What does the Cambridge Analytical scandal have to do with taxation? Nothing, or maybe something, or maybe a lot. In any case, last week, while Cambridge Analytica’s troubles dominated headlines, the OECD and the EU Commission both published their work on digital economy taxation.
Part 1: The Cambridge Analytica scandal
A friend shared this video with me a while ago. You can see Cambridge Analytica’s CEO saying loud and clear that they harvest data, profile everyone, and send targeted messaging.
Everyone is doing it for commercial purposes, but they are also doing it for political campaigns, based on psychographics. At the end of the journey, the big question is about that.
I was watching this mainly for my interest in the debate on digital economy taxation and was like, “uh, and is this acceptable”?
Disclaimer: I would like to stress that my interest is fully personal as I am not involved in any official work on this.
The video is from March 2017, one year old!
So, what exactly is the ‘scandal’?
Now, the scandal is that data were “kind-of-taken” from Facebook via an app developed by a university researcher (uh? For what purposes?) containing a psychological test that sucked up all your data. Uh?
But what about all the apps we have to play games, read newspapers, books, watch movies or porn, buy stuff, check directions, do sports, and, the one that has always amazed me the most, the torch app that helps you see in the dark that even gets access to your microphone, and so on.
So, the issue really is not Cambridge Analytica, it is the fact that we are not conscious of where our digital traces go, who has access to them, and for what purposes.
“So, the issue really is not Cambridge Analytica, it is the fact that we are not conscious of where our digital traces go, who has access to them, and for what purposes.”
Where do they get my data?
They have access to everything: the devise itself, chiefly the cell phone (uh? How long will it take before it becomes a chip to implant?), wearables, smart TVs, home devices, and now, connected cars, and so on.
For what purposes are my data used?
Profiling and micro-targeting at best, which means sending different messages to different people.
To be fair, when Barack Obama’s former campaign manager, Jim Messina, was talking about this back in 2013, he mentioned that in the first election they had 18 data analysts, in the second 165! This allowed Obama to have a direct relationship with the electorate.
We all watch a movie and then are offered a suggestion for another one based on what the algorithm-plus-the-big data has decided it is best for you. We buy something and we are told what others that bought the same have bought, so it really is common daily practice to use our data for commercial purposes and, in case you were wondering, we have all given consent to it, basically in exchange for a “free” service.
So now the “scandal” is actually about a commercial dispute between companies. The tweak is the fact that one company has put a great amount of personal data together and found ways, based on our psychological profile, to convince (manipulate?) us to take a decision.
Basically, this is the question: are psychographics good or bad and what protection do we have against it?
“Basically, this is the question: are psychographics good or bad and what protection do we have against it?”
And frankly, that has little to do with the results of Brexit, US elections, and, I tend to think (to be seen), Italian elections, too. Or, at least that should not pollute our judgement.
In other words, the answer should be the same if the strategy had been used by the Remain campaign (if you were against Brexit), Hillary (if you were a Clinton supporter), or otherwise.
It is a matter of principle, it is about the tool, not about the objective. Assume, for purposes of the analysis, that it is used to free kids abducted by terrorists that have brain-washed them to become fighters, just to make an example.
Do we really think that only Cambridge Analytica does this?
Prof. Kosinski from Stanford is one of the most recognised scholars on the topic and his list of publications is long, freely accessible online, with the most quoted piece being a 2013 article: Private traits and attributes are predictable from digital records of human behavior.
Part 2: Taxation of the digital economy
The OECD interim report
Rushing to get it out ASAP, the OECD’s interim report was published on 16 March. It honestly shows lack of consensus on what to do regarding the broader tax challenges of the digital economy.
Importantly, however, the US appears to be now ready to discuss permanent establishment and income allocation rules. And this is, in a way, big, very big, because the deal in BEPS was it would only deal with double non-taxation not with “source vs. residence”.
Also, some appear to be actually backtracking from the BEPS Action 1 report. Some countries say we can indeed ring fence the digital economy and have special income tax rules on it, plus, in the short run, some want a temporary measure outside the income tax space to redress the situation and focus on the specificities of large digital players (basically, user contribution and uh? data). The report hence puts forward a number of principles that such a new levy should respect.
So, the bottom-line here may well be that willing countries will likely move forward on a revenue-based levy while the work on rewriting the fundamentals of international tax rules will start soon within the Inclusive Framework and is expected to be finalised by 2020.
The European Commission package
The EU Commission published its package of proposals on 21 March. The package contains a draft directive on permanent establishment (PE) based on significant digital presence and new income attribution rules, a draft directive on a levy on revenues from certain digital services, and a recommendation to EU members to implement in their tax treaties with non-EU members the new definition of significant digital presence PEs.
Interestingly, the draft directive on significant digital presence PEs is meant, once implemented, to also override treaties between EU members.
The EU’s significant digital presence proposal covers the supply of digital services (broad definition) through a digital interface, when at least one of three conditions are met (calculation at group level, not entity level):
- Total revenues from digital services provided to local users in the country (proportion based on number of worldwide users) is above EUR 7 million.
- Number of is users above 100.000.
- Number of business contracts to supply digital services to local business is above 3.000.
Interestingly, sourcing rules for 1 and 2 are the location of the device that accesses the digital interface, with IP as proxy unless geolocation works better. At the same time, for 3, the source is the residence of the paying client (but only for companies, not others).
Income would be attributed to a significant digital presence PE based on the arm’s length principle, meaning that functions undertaken, assets used, and risks assumed via the digital interface are to be taken into account for transfer pricing purposes.
Activities related to data and users are, by definition, economically significant and shall play a role in analysing the DEMPE functions for intangibles. Profit split is the method to be used unless the taxpayer proves that another method is better. Research, development, marketing, data, and users are typical factors for split profits.
Revenue-based levy, aka the digital sales tax
This is conceived as a temporary, broad-brush measure to extinguish the fire.
- Scope: Limited to internet advertising, platform intermediation, data transfers. So relatively narrow scope.
- Sourcing rules Sourcing rules are as follows: ads where the device is when they are displayed; digital intermediation where the devise is when the transaction is concluded or where the account has been opened; data transfers where the devise the data comes from is located. So, it boils down to where you are walking while checking your smartphone! IP address is the favored proxy unless geolocation is more accurate. The sourcing rules are the most interesting part in my view — very, very, challenging.
- Thresholds: EUR 750 million global revenue and EUR 50 million from the EU in a calendar year. If you are below one or the other, you are out.
- Rate: 3%.
- Assessment: Self-assessment with tax return sent to one EU State (elected by the taxpayer as State of identification), indicating total revenues from EU and country-by-country allocation based on the sourcing rules, above.
- Collection: Pay to the State of identification and then the State will transfer to fellow EU states. Mimics the MOSS for VAT.
- Controls: According to national law. Good luck. This is the other very, very, interesting part.
Part 3: My (very, very personal) take
We need to increase awareness of our digital footprint and what is done with it, starting with our 5-year-old.
At the same time, we need to remember that so many good things come out of algorithms and big data and that there are differences between convincing vs. manipulating someone. Like many other things (the famous steak-knife), depending on the use, the tool may be a very good or a very bad one. The question is who should decide what is good and what is bad.
We need to increase awareness of our digital footprint and what is done with it.
On the tax debate, trying to connect the dots is not easy at this stage. BUT, the EU Commission proposal on digital permanent establishment and related income attribution rules is on the table, the OECD/G20 Inclusive Framework has a report to write for 2020, so a deal could be new rules on permanent establishment and income attribution rules to be signed in the multilateral convention in 2020, conditioned on a sunset clause for the digital sales tax.
“On the tax debate . . . a deal could be new rules on permanent establishment and income attribution rules to be signed in the multilateral convention in 2020, conditioned on a sunset clause for the digital sales tax.”
Interesting experiments in front of us.
The views expressed by the author are personal and do not represent the views of any affiliated institution