Identifying Trade Mis-Invoicing Through Customs Data Analysis

Abstract

This paper outlines a methodology for identifying trade transactions where undervaluation or over-valuation is highly suspected. As a first step, the methodology identifies trade transactions with an abnormal unit price. Secondly, it identifies trade transactions with values different from those noted in the records of trading partners. Finally, it presents the trade transactions that were commonly selected from the previous steps. The logic underlying the methodology is that if a trade transaction has an abnormal unit price as well as irreconcilable differences in the trade value ascribed by the trading partner, it would be reasonable to suspect under-valuation or over-valuation in the transaction. In a simulation using actual customs data, this methodology proved effective in detecting fraudulent imports. Of the imports suspected of under-valuation using this methodology, 18 per cent had been penalised and obliged to pay fines, more taxes or duties following customs interventions. This figure is much higher than the share of illicit imports in the test data (4.5%) and the targeting accuracy of the physical inspection of the country (12%). When this methodology was verified using only the imports that had been selected for physical inspection, the targeting accuracy increased to 36 per cent. The result suggests that this methodology could contribute to enhancing the targeting accuracy of existing Customs risk management tools.