Determining the locus of ownership of intangible property related to the development of new products and services is one of the principal challenges that tax managers in multinational enterprises (MNEs) face on an ongoing basis. Tax managers must balance non-tax corporate objectives, tax efficiency, tax risk management, and financial reporting considerations.
Since the mid-1990s, many MNEs have turned to cost-sharing (sometimes referred to as cost contribution) arrangements as a way to achieve these objectives. The cost-sharing rules (Treasury Regulations Section 1.482-7) adopted by the US Internal Revenue Service (IRS) in 1996, and the principles outlined in Chapter VIII of the 1995 OECD Guidelines, provide a flexible framework. MNEs used this framework to build corporate structures that located the ownership of highly valuable intangibles in tax-advantaged jurisdictions, in entities referred to as "intangible holding companies," or "IHCoís," while limiting exposure to withholding taxes and maintaining ease of internal administration. Some MNEs have achieved significant reductions in their worldwide effective tax rates by shifting the ownership of intangibles. For example, a recent article in the Wall Street Journal noted that Microsoft had reported a seven-percentage-point reduction in its worldwide effective tax rate in 2004, in part due to the movement of ownership of intangibles to Ireland (Glenn R. Simpson, "Wearing of the Green: Irish Subsidiary Lets Microsoft Slash Taxes in U.S. and Europe," Wall Street Journal, November 7, 2005, page A1).
Of course, when the ownership of already-existing intangibles is shifted, whether in connection with the establishment of a cost-sharing arrangement or otherwise, an armís-length price must be paid. It was perhaps inevitable that some MNEs would assert aggressively low values for pre-existing intangibles in connection with the establishment of cost-sharing arrangements, minimizing the "buy-in payment" required from the other participants to the initial developer of the intangibles and maximizing the benefits from the IHCo structure. It was equally inevitable that once the magnitude of the revenue losses from the widespread adoption of IHCo structures became apparent, tax authorities would move to close this supposed "loophole."
Thus, on August 29, 2005, the IRS issued proposed amendments to Regulations Section 1.482-7. A detailed description of the proposed amendments is beyond the scope of this article, but in the Preamble the IRS made its intentions clear:
Ö the valuation of the rights associated with [pre-existing intangibles] cannot be artificially limited by purported conditions or restrictions. Rather, the armís length compensation Ö must reflect Ö a "reference transaction" by which the benefit of exclusive and perpetual rights [is] provided.
(Notice of Proposed Rulemaking and Notice of Public Hearing, 70 Federal Register 51116, 51118 (August 29, 2005) (emphasis added)). The proposed amendments posit an "investor model" for determining payments for pre-existing intangibles, which would seemingly limit all participants in the cost sharing arrangement other than the original developer to a financierís return, with all entrepreneurial profit continuing to inureóseemingly for all time-- to the original developer. The investor model denies the validity of the premise on which the majority of past buy-in valuations were based, that is, that first-generation technology has a finite economic life, after which the entrepreneurial profit from subsequent generations of technology should be shared in proportion to the sharing of development costs.
The proposed amendments are highly controversial. It is likely that the vast majority of comments to be received by the IRS (as this is written, the deadline for submission of comments has not yet passed) will be unfavorable, some harshly so. Yet there seems little reason to believe that the IRS will back down.
If the proposed amendments are finalized in substantially their current form, and if they become the model for regulatory initiatives in other industrialized countries, then MNEsí ability to achieve substantial gains in tax-efficiency through IHCos and similar structures will be greatly inhibited, if not eliminated altogether. Thus, it seems appropriate to consider alternative ways of locating (or relocating) the ownership of intangibles in jurisdictions other than the jurisdiction in which the development of the intangibles take place. Some of these alternatives are well known, and therefore do not require an extended discussion at this time. Accordingly, the bulk of this article is devoted to an overview of a relatively new potential alternative to cost-sharing: the use of "real options" methods to determine transfer prices for intangibles.
Real Options and Transfer Pricing
"Real options," simply put, are options to buy or sell a "real," as opposed to a financial, asset. One example of a real option would be an option to buy the rights to a newly discovered chemical compound that may or may not be able to be developed into a useful drug. Many recent articles have speculated on the possible applications of real options in transfer pricing. Of particular interest to transfer pricing practitioners are applications of real options in pricing R&D activities and the resulting intangibles. A number of firms have adopted real options approaches to evaluating proposed research projects and structuring armís-length co-development agreements. The 1997 Merck Ė BioGen co-development agreement covering a class of drugs known asVLA4 inhibitors, described by Amram and Kulatilaka in their book Real Options: Managing Strategic Investment in an Uncertain World (1999), is an example of such an agreement. Although most discussions of real options as a framework for transfer pricing have involved drug development, there is no inherent reason why its use would be so limited.
All options, real or financial, are based on the premise that the right to purchase an asset at a predetermined value (the exercise or "strike" price) at some date in the future has an intrinsic value. That value is the price of the option. A number of methods exist to calculate what the price of an option should be, given complete information, and these methods can be readily extended to the pricing of real options. The primary methods are:
- Dynamic programming methods, such as the binomial tree method;
- Simulation methods, such as the Monte Carlo simulation; and
- Partial differential equations, such as the Black-Scholes equation.
EireCo and USPharm
For example (this example draws from an example originally developed by Faiferlick et al in their article, Using Real Options to Transfer Price Research-Based Intangibles, International Tax Journal, Spring 2003, p. 64), an Irish subsidiary (EireCo) of a U.S.-based pharmaceutical company (USPharm) could purchase a so-called "European" call option (that is, an option that can only be exercised on a specific future date) to buy the rights to make and sell drugs based on a particular compound or family of compounds. EireCo would buy the option for a price determined using one of the pricing methods discussed above; it could (but would not be obligated to) purchase the make-and-sell rights on the exercise date for the strike price. If EireCo chooses to exercise the option, it would then fund the further development and commercialization of the drug itself, and would use USPharm as a distributor in the U.S. market.
In this way, EireCo limits its risk by fixing its share of development costs (the option price), rather than being obligated to fund an uncertain stream of development costs. USPharm bears all other development expenses until the option is exercised; those expenses would be deductible against income subject to the higher U.S. income tax rate. Further, because USPharm bears all of the entrepreneurial risk associated with the research project unless the option is exercised, it need not earn the "guaranteed" return of a contract researcher.
As an alternative to a single option, a series of options with expiration dates at key development milestones could be acquired, which would further limit risk by allowing for more decision-making opportunities. Such an arrangement would permit more precise management of business risks, and also permit more precise management of the amount and timing of realization of income and expenses by USPharm.
That real options are economically superior as a valuation model compared to the discounted cash flow (DCF) approach is well known. However, there are potential obstacles to the use of real options as a transfer pricing method. Consider the example presented above. Assume that the USPharm wants to sell EireCo an option to buy the rights to a drug currently in pre-clinical development. In order to price the option, USPharm uses the Black-Scholes equation, which takes the form:
To determine the option price, USPharm estimates the expected net present value of the drug, A, using a probability-based DCF analysis. Assume that USPharm determines that A is equal to $120 million. It then estimates the future expenditures necessary to develop and commercialize the drug, discounting appropriately for the time value of money, and uses this value as X, the strike price, setting X equal to $200 million. T is set to 12, the typical number of years from pre-clinical development to FDA approval. The risk-free rate is set to 2.14 percent, the current rate on 10-year Treasury Inflation Protected Securities ("TIPS"), and s is estimated, based upon historical yields on pre-clinical stage products, to be 35 percent. The value of this option, using the Black-Scholes equation, is $46.4 million.
If EireCo buys this option, it effectively contributes $46.4 million up-front to the development of the drug, and has the right to pay another $200 million in 12 years to acquire it (which it is unlikely to do unless the drug passes all testing and approval processes). In this way, EireCoís expenses are capped at $246.4 million, regardless of the eventual profit or loss from the drug. What EireCo gets, effectively, is the right to buy the drug, at a future time when its commercial prospects are less uncertain, at a price that reflects the present riskiness of the development project, without adjustment for the reductions in commercial risk that occur as successive hurdles in the development processs are cleared. The strike price is virtually certain to be lessóin some cases, very much lessóthan an armís-length price, or an armís-length buy-in payment, determined at a later stage in the development process.
Viewed in this light, a real options approach to transfer pricing can be seen to have significant potential advantages. However, there are issues to consider.
First and foremost is the problem of demonstrating to the satisfaction of the taxing authorities (in our example, the IRS and the Irish Revenue) that the option price and the strike price are armís-length, and the further problem of demonstrating to the IRS that the strike price is commensurate with income. The use of the real options framework and option pricing models to determine the strike price would be an unspecified method under the U.S. transfer pricing regulations, and would bring with it additional documentation burdens.
While the IRS has repeatedly indicated its willingness to consider the use of real options approaches to transfer pricing, to date there is no published guidance whatsoever. To pursue such an approach without the protection afforded by an advance pricing agreement would require the assumption of an amount of tax audit risk that, while difficult to precisely quantify, is surely more than trivial.
If, in our example, the IRS tries to attribute more income to USPharm, it would have a number of avenues of attack. It could accept USPharmís pricing framework but challenge one or more of the values assigned by USPharm to the variables in the Black-Scholes formula. Relatively small movements in those variables can yield surprisingly large changes in the answer. For example, the IRS could argue that the yield on long-term U.S. Treasury bonds (currently 4.61 percent) should be used as r, the risk-free interest rate, instead of the TIPS rate. That simple change would yield an option price of $56.1 million, a change of $9.7 million. It could also argue (at least in theory, although we consider this to be an extremely weak argument) that the arrangement is in substance a cost-sharing arrangement, and that the option price is less than an armís-length buy-in payment.
In summary, real options provide a potentially valuable tool for managing risk and allocating intangible development costs across a MNE. However, the novelty and complexity of real options as a transfer pricing method suggest that MNEs considering it would be well advised to proactively manage their tax audit risk (and the associated financial reporting issues).
It is probably fair to say that cross-licensing, as a way of managing and determining transfer prices for the use of intangibles within MNEs, has gone somewhat out of fashion. However, depending on certain factors, an arrangement of this type may provide a viable alternative to cost sharing.
Intercompany cross-licensing involves two or more related parties, both owning valuable intangibles. Each party has the desire to exploit the otherís intangibles. In an intercompany context, this type of situation is somewhat unusual, because typically one group company is the "legal" owner of all intellectual property (IP). The ability to cross-license may occur at any time, but may be more common in situations where (1) a MNE acquires a company with existing intellectual property, or (2) a MNE has multiple R&D facilities around the world, each developing technologies complementary to one another. This type of arrangement may also be attractive for some MNEs from a management perspective. For those companies, it may not be ideal to have one subsidiary bear all the costs and risks associated with intangible development.
Pricing associated with cross-licensing can be tricky. In an unrelated-party context, companies often mutually share intellectual property without the payment of a license fee (when both sets of intangibles are of approximately equal value.) For example, in May 2005 Toshiba Corporation and Microsoft Corporation signed a patent cross-licensing pact that enabled the two companies to use each otherís patents freely in the computer and audiovisual equipment sectors. A royalty-free exchange of intangibles between related parties would likely be more problematic. A MNE could find itself defending the valuations of each of the shared technologies.
There may also be significant administrative complexity associated with cross-licensing in large multinational groups. Imagine a group made up of n companies, each owning a single intangible that is used by all group members. The number of license agreements needed, and the number of documentation studies required if each country in which a group company is resident requires documentation, is equal to [n (n Ė 1)], that is, 30 in the case of a group of six companies. Add a seventh company to the group, and the number of agreements and studies rises by 2n, to 42. Also, unless group members are located in jurisdictions with extensive tax treaty networks, withholding taxes on royalties become an issue. However, to the extent that group companies are located in EU member states, the proposed "masterfile" approach to intra-EU documentation will lessen the documentation burden.
Contract research arrangements are another alternative to cost sharing. Contract research involves one company engaging another party to perform all of its R&D activities. Under such arrangements, one party (the IP owner) owns all of the intellectual property associated with the product under development. The IP owner wishes to retain sole ownership of its IP, but does not have and does not want to acquire the resources necessary to develop it. The contract research organization has all of its costs (plus a profit element) paid by the IP owner, but has no interest in the IP once it is developed (and has no downside risk if its research is unsuccessful.) Arrangements of this kind between unrelated parties are common in a variety of industries, notably in the life sciences area (pharmaceuticals, biotechnology, and medical devices).
In the related-party context, it is important that any contract research agreement is structured to confer all legal and economic ownership to the party that is contracting out the research and development. Not clearly demarcating economic ownership can lead to confusion (and painful disputes with tax authorities) over who the economic owner of the IP may be, particularly if the R&D is being performed by a related party.
As an alternative to cost sharing, a contract research arrangement can be very advantageous, especially for certain industries. The typical scenario would involve a sale of the IP in development followed by a contract research agreement. Some of the key factors to consider are:
- The sale price of the IP (a company must be able to demonstrate it was armís length);
- A companyís risk management strategy (selling the IP early in its development cycle means a lower sales price, but confers more risk on the new IP owner); and
- The potential impact of the IRSís proposed services and cost sharing regulations (will a workforce-in-place be deemed to be an intangible in itself and thus be expected to earn something above a "routine" return?).
The principal advantage of this type of arrangement (compared to more traditional licensing arrangement), is that an outright sale a company is not subject to uncertainties associated with the commensurate-with-income rule.
While the potential demise of the current, highly flexible U.S. cost-sharing rules may make it more difficult for taxpayers to achieve desired outcomes in shifting ownership of highly valuable intangibles to tax-advantaged jurisdictions, the opportunity for legitimate tax planning will still exist. The three alternative methods of transferring ownership outlined here stay consistent with the armís-length standard. No technique will be appropriate for every company or in every case. However, they should be carefully considered by any multinational that seeks to optimize its global tax position.
About the Authors
D. Clarke Norton is a senior managing director in the New York office of FTI Consulting, and heads the firmís Transfer Pricing Services practice. Paul B. Burns is a managing director in the Transfer Pricing Services practice of FTI Consulting, based in the firmís Los Angeles office. The views expressed in the article are held by the author and are not necessarily representative of FTI Consulting. †
First published in International Tax Review December 2005
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.