Medtronic And The Role In Tax Of Unspecified Methods

Although the regulations offer straightforward grounds for challenging the fanciful transfer pricing method endorsed in Medtronic II, a potential appeal could shed light on broader questions about the proper role of “unspecified” transfer pricing methods.

The outcome in Medtronic Inc. v. Commissioner, T.C. Memo. 2022-84, was no surprise to those who followed the U.S. Tax Court’s Medtronic II trial, which was held in June 2021 on remand from the Eighth Circuit (Medtronic, 900 F.3d 610 (8th Cir. 2018), vacating T.C. Memo. 2016-112).

Tax Court Judge Kathleen Kerrigan signaled her intentions during the trial, expressing misgivings about each party’s preferred method for pricing Medtronic’s 2001 license of cardiological and neurological device patents and related intangibles to Medtronic Puerto Rico Operations Co. (MPROC).

Kerrigan’s reassessment of the parties’ competing methods was necessary after the Eighth Circuit held that her 2016 opinion, which largely sided with Medtronic, accepted the comparable uncontrolled transaction method without applying the regulations’ comparability factors.

Although Kerrigan expressed strong skepticism regarding the IRS’s characterization of MPROC as a routine manufacturer entitled to no more than the return on assets determined using the comparable profits method, she also voiced concerns about Medtronic’s use of its 1992 litigation settlement agreement with Siemens Pacesetter Inc. in applying the CUT method.

To address those concerns, Kerrigan invited the parties to propose unspecified methods during post-trial briefing.

Although the stated goal was to bridge the gap between the parties’ opposing methods, the unspecified method outlined by Kerrigan was distinctively CUT-like in that its starting point would be the Pacesetter agreement’s royalty rate.

Unlike the IRS, which stuck to its original position, Medtronic obliged Kerrigan’s request and proposed a CUT-like unspecified method in its post-trial briefs as an alternative to its primary CUT method analysis.

Although Kerrigan disagreed with some aspects of Medtronic’s alternative method, she accepted its general features and used it as the basis for deciding the case.

Some of the grounds for appealing Medtronic II are straightforward: The accepted method can’t be the best method under reg. section 1.482-1(c) and leads to results that clearly violate the realistic alternatives principle as cited in reg. section 1.482-4(d) and generally stated in reg. section 1.482-1(f)(2)(ii)(A).

However, the case also raises broader questions concerning the appropriate use of unspecified methods. Although the Medtronic II method doesn’t present itself as such, it is essentially a variation of the CUT method that openly disregards the CUT method’s comparability and reliability standards.

Whether taxpayers, judges, or the IRS are free to sidestep the standards and requirements associated with a specified method simply by referring to it as an unspecified method may be a question that warrants appellate review.

Another Rube Goldberg Machine

Following the Tax Court’s practice of referring to the method applied in Medtronic II as unspecified would be misleading.

The Medtronic II method is better understood as a variation of the CUT method that, like the approach derided as a “Rube Goldberg machine” in Coca-Cola Co. v. Commissioner, 155 T.C. No. 10, relies on a series of questionable assumptions and adjustments designed to generate the desired result.

In Medtronic II, the desired result was one that bridged the gap between the results of Medtronic’s CUT method analysis and the IRS’s CPM analysis.

The method applied in Medtronic II, which is a modified version of the method proposed by Medtronic in its 2021 opening post-trial brief, consists of a series of steps that echo the mechanics of a residual profit split.

In step 1, Medtronic receives a 17.3% royalty for the patents and other technology licensed under the MPROC license, plus a 5.4% trademark royalty that the parties had previously accepted as arm’s length. Those royalties, which applied to third-party retail sales, were equivalent to a combined intercompany wholesale royalty rate of about 33.1% at the wholesale conversion rate of 68%.

The 17.3% retail royalty rate was drawn directly from Medtronic’s CUT method analysis: It’s the sum of the 7% royalty stipulated in a 1992 litigation settlement agreement with Pacesetter and a series of upward adjustments approved in Medtronic I.

Although each upward royalty adjustment represented the high end of the range of appropriate adjustments estimated by Medtronic’s expert witness, the 17.3% royalty rate excludes the 3.5% profit potential adjustment adopted by the Tax Court in Medtronic I. The method attempts to account for profit potential in step 3.

Under step 2, MPROC receives a 41.3% return on its adjusted asset base, determined by applying the CPM. MPROC’s return is then reduced, and Medtronic’s profit is correspondingly increased, to reflect Medtronic’s distribution activities and manufacturing of components used by MPROC.

For reasons not disclosed in the opinion, the Tax Court concluded that step 3 should allocate 80% of the remaining unallocated profit to Medtronic and 20% to MPROC. That compares with the 50-50 and 35-65 splits proposed by Medtronic in its opening post-trial brief.

According to the court, an 80-20 split in favor of Medtronic properly accounts “for the imperfections of the CUT method, such as ‘know-how,’ having only one comparable, and differences in profit potential, and imperfections of the CPM, such as the inadequacy of the comparables and an unrealistic profit allocation to MPROC.”

The method results in an intercompany wholesale royalty rate of 48.8% (equivalent to a retail royalty of 33.2%) for the MPROC license, which falls between the 33.1% wholesale royalty derived from Medtronic’s CUT method analysis and the 62.2% wholesale royalty derived from the IRS’s modified CPM analysis.

It also represents a meaningful — if incremental — increase to the 38% wholesale royalty determined by the Tax Court in Medtronic I. The court concluded that the resulting allocation of 68.7% of Medtronic’s implantable cardiac and neurological device profit to Medtronic and 31.3% to MPROC was a reasonable, if theoretically imperfect, result:

“Respondent’s expert [Brian] Becker testified that you may not like the logic of a method but ultimately the answer is fine. Because neither petitioner’s proposed CUT method nor respondent’s modified CPM was the best method, our goal was to find the right answer. The facts in this case are unique because of the complexity of the devices and leads, and we believe that our adjustment is necessary for us to bridge the gap between the parties’ methods.”

Whether the answer is ultimately fine in Medtronic II is doubtful. In cases that involve transfers of high-value intangibles (like Medtronic), the only feasible way to assess the reasonableness of the result is to evaluate the reliability of the underlying transfer pricing method. And the unspecified method accepted by the court is difficult to defend under the regulations, economic theory, or anything else.

Although the Medtronic II method’s most fundamental flaw is that it begins with the royalty rate from an agreement that the same opinion rejects as a comparable for the CUT method, it also misleadingly mimics the mechanics of the residual profit-split method without adhering to the latter’s conceptual underpinnings.

The royalties allocated to Medtronic for its trademark and technology licenses represent compensation for the transfer of rights to nonroutine intangible property. It therefore makes little theoretical sense to allocate royalty income in the same way that the residual profit-split method allocates routine returns or to treat the amount that remains unallocated after step 2 as residual profit.

And unlike the residual profit-split method, which generally allocates residual profit based on the parties’ capitalized intangible development costs, the Medtronic II method simply declares that an 80-20 split of unallocated profit in step 3 leads to a reasonable result.

It’s unclear which of the parties’ costs would qualify as investments in intangible development under the residual profit-split method, but Medtronic apparently incurred closer to 90% of total costs. As the party solely responsible for research and development, Medtronic’s proper share of residual profit may materially exceed 80%.

In any case, Medtronic II presents the 80-20 split as a justifiable way to address the differences between the MPROC license and Pacesetter agreement regarding profit potential and other differences in covered intangibles without any quantitative corroboration.

If there’s some meaningful link between the value of the intangibles licensed to MPROC and the right to 80% of an arbitrarily determined pool of unallocated profit, an explanation of that link never made its way into the Medtronic II opinion.

Not the ‘Best Method’

Posing as some variation of the residual profit-split method doesn’t by itself condemn the method endorsed by the Tax Court. But the Medtronic II opinion misconstrues or disregards the key reliability conditions for unspecified methods under reg. section 1.482-4(d), and it should be vacated or reversed on that basis alone.

Assuming that it’s possible to sidestep a specified method’s reliability standards by calling it an unspecified method, the option to apply an unspecified method doesn’t give taxpayers, courts, or the IRS carte blanche to cobble together novel methods designed to reach a specific result.

Recognizing the limitations of the CUT method, CPM, and profit-split method, the list in reg. section 1.482-4(a) of potential transfer pricing methods includes “unspecified methods described in paragraph (d) of this section.” The regulations thus allow the use of methods that aren’t expressly acknowledged or described, but only if they comply with the standards of reg. section 1.482-4(d).

By their nature, unspecified methods can’t be addressed by the kind of prescriptive rules that appear, for example, in the CUT method regulations under reg. section 1.482-4(c).

However, reg. section 1.482-4(d)(1) establishes general reliability conditions that apply to any unspecified method. The first is that the method comply with the general best method rule in reg. section 1.482-1(c).

Under reg. section 1.482-1(c)(2), the two key reliability considerations under the best method rule are “the degree of comparability between the controlled transaction (or taxpayer) and any uncontrolled comparables, and the quality of the data and assumptions used in the analysis.”

The most important comparability considerations vary by method, but reg. section 1.482-1(c)(2)(ii)(C) notes that the reliability of transactional methods tends to be highly sensitive to differences between the property transferred or services performed in the controlled and uncontrolled transactions.

Unless they can be addressed by appropriate comparability adjustments, those differences generally reduce the reliability of transactional methods more than that of profit-based methods like the CPM.

If there are material differences between the controlled and uncontrolled transactions, and the effect of those differences can be quantified with reasonable accuracy, comparability adjustments may improve the reliability of some methods.

However, reg. section 1.482-1(d)(2) provides that any adjustments like that must generally be based on “commercial practices, economic principles, or statistical analyses,” and reg. section 1.482-1(c)(2)(i) says “the number, magnitude, and reliability of [comparability] adjustments will affect the reliability of the results of the analysis.”

The CUT-like unspecified method approved in Medtronic II fares poorly even under those general standards. It falls squarely within the transactional category, as acknowledged by the court’s observation that “there are enough similarities that the Pacesetter agreement can be used as a starting point for determining a proper royalty rate” and use of the adjusted Pacesetter royalty in step 1. And as a transactional method, its reliability is highly sensitive to any material differences in the controlled and uncontrolled transactions — including differences in the property transferred.

It is undisputed that the MPROC license materially differs from the Pacesetter agreement regarding the licensed intangibles and circumstances of the transaction, and one of the most significant differences concerns profit potential.

Profit potential plays an important, if somewhat confusing, role in Medtronic II. Although similarity of profit potential is necessary to satisfy the two-part definition of “comparable intangible property” in reg. section 1.482-4(c)(2)(iii)(B)(1), Medtronic II addresses the issue as it relates to economic conditions — one of the generic comparability factors applicable to all controlled transactions under reg. section 1.482-1(d)(1). It’s not the case that, as stated in the opinion, “generally, intangible property is considered comparable if it is used in connection with similar products.”

Regardless, the court rightly acknowledged that there was a major imbalance in profit potential that undermined the Pacesetter agreement’s reliability as a comparable.

It was that difference in profit potential, along with differences in the functions carried out by the parties under the two licenses and the much broader scope of licensed intangibles under the MPROC license, that led the Tax Court to reject the Pacesetter agreement as a comparable for the CUT method in Medtronic II.

Even if the regulations allow transactions that can’t be used as comparable uncontrolled transactions for CUT purposes to be used as comparable uncontrolled transactions in CUT-like unspecified methods, the irregular, drastic, and numerous adjustments necessary to apply the Medtronic II method defy the reliability standards in reg. section 1.482-1(d)(2) and 1.482-1(c)(2)(i).

As tacitly acknowledged by the court’s statement that it agreed with Medtronic that “profits should not be addressed in step one because they are instead addressed in step three,” the split of unallocated profit in step 3 primarily serves as a profit potential adjustment.

But that adjustment, which took the form of an 80-20 split of unallocated profit, was supported by nothing more than the court’s ipse dixit declaration that an allocation of that kind leads to a reasonable result. Like the arbitrarily determined adjustments in Medtronic I, a judicial fiat that an 80-20 split leads to a reasonable result cannot possibly conform to any accepted commercial practices, economic theory, or statistical principles.

The magnitude of the profit potential adjustment, which dictates the allocation of over $1.3 billion in profit that remained after applying steps 1 and 2, is also beyond dispute.

Under step 1 of the Medtronic II method, Medtronic receives the adjusted Pacesetter royalty of 17.3% of third-party retail sales, which itself incorporates a series of additions to the unadjusted royalty rate of 7%. The method’s subsequent machinations yield a final retail royalty of 33.2% (equivalent to a 48.8% wholesale royalty), almost double the reference point set by the 17.3% adjusted Pacesetter royalty and nearly five times the unadjusted 7% royalty.

Whether the Medtronic II method complies with the best method rule is a mixed question of law and fact, and as noted by the Eighth Circuit, such determinations by the Tax Court are reviewed de novo on appeal.

The comparability adjustments required by the Medtronic II method are great in both number and magnitude and — especially for the profit potential adjustment — wholly unsupported by accepted commercial practices, economic concepts, or statistical principles. The adjustments therefore fail to reliably account for major differences between the MPROC license and Pacesetter agreement, and that alone should be enough to reject the method under reg. section 1.482-1(c).

‘Substantial’ Isn’t a Number

But the more clear-cut reason that Medtronic II shouldn’t be affirmed in a potential appeal is its failure to properly evaluate the selected method in accordance with the realistic alternatives principle.

Although the regs characterize the realistic alternatives principle as the basis for all potential methods, reg. section 1.482-4(d) places particular emphasis on the concept as the reliability standard for unspecified methods. In relevant part, reg. section 1.482-4(d)(1) reads:

“Consistent with the specified methods, an unspecified method should take into account the general principle that uncontrolled taxpayers evaluate the terms of a transaction by considering the realistic alternatives to that transaction, and only enter into a particular transaction if none of the alternatives is preferable to it. For example, the comparable uncontrolled transaction method compares a controlled transaction to similar uncontrolled transactions to provide a direct estimate of the price the parties would have agreed to had they resorted directly to a market alternative to the controlled transaction. Therefore, in establishing whether a controlled transaction achieved an arm’s length result, an unspecified method should provide information on the prices or profits that the controlled taxpayer could have realized by choosing a realistic alternative to the controlled transaction.”

That language establishes the realistic alternatives principle as the primary basis for assessing an unspecified method’s reliability under the best method rule. It also explains that the principle compares the prices or profits associated with the controlled transaction with those that would have been realized if the taxpayer had chosen a realistically available alternative. The test in reg. section 1.482-4(d) is thus a quantitative comparison, not an assessment of subjective difficulty.

The objective, quantitative interpretation of the realistic alternatives principle in reg. section 1.482-4(d)(1) is consistent with the regulations’ general formulation of the concept. Under reg. section 1.482-1(f)(2)(ii)(A), the IRS may make transfer pricing adjustments to ensure that the controlled transaction is arm’s length in that its terms would be acceptable to an uncontrolled taxpayer operating in similar circumstances faced with the same available alternatives.

Reg. section 1.482-1(f)(2)(ii)(A), like reg. section 1.482-4(d)(1), frames the concept as a quantitative comparison:

“The Commissioner may consider the alternatives available to the taxpayer in determining whether the terms of the controlled transaction would be acceptable to an uncontrolled taxpayer faced with the same alternatives and operating under comparable circumstances. In such cases the Commissioner may adjust the consideration charged in the controlled transaction based on the cost or profit of an alternative as adjusted to account for material differences between the alternative and the controlled transaction, but will not restructure the transaction as if the alternative had been adopted by the taxpayer.”

Read together, reg. section 1.482-4(d)(1) and -1(f)(2)(ii)(A) defines the realistic alternatives principle as a comparison between the revenue, profit, or cost savings associated with the controlled transaction and the forgone revenue, profit, or cost savings associated with a realistic alternative.

If the forgone economic benefits exceed the economic benefits actually realized, then the controlled transaction does not satisfy the arm’s-length standard.

The same forgone economic benefits interpretation of the realistic alternatives principle is the basis for the only example in reg. section 1.482-4(d)(2).

In the example, the IRS uses an unspecified method that compares the royalty a U.S. parent (USbond) company charges for licensing its manufacturing technology to a European manufacturing and regional distribution subsidiary (Eurobond) with the profit the parent could have earned under the realistically available alternative of selling directly into the European market.

USbond receives a royalty of $100 per ton of industrial adhesive sold by Eurobond, which sold the product to related and unrelated customers at a price of $550 per ton.

Because a price of $300 was assumed to be sufficient to cover USbond’s costs plus an appropriate routine return for its functions, risks, and assets, the parent was effectively surrendering $250 to Eurobond in exchange for a royalty of only $100.

Although the example doesn’t specify the amount of the necessary adjustment, it concludes that the disparity between the royalty received, and profit forgone, by USbond justifies a section 482 adjustment.

The transaction described in the example is simpler than the parties’ arrangement in Medtronic, which involved a trademark license, component sales, and distribution functions on top of the MPROC technology license. But as the IRS noted in its post-trial briefs, the example suggests that a reliable unspecified method would compare Medtronic’s profit under the MPROC license with the profit the company would have earned if it had opted to exploit the licensed technology directly.

According to the IRS, the relevant alternative to consider was Medtronic’s option of replicating an MPROC-like manufacturing facility over the course of 1.5 to 2.5 years at an almost trivial cost of $33.6 million.

If the IRS’s cost estimates and assumptions are anywhere near accurate, the result of applying the realistic alternatives test of reg. section 1.482-4(d) is clear: The $1.04 billion in profit allocated to MPROC for 2005 and 2006 under the Tax Court’s method far exceeds the one-time cost to Medtronic of establishing its own manufacturing plant. For the Tax Court’s selection of method to properly withstand appeal, it would have to be supported by a factual finding that the IRS’s replication cost estimate is off by orders of magnitude or suffers from some other massive flaw.

Instead, the Tax Court reiterated its earlier finding, which in essence was that replicating MPROC would have been a big hassle:

“In Medtronic I respondent made the same argument that MPROC was easily replaceable because it performed standard manufacturing activities expected of any manufacturer in the medical device industry. . . . Respondent has not provided evidence that disproves our initial conclusion that MPROC could not be replaced without substantial time and costs.”

But the notion that MPROC couldn’t be replicated with ease and that the time and costs of replication would be substantial doesn’t answer the key question posed by reg. section 1.482-4(d)(1) and -1(f)(2)(ii)(A), which is whether the total costs that Medtronic would have had to incur to create its own MPROC-like manufacturing plant exceed the total profit that Medtronic gave up under the MPROC license.

It’s hard to see how they could when the forgone profit exceeded $1 billion for 2005 and 2006 alone, but the Tax Court didn’t make the findings necessary to answer the question.

What’s In A Name?

The reliability of the method endorsed in Medtronic II under reg. section 1.482-4(d) is highly doubtful, both under the best method rule and the realistic alternatives principle. But evaluating the method’s reliability according to reg. section 1.482-4(d) assumes that the method should be considered an unspecified method in the first place.

One of the most important questions raised by Medtronic II is whether the regs really allow taxpayers and courts to disregard a specified method’s requirements simply by slapping an unspecified method label on what would otherwise be an unreliable application of the CUT method.

The court appropriately rejected the Pacesetter agreement as a valid comparable for the CUT method. However, it then defiantly declared that “there are enough similarities that the Pacesetter agreement can be used as a starting point for determining a proper royalty rate” – that is, that the agreement is a valid comparable uncontrolled transaction after adjusting for differences between it and the controlled transaction.

Can it really be that using Control+H to replace “CUT method” with “unspecified method” allows taxpayers, courts, or even the IRS to disregard the CUT method’s comparability and reliability standards? Or does Medtronic II represent an abuse of the flexibility that the option to apply an unspecified method was intended to offer?

According to reg. section 1.482-4(c)(1), the CUT method “evaluates whether the amount charged for a controlled transfer of intangible property was arm’s length by reference to the amount charged in a comparable uncontrolled transaction.”

The nature of the Medtronic II method may be obscured by a uniquely convoluted and unorthodox series of calculations, but it clearly falls within the scope of the regulatory definition: It uses the royalty charged in an uncontrolled license as the initial reference point for deriving the royalty for a controlled license. The Medtronic II method is thus still the CUT method — it’s just a version that makes royalty adjustments, including a profit-potential adjustment, that the CUT regulations don’t allow.

Unspecified methods still have a useful role — for example, by allowing the use of well-established financial valuation methods before they can be added to the regulations or recognizing that methods specified for one kind of transaction might be appropriate for other kinds of transactions.

But at least under the regulatory scheme finalized in 1994, there’s little policy justification for construing unspecified methods so broadly that they can accommodate specified methods that don’t adhere to the relevant reliability standards.

That hasn’t necessarily always been the case, especially for transfers of intangible property subject to the 1968 version of the section 482 regulations.

The closest thing to a specified transfer pricing method for intangible transfers in the 1968 regs is the approach in reg. section 1.482-2(d)(2)(ii). Although the 1968 regulations didn’t assign it a name or devote more than one sentence to its description, the method is a simplistic approximation of what is now the CUT method.

The 1968 regs recognized that CUT-like method as the presumptive best method as long as uncontrolled transfers of similar intangible property under similar circumstances were available as comparables. They reinforced that transactional preference by listing factors that “may be considered,” but only when “a sufficiently similar transaction involving an unrelated party cannot be found.” Reg. section 1.482-2(d)(2)(iii) vaguely alluded to the use of profit-based methods by including the transferee’s prospective profits, cost savings, start-up costs, and capital investment in the list of relevant factors. But none of those considerations by themselves constitutes anything that approximates a transfer pricing method.

The natural consequence of adopting a regulatory scheme that revolves around one actual method and provides only general factors that may be considered when that one method can’t be reliably applied was that it often became necessary to apply methods that didn’t appear in the 1968 regulations. Unspecified methods therefore became indispensable, and courts often used them to decide section 482 cases involving intangible transfers.

Relying on the inclusion in reg. section 1.482-2(d)(2)(iii) of the transferee’s prospective profits and capital investments, the Tax Court and Second Circuit endorsed a crude 50-50 profit split in Bausch & Lomb Inc. v. Commissioner, 92 T.C. No. 33 (1989), aff’d 933 F.2d 1084 (2d Cir. 1991).

Under the 1994 regulations, the profit-split method in reg. section 1.482-6 would likely have been more reliable. DHL Corp. v. Commissioner, T.C. Memo. 1998-461, rev’d in part on other grounds, 285 F.3d 1210 (9th Cir. 2002), is another high-profile example.

In DHL, the Tax Court and Ninth Circuit agreed with the IRS’s selection of a method that resembled the acquisition price method that became a specified method for valuing platform contributions with the introduction in 2009 of temporary cost-sharing regulations (T.D. 9441).

The acquisition price method would still technically be an unspecified method under the current regs, but reg. section 1.482-4(g) provides that the specified methods for cost-sharing arrangements may be appropriate as unspecified methods for other intangible transfers.

Medtronic II cites other examples of judicial approval of unspecified methods, including Sundstrand Corp. v. Commissioner, 96 T.C. No. 12 (1991), in which the Tax Court rejected uncontrolled licenses as transactional comparables but still used the royalty rate “as a base from which to determine the arm’s-length consideration for the intangible property involved in this case.”

However, all those cases were subject to a very different regulatory scheme than the one that applies in the Medtronic litigation. The CUT regs finalized in 1994 imposed stricter transactional comparability standards for intangible transfers than did the 1968 regulations, but the 1994 overhaul also (at least partially) filled the resulting gap with a list of new profit-based methods.

The need to resort to unspecified methods under a 1968 regulatory scheme that identified only one method says little about their role under the more expansive 1994 regulations, and it says even less about allowing what are now specified methods as unspecified methods to avoid inconvenient reliability standards.

If the Tax Court was concerned about the CPM’s inability to capture MPROC’s unique contributions and risks, the natural solution would have been to take advantage of one of the specified methods in the 1994 regs. Why — after rejecting both parties’ methods — the court applied a twisted variant of the CUT method that artificially resembles the profit-split method instead of the actual profit-split method is a mystery.

In any case, it is highly unlikely that an unspecified method that closely resembles a specified method without fully complying with its requirements would fare well under the general standards of reg. section 1.482-1.

The value of securing judicial confirmation that the section 482 regs’ standards for reliability of methods can’t be circumvented by manipulating labels may be the most compelling reason for a government appeal of Medtronic II.