Detecting profit shifting in administrative data: A South African perspective

This paper develops a data-driven methodology for detecting potential profit shifting by multinational enterprises in South Africa. Drawing on detailed administrative tax and customs data, we construct a series of red-flag indicators based on firm-level profitability, intra-group transactions, and trade with tax haven jurisdictions. A firm is flagged only if it persistently reports profits below its industry average while exhibiting at least one additional red flag related to related-party debt, service payments, or haven-linked trade. Although fewer than 1% of multinational enterprises are flagged, this small subset accounts for a disproportionately large share of potential tax revenue at risk. The approach offers a practical and replicable framework for tax administrations in developing countries seeking to prioritize audits and strengthen domestic resource mobilization using existing administrative data.