Abstract

International trade statistics are usually reported from two complementary perspectives. Reporter data, provided by the declaring country and the partner data, reported by the corresponding trade partner. While, in principle, these figures should match, in practice, substantial differences arise. The main sources of divergence include valuation conventions (such as CIF vs FOB), incomplete or delayed reporting, lack of reporting by all countries, product misclassifications and sparing recording of cross-border transactions. These discrepancies create persistent asymmetries across international datasets such as FAOSTAT, UN Comtrade and ITC Trademap.

This paper statistically investigates discrepancies between reporter and partner data using bilateral trade data. The analysis focuses on cases where the list of trading partners reported on both sides of the data closely match and major trading partners are included on both sides of the data (that is by reports and their partners). The paper presents a statistical analysis of observed asymmetries. In particular, it focuses on whether reporter or partner data exhibit greater stability over time, using simple measures of variance and persistence; whether discrepancies display systematic distributional patterns across products, regions, or trade relationships; and whether thresholds can be identified to separate structural differences from reporting errors. Building on this, the study will explore the development of a statistical framework aimed at evaluating the relative reliability of the two perspectives. The objective is not to assess, in a systematic and evidence-based manner, whether one source tends to provide a closer approximation to the underlying trade flows than the other or if a combination of the two sides of data could be used. This will involve examining the directionality and consistency of discrepancies and considering the conditions under which one perspective may be regarded as more accurate.

By pursuing this agenda, the paper aim to make empirical as well as methodological contribution to the existing literature: empirically, by presenting an analysis of the extent and nature of reporter–partner discrepancies; and methodologically, by proposing statistical tools that can guide decisions about which source, or what combination of sources, provides the closest approximation to underlying trade realities. The results are expected to be of value for improving the accuracy and policy relevance of international trade statistics, with implications for agri-food trade and food security monitoring.