The Offshore Stack

A typical multinational offshore structure uses between three and seven entities across multiple jurisdictions, each selected for a specific legal or tax attribute. The canonical chain runs as follows: a parent corporation incorporated in the United States (or another high-tax home jurisdiction); a holding company in the Cayman Islands or Bermuda, where there is no corporate income tax; an intermediate entity in the British Virgin Islands, where formation costs are minimal and disclosure requirements are negligible; an operating subsidiary in Ireland, where the corporate tax rate is 12.5 percent and where favorable intellectual property regimes further reduce the effective rate; a financing entity in Luxembourg or the Netherlands, where tax treaties eliminate or reduce withholding taxes on inter-company payments; and, where needed, additional subsidiaries in Singapore, Hong Kong, or Switzerland for regional treasury functions.

Each entity in the chain serves a purpose. The Cayman holding company owns the group's most valuable intangible assets — patents, trademarks, algorithms, brand rights — and licenses them to operating entities in high-revenue markets. The license payments flow from the operating entities (which earn revenue in the United States, Europe, and Asia) to the Cayman entity (which pays no tax). The Irish subsidiary functions as the group's non-U.S. operational headquarters, benefiting from Ireland's low statutory rate and its network of double taxation treaties with over 70 countries. The Dutch or Luxembourg entity routes financing and royalty payments, exploiting treaty provisions that eliminate withholding taxes on cross-border flows that would otherwise be taxable.

The result is that revenue generated from customers in high-tax markets — the United States (21 percent corporate rate), Germany (30 percent), France (25 percent), Japan (30 percent) — flows through the chain and accumulates in jurisdictions with effective rates approaching zero. The operating entities in high-tax markets report thin margins because their costs include large royalty and license payments to related entities in low-tax jurisdictions. The low-tax entities report enormous profits because they own the intellectual property generating the value, even though the IP was originally developed by employees in the high-tax markets.

The Double Irish arrangement was the single largest corporate tax avoidance structure in history. Developed in the late 1980s and used primarily by U.S. technology multinationals, it exploited a mismatch between Irish and U.S. tax residency rules. Under Irish law prior to 2015, a company incorporated in Ireland could be tax-resident elsewhere if its central management and control was exercised outside Ireland. Under U.S. law, a company incorporated in Ireland was tax-resident in Ireland. A company that was incorporated in Ireland but managed from Bermuda was, therefore, resident nowhere for corporate tax purposes — stateless income in the precise sense.

The structure required two Irish companies. The first — typically called the "head" company — was incorporated in Ireland but managed and controlled from Bermuda or the Cayman Islands. It owned the group's non-U.S. intellectual property rights. The second Irish company was incorporated and tax-resident in Ireland, functioning as the operational entity that conducted the group's non-U.S. business. Revenue flowed into the second Irish company, which then made large royalty payments to the first Irish company for the use of the IP. These payments were deductible against Irish income, reducing the second company's taxable profit in Ireland to near zero. The first company, being tax-resident in Bermuda under Irish rules, paid no Irish tax on the royalties received — and Bermuda has no corporate income tax.

The "Dutch Sandwich" variation added a Netherlands entity between the two Irish companies to avoid Irish withholding tax on the royalty payments. Royalties from the operating Irish company were paid first to the Dutch entity (no withholding tax under the Ireland-Netherlands treaty), then from the Dutch entity to the Bermuda-managed Irish company (no withholding tax under the Netherlands-Ireland treaty). Google used this structure to shift $75.4 billion in profits out of Ireland in 2019 alone, the last year it used the arrangement. Over the preceding decade, Google's overseas effective tax rate was reported at approximately 2.4 percent — against a weighted average statutory rate in its operating markets well above 20 percent.

Ireland closed the arrangement to new users in 2015 under pressure from the European Commission, but granted existing users — including Apple, Google, Facebook, and Pfizer — a transition period until January 2020. By the time it closed, the Double Irish had shielded an estimated $100 billion or more per year in U.S. multinational profits from taxation. The structures that replaced it — including Ireland's Knowledge Development Box and various patent-box regimes across Europe — accomplish similar objectives through different mechanisms.

Estimating the total cost of offshore profit shifting to public treasuries is methodologically difficult and politically contested. The most widely cited academic estimate comes from Gabriel Zucman, professor of economics at the Paris School of Economics and the University of California, Berkeley, and winner of the 2023 John Bates Clark Medal. Zucman's research, developed through a series of papers from 2014 onward and synthesized in the EU Tax Observatory's Global Tax Evasion Report 2024, estimates that approximately 37 percent of multinational profits booked outside firms' home countries are shifted to tax havens. The revenue cost to governments ranges from $200 billion to $600 billion annually, depending on the methodology, the definition of "tax haven," and the counterfactual tax rate assumed.

The Tax Justice Network's State of Tax Justice report, published annually since 2020, provides a complementary estimate. The 2021 edition found that cross-border corporate tax abuse costs governments approximately $312 billion per year, with OECD member countries and their dependent territories responsible for 78 percent of the losses. The concentration is notable: a small number of jurisdictions — the Cayman Islands, the British Virgin Islands, Bermuda, the Netherlands, Luxembourg, Ireland, Singapore, and Switzerland — account for the majority of global profit shifting. These jurisdictions have populations totaling less than 40 million people but host trillions of dollars in booked multinational profits.

Apple's structure illustrates the scale at the firm level. The European Commission ruled in 2016 that Ireland had granted Apple illegal state aid worth up to 13 billion euros by allowing the company to attribute almost all of its European profits to a "head office" that existed only on paper, had no employees, and no physical presence. Apple Sales International, the subsidiary at the center of the structure, generated more than $120 billion in profits over five years — close to 60 percent of Apple's worldwide earnings — while reporting an effective Irish tax rate that the Commission calculated at 0.005 percent in 2014. Apple relocated its offshore IP holding to Jersey, a British Crown Dependency with no corporate tax on foreign profits, when Ireland changed its residency rules in 2015. In September 2024, the European Court of Justice ruled that Apple must pay Ireland the full 13 billion euros in back taxes.

The OECD's Pillar Two initiative, which introduces a 15 percent global minimum tax for multinational enterprises with consolidated revenues above 750 million euros, is projected to generate approximately $150 billion in additional annual tax revenue when fully implemented. That the expected additional revenue from a 15 percent floor is $150 billion per year is itself a measure of the current scale of undertaxation: it represents the revenue that governments are currently forgoing because effective rates in major booking jurisdictions are below 15 percent. The gap between the minimum and the statutory rates in the markets where value is actually created — 21 to 30 percent — remains outside the reform's scope.

Reform efforts in international corporate taxation follow a consistent pattern: identification of abusive structures, multilateral negotiation, partial closure of the identified structure, and rapid adaptation by the same professional services firms that designed the original structure. The Double Irish was identified as abusive by the mid-2000s. It took until 2015 to close it to new entrants, and until 2020 to close it entirely. By that time, replacement structures — patent boxes, cost-sharing arrangements, and principal structures — were fully operational and serving the same function through different legal mechanisms.

The OECD's Base Erosion and Profit Shifting (BEPS) project, launched in 2013 with strong political support from the G20, produced 15 "action items" designed to address the most egregious forms of profit shifting. Implementation was voluntary for most measures, and the results have been mixed. Country-by-country reporting (Action 13) increased transparency but did not change tax outcomes. Transfer pricing documentation requirements (Actions 8-10) increased compliance costs without fundamentally altering the ability of firms with valuable IP to allocate profits to low-tax jurisdictions through arm's-length pricing of intangible assets — because the arm's-length standard itself permits the outcomes it was supposedly constraining.

The Pillar Two global minimum tax represents the most significant structural reform in decades. Approximately 140 jurisdictions have endorsed it, and implementation began in 2024. Its 15 percent minimum rate will capture some of the most extreme undertaxation — the sub-5-percent effective rates achieved through structures like the Double Irish. But 15 percent is itself a compromise. It is below the statutory rate of every major economy. It exempts the first 5 percent return on tangible assets and the first 5 percent return on payroll costs, creating new planning opportunities. And the United States, the home jurisdiction of the multinationals most responsible for global profit shifting, has not adopted domestic legislation to implement Pillar Two — creating uncertainty about the regime's long-term effectiveness for the firms that matter most.

The professional services industry that designs offshore structures — dominated by the Big Four accounting firms (Deloitte, PwC, EY, and KPMG) and a network of specialized law firms — generates estimated annual revenues exceeding $20 billion from international tax planning services. These firms simultaneously advise governments on tax policy design and advise their corporate clients on how to minimize tax within the policies they helped design. The structural conflict is not alleged; it is disclosed in the firms' own marketing materials, which advertise both "tax policy advisory" and "tax optimization" services. The reform cycle generates demand for both.

The Offshore Stack operates through a principle that is simple even when its implementation is complex: intellectual property and other intangible assets are the primary drivers of profit in the modern economy, and intangible assets can be located anywhere because they have no physical form. A factory must be in a specific place. An algorithm does not. A patent developed by engineers in California can be owned by an entity in Bermuda. The profits attributable to that patent — which may represent the majority of the enterprise's total profit — flow to the entity that owns it, not to the jurisdiction where it was created or where the customers who paid for it reside.

Transfer pricing rules nominally require that transactions between related entities occur at arm's-length prices — the price that would be charged between unrelated parties. In practice, there is no arm's-length market for unique intangible assets. There is no comparable transaction for the licensing of Google's search algorithm or Apple's iOS ecosystem. The arm's-length standard, when applied to assets for which no arm's-length market exists, becomes an exercise in modeling assumptions — and the assumptions can be calibrated to support a wide range of outcomes. The professional services firms that prepare transfer pricing documentation are the same firms that designed the structures being documented. The documentation consistently supports the conclusion that the allocation of profit to the low-tax IP-holding entity is arm's-length.

The Offshore Stack is the Corporate Shell's third layer and its largest in terms of aggregate fiscal impact. Where the Delaware Design establishes the governance container and the Loan Architecture converts individual wealth to non-taxable borrowing, the Offshore Stack converts corporate revenue into minimally-taxed profit. The mechanism is jurisdiction arbitrage: the deliberate selection and sequencing of legal domiciles to route economic activity through the combination of treaty provisions, IP regimes, and statutory rates that produces the lowest effective global tax burden. No individual step in the chain is illegal. No individual jurisdiction's laws are violated. The aggregate effect is a global corporate tax system that collects a fraction of the revenue its statutory rates imply.