No, that’s not the math problem Judge Cary Douglas (“C-Dough”) Pugh has to solve in Facebook, Inc. and Subsidiaries, 167 T. C. 12, filed 9/29/26. That’s the twenty (count ’em, twenty) lawyers for Facebook and Subs and thirteen (ditto) for IRS, to ascertain the correct calculation of a sixteen year old Cost Sharing Arrangement per Reg. Section 1.482-7T.
For the backstory, see my blogpost “Facebook Faceoff Draw – (Sort Of),” 5/22/25.
Sent to the Rule 155 beancount, the 33 (count ’em, 33) lawyers couldn’t agree how to do it. I’m shocked…shocked. Aren’t you? And of course they don’t agree with Judge C-Doug Pugh’s opinion; this fight is only about how to compute the number in the decision from which one or both are going to appeal. So Judge C-Doug Pugh scraps the Rule 155 beancount and, contrary to my oft-repeated jibes, does the work and shows she can add.
The deal was the swap of Facebook’s IP for everywhere in the world but US and Canada to its Irish sub. What the Irish paid Facebook for these goodies (and what Facebook had to recognize as income therefrom) is the name of this tune. The good news is that the record is sufficient for the purpose; no more dueling experts.
“The largest issue separating the parties is the structure of the contingent annual royalty payments from Facebook Ireland to Facebook US. As we observed in Facebook I, respondent does not dispute that the royalty payments should be contingent and annual. But the parties disagree over whether respondent’s aggregate 6.29-year flat-rate royalty should be considered reasonable or instead we should leave undisturbed the form adopted by petitioner in the PCT [Platform Contribution Transaction] report, that is, multiple royalties over different periods (or instead adopt a proportionate increase as petitioner proposed). This dispute is legal in that the parties disagree over what the regulations require and factual in that respondent argues that Facebook did not satisfy those requirements.” 167 T. C. 12, at p. 11.
The parties do agree that contingent annual royalty payments should be the form the PCT takes.
“To compute the royalty percentage respondent divided his PCT Payment by the sum of the NPVs of the ROW revenue he projected over his proposed royalty period (2010 to 2016).12 Petitioner argued that we should respect the separate royalties over different periods in the PCT report which formed the basis for its tax return position. And petitioner challenged respondent’s aggregate value and urged us to adopt the separate valuations that Dr. Unni offered. The focus was on aggregation and valuation rather than payment structure, but the royalty payment structure and period were disputed at least implicitly.” 167T. C. 12, at p. 12. (Footnote omitted, but it says the disagreement is what are the variables and whether separate royalties are appropriate).
“…the better reading of the regulation is that, at a minimum, the form of payment must be adopted in the CSA agreement, but the base and structure may be left to CSA documentation that satisfies the requirements of paragraph (h)(2)(iii). This gives effect to the requirement of paragraph (k)(1) that the taxpayer must specify the form of payment in the CSA agreement and accommodates the timing rules in paragraph (h).” 167 T.C. 12, at p. 16. IRS’ reading of the Reg is too narrow.
Of course it’s never that simple.
“The regulation does not answer the question of the appropriate structure or period for royalty payments. It only requires the taxpayer to choose before outcomes are known. This, of course, is because the timing of the payments affects the allocation of risks between payor and payee. Dr. Newlon expressly recognized this too. In his rebuttal expert report he pointed out that the income method was simply a method for determining the arm’s-length compensation to Facebook US, expressed in present value terms, for its contribution. He explained that converting from an upfront lump-sum payment to a royalty rate means that ‘Facebook Ireland and Facebook US to some extent share the impact of deviations of actual revenue from forecast revenue over the period during which royalties are paid to Facebook US.’” 167 T. C. 12, at p. 19. (Footnote omitted). And there is no arm’s-length analogy.
If actual revenue differs materially from the estimates, YMMV puts it mildly. That’s why the Reg eliminates hindsight; both parent and sub are sitting big blind. And so, to some extent, is Judge C-Doug Pugh.
Getting to the bottom of this rabbithole, Judge C-Doug Pugh opts for the PCT report.
“The variable royalty rates for the PCT contributions reflect a specific allocation of outcome risks between the CSA participants and can be replicated following the methodology in the PCT report. Of the options presented by the parties for conversion of the PCT Payment to contingent annual royalties, the PCT report methodology is the closest to the economic substance of the transaction. It is the most faithful to the agreement between the CSA participants, the actual allocation of risks between them before the outcome of the CSA is known, and their actual conduct. It therefore is the method that should be applied to convert the PCT Payment into contingent annual royalty payments for purposes of computing petitioner’s corrected tax liability.” 167 T. C. 12, at p. 24.
So maybe so might could be hindsight isn’t entirely out of the picture.
As for reductions in Other Revenue, the Long Range Plan is the only guideline.
“The record includes no other basis for reducing the projected expenses in the LRP… beyond the credit card processing fees expressly associated with Other Revenue. Consistent with the LRP, we likewise find that those expenses should be the only ones excluded.” 167 T. C. 12, at p. 26.
And finally, the magic number: the WorldWide Discount Rate.
“… in Facebook I after considering all of the discount rates in the record and the flaws in each, we selected a 17.7% discount rate for valuing the cost sharing alternative because it fell within the range we considered reasonable and could be considered a concession by petitioner that the rate should be no higher than that. (Petitioner argued for a higher rate and respondent for a lower one.) That rate applied to ROW revenue; we did not adopt a rate for worldwide revenue.
“At the hearing on the Rule 155 computations, petitioner stated that a lower worldwide discount rate favored it. We construe this as acceptance of respondent’s position. We also note that 17% is the rate that EY also used in the transfer pricing documentation that computed Facebook Ireland’s RAB share and corresponding CST payments. Therefore we adopt a worldwide discount rate of 17% as a concession—in effect—by petitioner.” 167 T. C. 12, at p. 26.
And to help all y’all sort through this abbreviation salad, Judge C-Doug Pugh has annexed a schedule of Defined Terms.
There’s also an exhibit in reply to computation from Facebook’s twenty that ascends to such rarefied heights of pure mathematics that it is said that there is no woman or man in the scientific press capable of criticizing it.
