A possible explainer for Google’s recent tax settlement?

The 2020 accounts for Google Ireland Limited were published last week and showed that the company had a €218 million tax charge that arose from “the resolution of certain tax matters relating to prior years” (and there was a further €127 million of associated interest).  The group’s parent company, Alphabet Inc., had stated in its own 2020 annual report that its “tax years 2011 through 2019 remain subject to examination by the appropriate governmental agencies for Irish tax purposes.”

It’s possible we can get some insight into the nature of this issue from the table of the annual profit and loss statements which were summarised in the previous post which looked at Google’s footprint in Ireland since 2003.  Here is the relevant table. Click to enlarge.

In particular we are drawn to the column for operating profit, which forms the bulk of the company’s profit before tax.  It looks like Google Ireland Limited’s operating profit can be broken down into three time periods:

• 2003-2011
• 2012-2015
• 2016-

A post a number of years ago looked at the determination of Google’s profit in Ireland and looked at the period 2012 to 2014 which ties in with the middle time period above.  As the table shows, in 2012 there was a significant step-up in Google Ireland Limited’s operating profit compared to the previous decade – or at least an increase above that which could be explained by the expansion of the company.

The earlier post reached the conclusion that “Google Ireland’s operating profit has been around 6 per cent of its expenses excluding the license [or royalty] payment.”  Prior to 2012, it seems that the cost-plus arrangement for determining Google Ireland Limited’s operating profit only included the costs incurred in Ireland.  It seems likely that the step-up in 2012 was due to the inclusion of the costs Google Ireland Limited incurs in the payments it makes for the sales and marketing efforts to the local Google subsidiaries in the markets in which it sells.  The earlier post sets this out for a sample of countries. 

The earlier post should have been updated because in 2016 there was another step up in Google Ireland Limited’s operating profit.  It looks like the profit margin is still in the region of six to seven per cent of expenses but it now includes all expenses, most notably the royalty payment.

Here is Google Ireland Limited’s operating profit as a per cent of its administrative expenses since 2005.

The step-ups in 2012 and particularly 2016 are clear.  The outcome for the period 2012 to 2015 would be around 6.5 per cent if the royalty payment is excluded from the base, which is pretty much what it has averaged in the period since 2016 with the royalty payment included.

It is only supposition, but it is possible that the tax issue that Google Ireland Limited revealed in its accounts related to the exclusion of the royalty payment from the base for determining its operating profit using a cost-plus margin in the years before 2016.  The fact that there was €127 million of interest linked to the €217 million tax charges lends credence to the conclusion that it relates to tax due a number of years ago (with Revenue applying an rate of eight per cent per annum to such amounts).   

It is possible that Revenue began this review back in 2015 or 2016 with Google then deciding to include the royalty payments in the base for the cost-plus determination from then on.  Thus what was in dispute was the operating profit figures for Google Ireland Limited for years prior to 2016.  Of course, the change in 2016 also coincides with the time when the cage-rattling by the European Commission using state-aid cases into tax was going strong and that may have influenced the company’s decisions.

So, can we get numbers to fit the hypothesis that the tax settlement is linked to the inclusion of royalties in the case for the cost-plus assessment of Google Ireland Limited’s profit? Perhaps we can.

Looking at the amount of royalties paid, figures for which are available in the accounts of a subsidiary in The Netherlands, Google Netherlands Holding’s BV, shows that these came to €29.7 billion from 2013 to 2015.  While we don’t know the cost-plus margin applied, using 6.5 per cent gives a return of €1.8 billion.  The following table shows what happens if these returns are taxed at 12.5 per cent and the interest that would have accrued if the tax should have been paid seven, six and five years ago.

Maybe it is little more than coincidence that the figures from this scenario are close to “adjustment for corporation tax of prior periods” that Google Ireland Limited disclosed in its 2020 accounts.  These were an additional tax charge of €218.2 million and €127.3 million of related interest.  In fact, the figures are more than just close; they are almost identical.

One fly in the ointment is that the above table only includes the years 2013 to 2015, whereas the note in Alphabet’s 10k said that all years back to 2011 were under review.  But years being under review does not mean there will be a revised tax assessment for them.

We do know there has been a change since 2016 and it is possible that the move to include the royalty expense in the cost-plus calculation has added around €700 million to Google Ireland Limited’s Irish tax bill in the five years since.  The first table above shows a step-up in Google Ireland Limited’s tax charge in 2016. If this change to the cost-plus methodology hadn’t been implemented, and Revenue’s assessment held up, then Google could have been announcing a €1 billion tax settlement last week – assuming the hypothesis here is correct. 

So, as has often been the case the issue was not the tax rate applied to profits but the amount of profit to be subject to Ireland’s 12.5 per cent rate.  The suggestion here is that, up to 2016, the royalty paid by Google’s Irish subsidiary for the right to sell advertising using Google’s platform and technology was excluded from the cost-plus assessment used to calculate its taxable profit in Ireland. 

The Revenue position looks to have been that while the change in 2016 was fine it should also have applied to a number of earlier years leading to an increased tax charge for those years.  Is this a guess? Yes. But probably not a bad one.

Google’s footprint in Ireland since 2003

Google set up its EMEA headquarters in Dublin in 2003, a year before its IPO.  Initially it was a modest operation and during 2003 the average headcount was 21. By the 2020 annual report of Google Ireland Limited this had increased to 4,314.

The accounts also give the staff costs incurred.  These have increased from €0.7 million to just over €750 million in 2020.  All told, Google Ireland Limited has had €4.8 billion of staff costs since it was established in 2003, with almost three-quarters of that being wages and salaries.  For its most recent year, 2020, the average of wages and salaries per person employed was €120,000

The column for “social welfare” is almost certainly employer’s PRSI and the company has paid €360 million of this over the past 18 years.  The company has also made €121 million of payments into its defined contribution pension plan where “the company matches the employee’s contributions up to a maximum a seven per cent.”

We now turn briefly to the financial outturns for Google Ireland Limited with the numbers and the table getting a bit bigger.  Click to enlarge.

Turnover has grown from just €7.5 million in 2003 to €48.4 billion in 2020.  In cumulative terms, the company has had more the €300 billion of turnover but more than half of that was in the last four years.

The company has had a cumulative pre-tax profit of €10.2 billion and incurred a tax charge of €1.7 billion.  Of this €0.1 billion was for foreign withholding taxes leaving a charge to Irish Corporation Tax of €1.6 billion.  That gives a charge to Irish tax of 15.8 per cent of profit before tax.  This is higher than the standard 12.5 per cent that applied during the period for two reasons.

First, the company incurred expenditure that was not allowable as a deduction for tax purposes meaning its taxable income was larger than the profit reported in the accounts. Second, the company had some non-operating income, such as income received, and this is taxed at the 25 per cent of Corporation Tax for non-trading income.

There does appear to be a couple of noticeable step-changes in the accounts for Google Ireland Limited, most noticeably for operating profit.  Could these be linked to the recent tax settlement that was revealed last week.  Perhaps.  And we will look at that in a subsequent post.

What impact did the end of the ‘double irish’ have on Google Ireland Limited? None

Google ended its use of the high-profile ‘double-irish’ tax structure in 2019.  We have been tracking the impact of the revised structure in Ireland’s balance of payments data and in the consolidated accounts of Alphabet, the name of Google’s parent company.  A detailed examination of these changes is provided in this technical paper.

The financial accounts for Google’s subsidiary in Ireland, Google Ireland Limited, are now available and this allows us to see the impact the ending of the ‘double irish’ had on Google in Ireland.

And we can see that it had no impact.  The outcomes in 2020 are pretty much inline with the outcomes in 2019.  Turnover rose to reach €48.4 billion with a reduction in the cost of sales leading to a gross profit of €36.4 billion.

The main cost of sales for the company are payments to third-parties with websites on which Google’s advertising is displayed.  These increased in 2020.  The reason for the reduction in cost of sales was “a reduction in certain operating fees paid to fellow group undertakings.”

From gross profit, €33.6 billion of administration expenses are deducted which, after other operating income and expenses, leaves an operating profit of €3.0 billion.  The administration expenses include the operating costs of the company such as €750 million of staff costs and will include other running costs.

However, the main element in this item is the expenses Google Ireland Limited incurs for the right to sell advertising using Google’s platforms and technology. This technology is developed elsewhere and the Irish subsidiary pays for the right to use that technology.  This right is transferred through a licensing agreement and in return for that right Google Ireland Limited pays a royalty fee to the owner of the intellectual property. 

A breakdown with the royalty payment is not provided.  The accounts of a Google subsidiary in The Netherlands, the “dutch sandwich”, show that the royalty payments made by Google Ireland Limited in 2019 were €19.4 billion.  Given the increase in administration expenses shown in the accounts, the royalty payments in 2020 were probably around €22-23 billion.

At the 12.5 per cent of Corporation Tax, the tax on Google Ireland Limited’s €2.85 billion of profit would be €357 million.  The total tax charge for the year was €622 million with the following table setting out the reasons for the difference.

In its financial accounts, Google Ireland Limited had around €250 million of expenses which are not deductible for tax purposes.  This increases the tax charge by €32 million relative to what it would be if tax was levied on financial profit rather than taxable income.  The company also incurred withholding taxes, possibly in other jurisdictions, of €15 million.

The largest item is the result of the conclusion of a tax audit which resulted in an additional tax liability of €218 million (with a further €127 million of interest).  In its 2020 annual report, Alphabet Inc. noted that its “tax years 2011 through 2019 remain subject to examination by the appropriate governmental agencies for Irish tax purposes.”  It is likely that the above figures represent the conclusion of this examination and thus are unrelated to the ending of the ‘double-irish’ structure.

The ‘double-irish’ has ended and Google Ireland Limited continues to receive tens of billions in revenue from the sale on online advertising in markets across Europe, the Middle East and Africa (EMEA).  Some of this revenue goes to third-party sites that host the advertising, some goes to cover the staff and running costs of the Dublin office but the main cost of Google Ireland Limited continues to be the royalty expense it incurs for the right to sell advertising using Google’s technology.  None of this has changed with the ending of the ‘double-irish’.

There has been no impact, or additional tax liability, in market countries and there has been no change in how the Irish subsidiary operates.  From our previous analyses we do know that what has changed is where the royalty payments are going to.  Previously they went to Bermuda, via The Netherlands.  The 2020 accounts of the company that was in Bermuda are also now available.

Here there is a change.  The company had a turnover of nil in 2020.  In 2019, the company in Bermuda had a turnover of $26.5 billion comprising the royalties paid out by Google Ireland Limited in Dublin and also by Google Asia Pacific Pte. Limited in Singapore which covers markets in Asia for Google.  After administration expenses of $14.1 billion (the bulk of which was a $10.4 billion contribution to the R&D costs of the group’s US parent) the company in Bermuda had a profit of $13.7 billion.  With no turnover, this was not repeated in 2020.

The reason is that the royalties are now paid to the United States.

There is only one thing that has been impacted by the end of the ‘double-irish’. That is in how the company is taxed in the US.  And that is really only a change in the provisions under which the company is taxed (from GILTI to FDII) rather than a dramatic increase in the amount of tax paid by the company.

There has been a decade of headlines about the ‘double-irish’.  The pantomime villain stopped using the structure almost two years ago.  There has been no impact on the taxation of Google in Ireland or on the taxation of Google in the markets where it sells. Doubtful we’ll see a slew of headlines about that though.

Ireland in the Global Income Distribution

Here is a website that has an interactive chart showing the position of a country’s income distribution within the global distribution.  The ventiles (one-twentieths) for each country ranked along the horiontal axis where their position in the world income distribution given on the vertical axis.  The example below shows Ireland.

It shows that most incomes in Ireland are towards the top of the global income distribution (in price-adjusted terms).  Ireland’s first ventile (the bottom 5 per cent) were located at around the 65th percentile of the global income distribution with the second ventile at around the 80th percentile.

As the chart title indicates the estimates, which are based on the work of Branko Milanovic, are based on data that is around ten years old.  There have been significant changes in Ireland since then – the country is no longer facing the teeth of a deep recession – and these have led to large changes at the bottom of the income distribution.

The post-2008 crash resulted in large drops for the cut-off points of the percentiles that make up the lowest ventile of Ireland’s income distribution.  The first chart compares Ireland’s position in the global income distribution at a time when the income of the lowest ventile was unusually low.

After 2014, there was very strong income growth for these percentiles with a doubling of the incomes shown in just five years.  It is possible that average income of the bottom 5 per cent in Ireland would be placed at around the 85th percentile in the current global income distribution.

The at-risk-of-poverty threshold at 60 per cent of the national median is currently around €15,000 for a single person (or €32,000 for a 2+2 family).  This income would be at around the 90th percentile of the global income distribution.

The interactive tool can be used to make find some unusual comparisons.  There is the example of South Africa where, in the 2011 data, the average income of the top five per cent was comparable to the income of the top 5 per cent in Ireland but where the bottom five cent were amongst some of the lowest incomes in the world.  Or Nigeria where the average income of the top five per cent is lower than the income of the bottom five per cent in Ireland.

The Domestic Non-Financial Corporate Sector in 2020

The distortions in Ireland’s national accounts means that the contribution of some sectors can be overlooked as it cannot be disentangled from huge flows in the overall data due to the presence of US MNCs here.  One such sector is the domestic business sector which is lumped in with the MNCs in most national accounts releases.

The annual institutional sector accounts remedy this as they provide a very useful sectoral breakdown with one of these being for a “domestic non-financial corporate sector”.  We have previously examined the contribution of this sector to pre-pandemic growth and, while there have been some revisions to the figures for earlier years, here we will focus on the 2020 changes. First, the current account.

Summary: pretty much everything went down.  In 2020, output was down, wages paid was down, profit was down. Pretty much the only thing that went up was subsidies received, without which the drop in wages paid would have been even greater.

But the above breakdown misses some of the significant differences by economic sector. Here is a breakdown by economic sector the of compensation of employees paid by the domestic corporates (with domestic financial corporates (NACE K) also included).

The stand-out figure is the 60 per cent drop in COE paid by the Accommodation and Food Services sector.  This contrasts with the three best-performing sectors in the above table which saw their COE expenditure increase by more than five per cent last year.

If also worth noting that the domestic NFC sector includes publicly-owned non-financial corporations.  These include semi-states such as the ESB, An Post, Coillte, Bord na Mona, Rehab, Gas Networks Ireland, RTE, the Dublin Airport Authority, Dublin Bus, Bus Eireann and others. Included in others are the seven universities as they are classified in the non-financial corporate sector and are likely a significant contributor to the COE from the Education sector shown in the above table.

The impact of the pandemic on the capital account of the domestic NFC sector seems to have been a bit more muted and as with the current account there have also been revisions to earlier years.

All told, the bottom line of the non-financial accounts is that the domestic NFC sector has been a consistent net lender in recent years.  The pandemic seems to have had only a modest impact on this.

The next step would be to move to the financial accounts to see what the sector is doing with this net lending.  We don’t have a financial transactions account so we move beyond that to the financial balance sheet.  We will show all years for which data is available. Click to enlarge.

As noted before the financial balance sheet for Irish-owned NFCs has been exploding in recent years.  Total financial assets went from €167 billion at the end of 2012 to €529 billion at the end of 2020.  That far outstrips anything that could be explained by the net lending shown in the capital account or any revaluation effects.

At the same the total liabilities of the sector rose from €275 billion to €705 billion with the most significant changes on both sides of the balance sheet being for equity items. And, as can be seen with the bottom line, the financial net worth of the domestic NFC sector has significantly deteriorated in recent years. Financial net worth excludes real assets so this is not an indication that the sector is insolvent.

At the end of 2012, financial net worth was minus €108 billion.  By the end of 2020, this had deteriorated to minus €175 billion.  Again, this is in contrast to the net lending position shown in the non-financial accounts.  It be could that there activities of a small number of large Irish NFCs are distorting the financial balance sheet of the domestic NFC sector – not dissimilar to the type of thing we see with US MNCs on other parts of the national accounts.

Is something mucking up the modified current account?

The current account of the balance of payments is an important indicator of imbalances in an economy.  The historical estimates of Ireland’s current account show significant deteriorations in the late-1970s and mid-2000s which were followed be periods when real growth turned negative (the shaded regions).

The concern with the latest estimates of Ireland’s current account is not that it shows deficits that are too large but of surpluses that are too large.  When published back in the summer the modified current account balance for 2020 was put at 11.5 per cent of GNI*, with comparable surpluses only recorded for the period of The Emergency (WWII to the rest of the world) in the early 1940s.

And the large surplus in the modified current account wasn’t just something that emerged with the COVID pandemic it is something that has been growing over recent years.

We can examine some of the sectoral developments underpinning the current account with the 2020 Institutional Sector Accounts which have been published by the CSO.  We will focus on the modified current account which strips out the impact of a number of globalisation effects (such as IP transfers, aircraft leasing and redomiciled PLCs). 

Some of the extra breakdowns provided by the CSO are useful in examining the improvement in the current account since 2013.  Here is a breakdown of savings minus investment by institutional sector with the combined sum representing the total economy current account.

The published figures are taken for all sectors bar that for foreign corporations (financial, non-financial and redomiciled PLCs).  The figures for foreign corporation is a residual to fit with the estimates of the modified current account, CA*.

The improvement from 2013 to 2019 can be attributed to two factors: an improvement in government’s position from large deficit in 2013 to modest surpluses in 2018 and 2019 and a switch in the impact of foreign corporations on CA* (through the residual component) from being negative in 2013 to positive in recent years.  It is not clear what has driven this change.

Clearly, there were a lot of sectoral changes in 2020.  The government moved into significant deficit while the surplus of the household sector increased considerably with these two effects largely netting out (in aggregate terms).

Domestic non-financial corporations have been in a surplus position in recent years while the impact of domestic financial corporations has always been relatively small though did become negative in 2020 for the time in the series which goes back to 2013.

A current account surplus of 11.5 per cent of national income is unusually large for Ireland.  Are there unusual factors which could explain it? Perhaps.  Corporation tax revenues have been soaring recent years and most of this is paid by foreign-owned corporations.  But the period that has seen Corporation Tax revenues go from €4 billion to €12 billion has been the CA* improve from a deficit of €1 billion to a surplus of €23 billion.  And of course the impact of the Corporation Tax on the current account will be tempered to the extent it is used to fund (import) spending elsewhere.

Is there an adjustment missing for the foreign-owned sector?  Possibly.  Though what that might be cannot be ascertained from this data.  And what about domestic companies?  Some of the balance-sheet developments for this sector have been extraordinary but it is not clear these have impacted the current account.

There is little doubt that Ireland is running a balance of payments surplus, and a significant one at that.  But the latest estimate of 11.5 per cent of national income is certainly close to the upper limit of the range of plausible estimates, and possibly even above it. 

Anyway, concerns about the precise level of an extant balance of payments surplus are a long way from the concerns raised by the balance of deficits of the late 1970s and early 2000s.  No one will be going on television to say we are living a way beyond our means.  Indeed, there is scope to increase spending if the things that people might like to buy (such as new houses) were to be made available.