Getting To The Core Of The Apple State Aid Decision

In this episode of Tax Notes Talk , Tax Notes contributing editor Ryan Finley breaks down the EU Court of Justice's recent ruling that Ireland granted €13 billion of unlawful state aid to Apple and its implications for state aid analysis.

Tax Notes Talk is a podcast produced by Tax Notes. This transcript has been edited for length and clarity.

David D. Stewart: Welcome to the podcast. I'm David Stewart, editor in chief of Tax Notes Today International. This week: another bite at the Apple. On September 10 the EU's Court of Justice ruled in European Commission v. Ireland that Ireland had granted €13 billion in unlawful state aid to Apple.

This highly anticipated ruling came as something of a surprise to many in the tax community since it was expected that the decision would align with previous state aid cases. So how did Apple and Ireland respond to this loss, and what impact does this have on future state aid analysis? Here to talk more about this is Tax Notes contributing editor Ryan Finley. Ryan, welcome back to the podcast.

Ryan Finley: Thanks for having me.

David D. Stewart: So first let's lay down some groundwork. What is state aid?

Ryan Finley: Yeah, so state aid is a unique feature of EU law based on the Treaty for the Functioning of the European Union, which is sort of the foundational document of the EU, specifically 107 of the treaty. And it generally prohibits any policies or actions by member states that distort competition by selectively favoring one business or a subset of businesses.

So in the tax context, something like an advance pricing agreement or tax ruling, any taxpayer-friendly deviation from the member state's normal tax rules would be illegal unless the deviation can somehow be justified by the sort of logic of the tax system. What happens if a member state is found to have granted state aid? Then they have to recover the aid from the beneficiary.

David D. Stewart: So when it comes to Apple and Ireland, what did the commission find?

Ryan Finley: The Apple investigation was part of a whole wave of investigations that were launched after the LuxLeaks scandal in 2014 that basically targeted these various advance pricing agreements granted to large multinationals. The Apple investigation stood out because the amount of the alleged aid was so much greater, about €13 billion.

But the specific circumstances of the APA — there was an APA granted by Ireland in 1991, and it was renewed in 2007 — involved these two Apple group entities, Apple Sales International, or ASI, and Apple Operations Europe, or AOE, and these two entities participated in a cost-sharing arrangement with Apple Inc., in the U.S., which gave them the rights to exploit the Apple group's core technology and marketing intangibles outside the Americas.

Now this was in a variation of the double Irish structure. So these entities, they were incorporated in Ireland, but effectively managed from the United States. So under the law at the time, they weren't tax resident in either country.

Each of them did have an Irish branch or permanent establishment that performed important but fairly routine functions like manufacturing, procurement, marketing, and support services. But other than these branches, the entities were essentially cash boxes designed just to participate in this cost-sharing arrangement.

The APAs, they specifically approved a method for attributing profit to each of these branches, and it's important to remember that this is a PE profit attribution case rather than a traditional transfer pricing case because that distinction — I think to many people's surprise — ended up being important.

Basically because these branches' activities were routine, the profit attribution method allowed the vast majority of each of these companies' income to stay with the head offices, as they called it, basically the rest of these companies other than their Irish branches, which essentially meant that it went untaxed.

So the commission looked at this arrangement and said there's really no way that a so-called head office that has no employees, basically just holds board meetings and signs off on documents, should be entitled to these massive IP returns from Apple's group IP. The way they saw it, the only part of either company that was capable of doing anything was its Irish branch, and therefore, the right part of the enterprise to attribute these cost-shared IP rights and the returns was the branch.

They presented their argument in terms of the authorized OECD approach, or AOA, to profit attribution, although we can debate whether they applied it correctly, but that's another important feature of the case. But they eventually issued a 2016 decision that said that based on their interpretation of the AOA, which they saw as sort of inherent in Irish law, even though it wasn't actually adopted by Irish law, that Ireland should have taxed all the profit reported by ASI and AOE going back to 2003 all the way through 2014, and not just the small amount that was attributed to the branches under the APA, so hence the €13 billion recovery order.

David D. Stewart: So what did the General Court decide on this case?

Ryan Finley: Ireland and Apple were not happy about this. They promptly appealed, and their arguments — they made a bunch of arguments, but I think you could say the two main ones are first that Ireland, obviously it could not possibly have had the AOA in its domestic law in 1991 because the AOA wasn't formalized until 2008. It would defy the laws of physics for them to have adopted the AOA. And in theory, the selectivity assessment for state aid is supposed to be based on domestic law. If the AOA is not in Irish domestic law, they argued, then deviating from the AOA can't be illegally selective state aid.

But they also argued that even if you do sort of equate the AOA with domestic law or however you get that into the analysis, that they misapplied it. Just a little background, the AOA, it attributes profit to a PE using the arm's-length principle, which is usually reserved for pricing transactions between separate legal entities, but it has to be adapted for the purpose of profit attribution.

And the adaptation is you have to essentially allocate the enterprise's assets and risks based on what are called significant people functions. And that's basically the activities that have to do with actively making key decisions about acquiring and managing assets and risks.

For Apple's cost-shared IP, it's pretty clear that all of those significant people functions happen in Cupertino, California, and for that reason, Ireland and Apple said the commission was wrong to attribute these IP rights to Irish branches that did not perform any of the necessary significant people functions.

So these are the arguments they came to the General Court with in the appeal, and they were successful at that stage. In a 2020 judgment, the General Court annulled the commission's decision. The court rejected that first argument that you can't even use the AOA because it wasn't part of domestic law. A controversial part of the judgment essentially said that the AOA is close enough basically to what Irish law required at the time. So it was reasonable for the commission to use that as its means of assessing selectivity.

However, the court agreed with Ireland and Apple that the commission misapplied the AOA and on the grounds that again, the significant people functions were carried out in the United States, not in Ireland. So on that basis, they ruled for Ireland and Apple.

David D. Stewart: So while all of this was going on, there were a bunch of other state aid cases that the commission was pursuing. So what was happening on the sidelines of this case?

Ryan Finley: Right. So between 2020 and earlier this month when the Apple judgment came out, there were two, one in particular, one very important decision by the Court of Justice in late 2022 in the Fiat case. And the Fiat case was distinct from a lot of these other cases in the sense that the commission actually won at the General Court level. In the state aid case involving Amazon, obviously in the case with Apple, the General Court almost in every single case sided with the member state and/or the beneficiary against the commission.

But not in Fiat. So it made it all the more noteworthy that the Court of Justice reversed the General Court in its 2022 judgment. And it really came out emphatically saying that the commission cannot come in and assess selectivity based on OECD guidance unless the member state's law expressly incorporates that guidance into domestic law.

This Fiat judgment was very clear that express incorporation of OECD guidance was necessary for it to enter into the selectivity analysis at all. And as recently as December 2023 in the Amazon judgment, the Court of Justice affirmed this, basically came down the same way, said that the European Commission has to base its state aid case entirely on the provisions of domestic law because under EU law, member states have autonomy in direct taxation, and unless they decide to harmonize their laws with the OECD, then the OECD guidance has no place in the analysis.

So almost everyone assumed that the commission was doomed in Apple as well because, as I explained that the commission's argument was based on the AOA, there's even less international consensus over the AOA than there is over the OECD transfer pricing guidelines. So the prospect of holding a member state to the AOA decades before the AOA was essentially formalized did not seem like a realistic possibility, but everyone was wrong.

It turned out that the Court of Justice came down in favor of the commission in this case. Its reasons for doing so were a little odd, basically because the General Court had said that the AOA and the arm's-length principle were a valid basis for the commission to assess selectivity, and because Apple and Ireland never specifically cross-appealed those holdings, that on the basis of res judicata, the Court would refuse to consider the arguments on that basis. So that throws out the whole argument that you're not allowed to sort of foist OECD guidance onto a member state. They said, "No, we're not going to consider that because you never filed a cross-appeal." Then after dismissing those objections, they said that under the AOA and the arm's-length principle, the allocation of these cost-shared IP rights has to be entirely based on the functions carried out within that individual enterprise.

So for ASI and AOE, you can't look at any significant people functions that take place outside of that entity. The IP rights have to either go to the head offices or the branches. There's nowhere else under this interpretation. So you have to basically ignore the reality that Apple Inc. in the U.S. performed all the significant people functions, and you have to find significance somewhere within ASI and AOE, whether it's there or not, because the Irish branches, even though they didn't do anything that would ordinarily qualify as significant people functions, decide what technology to develop or set global marketing strategy, at least they performed real marketing and manufacturing activities that were necessary to exploit the IP. So something is more significant than nothing.

And so the Court agreed with the commission that the IP rights were properly allocable to the Irish branches, and it followed from that that all the income should have been taxed in Ireland and that Ireland's failure to tax it was illegal state aid. And this was a big surprise, the outcome and the reasons for the outcome. It was especially, I think, striking that the Court did not even try to resolve the conflict with Fiat and Amazon, which were landmark cases decided within the last two years, just basically cited res judicata and then moved on. And it did not bother to really justify its interpretation of the AOA and the arm's-length principle, which I'd say is at the very least debatable.

David D. Stewart: Did the Court give reasons for why it was breaking so harshly with the General Court and even the recommendations of the advocate general?

Ryan Finley: Yeah, well, so the advocate general, the Giovanni Pitruzzella opinion was itself kind of a surprise in that it urged the Court to set aside the General Court judgment and then remand the case for further proceedings. Procedurally, if the Court of Justice, which is the highest court, if they're not in a position to issue a definitive ruling, then they can set aside the judgment and remand it, which is what the advocate general suggested they do. But he did make the same argument that under the AOA, the IP rights are trapped in that one entity and you have to either put it in the head office or the Irish branch, and the Irish branch is obviously going to win that contest.

So in that sense, the Court did file the opinion; it just went even further and said, "We don't need to remand it. This is enough for us to definitively rule and say the commission wins." They did not make any discernible attempt to reconcile the outcome with earlier precedent. We can speculate as to distinctions, but it's not easy to figure out what distinction between this case and Fiat and Amazon would justify the completely different outcome.

David D. Stewart: So there's a lot of money at stake here that needs to be collected from Apple. So how does that happen, and where does the money end up?

Ryan Finley: So the money has been held in escrow for years pending the final outcome of this case. So the money's sort of sitting there, and apparently, according to Irish government officials, it's going to take some time, as long as six months possibly, for the funds to be transferred from this escrow to the Irish government's coffers. So that's a whole process unto itself, but at the end of the day and however long it takes, the effect will be that the €13 billion plus whatever interest has accrued will have to be recouped by Ireland and paid over by Apple.

David D. Stewart: So how did Apple and Ireland react to the loss? Ireland clearly wanted to win this case, but they ended up with a lot of money anyway.

Ryan Finley: Well, apparently the €13 billion was not worth the sort of blemish on their national honor. They did fight it tooth and nail. They said they were disappointed with the outcome, but obviously they'll comply with the ruling. Apple also came out with a statement saying about what you'd expect: "We disagree with the ruling." They said what people normally say when they lose a huge case like this.

David D. Stewart: So on the other hand, what is the EU saying, and does this embolden them for future state aid enforcement?

Ryan Finley: Right. So Margrethe Vestager, the executive vice president of the commission who was the individual largely responsible for launching and spearheading these investigations, understandably welcomed the outcome. She said it was very encouraging. It encourages the commission to do more, both in terms of enforcement and legislative measures, but whether it will really help the commission in other cases, ongoing cases and future cases, is unclear.

We still don't really understand how this holding sits with Fiat and Amazon and when one would apply and when the other would. As I said, it was important that this was a profit attribution versus a traditional transfer pricing case because it's only in the profit attribution context that you have to pick one part of an entity to allocate assets to potentially whether there are any significant people functions in either part of the entity, or at least according to the commission's interpretation.

If you have, in the more common scenario, where you have separately incorporated subsidiaries with their own contractual obligations and their own asset registrations, you don't really have the situation that came up in Apple, and that seemed to be very important in the outcome.

The other thing I'd point out is that it seemed like it was very important that Ireland and Apple never formally cross-appealed the holding that the AOA was essentially equivalent to Irish domestic law. In a case where the member state and the beneficiary do file that cross-appeal, maybe the outcome turns out differently. And the other thing is that this exact situation will become less and less common over time because Ireland has changed its residency laws to prevent stateless entities. There have been reforms to national tax ruling practices that should prevent these kinds of controversial APAs.

There's also a trend toward harmonization and transfer pricing rules across the EU. There's a proposed directive that if it's ultimately agreed on, you're not really going to have these big interpretive differences necessarily, and maybe you won't have these cases either. But having said all that, it's still too early to say which of these differences, which of these factors are going to have what effects for ongoing cases. People are still trying to make sense of it because, frankly, the Court of Justice didn't really explain it.

David D. Stewart: So are there any cases out there that we should be looking to to start to figure out what this new world will be?

Ryan Finley: Yes. I think the cases to pay particular attention to are the ones involving Nike and IKEA. Those are not branch profit attribution cases. So it's subject to the caveat I mentioned before. They're more along the lines of the Amazon state aid case. So since you don't have a branch profit attribution situation, you have basically a cash box cost-sharing participant that sublicenses rights to local affiliates.

Maybe those cases still fall within the Amazon and Fiat precedent, or maybe not. Or maybe there's some procedural peculiarity like the failure to file a cross-appeal that may end up dictating the outcome. What happens will be closely watched because we really don't know what the effect of Apple's going to be on these in other cases.

David D. Stewart: Are there any long-term lessons that we should be taking from this?

Ryan Finley: I think the main lesson that I would take is that you never know what you're going to get from the Court of Justice in a tax case. And I guess I'd also say make sure you don't forget to file your cross-appeal. Other than that, I think the lessons to be drawn from this will not be clear for some time.

David D. Stewart: All right. Well, I think those are some good lessons to draw from, and Ryan, thank you so much for helping us understand this case and what it means.