GRADE Simulations
Government Revenue and Development Estimations · Universities of St Andrews and Leicester
Underlying profit-shifting estimates from the Atlas of the Offshore World, a joint initiative of the International Tax Observatory (formerly the EU Tax Observatory) and Skatteforsk – Norwegian Centre for Tax Research

If profit shifting ended

Multinational companies can book profit in jurisdictions that tax it lightly rather than in the countries where the activity that produced it took place. The Atlas of the Offshore World estimates that US$5.4 trillion of profit was shifted in this way between 2015 and 2021, in current US dollars. The corporate tax revenue lost to it came to $1,280bn over the same seven years — about $183bn a year — in constant 2015 US dollars. That revenue is not incidental: it is the money that funds the clinics, schools, water systems and cash transfers on which children’s rights depend, and it is forgone in the countries least able to replace it.

We use revenue-losing countries for those that lose revenue to profit shifting and tax havens for those that gain it. The gain is concentrated: over 2015–2021 Switzerland took $160bn, the Netherlands $123bn, Ireland $57bn and Belgium $46bn — $385bn of the $522bn total, 74%, among four countries with a combined population of 43 million in 2021. All money figures on this page are cumulative over 2015–2021 in constant 2015 US dollars unless stated otherwise. Havens collect far less than is lost, because the shifted profit is taxed at a much lower rate where it lands than where it was earned.

Most of the loss is not transferred to another government — it is simply never collected, and what is collected goes to very few countries
Where the $1,280bn of tax revenue lost by 154 revenue-losing countries over 2015–2021 ends up. Constant 2015 US dollars.
TAX REVENUE LOST BY 154 REVENUE-LOSING COUNTRIES $1,280bn $522bn 41%, collected by 41 tax havens $758bn 59%, collected by no government OF THAT $522bn, FOUR COUNTRIES TAKE $385bn — 74% $160bn $123bn $57bn $46bn $137bn Switzerland 31% Netherlands 24% Ireland 11% Belgium 9% all other 37 havens 26%
Figure 1. This is not a transfer between governments. Profit is taxed far more lightly where it is booked than where it was earned, so $758bn of the loss becomes revenue for no government — it stays with the companies instead.

Why the same revenue does more in some countries than others

The developmental impact of revenue varies by country. Several mechanisms contribute to this and include:

  1. Additional revenue has a larger fiscal impact in low-income countries.
    Government revenue per capita varies significantly among countries. For example, if a low-income country with a population of 10 million and revenue of $100 per person receives an additional $200 million, revenue per person rises by $20 — a 20% increase. The same $200 million in a high-income country with a population of 10 million and revenue of $20,000 per person would increase revenue by just 0.1%.
  2. The development return on revenue diminishes as countries develop.
    Countries in the earlier stages of development often experience rapid increases in the coverage of development indicators with additional revenue because early improvements are less expensive to achieve. Countries at the plateau phase of development (most tax havens) will see minimal benefit from additional revenue on these indicators (Figure 2).
  3. Revenue improves governance. Additional revenue strengthens governance, which, in turn, amplifies the development impact of that revenue, as illustrated by the steeper dashed curve in Figure 2.
S-curve: Government Revenue per capita vs Coverage of Development Indicators A sigmoid curve showing that coverage of development indicators rises slowly at low revenue levels, accelerates rapidly in the middle range, then plateaus near 100%. Quality of governance steepens the gradient. Coverage of Development Indicators (water, sanitation, education, healthcare, social security) Government Revenue per capita 100% Low initial coverage A rapid increase Quality of governance increases the gradient Plateau phase Baseline trajectory Higher governance quality

Figure 2. The relationship between government revenue per capita and the coverage of development indicators follows an S curve. Early gains are large and relatively cheap; the rate of improvement then accelerates through a middle range, and coverage levels off as it approaches universal. Better governance steepens the curve for all income levels.

Aims of this visualisation

This visualisation models one counterfactual: profit shifting ends. Revenue-losing countries recover what they lost; tax havens forgo what they gained. Its purpose is to express that revenue in terms of people — deaths averted, children in school, people reaching basic water, sanitation and clean fuels — rather than in dollars, and to show where that development would accrue and where it would instead be forgone.

Both sides come from a single model run, and each is measured against that country’s own position over 2015–2021, so the two can be compared directly. Use the toggle to move between them.

Methods

Estimates of tax revenue lost to corporate profit shifting, and of the revenue gained by the jurisdictions the profit is shifted to, were taken from the Atlas of the Offshore World for 2015–2021 and entered into GRADE (version 3.24.2). Tax revenue lost is the value of profits shifted multiplied by the source country’s statutory corporate income tax rate; tax revenue won is its counterpart for the haven. Loss was available for 158 economies, of which 154 were entered: three reported zero in every year (San Marino, Palau and Vanuatu) and Taiwan has no World Bank conversion factor. Gain was available for 41 havens, all of which were entered.

Each country-year estimate, published in current US dollars, was converted to constant 2015 US dollars — the unit of government revenue in GRADE — using a country- and year-specific World Bank factor (GDP in constant 2015 US dollars, NY.GDP.MKTP.KD, divided by GDP in current US dollars, NY.GDP.MKTP.CD; World Development Indicators, 13 July 2026). Six havens have no World Bank national accounts and were converted with the US GDP deflator instead: for the British Virgin Islands, Bonaire/Sint Eustatius/Saba and Anguilla that is close, since they use the US dollar or a currency pegged to it, while Jersey, Guernsey and Gibraltar are sterling-linked and the substitution is looser. The mean across the years the Atlas reports was entered as the revenue change: years it does not report are excluded from the average, while reported zeros are kept. Nine of the 154 revenue-losing countries have fewer than seven reported years; all 41 havens have seven.

Direction. GRADE is modelled on observed government revenue, which already reflects profit shifting: a losing country’s revenue is net of its loss, a haven’s contains its gain. The counterfactual is therefore — revenue added back for losing countries, removed for havens.

GRADE accounts for the non-linear relationship between revenue and outcomes and for governance quality, measured by the six Worldwide Governance Indicators.

Findings

Outcomes are child-rights indicators across four domains: child survival and health, education, basic services (water, sanitation, electricity, clean fuels), and social protection. Cumulative figures are computed only for flow variables — deaths averted and additional child school years — which can be summed across years. For stock variables, such as the numbers with access to basic water, the value in the latest year (2021) is reported, so the same person is counted once.

Coverage. Outcome data are not available for every country. All 154 revenue-losing countries carry results, but only 27 of the 41 havens: the rest have no government revenue series in the base data, among them Bermuda, the Cayman Islands, Jersey and the British Virgin Islands. Social protection is the narrowest indicator of all, modelled for 59 of the losing countries and a handful of havens, because the ILO publishes no country-level value for it before 2015. Those 59 are weighted towards high-income countries — 30 of 41 high-income against 5 of 22 low-income — so the social protection figures should not be read as a fixed share of a global total.

forgone by tax havens   gained by revenue-losing countries
Both sides are drawn to one shared scale, so the lengths are directly comparable. Values are labelled because the haven bars are short by a wide margin, which is the point rather than a defect of the chart.
🔴 154 countries would gain
—
vs
🔵 27 havens would forgo
—
Outcome
group
Country
group
Timeframe
Income
Region
Country
Tip: click any country bar to see its full development profile.

What’s changed — revision of 19 September 2026

The figures on this page were revised in September 2026. They differ, in places substantially, from the version published earlier in 2026. Two things changed, each a correction rather than a change of preference.

  1. Conversion to constant 2015 US dollars.
    The earlier version deflated every country with a single US GDP deflator. Each country is now converted with its own World Bank factor for each year (GDP in constant 2015 US$ divided by GDP in current US$), which is the measure GRADE’s revenue variable is built on.
  2. Social protection now covers far fewer countries, and that is the correction.
    Social protection is the ILO’s SDG 1.3.1 series. The ILO publishes no country-level value before 2015, so only the countries with a genuine 2015 observation carry a result. Those countries are weighted towards high-income ones, so the social protection figures should not be read as a fixed share of a global total.

Six havens have no World Bank national accounts and so no country factor: Anguilla, Bonaire/Sint Eustatius/Saba, Guernsey, Gibraltar, Jersey and the British Virgin Islands. All six are retained, converted with the US deflator. For the British Virgin Islands, Bonaire/Sint Eustatius/Saba and Anguilla that is a close approximation, since they use the US dollar or a currency pegged to it; for the three sterling-linked territories it is a looser one. Together they are 2.1% of the cumulative haven gain.

Source: GRADE simulations (Universities of St Andrews and Leicester) applying published outcome elasticities to the additional government revenue that would be available if revenue lost to corporate profit shifting were recovered, or foregone by tax havens. Profit-shifting estimates from the Atlas of the Offshore World. Country-year panel: 154 revenue-losing countries and 41 tax havens, 2015–2021; modelled outcomes for 154 and 27 respectively. Both sides come from one GRADE run in which the revenue is restored to losing countries and removed from havens.