent with the main trading partners). The identification of counterparts within Orbis Historical will enable analysis and research by incorporating their financial, economic, and ownership characteristics.
