Go back to menu

Missing the target?

The surprising scope of the proposed new EU digital services tax

21 March 2018

The European Commission has just published a proposal for a new pan-EU Digital Services Tax (DST). The proposal would apply a 3% tax on the gross revenues of a wide range of businesses. The intended target is primarily the large US digital businesses, but the scope of the tax is surprising. Many large digital businesses would not be subject to the DST, but many more traditional businesses would be, both B2B and B2C. In particular, all internet advertising sales by large businesses would be subject to the DST - and the broad scope of the DST charge on "multilateral interfaces" means that it may also apply to many financial services. This briefing summarises the new proposal and its impact on business. We also identify a potentially serious impact of the DST on the privacy of internet users.

Why now?

The last few years have seen increasing controversy over the way in which many large US-headquartered digital companies are taxed on their worldwide non-US profits. 

Typically they pay very little tax. Often that results from carefully structured arrangements involving holding intellectual property offshore. In some cases it results from the groups in question making very little profit. 

But there is a more fundamental issue. The OECD international tax framework for taxing corporate profits means that companies based on one country are generally only taxed in another if they have a physical presence there (a "permanent establishment"). The digital economy means that a company can undertake very significant activity without such a presence, and therefore escape taxation. In principle this should mean that the large US digital companies would be taxed predominantly in the US and not the other countries where they undertake business. However in practice the US tax system has historically enabled companies to pay only very limited tax in the US. The long term implications of US tax reform remain unclear, our initial view is that it seems unlikely to fundamentally change that dynamic. 

This has attracted considerable political, media and popular attention, and the Commission clearly feels under pressure to do something about it.

However the Commission's immediate proposal is not an anti-avoidance rule to counter tax planning by digital businesses – they are proposing an entirely new tax targeting digital. That will be controversial. The OECD concluded in 2015 that, because the digital economy is increasingly becoming the economy itself, it would be difficult, if not impossible, to ring-fence the digital economy from the rest of the economy for tax purposes. Yet that is, to a great extent precisely what the DST does.

What is the proposal?

The Commission are proposing an "interim" Digital Services Tax: a 3% tax on the gross revenues from providing any of the following services to users in the EU:

  1. advertising: the making available on a "digital interface" of advertising space for advertising that is aimed at users of that interface; 
  2. multilateral interfaces: the making available to users of a "multi-sided digital interface" which allows users to find other users and to interact with them, and which may also facilitate the provision of underlying supplies of goods or services directly between users; and
  3. selling user data: the transmission of data collected about users and generated from users' activities on digital interfaces.

The Commission estimates the DST would raise €5 billion, which would be split between EU Member States pro rata to the number of users in each Member State.

The tax would only apply to companies or consolidated groups which have total global revenues over €750 million and total EU digital revenues, taxable under the DST, over €50 million. 

The liability to pay the tax falls on the business providing the service – although they may of course pass on the cost to customers. Where the service consists of advertising, and the advertising is placed by an entity that does not own the digital interface in question, then the tax liability falls on that entity (e.g. the online advertising platform).

An important point is that, where a marketplace connects users who subsequently transact, the transaction between the users is not generally subject to the DST: only the gross revenues of the service provider. It may, however, be that in some cases a service provider books the gross transaction volumes as revenues – in such a case the full amount would potentially be taxed.

Why is the tax interim?

The Commission's intention is to move towards a comprehensive long-term solution, where countries tax foreign digital businesses on the basis they have a "virtual permanent establishment". The DST is intended to be an interim measure that is repealed at the point that the long-term solution comes into effect.

Interim taxes have a habit of remaining in place longer than expected - UK income tax was introduced as a temporary measure to fund the Napoleonic Wars.

There are good reasons to think that will be the case here. Any change to permanent establishment rules requires amending Member States' double taxation treaties and, in particular, their treaties with the United States (given that is where the most significant digital businesses are based). We see no sign that the US Administration would agree to such amendments – and in recent times the Senate has been unwilling to ratify even uncontroversial tax treaties.

Hence our assumption is that, if introduced, the DST would remain in place for the long term, with the "virtual permanent establishment" proposal making little headway. 

Are there exemptions?

The key exemptions are for:

  • "Multi-sided digital interfaces" with the sole or main purpose of supplying digital content, communication services or payment services to users. Hence video and audio streaming services are not subject to the DST (unless they make revenue from advertisements), and neither are cash payment services.
  • The supply by a trading venue or a systematic internaliser of any of the services as referred to in points (1) to (9) of Section A of Annex I to Directive 2014/65/EU. So most trades on EU regulated exchanges would not be subject to the DST. But this raises the question of the treatment of other similar services which are not EU regulated – for which see further below.
  • The supply by a regulated crowdfunding service provider of any of the services referred to in points (1) to (9) of Section A of Annex I to Directive 2014/65/EU, or a service consisting in the facilitation of the granting of loans. Regulated crowdfunding and peer-to-peer lending would therefore not be taxed – however other similar services could be. Again, see further below.
  • The transmission of data by a trading venue, systematic internaliser or regulated crowdfunding service provider.
  • Services provided between entities within a single consolidated group.
What are the intended targets of the tax?

An early Commission draft specifically identified businesses that would be taxed under the "advertising" concept: Facebook, Google Adwords, Twitter, Instagram and "free" Spotify, as well as under the "multilateral interface" concept: Airbnb and Uber.

It is reasonably clear that these businesses will indeed be caught. By contrast, a digital retail store selling physical or digital products directly to consumers or business will not be taxed. That is a surprising result given that much of the political and media focus has been on businesses of this kind. The Commission justify this on the basis that the digital interface is simply used as a means of communication – but that seems to understate the transformational nature of digital retail.

What other businesses would be caught by the "advertising" concept?

Any business which sells online advertising will be within scope of the tax, if operated by a company/group which meets the financial thresholds. 

It follows that, for example, large newspaper groups will now be taxed on their internet advertising revenue. Indeed this will be the case for all businesses relying on the common internet and mobile app model of a "free" product which is funded by advertisements (subject again to the financial thresholds).

A video or audio streaming service which runs on a subscription-only basis would not be taxed – but a service funded (in whole or part) by advertising would be.

Commercial broadcast television, funded by advertisements, is not taxed – but exactly the same service broadcast over the internet would be subject to the DST.

These seem, at least to us, surprising results. Many of those taxed by the "advertising" head of the DST are already fully subject to tax on their corporate profits. What is the policy objective of imposing an additional tax? 

The Commission's explanatory notes say that the business models taxed by the DST are those which would not be able to exist in their current form without user involvement. That is clearly true for Facebook – but how is it true for an online newspaper carrying advertisements? Or, at least, how is it any more true than for a printed newspaper that carries advertisements?

But this result may be an inevitable consequence of choosing to tax Google's main revenue stream – advertisements. Once that decision was made, then either a distinction had to be made between Google and other businesses (difficult, particularly given the wide range of Google's activities), or the Commission had to accept that other businesses would be caught. 

What other businesses would be caught by the "multilateral interface" concept?

The scope of the "multilateral interface" concept is novel:

"the making available to users of a "multi-sided digital interface" which allows users to find other users and to interact with them, and which may also facilitate the provision of underlying supplies of goods or services directly between users"

Whilst not entirely clear, we would say from the overall scope of the draft Directive that the "finding" and "interaction" with other users could occur behind the scenes – with users potentially unaware of each other's identity or even their existence. 

If that is right then a wide variety of services could be taxed as "multilateral interfaces", particularly financial intermediaries and disintermediation services:

  • Third country trading venues / dealers: the definitions of "trading venue" and "systematic internaliser" only cover EU venues and firms. Some EU firms are members of third country trading venues or other securities and derivatives trading platforms, they may be within the scope of the DST. Given the volumes traded on these platforms, the impact could be very significant.
  • Other trading venues such as betting exchanges and cryptocurrency exchanges.
  • Order routing services: Would an investment firm which provides a service which routes client orders to trading venues, affiliates or third parties for execution be considered to be making available a multi-sided digital interface? 
  • CCPs and other post-trade service providers (e.g., providers of portfolio compression services) on the basis they match "users" with each other.
  • Peer-to-peer marketplaces of all kinds – so, for example, crowdfunding which does not relate to regulated financial instruments would seem to be taxed.

It is unclear which, if any, of these is intended to be subject to the DST. 

The design of the DST creates an immediate additional cost for any business which is seeking to intermediate between transactions that are ordinarily bilateral. To take an obvious example: a traditional taxi firm, engaging drivers and supplying them to customers, is not subject to the DST. A platform which introduces customers to drivers is subject to the DST. 

How is user location determined?

The concept of user location is central to the DST. The DST would only apply in respect of the relevant services when supplied to users located in the EU. And then DST revenues would be allocated between Member States pro rata to users' locations.

Under the proposal, a user's location is determined by looking to the location of a user's device, as determined by their IP address or, if more accurate, any other method of geolocation. In principle this is simple enough – and it should be reasonably straightforward for marketplaces, and other services where users identify themselves and log in.

For services delivering advertising - for example search engines and newspapers - the concept quickly becomes problematic from both a technical and a privacy perspective. There will need to be considerable further analysis undertaken by all those affected, but our immediate thoughts and questions are:

  • Taxing by reference to the location of users relies upon identifying individual users. But how can this be achieved, when it is common for users to have multiple devices, each accessed over multiple sessions over the course of a year?
  • Technically it is straightforward to track across multiple sessions on one device (using "cookies"), but some businesses make a point of not tracking users in this way – are they to be told this is an unacceptable business model? 
  • Some businesses operate across multiple domains – they may technically be unable to track users across those domains (as this requires "3rd party cookies", which are blocked by many browsers). 
  • Whilst users can be tracked without cookies (for example using "device fingerprinting") this raises significant privacy concerns from an ethical and legal perspective.
  • Even more problematic is the question of tracking users across multiple devices. Yet if this is not done, then the DST would not be taxing users at all – it would be taxing devices. Member States where users have multiple devices (i.e. typically wealthier countries) would then receive a greater allocation of revenues than countries where users have fewer devices.
  • What if a new version of a major browser were to disable user tracking altogether? 
  • The "user" concept requires keeping extensive data on users and their location, and retaining it - potentially for many years. That heightens privacy concerns.
  • "User" is defined to mean an individual or business. But which? When a person accesses google from a computer owned by their employer, who is the user? This could make a very significant difference to the count of users, and hence DST computation and allocations.
  • Most fundamentally, the mandatory use of geolocation/tracking data is contrary to the General Data Protection Regulation (GDPR) principles of data subject control and authority over personal information. 
  • In principle the DST Directive can override GDPR, but that seems to us to be a potentially very controversial approach for the Commission to take.

It is disappointing that the Commission does not seem to have taken account of privacy considerations. The word "privacy" does not even appear in the DST documentation.

Why tax gross revenues?

The DST represents a radical departure from most existing corporate taxes, as it applies to a business' gross revenues. VAand other similar taxes are excluded, but no deduction is permitted for expenses. 

The decision to tax on a gross basis is not justified in the Commision's documentation. We can, however, speculate as to the reasons:

  • Some high profile digital companies have very low profit margins, or are even loss-making (e.g. Uber). There is perceived to be a popular demand that such companies be subject to new taxes – and clearly a tax on net profits would not achieve that.
  • There is a view that any tax on net profits would be contrary to existing double tax treaties. The Commission's explanatory memorandum refers to " several legal constraints" – this may be what they are referring to. However, if so, we are unsure that view is correct as a technical matter.
  • It might be thought that taxing net profits creates the possibility of tax avoidance by "profit shifting" or other arrangements which artificially reduce the net profit. This is a legitimate concern where taxes are applied on an entity-by-entity basis. However where taxes are applied against the net profits in a consolidated group accounts, it would ordinarily not be possible to artificially reduce the net profit in that manner. If there were specific concerns around, for example, shareholder debt, then appropriate anti-avoidance provisions could be introduced. We would therefore not agree that a gross revenue tax is necessary to prevent avoidance.

An obvious consequence is that a highly profitable company will be taxed at a lower effective rate than a profitable company. Google, with gross profit margins not far short of 50%, may easily be able to absorb the cost of the DST. A hypothetical Google competitor, either loss-making or with slim margins, would find that more difficult. Less hypothetically, Airbnb, with gross profit margins closer to 2%, may feel it has no choice other than to pass the cost on to consumers. Indeed the history of gross revenue taxes in Europe – insurance premium tax and air passenger duty in particular – suggests that such taxes are often passed on.

How do the €750 million and €50 million thresholds work?

The DST will only apply to entities or groups with total annual worldwide revenue above €750 million (from all business lines) and total annual revenues taxable under the DST of €50 million (i.e. EU digital revenues). "Groups" are determined on the basis of accounting consolidation.

The intended effect is to prevent start-ups and smaller businesses being subject to the tax.

The actual effect seems to us to create significant distortions.

Take two identical digital businesses, each with a €50 million turnover. One is a standalone business; the other is owned by a large traditional business with a €750 million turnover. The first is outside the DST; the second is taxed. The principled justification for this is unclear – but there is obvious potential for economic distortion.

How are other countries likely to react?

There is no way to avoid the conclusion that the businesses subject to the DST will mostly be headquartered in the US. Indeed there are remarkably few European internet businesses with a €750 million turnover.

The US has already indicated that it is unhappy with the EU proposal. Other countries whose businesses are taxed – and China in particular – may also be less than supportive.

The difficulty is that, for a century or so, most European countries have had tax systems that levied tax based on the concept that the headquarters of a company is a significant source of its profit - so that a European company selling to the wider world would be taxed in Europe. Many other governments are likely to see the EU's proposal as a convenient rejection of this orthodoxy for one particular sector - where EU companies have been conspicuously unsuccessful - whilst maintaining it in others.

These countries could react with their own taxes targeting prominent EU businesses. The US could even utilise its long dormant "Section 891" power, which allows the president to double tax rates for foreign citizens and businesses of any country the administration considers to be discriminating against US companies. In the past that would have been thought far-fetched but, in the current environment it cannot be ruled out.

How will the DST interact with existing taxes and treaties?

The DST will in general not be impacted by existing double tax treaties – it is simply outside the scope of the standard double tax treaty model.

The Commission say that Member States should permit companies resident in a Member State to deduct their DST cost from their profits for local corporate income tax/corporation tax purposes. That means that small countries where many digital businesses are based (such as Ireland) could potentially suffer a net revenue cost from the DST.

Could the DST be subject to legal challenge?

It is not entirely clear that the DST is within the scope of the EU's competence over indirect tax. Article 113 of the Treaty for the Functioning of the European Union permits EU legislation on indirect tax to the extent necessary for the functioning of the internal market and to avoid distortions of competition.

The DST documentation asserts that eleven Member States are planning to implement (or have implemented) unilateral measures, and that this risks fragmenting the single market and distorting competition. However "fragmentation" in this context is just another word for different Member States having different taxes (as they are permitted to do: only VAis harmonised). No case has been made that competition is distorted.  

What happens next?

The DST will need unanimous support from Member States before the Directive can be adopted. It is not clear at this point whether that will be forthcoming. 

If unanimity cannot be achieved then it is possible that nine or more Member States could proceed together under the "enhanced cooperation procedure" (ECP). However whether it is appropriate and lawful to use the ECP for a measure that will impact the cross-border provision of services is, in our view, subject to considerable doubt. 

Even if the use of the ECP is in principle lawful, it is doubtful whether an ECP DST Directive could override the GDPR, calling into question the fundamental DST requirement to track the location of users. 

What should business and stakeholders be doing now?

Under the EU "better regulation" initiative, the draft Directive will be subject to an eight week consultation period. 

During that time, any business concerned they could be inadvertently affected may wish to consider their position and make representations. Any stakeholders with wider concerns around privacy may also wish to input into the process.