In theory, the exports of country A to country B should mirror the imports of country B from country A. In practice this is however rarely the case, due to a variety of factors including for example differences in valuation (CIF for imports and FOB for exports), in partner country classification, in customs regimes, and in compilation and dissemination methodologies. The asymmetries between the export and import values for the same trade flow have long been recognised as an important factor that limits the analytical and policy use of international merchandise trade statistics.
To tackle this issue, the OECD, through its Working Party on International Trade in Goods and Services Statistics, bringing together over 40 countries, has developed an approach to reconcile international merchandise trade statistics. The approach consists of four steps. First, detailed statistics reported in different versions of the Harmonized System (HS 1988, HS 1996, HS 2002, HS 2007, HS 2012 and HS 2017) are collected and organised, and imports are converted to FOB prices to match the valuation of exports. Secondly, data are adjusted for several specific large problems known to drive asymmetries. Presently these include “modular” adjustments for unallocated and confidential trade; for exports by Hong Kong, China; for Swiss non-monetary gold; and for clear-cut cases of product misclassifications. The list of modules is expected to grow over time. In the third step, adjusted data are balanced using a “Symmetry Index” that weights exports and imports. As the final step, the data are also converted to Classification of Products by Activity (CPA) products to better align with National Accounts statistics, such as in national Supply-Use tables.
It is important to consider that the balanced data, while typically in between reported exports and imports, may sometimes be higher or lower than both, due to the adjustments that are made to the data. For example, when one country reports significant confidential trade flows (which are first distributed across products/partners concerned), the final balanced values may be higher than what reported by each partner. Aggregation from HS6 digit (the working level of the database) to CPA 2-digit also plays a role, although minimally so since large differences, which typically point at product misclassifications, have been treated manually.
This dataset is regularly updated and continuously improved. Further work to reduce asymmetries in officially reported data, including through bilateral and multilateral meetings, is under way in collaboration with national statistical offices and other international organisations. At the moment, the manual (transparent and replicable) adjustments to underlying data reduce global asymmetries by 20% (prior to mathematically balancing bilateral flows) in the current database, which covers 160 countries and the years 2007 to 2018. The dataset is designed to be used both for stand-alone analyses, providing a granular view on global value chains, and as input to the construction of international supply-use and input-output tables such as those that underpin The OECD-WTO Trade in Value Added (TiVA) database and various regional TiVA initiatives such as the European FIGARO TiVA initiative, North American TiVA and APEC-TiVA.