e600 Billion and Counting: Why High-Tax Countries Let Tax Havens Flourish

e600 Billion and Counting: Why High-Tax Countries Let Tax Havens Flourish Thomas Tørsløv (U. of Copenhagen) Ludvig Wier (U. of Copenhagen) Gabriel Zucman (UC Berkeley) November 2017

Introduction How big is the artificial shifting of profits by multinational companies to tax havens, and who benefits/loses from it? An important question for: Study of the redistributive effects of globalization Measurement of global economic activity Tax policy and tax enforcement

Our contribution: we analyze new data capturing profit-shifting to tax havens Systematic analysis of balance of payments & national accounts data published by tax havens & counterparties EU tax havens report detailed data to Eurostat enable to estimate amount of profits artificially shifted Recent improvement in service trade coverage enable to estimate which countries lose revenue Comprehensive estimate of size of global profit-shifting and revenue implications for governments worldwide

Our results: artificial profit-shifting redistributes tax revenue massively 45% of multinationals profits are artificially shifted to tax havens more than e600bn in 2015 Global corporate tax revenue loss around e200bn per year ( 12% of global corporate tax revenue) Under most sensible apportionment rule, European Union is the main loser (loses 20% of its revenue) Main winners: Ireland, Netherlands, Luxembourg (impose low rates of 2 3%, but on huge artificial base)

The E.U. loses about 20% of its corporate tax revenue in tax havens 35% Lost corporate tax revenue due to artificial profit-shifting (% of corporate tax revenue collected) 30% 25% 20% 15% 10% 5% 0% Germany France Hungary Italy EU22 United Kingdom Profits shifted to non-eu tax havens Profits shifted to EU tax havens Spain Sweden Austria Finland Estonia Denmark Portugal Poland Latvia Croatia Greece Slovenia Lithuania Czech Republic Romania Slovakia Note: This figure shows the amount of tax revenue lost because of the artificial shifting of multinationals' profits to tax havens, as a share of total corporate tax revenue collected in 2015. Bulgaria

The failure of enforcement Why high-tax countries fail to stop profit-shifting: Tax authorities have incentives to go after transfer mispricing involving other high-tax countries This crowds out enforcement on tax havens We analyze new data showing that in practice, almost all enforcement is against other high-tax countries In effect, high-tax countries are stealing from each other while letting tax havens flourish

Implications for policy BEPS reinforces perverse incentives of current system: Makes it easier to go after profits shifted to other high-tax countries...... Further crowding out enforcement on 0-tax countries where bulk of shifting tax place There is a simple fix to this problem: Sales apportionment of global profits Can be done unilaterally

Methodology to estimate the size and cost of profit-shifting

Main challenge in the literature: little data on what happens in tax havens Widely used source to study profit-shifting: financial accounts micro-data (e.g., Orbis) and customs data Face two key challenges: Orbis: misses most of the subsidiaries in tax havens Customs: miss service trade (e.g., intangibles) Big disconnect between public debate (which focuses on 0-tax countries and intangibles) and economic research

Example: Google Alphabet Google transferred its intangibles to hybrid Irish Bermuda subsidiary in 2003 In 2015, made $15.5bn in profits in 0-tax Bermuda invisible in Orbis Which country loses tax revenue: US? EU? Impossible to tell with available micro data Which country wins: Bermuda? Ireland? None? Why do high-tax countries fail to tax these earnings?

Most of Google s profits are invisible in financial accounts data Bn. Google's profits in Orbis 25 20 True global profits Sum of observable profits 15 10 5 0 2013 2014 2015 2016 Note: This graph shows Google's global consolidated profits, and the sum of the profits made by Google's subsidiaries, as recorded in Orbis. The difference is due to the fact that the subsidiaries where Google makes the bulk of its profits are not visible in Orbis.

Most of Apple s profits are invisible in financial accounts data Bn. Apple's profits in Orbis 70 60 50 True global profits Sum of observable profits 40 30 20 10 0 2013 2014 2015 2016 Note: This graph shows Apple's global consolidated profits, and the sum of the profits made by Apple's subsidiaries, as recorded in Orbis. The difference is due to the fact that the subsidiaries where Apple makes the bulk of its profits are not visible in Orbis.

None of Facebook s profits are visible in financial accounts data Bn. Facebook's profits in Orbis 12 10 8 True global profits Sum of observable profits 6 4 2 0 2013 2014 2015 2016 Note: This graph shows Facebook's global consolidated profits, and the sum of the profits made by Facebook's subsidiaries, as recorded in Orbis. The difference is due to the fact that the subsidiaries where Facebook makes the bulk of its profits are not visible in Orbis.

Most of Nike s profits are invisible in financial accounts data Bn. Nike's profits in Orbis 6 5 True global profits Sum of observable profits 4 3 2 1 0 2013 2014 2015 2016 Note: This graph shows Nike's global consolidated profits, and the sum of the profits made by Nike's subsidiaries, as recorded in Orbis. The difference is due to the fact that the subsidiaries where Nike makes the bulk of its profits are not visible in Orbis.

How we track artificial profit-shifting We focus on macro data (more comprehensive than financial accounts micro-data) Consider two key macro statistics in each country: π = taxable corporate profits / compensation of employees φ = taxable corporate profits accruing to foreigners / national income High π: abnormally high profits. High φ: profits are made in foreign-owned subsidiaries.

Abnormal profitability π π is related to the capital share in the corporate sector α If no net interest paid by corporations, π = α/(1 α) With identical technology and α = 25%, all countries should have π = 33% If net interest paid = p% of operating surplus, π = α/(1 α) (1 p) If π >> 33%: inward profit-shifting (either through real transactions or interest payments)

Corporations in tax havens are abnormally profitable 300% 250% 200% 150% 100% 50% 0% Luxembourg Taxable corporate profits (% of compensation of employees) Average among non-havens: 34% Ireland Puerto Rico Malta Netherlands Denmark Belgium Germany Italy U.K. Spain Sweden Finland United States Switzerland Note: This figure shows the ratio of corporate profits (net of interest paid and depreciation) to compensation of employees, as recorded in the national accounts, in 2015. France

Corporate profits accruing to foreigners φ φ is related to current account balance ca, trade surplus t, and net interest received from abroad i (all in % of NI) ca = (t + i) φ Inward profit-shifting (Ireland, Luxembourg,...): (t + i) > 0 and φ > 0 Outward profit-shifting (U.S., France,...): (t + i) < 0 and φ < 0 These identities summarize how profit-shifting affects balance of payments and national accounts statistics

Tax havens run huge trade surplus, all paid back to foreign parents United States United Kingdom France Spain Italy EU22 Belgium Germany Netherlands Malta Puerto Rico Ireland Luxembourg Current account balance (% of gross national income) Net trade surplus Net foreign interest received Net foreign corporate profits -200% -150% -100% -50% 0% 50% 100% 150% 200% Note: This figure shows the current account balance of a selection of countries, as a share of their Gross National Income in 2015. EU22 is the Euoropean Union minus the 6 EU tax havens (Belgium, Cyprus, Ireland, Luxembourg, Malta and Netherlands).

How we measure artificial profit-shifting We compute artificial profits by setting π i = π in all tax havens i Assumption: all profitability in tax havens above world average π reflects inward profit-shifting Potential limitation: high π i could be due to other factors (technology, bargaining, etc.) But testable: correlation between π i and foreign-ownership φ i

Where profits are abnormally high, they are all within MNEs artificial Taxable profits accruing to foreigenrs / GNI Profits accruing to foreigners vs. abnormal profitability 200% Luxembourg 150% 100% Ireland 50% Puerto Rico Malta Netherlands 0% 0% 100% 200% 300% 400% Taxable corporate profits / Compensation of employees

How we allocate the artificially-shifted profits across countries Benchmark allocation: based on which countries import from (and pay interest to) tax havens Focus on high-risk service imports (IP, financial services, etc.), particularly conducive of shifting Mimics a sales-apportioned corporate tax base (= how California, New York, etc., tax profits) Alternative allocation: based on residence of owner Mimics a residence-based corporate tax base (= how U.S. has taxed profits so far)

Profit-Shifting to European Tax Havens

Data for E.U. tax havens 6 EU havens: Netherl., Ireland, Lux., Malta, Cyprus, Belg. Key advantage: report bilateral data to Eurostat Service exports more reliable than imports (services sold from LU to FR person unrecorded in FR: Spotify) Limitation: miss some profits (hybrid corp, inconsistent definition of residency) We fix this by forcing consistency with U.S. data on profits made by U.S. multinationals

Service exports recorded by havens are more reliable than imports rec by importer Bn. The missing service exports of Luxembourg % of Lux. GNI 60 200% 50 40 150% 30 100% 20 10 50% - Exports from Luxembourg to EU22 according to Luxembourg Imports of EU22 from Luxembourg according to EU22 Note: EU22 is the European Union minus the 6 EU tax havens (Netherlands, Ireland, Luxembourg, Cyprus, Malta, and Belgium). 0%

Service imports from tax havens are under-estimated by importers (B2C sales) Bn. The missing service exports of the six EU tax havens 70 60 50 40 30 Exports to EU22 recorded by exporter Imports recorded by EU22 20 10 - Luxembourg Ireland Belgium Netherlands Malta Cyprus

At least 30% of the services exported by EU havens go unreported by the importer 60% Missing service exports, % of total service exports 50% 40% 30% 20% 10% 0% -10% EU22 EU6 Luxembourg Ireland Belgium Netherlands Malta Cyprus Note: Service exports include exports to all EU22 countries (EU26 minus Luxembourg, Ireland, Belgium, Netherlands, Malta, Cyprus).

Some profits made by U.S. MNEs are missing in EU havens national accounts 400% 350% 300% 250% 200% 150% 100% 50% 0% Luxembourg Taxable corporate profits (% of compensation of employees) Missing profits of U.S. multinationals As reported in the national accounts Average among non-havens: 34% Ireland Puerto Rico Malta Netherlands Denmark Belgium Germany Italy U.K. Spain Sweden Finland United States Switzerland France Note: The blue bar shows the ratio of corporate profits (net of interest and depreciation) to compensation of employees, as recorded in the national accounts, in 2015. The red bar adds corporate profits missing in the national accounts, computed as the discrepancy between FDI income credits reported by the U.S. and total FDI income debits.

A growing amount of profits is artificially shifted to the EU havens 180% 160% 140% 120% 100% 80% 60% 40% 20% Taxable corporate profits in Ireland (% compensation of employees) 0% 1980 1985 1990 1995 2000 2005 2010 2015

By applying very low rates on a huge base, EU havens generate a lot of revenue 5.0% Corporate Income Tax Revenue (% Net National Income) Ireland 4.0% European Union 3.0% 2.0% 1.0% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Note: European Union is the average of France, Germany, U.K., and Italy.

By applying very low rates on a huge base, EU havens generate a lot of revenue % corporate tax revenue collected 80% Tax revenue gained by EU havens on profits artificially shifted Tax revenue on artificial profits (l.h.s.) Tax rate on artificial profits (r.h.s.) 15% 60% 10% 40% 20% 5% 0% Belgium Netherlands Luxembourg Ireland Note: This figure shows the amount of tax revenue collected on artificially shifted profits and the implied rate at which these profits are taxed. The revenue collected on artificially shifted profits are calculated as the amount of revenue collected above the average corporate income tax revenue in all non-haven EU countries (scaled by GNI). 0%

Global Profit-Shifting

Close to 20% of global profits are made by multinationals abroad 20% Multinational profits (% of global corporate profits) 15% 10% 5% 0% 1975-79 1980-89 1990-99 2000-09 2010-15 Notes: This figure charts the share of global corporate profits made by multinational corporations. Multinational profits are defined as the sum of portfolio equity and FDI equity income receipts across all countries. We subtract income received by tax havens to avoid double counting. Multinational profits were around 1.4 trillion in 2015, while global corporate profits were around 7.9 trillion.

45% of multinationals foreign profits are artificially shifted to tax havens Bn. 700 600 500 400 300 200 100 0 Profits artificially shifted to tax havens: the global view EU havens Non EU havens Malta Belgium Luxembourg Netherlands Ireland Hong Kong Singapore Carribbean Rest Total EU tax havens Non EU tax havens % of MNE profits 50% Switzer. P. Rico Bermuda Note: This figure shows the amount of taxable profits artificially shifted to tax havens in 2015. The total adds up to 627 billion euros, of which 334 billion is shifted to non EU tax havens, and 293 billion is shifted to EU tax havens. 40% 30% 20% 10% 0%

63% of the foreign profits made by US multinationals are shifted to tax havens The share of tax havens in U.S. corporate profits made abroad 60% Singapore % of U.S. corporate profits made abroad 50% 40% 30% 20% 10% 0% 1982 1984 1986 1988 Switzerland 1990 1992 1994 1996 1998 2000 2002 Bermuda (and Caribbean) Netherlands 2004 2006 2008 Luxembourg 2010 Ireland Notes: This figure charts the share of income on U.S. direct investment abroad made in the main tax havens. In 2016, total income on U.S. DI abroad was about $450bn. 16% came from the Netherlands, 8% from Luxembourg, etc. Source: author's computations using balance of payments data, see Online Appendix. 2012 2014 2016

Allocating the profits artificially shifted offshore: sales vs. residence % of total profits artificially shifted to tax havens 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Allocation of profits artificially shifted to tax havens Benchmark scenario: High risk imports from tax havens Residence scenario EU US Developing countries Rest of OECD Note: In the benchmark scenario, offshore profts are allocated proportionally to the sum of high-risk services imported from and FDI interest paid tax havens. In the "residence" scenario, offshore profits are allocated based each country's share of global FDI income credits.

EU and US lose almost 20% of their corporate tax revenue 30% Tax revenue lost due to artificial profit-shifting (% of current corporate tax revenue) 25% 20% Benchmark scenario: High risk imports from tax havens Residence scenario 15% 10% 5% 0% EU USA Developing countries Rest of OECD Note: In the benchmark scenario, offshore profts are allocated proportionally to the sum of high-risk services imported from and FDI interest paid tax havens. In the "residence" scenario, offshore profits are allocated based each country's share of global FDI income credits.

EU and US lose about e60bn annually due to the artificial shifting of profits 100 90 80 70 60 50 40 30 20 10 0 Tax revenue lost due to artificial profit-shifting (billion of euros) EU US Developing countries Rest of OECD Benchmark scenario: High risk imports from tax havens Residence scenario Note: In the benchmark scenario, offshore profts are allocated proportionally to the sum of high-risk services imported from and FDI interest paid tax havens. In the "residence" scenario, offshore profits are allocated based each country's share of global FDI income credits.

The higher the corporate tax rate, the more profits are shifted 35% 30% 25% 20% 15% 10% 5% 0% Lost corporate tax revenue due to artificial profit-shifting (% of corporate tax revenue collected) Germany France Hungary Italy EU22 United Kingdom Spain Sweden Austria Finland Estonia Denmark Portugal Poland Profits shifted to non-eu tax havens Profits shifted to EU tax havens Corporate tax rate (avg. 2010-2015) Latvia Croatia Greece Slovenia Lithuania Czech Republic Romania Slovakia Bulgaria Note: This figure shows the amount of tax revenue lost because of the artificial shifting of multinationals' profits to tax havens, as a share of total corporate tax revenue collected in 2015. The grey line shows the top statutory corporate tax rates.

The failure of tax enforcement

The perverse incentives involved in enforcing arm s length prices High-tax ctries have incentives to go after other high-tax: Danish tax authority (tax rate 24.5%) can go after mispricing involving Bermuda (0%) or Germany (30%) MNEs make it hard to go after Bermuda (they would lose revenue) and easy to go after Germany (they win) Mutual agreement procedures facilitate resolve of disputes among OECD countries (eg, Denmark Ger.)

Most transfer price enforcement is against other high-tax countries % of total 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Counterpart in Mutual Agreement Procedures in the EU EU high tax country EU tax haven Note: The graph plots the distribution of the value of mutual agreement procedures in the EU by counterpart. Mutual agreement procedures are cases in which the country conducting a transfer pricing correction (and thus raises the taxable income in the home country) will approach the counterpart country (the country accussed of having excessive profits) and ask them to lower their tax base. The counterpart is the country that the Danish tax authority argue have received excessive taxable profits. The graph shows that 65% of the value of transfer pricing corrections concerns a high tax country (Non tax haven).

E.U. tax authorities barely attempt to go after profits shifted to tax havens Pct. of total 70% 60% 50% 40% 30% 20% 10% 0% Distribution of Danish transfer price corrections (cases) Non-EU Non tax havens Tax havens Counterpart unknown Note: The graph plots the distribution of the number of transfer price corrections by counterpart. Transfer price corrections are cases in which the Danish tax authority have corrected an intra-group cross-border transfer price and as a result raised the taxable profits of firms operating in Denmark. The counterpart is the country that the Danish tax authority argue have received excessive taxable profits. The graph shows that the counterpart in 40% of the cases is a high tax EU country (Non tax haven) and in 24% of the cases is a non-eu high tax country. EU

E.U. collects negligible revenue by correcting transfer prices involving havens Pct. of total 70% Distribution of Danish transfer price corrections (value) 60% 50% 40% Non-EU EU 30% 20% 10% 0% Non tax havens Tax havens Counterpart unknown Note: The graph plots the distribution of the value of transfer price corrections by counterpart. Transfer price corrections are cases in which the Danish tax authority have corrected an intra-group cross-border transfer price and as a result raised the taxable profits of firms operating in Denmark. The counterpart is the country that the Danish tax authority argue have received excessive taxable profits. The graph shows that 65% of the value of transfer pricing corrections concerns a high tax country (Non tax haven).

As settlement is facilitated, high-tax to high-tax disputes are growing Number of Mutual Agreement Procedures in the OECD Number of cases 7,000 6,000 5,000 New OECD MAP cases globally Inventory of OECD MAP cases 4,000 3,000 2,000 1,000 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Note: The graph plots the development in the number of mutual agreement procedures (MAP cases) in the OECD. Mutual agreement procedures are cases in which the country conducting a transfer price correction (and thus raises the taxable income in the home country) will approach the counterpart country (the country accused of having excessive profits) and ask them to lower their tax base. New MAP cases are cases initiated within a given year. Inventory of MAP cases is the total of cases currently in process, that is both new plus cases from previous years that have not been convluded.

Conclusion

Profit-shifting and weak enforcement are leading to a race to the bottom 34 Global corporate tax rates 32 30 Africa 28 26 World Latin America 24 EU 22 20 Asia 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

The race to the bottom is accelerating Global corporate tax rates (%) 38 United states 34 30 Africa 26 World EU Latin America OECD 22 Asia 18 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Reforming the corporate tax Apportionment of global profits proportionally to where sales are made Removes any possibility to shift profits, and any incentives for real tax competition Works reasonably well for US States Can be done unilaterally Would increase corp tax revenue by about 20% in U.S. and Europe