Trade Misinvoicing and Illicit Financial Outflows from Pakistan: Product Level PPML Gravity Estimates with a GVC Exposure Map

Abstract

Illicit financial flows (IFFs) are difficult to observe directly; this paper develops a transparent, trade‑data screening framework for Pakistan that treats misinvoicing signals as upper‑bound risk proxies. Using UN Comtrade HS6 bilateral flows for 1995– 2024, we estimate product‑level PPML gravity models to generate conditional‑mean ‘expected trade’ benchmarks that accommodate zeros and heteroskedasticity. Robustness checks use BACI‑reconciled bilateral flows. We harmonize valuation by identifying CIF–FOB wedges to place imports and exports on a comparable basis, then construct directional discrepancy measures consistent with outward leakage (import overinvoicing and export underinvoicing). Results show an episodic, shock‑driven discrepancy pattern with a pronounced mid‑2010s spike and higher variance in recent years, rather than a smooth trend. Risk is highly concentrated in a small set of commodity/agri‑food sectors (notably cereals and oilseeds) and in specific partner– product clusters. Mapping HS products to input–output sectors links hotspots to GVC exposure, informing targeted customs risk management and audit prioritization.