John Christensen ■ Taxing Google (and Facebook, Twitter, Uber and the rest)


solGuest blog from TJN senior adviser Professor Sol Picciotto

The announcement of the deal struck by Google with Her Majesty’s Revenue and Customs (HMRC) has aroused much speculation. The term ‘permanent establishment’ has even been mentioned in some august media outlets, such as the Financial Times. These are waters which have been deliberately muddied, so let’s see if we can clarify what is going on.

Can we tax them?

It’s true that until now Google has maintained that its UK subsidiary is responsible only for marketing, and that the subsidiary to which sales of advertising are attributed is Irish (though resident in Bermuda), and hence has no taxable presence (Permanent Establishment, or PE) in the UK. However, Google now seems to say that it is because the law has changed that it is making these new tax payments.

This presumably refers to the Diverted Profits Tax (DPT), enacted last year, which was aimed at ‘diversion of profits’ through arrangements designed to avoid having a permanent establishment. The statute’s detailed rules defining such arrangements are highly complex, and I am sure could be challenged legally. So far, Google and other such companies have not done so. Both they and HMRC prefer to sort these issues out in private if possible. Indeed, the DPT was designed to encourage this: it asked companies themselves to come up with a figure for the profits they had ‘diverted’, while threatening them with taxation at the higher rate of 25% if they failed to do so. This of course would depend on whether the HMRC took them to the tribunals, and above all whether such a case could succeed. This set the stage for the bargaining which has resulted in the settlement just announced.

Google has said it will pay more in the future, but not how much. As our regular readers will know, back in 2013 the G20 world leaders asked the tax specialists of the OECD to come up with reforms to international tax rules to ensure that multinationals could be taxed “where their economic activities occur and value is created”. The OECD designated this as its project on Base Erosion and Profit Shifting (BEPS), and the final reports were delivered to the G20 leaders last November. The BEPS Monitoring Group, a network of independent researchers who followed the project closely, issued this Overall Evaluation.

The BEPS outputs included proposals which would make it a bit easier to find that Google’s UK affiliate acts as a PE in the UK of the Irish/Bermudan subsidiary. This would depend on whether the UK entity can be shown to be regularly involved in the conclusion of contracts, or activities preparatory or auxiliary to sales. These proposals require changes to tax treaties, which will take at least a couple of years (which was one reason for the DPT).

However, it was already possible under existing law for HMRC to challenge the fiction that the “marketing” activities were nothing to do with sales. Since the settlement apparently includes payments for periods prior to enactment of the DPT, presumably Google has now conceded this. Other such companies may prefer to wait, and even to brazen out HMRC’s threat from the DPT.

How much tax?

Having a taxable presence is only part of the story, you then have to determine how much profit should be attributable to that entity. The DPT gave no guidance on this, as we pointed out at the time, but left it to companies to bargain with HMRC. However, while playing their cards both will also be looking over their shoulders at the kibitzers, especially the USA. Google would prefer that any tax it pays to the UK should be deductible as a credit against its taxes in the US. The US Treasury has not yet stated whether it will allow a tax credit for tax paid under the DPT.

Of course, at present most of Google’s revenues from the UK and many other countries around the world, channelled through Ireland to Bermuda, are not taxed anywhere. But the tax liability is only “deferred” as long as they are kept offshore – the US has not given up its claim to the ultimate tax rights. At some point, perhaps under President Trump, the US may reform its corporate tax laws, and require these firms to repatriate such profits, or more likely entice them to do so by offering an amnesty and/or a low tax rate. In the meantime, they are part of Google’s war-chest, enabling it to finance, or more likely acquire, new innovations.

The BEPS project also tried to deal with this, by reforming the OECD Guidelines on Transfer Pricing. Although not formally binding, these are applied in practice by tax authorities all around the world, even non-OECD countries. The revisions have expanded the Guidelines from 370 to closer to 500 pages, which provides plenty of lucrative work for transfer pricing specialists. Essentially, however, they allow tax authorities to adjust the pricing of transactions between MNE subsidiaries to reflect the “functions performed, assets deployed and risks assumed” by each. In fact, the BEPS project did not finally decide the criteria for attributing profits to a PE, this is part of its continuing agenda, but the result is likely to adhere to this functions-assets-risk approach applied to subsidiaries.

Two points should be made about this approach. First, the tests to be applied are highly subjective, leaving plenty of room for discretionary judgements, guesswork, and bargaining. However, this process has to be dressed up to appear technically justified. This is where the army of transfer pricing specialists comes in, deploying econometric techniques to provide a spurious precision. In practice even these techniques generally result in a range, often wide enough that choosing the appropriate end can wipe out all the profit (as shown by Starbucks).

More importantly, this approach attempts to maintain the fiction that MNE affiliates should be treated as if they were independent entities, dealing “at arm’s length” with each other. So both Google’s UK “marketing” affiliate and even its sales operation are considered as if they were totally independent of all the other entities in the Alphabet soup.

A main aim of the MNEs and their tax advisers was to ensure that the BEPS project did not end this fiction. In this they were supported especially by the US negotiators, who inevitably succeeded in holding the line to defend the arm’s length method. However, although tax specialists ritually pay obeisance to this concept, it has become entirely nebulous. A rare occasion when this was frankly recognised occurred during the consultations in the BEPS process on “special measures”, when the Chinese delegate (Xiaoyue Wang, deputy director-general in the International Taxation Department) boldly stated that “the arm’s length principle does not work”. Subsequently, China has put forward new transfer pricing rules which, while genuflecting to the arm’s length principle, starts to introduce new criteria for value creation.

So the various tax authorities and MNE tax advisers are engaged in complex games, tweaking the rules and playing their cards to their best advantage, while also not wishing to knock the card table over. Although the UK’s DPT seemed an aggressive move, a signal was sent to the US that it would be applied cautiously, by projecting that it would recover only £300m in a full year, from all companies. Google has now settled with the UK, no doubt hoping that the deal will set a benchmark for other countries too, which also explains why it was publicly announced rather than kept confidential. However, it also seems formulated as an opening offer, to test out public opinion, since the basis of calculation has not been explained, nor how it would apply in the future.

On what basis?

Nevertheless, the cat is out of the bag. The debate is now about how much Google should pay in each country, not how much should be attributed to this or that subsidiary. Most people share the feeling that a payment of some £30 million in tax on sales attributable to the UK of some £4.6 billion seems low. However, some commentators aim to be more “realistic”, For example, Maya Forstater suggests that the profits attributable to the UK should indeed be based on the functions it performs in the UK. She accepts that these are limited to marketing activities., and that “the vast majority of the cool stuff at Google goes on in Mountain View California”. This essentially goes along with the conventional view, and not surprisingly supports Google’s deal with HMRC.

But hold on, what is the “cool stuff” done by Google, and where actually does it take place? Google itself claims that the bulk of the value it creates is due to the software behind its search engine. Conveniently, and no doubt following the best professional advice, Google transferred the rights to this software to an Irish affiliate more than a decade ago. This was before it had begun to generate the enormous revenues, and hence the transfer was at a low price. In 2006 the US IRS accepted that the price was reasonable, and that the “risk” was transferred. But for Google there was no actual downside “risk”, only the upside potential for profit – which has been channelled to Bermuda. Now, with the proposed changes to transfer pricing rules, Google will have to show that the Irish entity actually “controls” the software development, and manages the risks.

However, this ignores how this software evolves and what it actually does. Certainly, much is due to the efforts of the software engineers, who are probably mainly based in Mountain View, but many of whom could be anywhere in the world. But the software is really more like an ecosystem, which works with and draws in its users. Every search contributes data to the search engine; every message in Googlemail is harvested for language data for Google Translate; users of Maps contribute location data. These contributions may be regarded as relatively “passive” but users also actively contribute content in other Apps provided by Google and other internet companies. These range from reviews of products and services (Amazon, TripAdvisor), to personal details and photos (Facebook), to videos on Google’s own YouTube.

A global approach – but which?

It should be clear that Google and the others are global companies, and that all of their activities interact, so that their enormous profits result from the synergy of the whole. It is not possible to attribute particular parts of the profit to specific discrete activities. That is why we have long advocated that multinationals must be treated in accordance with the economic reality that they are unitary firms.

This principle is in fact accepted by most independent international tax specialists. However, they have differed on the most practicable, effective and fair system to apply. Some, notably Michael Devereux, Alan Auerbach and others, advocate a “destination based corporate tax”. This would actually allocate the tax base entirely to the country of destination of sales, so that the UK would indeed tax the £4b of Google’s UK sales.

Others favour residence-based worldwide taxation, which would mean the home country of each MNE applying its taxes to the consolidated worldwide profits of the MNE group as a whole, but allowing a credit for foreign taxes paid. Supporters of this are mainly US commentators, such as Ed Kleinbard, Jeff Kadet, and Cliff Fleming-Robert Peroni-Stephen Shay. This system could, indeed, have been achieved within the BEPS process, if the Action on controlled foreign corporations (CFCs) had accepted the arguments for full-inclusion, i.e. treating all affiliates as CFCs. This approach could even be adopted now, if the politicians in the EU and the US could be persuaded to do so which, however regrettably, is unlikely.

The version we have put forward is unitary taxation with formulary apportionment. This would apportion profits of integrated MNEs based on factors reflecting the actual activities in each country. In principle these would balance production and consumption factors, so should use payroll costs and sales revenues. So Google’s software engineers, whether based at Mountain View or elsewhere, would be counted based on their no doubt ample remuneration packages, while the contributions of users, including the data which fuels advertising sales, would also count. This doesn’t necessarily require the same formula for all economic sectors, suitable factors could be identified for different business models. The system could develop pragmatically, by building on the profit-split method, which is already accepted in the OECD Guidelines.

In fact, the specialists on the BEPS project, in its Task Force on the Digital Economy (TFDE), also accepted that a holistic approach is needed. The TFDE report recognised that digitalisation means that MNEs have come ‘closer to the economist’s conception of a single firm operating in a co-ordinated fashion to maximise opportunities in a global economy’ (para. 232). It also accepted that this entails a ‘substantial rewrite of the rules for attribution of profits’ (para. 286). The report canvassed several possibilities, including ‘fractional apportionment’, ‘deemed profit’ methods, a withholding tax on digital transactions, and an ‘equalisation levy’.

However, it could not agree even to recommend any of these, and left it to states to decide. Meantime, the Task Force will continue, aiming to produce a report in 2020.

Plainly, we cannot wait that long. Google’s domination is based not only on its technological dominance, which is deserved, but also on its tax avoidance, which damages fair competition. The cash mountain it has built up enables it not only to fund its own innovation, but also buy out the small start-up outfits, which are often much more innovative.

The UK government should not hide behind HMRC and its timid application of the existing minimalist consensus. We need clear criteria for determining the apportionment of the profits of firms such as Google, based on their activities and value created in each country.


Related articles

Comments • 5

  • January 28, 2016 - 5:28 pm

    Hello Professor Picciotto, Prem Sikka and others at Tax Justice Network

    Nice to ‘meet’ you and thank-you for the mention.

    I agree that the Google case demonstrates the challenge for national and international tax systems in working out how to tax digital trade. Rules and mental concepts shaped by thinking about trade in widgets are clearly not adequate.

    Just to be clear in the post you quote I am not ‘supporting’ Google’s deal with HMRC (or HMRC’s deal with the Google) but questioning whether the figure that has been widely quoted that it represents a 3% tax rate is credible or meaningful. Perhaps £30 million in tax on sales is too low. But £200m for the UK seems awfully high based on any kind of distribution which reflects economic activity.

    As I understand the £200m calculation and the view that Google is paying a 3% ETR (which Prem Sikka and other proposed and which TJN put forward in its letter to the European Commission) is based on an approach closer to something like Professor Devereux’s “destination based corporate tax” than the type of formulary apportionment approach that the tax Justice Network generally proposes.

    While figuring out how to tax Google (and Uber and Facebook etc…) may be particular complex, the underlying tension between national interests and international fairness is a more basic question: Everyone cannot tax 100% of the profits of multinationals, however much it might look attractive.

    I cannot understand how the conclusion that the Google should have paid all profits related to UK revenues only in the UK (which seems to be implied by the £200m / 3% calculation) can be reconciled with the position that Google should have paid profits in line with economic activity.

    If it was possible to set up an international tax system where countries could tax 100% of the profits earned by multinationals through serving consumers in their country (as in the case of the Maugham/Sikka/Murphy approach for Google in the UK) without countries giving up on taxing the profits earned by multinationals undertaking economic activity to serve consumers in other countries – then there wouldn’t be a problem of international tax at all!

    Of course none of these back-of-an-envelope calculations are likely to be ‘right’ and I am not asking you to defend it on that basis, but please could someone take up the challenge of explaining how you reconcile the principles behind the 3% calculation with principle that taxable profits should be distributed between countries in proportion to the scale of economic activity.


    • Sol
      January 29, 2016 - 12:38 pm

      Dear Maya

      Thanks for your reply. We would be delighted to propose a more accurate apportionment of Google’s profits, based on principles which we support, if we had the data. Unfortunately we don’t. Although the BEPS project agreed to a scheme for country-by-country reporting (which we at TJN have advocated since we began), the reports will be given only to tax authorities. So the secret bargaining I described in my blog will continue.

      May I in turn challenge you? Do you agree that multinationals should be treated as unitary firms for tax purposes? I think you are wrong to say that we need new rules only for the digital economy. In fact, unitary taxation with formulary apportionment was introduced very early in federal or confederal systems, such as Canada, the US, and Switzerland. Of course, it’s not a panacea, but in my view it offers a rational basis for dividing the tax base of integrated multinationals. As I pointed out, other unitary approaches are also possible. They could also be combined, e.g. the EU could adopt its CCCTB together with full-inclusion CFC rules.

      The main reason given for rejecting UT with formulary apportionment is that countries would not be able to agree on the apportionment factors. In fact, experience shows that states need to balance the effects on tax revenues with the impact on inward investment. This tends towards a higher weighting of sales in the apportionment formula. So applying Google’s overall profit margin to its UK sales revenues is not a bad starting point. I also like Richard Murphy’s point: “British ads sold by British sales people aimed at British customers on a UK website that get value by clicks on British computers are not exported to the UK: they are made here.” – See more at:

      But it’s clearly time to grasp this nettle. The issues are actually not complex. The international tax system has become highly complicated precisely because it has been built on a fiction, that multinationals consist of an agglomeration of independent entities. Maintaining this fiction provides very well paid work for armies of tax advisers. Revenue authorities then have to hire their own specialists. All of these people have invested a lot in their expertise, and their daily practice generates a view of the world which is hard to shake. A few have now begun to accept that a new approach is needed.

      The BEPS project tried to keep the patient alive with bandages and codeine. We now need to look for a cure.

      • January 31, 2016 - 5:09 pm

        Dear Sol

        Thank you for your reply. You have not answered really answered the question though. Perhaps Prem Sikka or John Christensen could (apologies if I am asking in the wrong place):

        My Question: TJN says Google says its analysis suggests Google is paying a corporation tax rate of 3% compared the profits that should be part of its UK tax base. This is based on a methodology which allocates 100% of estimated profits on UK sales to the sales destination. How do you reconcile this with the principle that taxation should be where economic activity takes place, or indeed with current law?

        In response to your ‘ challenge’, although I would perhaps call it a question: How do I feel about unitary taxation/ formulary apportionment? I am not a tax policy expert, and those much more qualified, including yourself, have argued the pros and cons of the different approaches. I am not yet convinced that it is either possible or better (I.e. less arbitrary/ less vulnerable to gaming). In other words I don’t know.

        What I do know is that often wishful thinking can creep into analysis and public understanding of the case for and against different policy proposals. The idea that Google is running rings around the current law, and getting away with paying 3% of what it really owes suggests that trying to fix the current system must be hopeless. The idea that a “fair” apportionment would mean the UK would get six times more from Google than what the current system delivers suggests a compelling case for switching to such a new formulary system. But if the 3% figure is nothing to do with either current law, or the formulary apportionment proposal then these are misapprehensions which TJN could help to clarify.

        I agree with you that taxing globally integrated companies is a difficult and increasing challenge. In fact I think the idea that Google is in the business of serving “British ads sold by British sales people aimed at British customers on a UK website that get value by clicks on British computers” is a fundamental misunderstanding of how Google/Youtube and other pure-play digital companies work. (I have written more here, thinking about Google’s business model )

  • January 29, 2016 - 8:46 am

    Thanks Solicitors. Good article. Just a couple of points. Firstly not all TP advisers operate with spurious precision…Secondly I think the US have proved that FA is even less reliable and open to abuse than the arms length principle. The solution therefore for highly integrated businesses at least is residual profit split. We are using this already in our post BEPS TP work and it seems to get us to do reasonable conclusion most of the time which is also acceptable ( with some negotiarion) to HMRC.
    Happy to discuss further
    Hope you are well.


  • Harry
    March 2, 2016 - 4:22 pm

    Professor Picciotto,

    I was wondering whether you could help me to get my head around a problem that I’m finding somewhat perplexing…

    Assuming the CCCTB were in place at this present point in time, what would be the implications for a firm like Uber? Would its profit generation within the EU be caught by the allocation formula’s sales factor, putting an end to its diversion of income away from the Member States where it conducts its ride-sharing activities? What I am finding most confusing is the fact that, having sent all income derived from Member States to the Netherlands, Uber is then channelling its revenue to Bermuda, which will of course be a third country for the purposes of the CCCTB… So I was wondering how this structure would look under the CCCTB? Would Bermuda be entirely removed from the equation, and MSs allocated a fair share of the company’s activities?

    I think that Uber makes for particularly interesting analysis of the CCCTB, given its lack of employees in the traditional sense, and physical assets owned, but I’m finding it hard to understand the exact implications of the proposed system for the firm…

    Many thanks (especially for inspiring my current research project!),


  • Leave a Reply

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.