What is Tax Treaty Shopping?

How it works

The basic mechanic: country A withholds 30% on dividends paid to country B. Country A withholds 5% on dividends paid to country C. Country B has no useful treaty with country A; country C does. An investor in country B inserts a holding entity in country C to receive the country A dividends, then pays them back to country B as a dividend (or interest, royalty, etc.) — saving 25 percentage points of WHT.

Country C's treaty was negotiated to relieve double tax for its own residents, not as a free pass for country B residents. Treaty shopping abuses this by routing income through a country with no commercial connection to the underlying business, purely to access its treaty network.

Historical structures

Prior to BEPS, common treaty-shopping structures included:

• Dutch BV / Cooperative — Netherlands has an extensive treaty network; the Dutch coop historically allowed dividends out without Dutch WHT. Combined with Netherlands-source treaty access, it was the conduit of choice for European M&A and US inbound investment.
• Cyprus holding — pre-2018, Cyprus's broad treaty network plus near-zero outbound WHT made it the preferred Russian / Middle Eastern conduit.
• Mauritius — India — Mauritius-India treaty zero capital gains made Mauritius the gateway for Indian portfolio investments. Renegotiated in 2016.
• US-Netherlands-Bermuda triangle — Apple's "double Irish with a Dutch sandwich" before its closure.

Post-BEPS landscape

The OECD/G20 BEPS Action 6 (2015) committed countries to introducing treaty anti-abuse provisions. The MLI (signed 2017) implemented these across 100+ countries' treaties simultaneously. Modern treaties now include:

• Limitation on Benefits (LOB) — objective tests for treaty residence (public company, ownership, active trade or business, derivative benefits).
• Principal Purpose Test (PPT) — subjective test denying benefits where one of the principal purposes was to obtain them.
• Beneficial owner requirements — treaty benefits available only to the beneficial owner of the income, not a conduit.
• Domestic GAAR — backstop in case the treaty mechanisms fail.

Combined effect: pure conduit holdings without commercial substance are routinely denied treaty benefits. Tax authorities issue substance-of-management challenges, deny WHT reductions, and often impose penalties on top.

What still works

Treaty access through a holding country remains legitimate when the structure has genuine commercial substance:

• Real headquarters / regional management — central decision-making, strategic functions, executives based in the jurisdiction.
• Asset management or operating substance — staff, premises, operations beyond just holding shares.
• Active trade or business connection — treaty income connected to the holding's actual business activities.
• Public company status — listed groups generally pass LOB without difficulty.
• Equivalent beneficiary qualification (derivative benefits) — beneficial owners are treaty residents who would have qualified directly.

The historical "letterbox + nominee director" structures don't survive review. Modern substance is full-time staff, premises with rent receipts, board meetings actually held in the jurisdiction, real decision-making documented contemporaneously.

Examples

• Russian investor uses Cyprus holding for UK property dividends, post-2017 MLI. Cyprus-UK treaty 0% on dividends. Cyprus holding = SPV with no employees, no premises, founded month before the deal. UK HMRC applies PPT (UK-Cyprus treaty as modified by MLI): one of the principal purposes was treaty benefit; benefit denied → UK 20% WHT applies. Russian investor sees no Cyprus benefit and faces UK denial.
• French operating multinational uses a Luxembourg holding for European subsidiaries. Luxembourg holding has a 5-person team, manages investments, makes acquisition decisions in Luxembourg, has been operating for 8 years. French sub's dividends to Luxembourg: France-Luxembourg treaty 5% applies. PPT challenge: substance defends the position; benefit granted.

Common mistakes

• Believing letterbox structures still work. They don't. PPT, LOB, beneficial-owner challenges, and domestic GAARs all neutralise paper-only conduits.
• Underestimating substance requirements. "Substance" today means full-time qualified staff, real premises with operating expenses, decision-making documented in board minutes held locally. Token directors and a virtual office fail every modern test.
• Forgetting Pillar Two. For €750M+ revenue groups, Pillar Two QDMTT and IIR can recapture top-up tax even when the source-country WHT is reduced. Treaty shopping returns less benefit than it used to even when it works.
• Ignoring source-country domestic GAAR. Even successful treaty positioning at the source country can fail under the source country's domestic GAAR (France abus de droit, UK GAAR, India GAAR).