Trade Flow Direction

Import

Goods entering the reporting country’s statistical territory. Recorded by the receiving administration from the import declaration.

Export

Goods leaving the reporting country’s statistical territory. Recorded by the despatching administration from the export declaration.

Re-export

Goods previously imported that leave again without having been substantially transformed. The reporting country handled them but did not produce them.

Re-import

Goods previously exported that return in essentially the state they left. Common where a consignment is rejected at destination or returned unsold.

The re-export category is the one that most affects agricultural analysis, because it breaks the intuitive link between a country’s exports and its production. A country with a major transhipment port or a large processing and redistribution trade can export substantial quantities of a commodity it does not grow. Treating its export figure as a measure of what it produced is a category error that the data itself does not prevent.

The most systematic reason an export figure and the corresponding import figure disagree is that they are conventionally valued on different bases. An export is customarily recorded at a value that stops at the exporting country’s frontier; an import is customarily recorded at a value that includes carriage and insurance to the importing country’s frontier. The difference between them is, broadly, the cost of getting the goods there.

This is a convention, not an error, and it means an import figure is expected to exceed the matching export figure by roughly the freight and insurance in between. For bulk agricultural commodities moving long distances, that margin is not small. Comparing an export value directly with an import value, and treating the gap as evidence of misreporting, mistakes the convention for a discrepancy.

Timing adds a second layer. A cargo shipped near the end of one period may be recorded as an export in that period and as an import in the next, so the two series can be out of phase even when both are accurate. The longer the voyage, the more pronounced the effect — which again bears hardest on bulk commodity trades.

Partner attribution is a third. An importing country typically attributes the flow to the country of origin, while an exporting country typically attributes it to the country of last known destination. Where goods move via a hub, or where the final destination changes after despatch, the two attributions name different partners for the same cargo.

Countries compile trade statistics against a defined statistical territory, and there is more than one convention for where its boundary falls. Under the general trade system the boundary is essentially the economic territory, so goods entering customs warehouses and free zones are counted on arrival. Under the special trade system the boundary is drawn more tightly around the free-circulation area, so goods held in those facilities may not be counted until they are released.

The choice affects what a country appears to import and when. It matters especially for economies with large free-zone or entrepôt activity, and for commodities that are routinely warehoused before onward sale. Two countries using different systems are not counting quite the same universe of goods, which is one more reason their figures for a shared flow can diverge legitimately.

The practical questions to settle before interpreting an agricultural trade figure are the same each time, and they are all questions about direction and basis rather than about the commodity.

• Which country reported it, and is the figure its own record or its partner’s?
• Is the direction an import, an export, a re-export, or a re-import?
• On what valuation basis is it recorded, and is it being compared with a figure on the same basis?
• What does the partner attribution mean in this direction — origin, consignment, or destination?
• Which trade system does the reporter use, and does free-zone or warehouse activity matter for this commodity?
• Which HS edition were the codes reported under, and does the period cross a revision boundary?

The last question links back to nomenclature versioning, and the combination is where most misreadings of long-run agricultural trade data originate: a series that changes direction convention, partner attribution, or code coverage part-way through will show a break that has nothing to do with trade.