The extra-ordinary tax payments of US MNCs in Ireland

The US tax collector, the Internal Revenue Service (IRS), has been publishing aggregate data from the country-by-country reports US MNCs have been filing with them.  The data is available here and the most recent year covered is 2018.

Here are a few categories from the 2018 data with the countries ranked by the amount of cash payments made for corporate income taxes to each of them.  The top 15 are shown.

As would be expected, the United States itself is the largest recipient of corporate taxes from US MNCs.  Of the $261.5 billion of cash payments for income taxes made by US MNCs in the IRS country-by-country statistics just over $140 billion (53 per cent) was paid to the US.

Next on the list is the UK which was the recipient of $10.9 billion of corporate tax payments from the US MNCs in this data.  And extraordinarily, Ireland is next.  The IRS statistics show that US MNCs paid close on $8 billion of Corporation Tax in Ireland in 2018.  This figure is likely to be higher when the IRS updates its data with subsequent years.

Ireland is the third-highest recipient of corporate taxes from US MNCs in the world.  The scale of the payments being made here is highlighted when they are put in terms of national income.  Here they are put in terms of Gross National Income (for countries with a GNI of more than $100 billion) with the highest 30 countries shown.

The $8 billion of tax paid by US MNCs in Ireland in 2018 was equivalent to 2.6 per cent of Ireland’s Gross National Income (if GNI* is used it moves above 3.0 per cent).  No country with a GNI of more than $100 billion comes close to collecting this amount of tax from US MNCs as a share of its national income. 

The are some smaller countries where the share in Ireland is apparently eclipsed.  These are the Cayman Islands($213 million of tax from US MNCs, 6.9 per cent of GNI), Bermuda ($485m, 6.5%), The Bahamas ($516m, 4.2%) and Luxembourg ($1,398m, 3.1%).  However, these are small payments relative to very small economies and is not clear if the tax payments attributed to US MNCs subsidiaries in those countries are made to those countries.  The GNI of the Cayman Islands is around $3 billion which is 100 times smaller than Ireland and it does not have a corporate income tax from which revenue can be generated.

It is worth noting that the US itself in included in the above chart.  In 2018, the corporate tax revenue from US MNCs in Ireland as a share of GNI was nearly four times higher than it was in the United States.

We know so much about Google’s tax structure, how is it possible to get the reporting so wrong?

The taxation of US multinationals continues to make headlines. It is right that these companies are subject to intense scrutiny but given how frequently they are reported on it is surprising how frequently significant errors arise in the coverage. A recent case in point arises in this piece from The Irish Times.  The piece begins:

Google shifted more than $75.4 billion (€63 billion) in profits out of the Republic using the controversial “double-Irish” tax arrangement in 2019, the last year in which it used the loophole.

The technology giant availed of the tax arrangement to move the money out of Google Ireland Holdings Unlimited Company via interim dividends and other payments. This company was incorporated in Ireland but tax domiciled in Bermuda at the time of the transfer.

It looks like there is headline worthy stuff going on here but in just the opening two paragraphs the piece manages to utterly contradict itself.

Paragraph 1: A Google subsidiary shifted profit out of Ireland.

Paragraph 2: This Google subsidiary was based in Bermuda.

How can a holding company in Bermuda shift profit out of Ireland? Sure, Google Ireland Holdings made substantial dividend payments and distributions in 2019 but it was a payment from a company in Bermuda to its immediate parent.  Neither party to the transaction was in Ireland.  And this is clearly set out in the company’s accounts.

Note 9 to the accounts tells us  “the Company is domiciled and tax resident in Bermuda” while Note 10 tells us that the distributions in 2019 comprised $15 billion of financial assets, $27 billion of debt securities, dividends of $30 billion and other net assets of $3 billion.  That is the $75 billion that generated the headline.

Of course, distributions and annual profit are different things.  We can use the very same accounts the dividends and distributions figure came from to get the company’s profit.   To do that all we have to do is look at its income statement.

Google Ireland Holdings certainly was a profitable company.  It had a pre-tax profit of $13.7 billion in 2019 from turnover of $26.5 billion.  Of course, as a company resident in Bermuda (which doesn’t have a corporate income tax) pre-tax profit and post-tax profit were the same.

The third paragraph of The Irish Times piece goes:

The move allowed Google Ireland Holdings to escape corporation tax both in the Republic and in the United States where its ultimate parent, Alphabet, is headquartered. The holding company reported a $13 billion pretax profit for 2019, which was effectively tax-free, the accounts show.

Hmmm. Didn’t we have a huge amount of coverage last week, including in The Irish Times itself, of proposed changes to the US minimum tax on the foreign profits of its MNCs – the so-called tax on GILTI, Global Intangible Low-Taxed Income. So, yes you can say that Google Ireland Holdings profit “was effectively tax free” once you ignore the tax that its ultimate parent, Alphabet Inc., is liable to pay on those profits to the US. 

Similarly, we could say Ireland is effectively a landlocked country – once you ignore the hundreds of miles of coastline.  We can say it but it means nothing.  

Moving on to paragraph five:

Google has used the double Irish loophole to funnel billions in global profits through Ireland and on to Bermuda, effectively put them beyond the reach of US tax authorities.

It seems somewhat odd given the above statement, that Google would have included the following note in its 2017 annual accounts:

One-time transition tax

The Tax Act requires us to pay U.S. income taxes on accumulated foreign subsidiary earnings not previously subject to U.S. income tax at a rate of 15.5% to the extent of foreign cash and certain other net current assets and 8% on the remaining earnings. We recorded a provisional amount for our one-time transitional tax liability and income tax expense of $10.2 billion. We have recorded provisional amounts based on estimates of the effects of the Tax Act as the analysis requires significant data from our foreign subsidiaries that is not regularly collected or analyzed.

As permitted by the Tax Act, we intend to pay the one-time transition tax in eight annual interest-free installments beginning in 2018.

This one-time transition tax, or deemed repatriation tax, was levied on the profits that contributed to the dividends and distributions made by Google Ireland Holdings in 2019 and will be paid in annual installments out to 2025. 

So, absolutely you can say that for these profits Google was able to “effectively put them beyond the reach of the US tax authorities”, once you ignore the $10 billion of tax that Google will pay on them.  And, as noted above, any such profits earned since 2018 have been subject to the tax on GILTI.  That tax is in place but as it was introduced in legislation called the Tax Cuts and Jobs Act it is not clear how much revenue is being generated by it.

In overall terms, we can assess Google’s tax payments by looking at its annual financial statements.  Here are the company’s income statements for all years from 2010 to 2020, as well as a line about actual tax payments extracted from the company’s cash-flow statements. Click to enlarge.

All told, in the 11 years since 2010 Google has reported a cumulative pre-tax profit of $264 billion.  The provision for income taxes sums to $54 billion giving a effective accounting tax rate of just over 20 per cent.

The provision for the one-time transition tax was made in 2017 and it can be seen that the effective tax rate for that year exceeded 50 per cent as the provision for taxes included $10 billion of tax for accumulated foreign profits.

On a cash basis the company paid $41 billion of tax.  This will increase to match the provision for income taxes as future installments of the deemed repatriation tax are paid.

We move on to paragraph six

Companies exploiting the double Irish put their intellectual property into an Irish-registered company that is controlled from a tax haven such as Bermuda. Ireland considers the company to be tax-resident in Bermuda, while the US considers it to be tax-resident here. The result is that when royalty payments are sent to the company, they go untaxed – unless or until the money is eventually sent home to the US parent.

Hmmm. Untaxed that is if you ignore the minimum tax that the US has on the foreign income of its subsidiaries.  The tax on GILTI was introduced in December 2017 which is surely enough time for the reporting to catch up.

The “double irish” was original designed around the “same country exception” in Subpart F of the US tax code.  As this IRS note states passive income, such as royalties, received by the foreign subsidiaries of US MNCs was immediately liable for US tax (under the pre-2018 regime).

Generally, the US shareholder of a foreign corporation is able to defer taxation of the corporation’s income until it has been distributed to the shareholder. However, in the case of a controlled foreign corporation (“CFC”), certain types of income are subject to current inclusion (“subpart F inclusion”) by the US shareholder under IRC 951. One such type of income is Foreign Personal Holding Company Income (FPHCI), which includes income of a CFC such as dividends, interest, rents, and royalties. The FPHCI rules eliminate the deferral of US tax on income earned by certain foreign corporations from portfolio types of investments, i.e., where the company is merely passively receiving investment income rather than earning active business income. Consequently, many of the exceptions to current inclusion of FPHCI focus on the recipients of income (i.e., whether the recipients meet certain criteria). 

As the note sets out there are some exceptions to the immediate inclusion of passive income for taxation in the US.

Another exception that looks to the payor of the income for eligibility requirements is the same country exception from FPHCI under IRC 954(c)(3). Under the same country exception, FPHCI does not include dividends and interest received by a CFC from a related CFC payor which is incorporated in the same country as the recipient CFC, and which has a substantial part of its assets used in its trade or business in that same country. Similarly, under this exception, FPHCI does not include rents and royalties received by a CFC from a related CFC for the use or privilege of using property within the same country as the recipient CFC’s country of incorporation. However, the FPHCI implications/treatment of rents/royalties are outside the scope of this Unit and will be covered in a separate Unit.

This is the reason for “double” part of the moniker. The key is to have two companies registered in the same country. Their residence doesn’t matter from a US perspective.  So, Google had a Irish-registered trading company operating in Ireland and an Irish-registered holding company based in Bermuda. 

The trading company paid for the right to use Google’s technology and because the royalty paid was received by another Irish-registered company, the income was not included when earned but the tax payment could be deferred until the profit was repatriated to the US.  The 2017 TCJA abolished the deferral on taxes on this income.

The company in Ireland is Google Ireland Limited and it is the fulcrum of Google’s international operations.

 

Some big numbers there. Google Ireland Limited had turnover of €45.7 billion in 2019.  Of this, around €14.3 billion went on the cost of sales which in this case is fees paid to website owners for hosting ads from Google’s services.

From gross profit if €31.4 billion the company incurred €29.7 billion of administrative expenses.  These include staff and premises costs in Ireland but by large the largest component is the royalty it must pay for the license to see advertising using Google’s technology.

After investment and other income, Google Ireland Limited had a pre-tax profit of just under €2 billion and a tax provision of around €250 million (c.12.5 per cent).  This, along with its €600 million of wage costs and purchases from Irish suppliers represents its contribution to the Irish economy.

The main supplier for Google in Ireland is the entity that produces the technology it uses.  Due to a Irish withholding tax in place at the time the structure was set up the royalties payments make a detour to The Netherlands (the “dutch sandwich”) before finding their way to Bermuda.  Under the EU’s Patents and Royalties Directive, Member States can levy a withholding tax on royalties paid to another EU Member State. 

At the time, Ireland would have levied this withholding tax if the payments went from Ireland to Bermuda but this was easily circumvented  by directing the payments first to The Netherlands and then on to Bermuda.

We have already looked the accounts of the Bermudan company above. It received very significant revenues but we can also saw that it incurred very significant costs.  The accounts give a breakdown of some of these.

The largest cost is R&D expenditure.  This is a holding company with no employees so it obviously isn’t doing the R&D itself. This is the payment for someone else to undertake the R&D on its behalf.  As the accounts note:

Administrative expenses

Administrative expenses increased from US$11.1 billion in 2018 to US$14.1 billion in 2019, an increase of US$3.0 billion. This increase is primarily due to an increase in research and development (“R&D”) expenses. The Company incurred US$10.4 billion (2018: US$9.6 billion) of R&D expenditure in the year pursuant to a cost sharing agreement with a fellow group undertaking. These expenses increased due to an increase in the worldwide spend on R&D.

Google Ireland Holdings was part of a cost-sharing agreement with Google in the US.  The R&D takes place in the US but the cost was split between the US parent and Google Ireland Holdings.  Google Ireland Holdings made a contribution to Google worldwide spend on R&D based on the relative size of the market it serves.  It look like Google Ireland Holdings paid for around 40 per cent of Google overall spend on R&D.

So while Google Ireland Limited did make a royalty payment of around €26.5 billion that ended up in Bermuda in 2019 around half of that was further transferred to the US to cover the costs of Google technicians, engineers and developers who are behind Google’s technology.  The residual that was left in Bermuda was subject to the US GILTI tax.

However, Google does nothing in Bermuda to justify the profits earned by Google Ireland Holdings.  In recent years, the OECD, through their ongoing BEPS project, have been proposed updates to transfer pricing guidelines that would ensure that the location where a company’s profit is reported better matches the location of the substance that generates that profit.  Google’s structure with $13 billion of profit in Bermuda in 2019 did not align with that.

The changes implemented through the OECD in relation to profits linked to intangible assets such as those produced by R&D are summarised as DEMPE – does a company do sufficient development, enhancement, maintenance, protection or exploitation of the intangible asset do justify the profit allocated to it?

Google Ireland Holdings does nothing so there is no way it had sufficient DEMPE functions to justify the profits allocated to it.  This means that payments made to Google Ireland Holdings would not be tax deductible.  It is these changes to transfer pricing guidelines that have brought an end to structures like these.  For Google this happened at the end of 2019.

Here a paragraph from The Irish Times article that is actually correct. It is a copy and paste of a statement from Google.

“In December 2019, in line with the OECD’s base erosion and profit shifting (BEPS) conclusions and changes to US and Irish tax laws, we simplified our corporate structure and started licensing our IP from the US, not Bermuda. The accounts filed today cover the 2019 financial year, before we made those changes.”

So, the Irish company still has to pay for the rights to use Google’s technology but now those payments go direct to the US instead of ending up in Bermuda.  This is as it should be.  The US should not have allowed a cost-sharing agreement that allowed the profits from activities undertaken within the US to leak out. 

When they did leak out the US wanted their MNCs to “exhaust all remedies” to minimise the amount of foreign tax they paid.  Locating the profits in Bermuda was US companies complying with the recommendations of the IRS. 

This is from an IRS note that sets out what US MNCs are expected to do if they wished to avoid double taxation:

The United States taxes income on a worldwide basis. To prevent double taxation, under the Internal Revenue Code (IRC) U.S. taxpayers are allowed a credit for foreign income taxes paid. However, the ability to credit foreign income taxes paid is limited. Pursuant to Treasury Reg. § 1.901-2(e)(1) U.S. companies may not obtain credits for foreign taxes paid in which they are not legally compelled to pay. If the U.S. Treasury were to allow foreign tax credits (FTCs) without requiring U.S. taxpayers to effectively and practically reduce their foreign tax payments as legally imposed, taxpayers would have no incentive to challenge any foreign tax, whether or not properly imposed. As a result, foreign tax costs may potentially be improperly shifted to the United States. 

As a general rule under Treas. Reg. Section 1.901-2(e)(5)(i), taxpayers cannot claim foreign tax credits for amounts paid to foreign taxing authorities where they have failed to exhaust all effective and practical remedies (including competent authority procedures where applicable).

So what can appear as egregious tax avoidance from, say, a European perspective can be seen as simply ensuring tax payments are maximised from a US perspective.  The US wanted their MNCs to locate their foreign profits in no-tax jurisdictions such as Bermuda.  This ensured that US payments were maximised, albeit that the US had a terrible record of actually collecting this tax.

The updated transfer pricing rules means that restrictions have rightfully been placed on companies reporting their profits in jurisdictions where they have no substance.  Google has responded this need to better align its profit with its substance by licensing its IP from the US and not Bermuda and this has changed how its profit is taxed.

What has been the effect of this in Ireland?  In reality, there is very change.  The Irish company is still paying for the technology it uses but the destination of those payments has changed.  We can see this from balance of payments data.

Whoa!  For pretty all of the last decade royalty payments from Ireland to the US were around €2 billion a quarter.  They skyrocketed in 2020 and, all told, around €52 billion of royalty payments went from Ireland to the US.  It is more than just Google that have moved to licensing their technology for international markets from the US.  Facebook have too.

This means that in 2020 the trading companies in Ireland continued to collect tens of billions in revenue from selling advertising on Google’s and Facebook’s platforms.  For the rights to sell the advertising, their Irish subsidiaries paid for the technology they are using and these payments are now being directed to the where the DEMPE activities that produce the technologies are located: the United States.  These payments may add 0.1 or even 0.2 per cent to 2020 US GDP.

But I wonder how the Irish reports will butcher their “analysis” of Google’s 2020 accounts in Ireland when they are published this time next year?  Because all the evidence is that butcher it they will. It seems “most read” or “most likes” is more important than “most accurate”.

The non-financial corporate sector in the institutional sector accounts

There are lots of reasons not the look at the figures for Ireland’s non-financial corporate sector in the quarterly institutional sector accounts (ISAs).  They are the figures that are most likely to be revised when the National Income and Expenditure (NIE) accounts are published in the summer.  And a much more information breakdown (by domestic, foreign-owned and redomiciled PLCs) will be published with the annual ISAs in the autumn. 

So, without expected it to be any more revealing that staring into a puddle here are the latest figures for the NFC sector.  First, the current account:

Lots of big numbers as can be expected given the scale of the MNCs that operate in the sector.  However, most of the 2020 changes are fairly modest (at least they are in this release anyway).

We can see that compensation of employees paid by NFCs held up fairly well in 2020 only showing a decline of 2.5 per cent or €1.7 billion.  However, this was supported by the government’s wage subsidy schemes and subsidies on production received were up almost 500 per cent. The (mainly domestic) firms receiving these subsidies used them to support the wages of their employees.

Amidst all the big number we see the continued rise in Corporation Tax payments from this sector.  In 2020, the NFC sector paid €9.6 billion of Corporation Tax up from €8.4 billion in 2019. 

Another item worth noting is the continuing rise in the amount of interest paid by the NFC sector.  The interest amount in the ISAs rose a further €2 billion in 2020.  This may be linked to the onshoring of tens of billions of IP assets by US MNCs.

The figures in the capital account are even more murky – particularly those in the lower panel showing capital formation, acquisition of non-produced assets and net borrowing.

The NFC sector in Ireland has been doing an enormous amount of capital formation expenditure in recent years.  This has been to the extent that a €50 billion drop in NFC investment is of concern to no one.  This is because the main reason for the wild fluctuations in the GFCF has been IP onshoring by US MNCs.  These transactions can be worth tens of billions but their impact on the overall economy is limited.

The lower panel shows that their has also been significant expenditure on the acquisition of non-produced assets.  Again this is linked to IP onshoring but is IP that is not the result of R&D activity.  The assets here include licenses and marketing assets such as customer lists.  This shows a large drop in 2020, but, like capital spending, is a figure that is subject to revision in the NIE.

This means that the bottom line is not informative either.  The net borrowing position of the NFC sector may have improved by more than €70 billion but most of this is due to reduced onshoring of IP by US MNCs.  We’ll get a much better picture of what is happening in the business sector in Ireland, and most importantly the domestic business sector, as the CSO works through it release schedule for the year.

The government sector in the institutional sector accounts

The CSO will publish the 2020 Government Finance Statistics (GFS) in the next few weeks but we can get a preliminary look at what they will say from the Institutional Sector Accounts (ISAs) that were published last week.

Here’s the government sector current account for the five years to 2020.

For 2020, the most significant changes were subsidies paid (+€4 billion), social benefits paid (+€6 billion) and consumption expenditure on goods and services (+€5 billion).  All told, these contributed to a €16 billion deterioration in the current account position of the government sector.

One of the remarkable things is that all of these changes were on the expenditure side with revenue holding up.  There was a decline in taxes on products and production.  Taxes on products (such as VAT and Excise Duty) were down €2.5 billion with taxes on productions (such as commercial rates) down just under €1 billion.  Some of this fall was just to reduced activity while some was due to forbearance and payment breaks.

Taxes on income and wealth rose in 2020. This was entirely due to increased Corporation Tax from non-financial corporates which rose another €1 billion but taxes on income from the household sector (Income Tax and USC) were essentially unchanged in 2020.  Social contributions received (mainly PRSI) were also largely unchanged in 2020.

On the spending side some of the other changes were the eight per cent rise in compensation of employees paid by the government sector with this almost reach €25 billion in 2020.  It might be surprising the the GDP of the government sector rose seven per cent in 2020 (schools closed etc.) but in the absence of prices the value added of the government sector is mainly measured by the wages paid.  Finally from the current account, we can also see that the interest bill continued to fall and the interest amount in the sector accounts falling by a further €1 billion in 2020.

There were also limited changes in the capital account.

The most significant change here was the further €1.5 billion rise in government gross capital formation in 2020 to reach €9.7 billion.

There was little change in most other items in the current account meaning that the overall change in the government’s non-financial position was a deterioration of €18 billion.  The government sector went from a modest net lending position of €1.5 billion in 2019 to a net borrowing position of €16.5 billion in 2020.

To the extent the the increased spending highlighted in the current account is temporary this deficit will fall as the need for emergency measures falls.  But if some of those represent permanent spending increases, as is likely, some of the deficit will be persistent.  It is that permanent increase in spending rather than any necessity to “pay the COVID bill” that will impact subsequent budgetary plans.

The household sector in 2020

The CSO have published the Q4 2020 update of the non-financial Institutional Sector Accounts.  While there may be some revisions when the annual accounts are published later in the year, the quarterly data can be used to get preliminary figures for the year as a whole. 

Here’s the household sector current account for the past four years as well as the annual change in 2020.

The bottom section tells us that the disposable income of the household sector rose by just over four per cent in 2020.  The biggest reflection of the restrictions that were in place for much of the year is the nine per cent reduction in consumption expenditure.  This meant that the gross saving of the household sector almost doubled in 2020, reaching €28 billion and giving an unusually high savings rate of 23 per cent.

Although it looks like there was relatively serene progression in the aggregate income of the household sector in 2020 there was obviously a lot going on above that.  In overall terms, the compensation of employees received by the household sector was essentially flat in 2020 showing a fall of just –0.3 per cent.  This, however, masks a difference across sectors with all sectors bar the government sector showing a fall in 2020.

Compensation of employees from the government sector rose a further eight per cent in 2020 to almost reach €25 billion.  The increase since 2015 is almost 30 per cent.

The nearly €2 billion rise in compensation of employees from the government sector almost fully offset the reductions elsewhere.  And those reductions would have been even larger were it not for the various wage subsidy schemes introduced which provided a subsidy to firms to support their wage payments to staff.

A little further down, we see that social benefits received by the household sector (excluding benefits in kind) from the government sector jumped from €24 billion in 2019 to €30 billion in 2020.  This increase was primarily driven by the Pandemic Unemployment Payment (PUP) though even the absence of this scheme there would have been an increase in these transfer as most, but not all, of the people impact would have been eligible for existing income supports.

And those are the most significant changes in the table.  It looks like there’s something going on with the FISIM adjustment but it seems to largely net out across interest paid and interest received.  There has been a reduction in contributions to private pensions in recent years. Social contributions paid to the financial sector have gone from €7.3 billion in 2018 to €5.9 billion in 2020.

This is perhaps surprising given the general increase in savings behaviour in 2020.  However, there was still a greater amount contributed to private pensions funds then paid out in benefits from those funds – the adjustment for pension entitlements in the second-last row was +€1.4 billion.

We can turn to the capital account to see what happened to see how the rest of the savings were used.

Gross capital formation by the household sector was down in 2020, falling from €7.2 billion in 2019 to €6.8 billion in 2020.  This is probably not a surprise given that the main capital expense of the household sector is housing (purchases of new units and refurbishments of existing dwellings).  Restrictions would have reduced output in the construction sector.

The bottom line shows that most of the savings was carried through the capital account.  Net lending by the household sector exceeded €20 billion in 2020.  This means that most of the additional saving made its way to the financial accounts: higher deposits, lower loans and maybe increased investment in other financial assets but higher deposits and lower loans are likely to dominate.

To conclude here is the savings minus investment positions of the household and government sectors.  The figures shown are four-quarter moving sums.

The turmoil of 2020 is clearly seen but is also noticeable is that as the year progressed the [S – I] deficit of the government sector was growing faster than the [S – I] surplus of the household sector.  The overall position across the two sector was still positive: if necessary the deficit of the government could be funded from domestic sources. 

Maybe the next 12 months will see an erosion of that savings behaviour possibly to fund a consumption boom which, in turn, will boost the government’s position.  It doesn’t even need the accumulated savings to be unleashed, just a reduction in the savings rate.