Baker & McKenzie Urges Withdrawal of Proposed Cost Sharing Regs

[1] On behalf of the Software Finance and Tax Executives Council ("SoFTEC"), we are submitting comments on the notice of proposed rulemaking under section 482, REG-106359-02, published in the Federal Register on July 29, 2002, regarding the application of the rules of section 482 to qualified cost sharing arrangements. These proposed regulations would require that taxpayers include the "cost" or value of stock-based compensation in the pool of costs to be shared under qualified cost sharing arrangements. As discussed below, these proposed regulations are inconsistent with the hallmark of the arm's length standard embodied in section 482, the regulations thereunder, the OECD Transfer Pricing Guidelines, and United States tax treaties. The proposed regulations should be withdrawn. In the alternative, we propose a "stock based compensation safe harbor" approach that would be created within the cost sharing regulations. This safe harbor would be an exception to the arm's length standard and would allow taxpayers to "opt in" and avoid the disputes that have been waged over the past several years regarding whether unrelated parties acting at arm's length would share the stock option value or "costs."

[2] SoFTEC is a non-profit trade association focusing on finance, tax, and accounting issues relevant to the software industry. SoFTEC's membership comprises many of the world's leading software companies. See http://www.softwarefinance.org. Over the years, many SoFTEC member companies have entered into qualified cost sharing arrangements in accordance with section 482 of the Internal Revenue Code and Treasury Regulation § 1.482-7 and have also entered into joint development agreements with unrelated parties. Over time, SoFTEC members have granted at-the-money stock options to their employees, including personnel involved in research and development activities covered under qualified cost sharing arrangements. Recently, SofTEC has made a motion to the U.S. Tax Court to file a brief amicus curiae in Xilinx v. Commissioner, T.C. Dkt. No. 4142-01 (Apr. 4, 2002), which involves the issue of whether taxpayers must include stock option value or "costs" in their cost sharing pools under Treasury Regulation § 1.482-7. That motion remains pending.

I. FOR NEARLY SEVENTY YEARS, THE HALLMARK OF INTERCOMPANY TRANSFER PRICING RULES IN THE UNITED STATES HAS BEEN THE ARM'S LENGTH STANDARD

[4] The Commissioner's authority to reallocate income first appeared in section 240(d) of the Revenue Act of 1921. Section 240(d) provided that where two or more related trades or businesses were owned or controlled directly or indirectly by the same interests, the Commissioner could "consolidate the accounts . . . in any proper case, for the purpose of making an accurate distribution or apportionment of gains, profits, income, deductions, or capital between or among such related trades or businesses."¹

[5] The consolidation requirement of section 240(d) was eliminated when the provision was re-enacted in modified form as section 45 of the Revenue Act of 1928.² The report accompanying the enactment of section 45 provided that it was to be applied to related party transactions "as may be necessary to prevent evasion (by shifting of profits, the making of fictitious sales, and other methods frequently applied for the purpose of 'milking') and in order to clearly reflect their true tax liability."³ The Revenue Act included these purposes -- prevent evasion of taxes" and "to clearly reflect income!" -- in the statutory language, where they remain to this day.

[6] In 1934, the Secretary issued regulations interpreting section 45. Like those of section 482 today, these regulations provided that the arm's length standard governed the application of section 45, so that a controlled taxpayer could be placed

[8] After more than 30 years of only nominal changes in the regulations,⁵ Treasury in 1966 issued proposed regulations under section 482 to provide more specific rules for the allocation of income and deductions among taxpayers (the "1966 proposed regulations").⁶ The arms length principle of the section 45 regulations remained unchanged under the 1966 proposed regulations. The proposed regulations articulated how the arm's length standard generally applied to five specific types of transactions between related parties (loans and advances; services; transfers of tangible property; transfers of intangible property; and cost sharing). For each category, the regulation required that the terms of the same transaction type between independent parties be examined to determine whether the transaction between the related parties was at arm's length, i.e., to determine the prices that unrelated parties charged, or would have charged at the time, in independent transactions under similar circumstances.⁷ The proposed regulations were adopted in 1968 (the "1968 regulations").⁸

[9] The next major development occurred in 1986 when Congress enacted the "commensurate with income" standard.9 Treasury again reaffirmed that the arm's length standard governs intercompany transfer pricing under section 482 in "A Study of Intercompany Pricing Under Section 482 of the Code" (the "White Paper"):¹⁰

II. THERE IS NO EXCEPTION TO THE ARM'S LENGTH STANDARD FOR COST SHARING ARRANGEMENTS

[18] As both the original 1968 regulations and, later, the 1988 White Paper recognized, "[C]ost sharing arrangements have long existed at arm's length between unrelated parties."²² Accordingly, related party cost sharing arrangements have always been evaluated under the arm's length standard. When first issued, the 1966 proposed regulations provided a detailed set of cost sharing rules,²³ much like those reflected in the cost sharing provisions of the 1992 proposed regulations and the final regulations under Treas. Reg. § 1.482-7.²⁴ Although they set forth significant administrative and disclosure requirements and certain empirical criteria for qualified cost sharing arrangements, the 1966 proposed regulations specifically required costs to be shared on an arm's length basis, that is, in the same manner that such costs would be shared between unrelated persons.²⁵ Furthermore, the Commissioner's discretion to make adjustments to these agreements was limited to adjustments required to reflect arm's length sharing of the risks, benefits, and costs.²⁶

[19] When the 1968 regulations were promulgated, however, the Service eliminated the detailed cost sharing rules of the 1966 proposed regulations and opted for the simpler approach of Treas. Reg. § 1.482-2(d)(4)(1968). What remained was significant. Section 1.482-2(d)(4) continued to provide for costs and risks to be shared on an arm's length basis and limited the Commissioner's authority to make adjustments to bona fide cost sharing agreements to only those adjustments necessary to reflect an arm's length sharing of risks and costs. Section 1.482-2(d)(4) also specifically provided that "the district director shall not make allocations with respect to such [a bona fide cost sharing agreement] except as may be appropriate to reflect each participant's arm's length share of the costs and risks of developing the property." In the Treasury Department Release that accompanied the 1968 regulations, Treasury explained its rationale for the changes in the cost sharing rules:

[21] After adopting the arm's length standard for cost sharing arrangements in its own regulations, the United States successfully exported this view to its global trading partners. The United States Model Income Tax Convention and the Technical Explanation to the Convention explicitly acknowledge that related party cost sharing arrangements, like any other related party transaction, will be respected only if the arrangement satisfies the arm's length standard.²⁹ The U.S. Model Technical Explanation to Article 9, the operative article of the U.S. Model Tax Convention dealing with transfer pricing issues, provides, inter alia, that the arm's length standard governs all transfer pricing issues, including cost sharing arrangements:

[25] The OECD has also recognized that the arm's length standard plays a vital role in promoting international trade:

[27] Taxpayers have steadfastly and vehemently disagreed. First and foremost, taxpayers maintained that unrelated parties acting at arm's length did not share such "costs." Taxpayers provided evidence of billions of dollars paid under research and development cost sharing agreements where the United States itself did not reimburse stock option "costs" yet companies willingly participated in such arrangements. The Service, for its part, never produced -- and is still yet to produce -- any evidence that unrelated parties acting at arm's length in fact do share stock option "costs" in their own cost sharing pools. In Tax Court, the Service has even several times admitted that it has no evidence of unrelated parties acting at arm's length ever sharing the "cost" or value of employee stock options.⁵⁶ Thus, taxpayers argued, the Commissioner had no authority to make any adjustments under section 482.

[28] The Service has since abandoned its position under the 1968 regulations and recently directed its agents to stop taking the position that stock option "costs" or value must be included in cost sharing agreements in all cases involving the 1968 regulations.⁵⁷ Likewise, the Service conceded the issue and entered into a stipulation of settlement to that end in Xilinx v. Commissioner.⁵⁸ Nevertheless, the Service continues to maintain its position under the 1995 cost sharing regulations. The Service continues as well to be unable to show that unrelated parties acting at arm's length include stock-based compensation costs in their cost sharing agreements.

[30] The arm's length standard of section 482 has never been based on a priori concepts of what the Service thinks taxpayers should do; it has been based on what arm's length parties actually do. Generally, uncontrolled parties do not negotiate their business transactions based on rigid, preset formulae, but rather on the basis of existing market conditions. Unrelated parties determine their pricing on a case-by-case basis by analyzing the available market data. It is a fact intensive exercise, based on specific circumstances surrounding the developed intangibles, the market, and the taxpayer. The OECD Guidelines conclude that preset formulae "are likely to be arbitrary since they rarely fit exactly the varying facts and circumstances even of enterprises in the same trade or business."⁵⁹ As further recognized: "No specific result can be provided for all situations, but rather the questions must be resolved on a case-by-case basis, consistent with the general operation of the arm's length principle."⁶⁰

[31] To assume that in arm's length dealings, unrelated parties would necessarily, in all markets, and at all times, include their stock option "costs" in a research and development joint venture arrangement is contrary to the basic mechanisms of arm's length dealings. As recognized by the OECD, "predetermined formulae are arbitrary and disregard market conditions, the particular circumstances of the individual enterprises, and management's own allocation of resources, thus producing an allocation of profits that may bear no sound relationship to the specific facts surrounding the transaction"⁶¹ -- nor any notion of an arm's length standard. Assistant Secretary of the Treasury Samuels made the same point in 1994: "[T]he inflexible results obtained under a predetermined formula would not resemble the results under the arm's length standard, where the method used is tailored to the individual facts and circumstances."⁶²

[32] Nevertheless, in the newly proposed cost sharing regulations, Treasury has chosen to redefine this long-standing, internationally recognized objective criterion of the arm's length standard by administrative fiat. In sum, Treasury has opted precisely for the type of "one size fits all" fixed formula that it encouraged the international community to reject in the OECD Guidelines. The proposed regulations state that:

[33] There is no evidence arising from the behavior of uncontrolled parties to support the proposition that parties dealing at arm's length currently share the "cost" or value of stock-based compensation under their joint development arrangements. Indeed, whether the issuance of stock options creates a "cost" to the company or to the shareholders in the first place under generally accepted accounting principles has been a matter of some dispute for many years.⁶⁵ One of the reasons for the lack of arm's length evidence and the accounting debate over the proper treatment of employee stock options is that it is unclear whether the employee stock options are an economic cost of the corporation issuing the options at all.⁶⁶ At issuance, the employee stock options create valuable incentive and agency effects that inure to the issuing corporation. Employee stock options align the interest of the employees and the shareholders (i.e., agency effects) and induce employees to work harder (i.e., incentive effects). The issuance of employee stock options has a potentially dilutive cost to the current shareholders, but this potential dilution occurs only if stock prices rise and employees exercise their options. To the extent that the benefit of the incentive and agency effects exceed the dilutive cost of the employee stock options, there is no net cost to the shareholders to grant the stock options. Since the adoption of Statement of Financial Accounting Standards No. 123 in 1995 (which coincided with the promulgation of the 1995 cost sharing regulations), the economic and accounting literature has continued the debate over whether the grant or exercise of employee stock options is an economic cost to either the corporation or the shareholders. The evidence to date shows that the benefits of agency incentive effects at least equals and may exceed the dilutive effect of employee stock options.⁶⁷ Thus, employee stock option costs do not appear to have any net economic cost to the firm.⁶⁸

[34] Moreover, the exercise date measurement of the stock option inclusion of the proposed regulations is indefensible as a matter of economics.⁶⁹ The exercise date measurement depends on the movements of the company's stock price, which may or may not relate to the success or failure of the research development project or the value of the underlying employee services. The effect of measuring the inclusion in the cost sharing pool by the spread on exercise of the options forces the cost-sharing participants to compensate their joint venture partners as if the cost-sharing participant wrote a call option on its joint venture partner's stock. Thus, the cost-sharing participant has full exposure to the increases in the joint venture partner's stock price no matter the results of the research development project.⁷⁰ This is akin to "shorting" your business partner's stock. There is simply no way that unrelated joint venture partners would agree to take such a risk; in fact, they do not.⁷¹

[35] Most fundamentally, unrelated parties in the marketplace, acting at arm's length, do not currently include stock option "costs" or value in their joint development agreements, regardless of how or when such "costs" or value are measured. No amount of hypothesizing by the Service or Treasury changes that empirical evidence.

[36] Therefore, rather than "clarifying" the current regulations, the proposed regulations actually are based on unsubstantiated assumptions of arm's length dealings. representing a fundamental change in what it means to reflect the manner by which unrelated parties act at arm's length. The arm's length standard has never been a subjective test; it always has been an objective review of actual market activity -- of actual observable results -- to determine what unrelated parties acting at arm's length do. Section 482 certainly never has been normative and has never attempted to dictate what arm's length parties would do. This concept is reflected as well in the "best method" rule of Treas. Reg. § 1.482-1(c), which recognizes that the determination of an arm's length price must be based on all of the facts and circumstances, not on a priori concepts of arm's length behavior. "Thus, there is no strict priority of methods, and no method will invariably be considered to be more reliable than others."⁷² Just as there can be no strict priority of methods, there can be no strict rule concerning what costs must be shared. Although the proposed regulations suggest a mere "amendment" to this best method rule, the amendment will eviscerate it.

[38] The proposed regulations also represent inappropriate, if not irresponsible, international tax policy. By imposing formulary standards and ignoring the relevant economic facts, the proposed regulations are antithetical to the objective criteria utilized throughout the international community. This unilateral change by the United States will lead to international double taxation. The United States and its treaty partners have consistently subjected cost sharing arrangements to the arm's length standard like any other related party transactions. Among other items, most U.S. Treaties provide that if an adjustment is made in one country in accordance with the arm's length standard, then, in certain circumstances, the second country will make a correlative adjustment to take into account the negative effect of the first country's adjustment.⁷³ Because the United States has never been able to establish that parties acting at arm's length share stock option "costs," United States treaty partners will likely reject any discussion of possible correlative adjustments.⁷⁴ Taxpayers in the United States (as well as the U.S. Competent Authority) would find themselves in the unenviable and untenable position of having to explain why "unrelated parties acting at arm's length" means one thing in the United States and something quite different in the rest of the world.

[39] The proposed regulations therefore create the real risk of placing the Service in a crossborder "whipsaw."⁷⁵ If a U.S. taxpayer cost shares with a subsidiary in a treaty jurisdiction and if the U.S. taxpayer includes stock option "costs" or value under the proposed regulations, then, following traditional arm's length principles, the treaty partner may well through Competent Authority force the reversal of that inclusion. On the other hand, if a foreign corporation from a treaty jurisdiction cost shares with a U.S. subsidiary, then the corporation may very well push the same stock option "costs" or value into the United States via the U.S. subsidiary. Because the United States will be bound by its own published guidance, it will have to abide by the taxpayer's decision. In the end, the U.S. fisc loses. The whipsaw potential of the non-arm's length proposed regulations "underscores both the need for consistency and the care required in the formulation of appropriate rules.⁷⁶

[40] In sum, the proposed regulations attempt to fit the "square peg" of stock-based compensation "costs" into the "round hole" of the arm's length standard. If promulgated as is, the proposed regulations will not be an amendment to (nor, most certainly, a "clarification" of) the arm's length standard. Rather, they will be directly contrary to the arm's length standard. For nearly 70 years, "In determining the true taxable income of a controlled taxpayer, the standard to be applied in every case is that of a taxpayer dealing at arm's length with an uncontrolled taxpayer. A controlled transaction meets the arm's length standard if the results of the transaction are consistent with the results that would have been realized if uncontrolled taxpayers had engaged in the same transaction under the same circumstances (arm's length result)."⁷⁷ Without any marketplace or real-world evidence that unrelated parties acting at arm's length share stock option "costs" or value, Treasury's proposal represents a formulary apportionment approach to section 482. That approach has been consistently rejected. The proposed regulations should be withdrawn.

[42] If Treasury insists on moving forward and requiring taxpayers to engage in conduct that is not arm's length, the proposed regulations should not "amend" the arm's length standard and jeopardize, among other things, the United States' relations with its treaty partners. Instead, if there is to be some change there should be, at most, a decidedly non-arm's length "stock based compensation safe harbor" within the cost sharing regulations. As discussed below, taxpayers who "opt in" to this safe harbor would receive benefits that would not be available to taxpayers who do not. If taxpayers wish to minimize the potential for tax controversies arising out of their cost sharing arrangements, then they may "opt in" to the safe harbor and include stock option "costs." If taxpayers wish to exclude stock option "costs" and are prepared to defend their non-inclusion of such "costs" under the arm's length standard, they may "opt out" of the stock-based compensation safe harbor. Taxpayers can then balance the benefits and burdens of utilizing the safe harbor and proceed under the general cost sharing rules of Treas. Reg. § 1.482-7. In either case, the decision would be the taxpayers, who would have the ability to plan their related party transactions accordingly.

[43] This represents a "carrot and stick" solution to this issue. The cost sharing regulations should include a stock based compensation safe harbor, with sufficient incentives to encourage taxpayers to utilize the safe harbor. The incentives to utilize the stock-based compensation safe harbor should be two-fold. First, taxpayers who elect this safe harbor will have the benefit of knowing that the Service will not challenge their cost sharing arrangements. Second, if a taxpayer elects to utilize the safe harbor, the taxpayer's cost sharing arrangements for years prior to the effective date of the regulations should not be subject to examination or reallocation under section 482 at all. Taxpayers who opt in to such a non-arm's length safe harbor are giving up quite a bit. If the Service wishes to encourage taxpayers to take advantage of the safe harbor, then it must be willing to provide a substantial enough "carrot" in return. These incentives should make the benefits sufficient for taxpayers to opt in to the new safe harbor thereby substantially improving the responsiveness of the new cost sharing regulations to the stated goal of reducing disputes over the development and exploitation of intangible property between related parties.

[44] Nevertheless, some taxpayers will wish to enter into cost sharing arrangements on terms that they believe to be arm's length terms which do not include stock option costs. These taxpayers have the right to do so under both section 482 and United States tax treaties. The costs that are shared under such arrangements must be on terms that are consistent with the behavior of unrelated parties acting at arm's length, as determined under the general arm's length principles of Treas. Reg. §§ 1.482-1 and 1.482-7.⁷⁸ These taxpayers will forego the repose of the safe harbor for stock-based compensation with the full knowledge that the Service may challenge the arrangement.

[46] In addition, the method by which the "cost" or value of stock-based compensation is included in the cost sharing pool should be amended. In a safe harbor context, the measurement and timing methods of Prop. Treas. Reg. § 1.482-7(d)(2)(iii) could perhaps be used -- notwithstanding the inherently flawed assumptions required that would make the use of such rules arbitrary, capricious, and unreasonable outside of a safe harbor context. Taxpayers should also be permitted to use any reasonable method to value the costs associated with their stock-based compensation. To the extent that taxpayers elect to value their stock options at grant date under the same terms as for financial statement purposes, the proposed regulations should also allow taxpayers to make an annual determination regarding whether individual employees actually provide services related to the intangible development area. The proposed regulations, which assign stock options to the cost sharing arrangement based on the employees' responsibilities on the grant date, do not result in a clear reflection of intangible development costs.⁷⁹ Employees frequently transfer back and forth between positions that involve the development of intangibles and positions that do not, and their position on the grant date can result in an arbitrary allocation of expenses. For financial statement purposes, companies may amortize the "cost" of stock options over the vesting period of the option, matching the "cost" to the services provided by the employees. Permitting taxpayers to make an annual determination will result in a more accurate determination of costs. Thus, under such methodology, a taxpayer is required to include the "cost" in the pool in an amount equal to the financial statement amortization only for options vesting while the employee works in the intangible development area. Taxpayers seeking to reduce the administrative burden of making annual determinations on an employee-by-employee basis should be allowed to elect a one-time determination on the grant date as to whether the employee provides services related to the intangible development area.

V. TRANSITION RULES

[47] The final regulations should also include appropriate transition rules involving the treatment of stock options for periods preceding the effective date of the regulations and permitting taxpayers to conform their existing cost sharing arrangements to the new regulations.

[48] The Treasury Department and the Service should make clear that all taxpayers, regardless of whether they elect the stock-based compensation safe-harbor, are not, and will not be, required to include the value or "cost" of employee stock options in their qualified cost sharing agreements for years subject to the 1995 cost sharing regulations, i.e. for years preceding the effective date of the final new cost sharing regulations. Treasury should explicitly recognize that the proposed regulations represent a fundamental change to its traditional approach to section 482. In fact, had the Service been able to prove that arm's length parties actually share these costs, it would not have needed these proposed regulations in the first instance. Taxpayers have relied on the known lack of any evidence that unrelated parties actually share stock option costs and any explicit statement in the regulations requiring that these costs be shared. For open tax years preceding the effective date of the regulations, the Service faces the risk that Xilinx, or another taxpayer, will prevail in court on the issue of whether arm's length parties share stock option "costs." An adverse court decision under the arm's length standard would effectively preclude the Service from achieving its objective of taxpayers sharing stock option "costs" or value under any theory. Consequently, the Service should use the final regulations as an opportunity to announce that it will no longer litigate this issue with respect to prior years.

[49] As currently proposed, the amendments to the regulations will generally be effective for taxable years beginning on or after the date that final rules are published in the Federal Register. It is not reasonable to expect taxpayers to implement the amended regulations on such short notice.

[50] It will take time for taxpayers to decide whether they wish to conform their cost sharing arrangements under the proposed safe harbor, chance controversy with the Service, or use some alternative means of developing and exploiting intellectual property. It will take time for taxpayers to amend their cost sharing agreements to conform to the proposed safe harbor or to implement other arrangements. Some taxpayers have established complex international structures that they may wish to restructure if they wish to avoid controversy with the Service but keep their related party arrangements arm's length (i.e., not arbitrarily include stock option "costs" in their cost sharing arrangements). This process cannot be done overnight and therefore the final regulations should provide an appropriate transition period. When the current cost sharing regulations were adopted, taxpayers were granted a one- year transition period in which to conform their cost sharing arrangements to the regulations.⁸⁰ As a result of the complexities of some taxpayer's structures, some taxpayers may require as long as two years to restructure. Therefore, taxpayers should be granted a two-year period after the finalization of the regulations to conform their existing cost sharing arrangements to include stock option "costs" if they wish to take advantage of the safe harbor being offered, or weigh their litigation risks to either continue their arrangements as currently structured or restructure their international operations to avoid controversy with the Service.

VI. CONCLUSION

[51] Requiring taxpayers to include the "cost" or value of employee stock options in a qualified cost sharing arrangement, despite no evidence that arm's length taxpayers engage in similar behavior, is simply incompatible with the arm's length standard. Joint research ventures, joint product development ventures, or joint marketing agreements are all examples of commercial activities that are undertaken at arm's length by unrelated parties every day in the marketplace. The arm's length standard requires no more and no less than that taxpayers conduct their related party cost sharing arrangements in the same manner as unrelated parties. The proposed regulations violate this core principle and, in so doing, contravene the arm's length standard as currently embodied in section 482, the Treasury regulations, the OECD Transfer Pricing Guidelines, and United States income tax treaties. Consequently, the proposed regulations are of dubious validity and should be withdrawn. If the Service wishes to encourage taxpayers to include stock-based compensation "costs" or value in their related party cost sharing arrangements, then it should encourage taxpayers to do so by creating a non-arm's length administrative safe harbor within the cost sharing regulations, while still recognizing that cost sharing is a real world commercial practice that is subject to general arm's length principles.

* * * * *

[52] We appreciate the opportunity to submit these comments.

² H. Rep. No.2, 70th Cong. 1st Sess. (Dec. 7, 1927), 1931-1 C.B. 384, 395; S. Rep. No. 960, 70th Cong., 1st Sess. (May 1, 1928), 1931-1 C-B. 409, 426; See also Nat'l Securities Corp. v. Commissioner, 137 F. 2d 600 (3rd Cir. 1943); Asiatic Petroleum Co. v. Commissioner, 79 F.2d 234 (2nd Cir. 1935).

³ House Ways and Means Committee Report on the Revenue Act of 1928, H.R. Rep. No. 2, 70th Cong., 1st Sess. 16-17 (1928).

⁴ Treasury Regulations 86, Under Revenue Act of 1934, Article 45-1 (emphasis added).

⁵See Treas. Reg. § 1.482-1, T.D. 6595, 1962-1 C.B. 43 (Apr. 14, 1962) (adoption of the section 45 regulations under Section 482 of the Internal Revenue Code of 1954).

⁶ Treasury first issued proposed regulations under section 482 on April 1, 1965. These proposed regulations were withdrawn on August 2, 1966, and a new notice of proposed regulations was published (31 F.R. 10394).

⁷See Prop. Treas. Reg. § 1.482-2, 31 F.R. 10394 (Aug. 2, 1966).

⁸ T.D. 6952, 1968-1 C.B. 218.

⁹ Tax Reform Act of 1986, P.L. 99-514, § 1231(e)(1).

¹⁰ Notice 88-123, 1988-2 C.B. 458, 458.

¹¹Id.

¹²Id., at 472 (footnotes omitted)(emphasis added).

¹³ Notice of Proposed Rulemaking (INTL-0372-88; INTL- 0401-88), 57 F.R. 3571 (Jan. 30, 1992), corrected by 57 F.R. 27716.

¹⁴ Intercompany Transfer Pricing Regulations Under Section 482, T.D. 8470, 1993-1 C.B. 90, 91 (emphasis added) (the "1993 temporary regulations").

¹⁵Id., at 92 (emphasis added).

¹⁶ T.D. 8552, 1994-2 C.B. 93, 98.

¹⁷Id., at 98-99 (emphasis added).

¹⁸Grenada Industries, Inc. v. Commissioner, 17 T.C. 231, 256 (1951), affd 202 F.2d 873 (5th Cir. 1953).

¹⁹Virginia Metal Products, Inc. v. Commissioner, 33 T.C. 788, 800 (1960), aff'd in part and rev'd in part on another matter, 290 F.2d 675 (3rd Cir. 1961).

²⁰Aiken Drive-In Theatre Corporation v. United States, 281 F.2d 7, 9-10 (4th Cir, 1960), quoting Commissioner v. Chelsea Products, Inc., 197 F.2d 620, 623 (3rd Cir. 1952).

²¹See, e.g., U.S. Steel Corp. v. Commissioner, 617 F.2d 942 (2nd Cir. 1980); Davis v. United States, 282 F.2d 623 (10th Cir. 1960); Simon J. Murphy Company v. Commissioner, 231 F.2d 639 (6th Cir. 1956); Bausch & Lomb, Inc. v. Commissioner, 92 T.C. 525 (1989) aff'd 933 F.2d 1084 (2nd Cir. 1991); Virginia Metal Products, supra.

²² White Paper, supra note 10 at 493.

²³ Prop. Treas. Reg. § 1.482-2(d)(4) (1966).

²⁴Compare Prop. Treas. Reg. § 1.482-2(d)(4) (1966) with Prop. Treas. Reg. § 1.482-2(g) (1992), and Treas. Reg. § 1.482-7.

²⁵ Prop. Treas. Reg. § 1.482-2(d)(4)(iv)(1966).

²⁶ Prop. Treas. Reg. § 1.482-2(d)(4)(i)(1966).

²⁷ Treasury Department Release F-1217 (April 16, 1968).

²⁸See Treas. Reg. § 1.482-1(b)(1).

²⁹ On September 20, 1996, the Treasury Department amended and restated the United States Model Income Tax Convention (the "U.S. Model Tax Convention") and the Technical Explanation to the U.S. Model Tax Convention (the "U.S. Model Technical Explanation").

³⁰ U.S. Model Technical Explanation, supra, note 29, Commentary to Article 9, Paragraph 1 (emphasis added).

³¹ Organization for the Economic Cooperation and Development, Commentary to the 1992 OECD Model Convention on Income and Capital (the "OECD Commentary") at C(9)-1 (1995).

³² Organization for Economic Cooperation and Development, Transfer Pricing Guidelines for Multinational Enterprises and Tax Administration (2001) (the "OECD Guidelines").

³³Id. at § 8.26 (emphasis added). See also OECD Guidelines, Chapter 8 (F), "Recommendations for structuring and documenting CCAs" at § 8.40 (requiring CCAs to conform to the arm's length principle and identifying conditions normally expected at arm's length).

³⁴Id. at § 8.8 (emphasis added).

³⁵ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, May 30, 1996, U.S.-Austria, S. Treaty Doc. No. 104-31, 1996 U.S.T. LEXIS 77; Dept. of Treasury, Technical Explanation of the United States-Austria Income Tax Treaty at 15 (September 19, 1996).

³⁶ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, August 19, 1999, U.S.-Denmark, S. Treaty Doc. No.106- 12, 1999 U.S.T. LEXIS 167; Dept. of Treasury, Technical Explanation of the United States-Denmark Income Tax Treaty at 29 (Oct. 27, 1999).

³⁷ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, January 15, 1998, U.S.-Estonia, S. Treaty Doc. No. 105- 55, 1998 U.S.T. LEXIS 193; Dept. of Treasury, Technical Explanation of the United States-Estonia Income Tax Treaty at 29-30 (Oct. 27, 1999).

³⁸ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect-to Taxes on Income and Capital Gains, Art. 9 ¶1, July 28, 1997, U.S.-Ireland, S. Treaty Doc. No. 10531, 1997 U.S.T. LEXIS 98.

³⁹ Convention for the Avoidance of Double Taxation and Prevention of Fraud or Fiscal Evasion, Art. 9 ¶, August 25, 1999, U.S.-Italy, S. Treaty Doc. No. 106-1 1; Dept. of Treasury, Technical Explanation of the United States-Italy Income Tax Treaty at 28 (Oct. 27, 1999).

⁴⁰ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, January 15, 1998, U.S.-Latvia, S. Treaty Doc. No.105- 57, 1998 U.S.T. LEXIS 195; Dept. of Treasury, Technical Explanation of the United States-Latvia Income Tax Treaty at 30 (Oct. 27, 1999).

⁴¹ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, January 15, 1998, U.S.-Lithuania, S. Treaty Doc. No.105-56, 1998 U.S.T. LEXIS 194; Dept. of Treasury, Technical Explanation of the United States-Lithuania Income Tax Treaty at 30 (Oct. 27, 1999).

⁴² Convention the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶, December 18, 1992, U.S.-Netherlands, S. Treaty Doc. No.103-6, 1992 U.S.T. LEXIS 194; Dept. of Treasury, Technical Explanation of the United States-Netherlands Income Tax Treaty at 22-25 (Oct. 27, 1993).

⁴³ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital, Art. 9 ¶, June 21, 1999, U.S.-Slovenia, S. Treaty Doc. No.106-9, 1999 U.S.T. LEXIS 169; Dept. of Treasury, Technical Explanation of the United States-Slovenia Income Tax Treaty at 27 (Oct. 27, 1999).

⁴⁴ Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital Gains, February 17, 1997, U.S.-South Africa, Art. 9 ¶, S. Treaty Doc. No. 105-9, 1997 U.S.T. LEXIS 117, Dept. of Treasury, Technical Explanation of the United States-South Africa Income Tax Treaty at 28 (Oct. 7, 1997).

⁴⁵ Convention for the Avoidance of Double Taxation with Respect to Taxes on Income, Art. 9 ¶, October 2, 1996, U.S.- Switzerland, S. Treaty Doc. No. 105-8, 1996 U.S.T. LEXIS 74; Dept. of Treasury, Technical Explanation of the United States-Switzerland Income Tax Treaty at 28-29 (Oct. 7, 1997).

⁴⁶Convention for the Avoidance of Double Taxation with Respect to Taxes on Income, Art. 9 ¶1, November 26, 1996, U.S.-Thailand, S. Treaty Doc. No. 105-2, 1996 U.S.T. LEXIS 71; Dept. of Treasury, Technical Explanation of the United States-Thailand Income Tax Treaty at 29-30 (Oct. 7, 1997).

⁴⁷Agreement for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶1, March 28, 1996, U.S.-Turkey, S. Treaty Doc. No. 104-30, 1996 U.S.T. LEXIS 78; Dept. of Treasury, Technical Explanation of the United States-Turkey Income Tax Treaty at 31 (Sept. 19, 1996).

⁴⁸Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital, U.S.-Venezuela, Art. 9 ¶1, January 25, 1999, S. Treaty Doc. No.106-3, 1999 U.S.T. LEXIS 162; Dept. of Treasury, Technical Explanation of the United States-Venezuela Income Tax Treaty at 30 (Oct. 27, 1999).

⁴⁹Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital, Art. 9 ¶1, U.S.-Finland, September 21, 1989, S. Treaty Doc. No. 101-11, 1989 U.S.T. LEXIS 209; Dept. of Treasury, Technical Explanation of the United States-Finland Income Tax Treaty at 13 (June 14, 1990).

⁵⁰Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital and to Certain Other Taxes, Art. 9 ¶1, August 29, 1989, U.S.-Germany, S. Treaty Doc. No.101-10, 1989 U.S.T. LEXIS 233; Dept. of Treasury, Technical Explanation of the United States-Germany Income Tax Treaty at 22 (June 14, 1990).

⁵¹Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶1, September 12, 1989, U.S.-India, S. Treaty Doc. No. 101-5, 1989 U.S.T. LEXIS 236; Dept. of Treasury, Technical Explanation of the United States-India Income Tax Treaty at 22 (June 14, 1990).

⁵²Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and Capital, Art. 9 ¶ 1, April 3, 1996, U.S.-Luxembourg, Sen. Treaty Doc. 104-33; Dept. of Treasury, Technical Explanation of the United States-Luxembourg Income Tax Treaty at 30 (Sept. 19, 1996).

⁵³Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income, Art. 9 ¶1, September 1, 1994, U.S.-Sweden, S. Treaty Doc. No.103-29, 1994 U.S.T. LEXIS 218; Dept. of Treasury, Technical Explanation of the United States-Sweden Income Tax Treaty at 2l (June 13, 1995).

⁵⁴OECD Guidelines, supra note 32, at § 1.7.

⁵⁵See FSA 200103024 (Oct. 17, 2000) (the 1995 cost sharing regulations); Seagate Technology, Inc. v. Commissioner, 80 T.C.M. (CCH) 912 (2000) (1991 & 1992 tax years under the 1968 regulations); FSA 2000003010 (Oct. 18, 1999) (same); 1997 FSA LEXIS 311 (Aug. 1, 1997) (the 1968 regulations); Adaptec, Inc. v. Commissioner, T.C. Dkt. No. 3480-01 (1997 tax year under the 1995 cost sharing regulations); Xilinx, Inc. v. Commissioner, T.C. Dkt. No. 4142-01 (1995, 1996 & 1997 tax years under both 1968 regulations and 1995 cost sharing regulations).

⁵⁶See Seagate, supra, at 914 (the Service admitting a lack of evidence of arm's length dealings supporting its position); Adaptec, T.C. Dkt. No. 3480-01 (Petition and Answer at ¶¶5(d)(19)-(21)); Xilinx, T.C. Dkt. No. 4142-01 (Petition and Answer at ¶¶5(a)(37)-(45), ¶¶5(a)(50)-(54)).

⁵⁷Large and Mid-Size Business Organization, Industry Directive on Stock Options and Cost Sharing Agreements, dated January 25, 2002, reprinted in 2002 TNT 21-45 (Jan. 30, 2002).

⁵⁸Xilinx v. Commissioner, T.C. Dkt. No 4142-01 (Apr. 4, 2002).

⁵⁹OECD Guidelines, supra note 32, at § 4.107.

⁶⁰Id., at § 8.15 (emphasis added). See also §§ 8.5, 8.13 and 8.14.

⁶¹Id., at § 3.67.

⁶²Treasury News Release, "Remarks by Leslie B. Samuels, Assistant Secretary for Tax Policy, Seventh Annual International Tax Institute, George Washington University (Dec. 16, 1994); See also Statement of Leslie B. Samuels Assistant Secretary (Tax Policy), Department of the Treasury Before The Committee on Foreign Relations, United States Senate (Oct. 27, 1993).

[Editor's note: Footnote marker for Footnote 63 does not appear in the full text.]

⁶³OECD Guidelines, supra note 32, at § 1.7.

⁶⁴Prop. Treas. Reg. § 1.482-7(a)(3), 67 F.R. 48997 (emphasis added).

⁶⁵For almost 50 years, mainstream accounting principles have not considered the grant or exercise of employee stock options to constitute a corporate "cost" or expense. See Accounting Research Bulletin No. 43 (1953); Accounting Principles Board Opinion No. 25 (1972); Financial Accounting Standards No. 123 (1995). Recently, however, the Financial Accounting Standards Board in the United States and the International Accounting Standards Board have begun to reconsider the treatment of employee stock options for financial statement purposes. To date, neither body has acted to require companies to treat the grant or exercise of employee stock options as a corporate expense.

⁶⁶See, generally, Report of Prof. William J. Baumol and Prof. Burton G. Malkiel, "Status of Stock Options in Shared-Cost Contracts," (April 8, 2002), Ex. F to SoFTEC's Brief Amicus Curiae filed in Xilinx, T.C. Dkt. No. 4142-01 (April 30, 2002) ("Baumol & Malkiel Report" attached as Exhibit A). Unlike most other corporate expenses, at issuance of the stock option, the corporation does not pay out any assets of the corporation to the employee. In theory, the employee takes less in cash wages when granted stock options; thus, the stock options granted to an employee save the corporation cash. At exercise of the option, the corporation receives the strike price from the employee. Baumol & Malkiel Report at 4-5, 12-13.

⁶⁷Baumol & Malkiel Report, supra note 66, at 3-6, 12-15, 25-36; see also, Lynn Rees and David Stott, "The Value-Relevance of Stock-Based Employee Compensation Disclosures," Journal of Applied Business Research Vol. 17, No. 2 105-116 (Spring 2001);

⁶⁸See Baumol & Malkiel Report, supra note 66, at 25-36.

⁶⁹Id., at 20-25.

⁷⁰Id., at 23-25.

⁷¹Id., at 36-45.

⁷²Treas. Reg. § 1.482-1(c)(1); Cf Treas. Dept. Release F-1217, supra note 27, at 2.

⁷³See, e.g., Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital, Art. 9 ¶2, December 31, 1975, U.S.-United Kingdom, TIAS 9682, 1975 U.S.T. LEXIS 605; Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital, Art. 9 ¶3, September 26, 1980, U.S.-Canada, TIAS 11087, 1980 U.S.T. LEXIS 93; Convention for the Avoidance of Double Taxation and the Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital, Art. 10 ¶2, August 29, 1980, U.S.-Egypt, TIAS 10149, 1980 U.S.T. LEXIS 325.

⁷⁴Louise M. Kauder, "The Unspecific Federal Tax Policy of Arm's Length: A Comment on the Continuing Vitality of Formulary Apportionment at the Federal Level," 60 Tax Notes 1147 (Aug. 23, 1993).

⁷⁵See Treasury Department News Release F-1069, "Remarks by the Honorable Stanley S. Surrey, Assistant Secretary of the Treasury, Before the National Foreign Trade Council Convention, The Waldorf-Astoria Hotel, New York, New York, Wednesday, November 1, 1967, 3:00 p.m., EST," at 19 (Assistant Secretary Surrey cautioning that transfer pricing rules have "two sides of the coin" such that one-sided rules create whip-saw potential.); See also, Surrey, "Reflections on the Allocation of Income and Expenses Among National Tax Jurisdictions," 10 Law and Policy in Int'l. Business, 409, 414 (1978).

⁷⁶Id.

⁷⁷Treas. Reg. § 1.481-1(b)(1) (emphasis added).

⁷⁸Safe harbor cost sharing arrangements that include the "cost" of employee stock options in the pool of costs to be shared among controlled affiliates are not uncontrolled transactions and cannot be used by the Service as evidence of arm's length dealing.

⁷⁹See Prop. Treas. Reg. § 1.482-7(d)(2)(ii).

⁸⁰See Treas. Reg. § 1.482-7(l).

In particular, we have been asked to consider the following issues:

Summary of Approaches and Conclusions

We will explore the pertinent issues in three ways in terms of economic analysis, via a review of relevant earlier studies and, finally, on the empirical evidence, that pertinent to the accuracy of valuation of options and that pertinent to arm's-length contracts in reality.

The analytical material will lay out, with the aid of the economist's approaches, the character and functions of stock options, the nature of any associated costs to the firm, their distinction from costs to current stockholders, and the appropriate role of stock options, if any, in a contract between two firms entailing their sharing of the costs of certain activities of one or both of those enterprises. This last item will play a key role in our analysis of the arm's-length evidence pertinent to the IRS contentions.

The examination of earlier studies will investigate two issues, first, whether the evidence those studies provide supports the contention that stock options do constitute a cost. At least equally important, we will also examine whether these studies, as well as the available methods of empirical analysis of options make it possible to obtain accurate and unambiguous estimates of the true values of those options and corresponding estimates of any associated costs.

The empirical materials, together with the analysis of the character of cost-sharing agreements and the pertinence of the issue of stock options to employees, will be used to draw conclusions on three related matters: first, why the arm's-length standard is a defensible basis for analysis of issues such as that before the court, second, whether the logic of the issue does or does not lead to the expectation that firms entering cost-sharing agreements at arm's-length will include stock options as entailing costs to be shared and, third, whether in practice there exists evidence that such inclusion is or is not universal or even usual.

Our analysis will lead us to the following conclusions:

1. The value of long-term stock options granted to employees cannot be estimated from the economic evidence with a reasonable degree of certainty. The disciplines of economics and finance do not provide a method by which the value of long-term employee stock options can be measured with any degree of accuracy, particularly given the long-term nature of such options. The Black- Scholes model, the most sophisticated tool available for the purpose, works extraordinarily well for periods up to three months in maturity. But even for exchange traded options, the Black-Scholes model works less well for options with maturities from six months to one year. And for longer periods (Xilinx's employee stock options had ten year lives) it is inherently unreliable and inaccurate for reasons that will be explained.

2. There is no measurable net economic cost to a firm related to the grant of employee stock options. Stock options, when provided to employees generally play at least one of two roles. They can constitute a partial substitute for payment of salary, wages or compensation, and/or they can help to eliminate or at least to alleviate one much studied problem in corporate governance, the possibility of divergence between the interests of the members of corporate management and those of the stockholders. For, by making the real earnings of the recipient of the options contingent on the performance of the firm's equity, stock options can help to align the interests of these two groups. Furthermore, the employee stock options may provide significant beneficial incentive effects. It is important to distinguish between a cost to a given body of stockholders and a cost to their firm. Even though the firm is the property of the body of its stockholders, a newly issued stock option, if it does nothing else, merely redistributes some of the firm's future earnings between the initial holders of its stocks and the new stockholders created by the options. Unlike an increased wage payment that, ceteris paribus, reduces the firm's net earnings, a new employee stock option that leaves all else unaffected preserves the firm's earnings unchanged. If there are any added returns that result from an incentive provided by the options they do not constitute a cost offset but a net addition to the firm's total future earnings. Where such enhanced earnings are present, all stockholders may benefit from the provision of the options.

[53] To the extent that employees accept lower cash compensation as a result of the grant of employee stock options, such grants help to preserve the firm's cash. To the extent that the employees later exercise their options after a rise in the stock price of the firm, the employees pay the firm the fair market value of the firm's stock price at the time of the option grants. In neither event does the firm incur any direct cost. Even more important, the incentive effects of employee stock options may improve the performance of the firm by causing employees to work harder or more efficiently and to align the interests of management and shareholders.

The grant of employee stock options, however, has a detrimental dilutive effect on the corporation's shareholders. If the shareholders and the marketplace at large were to consider the dilutive effects of employee stock option grants to outweigh the benefits to the firm of such option grants, then the price of the firm's stock would fall and this would indeed result in an economic cost to the firm -- an increase in the firm's equity cost of capital. On the other hand, if the shareholders and the market were to consider the positive benefits of the option to outweigh the dilutive effects, then the firm's stock price would not fall but would remain unchanged or possibly even increase. If this latter expectation were in fact shared by shareholders and the marketplace, then the grant of employee stock options would have no cost to the firm because the firm's equity cost of capital would remain the same or even decrease as a consequence of such grants.

The issue whether the grant of employee stock options affects stock prices is inherently empirical. There exists a considerable body of economic literature that investigates the effects of employee stock options on the position of stockholders. Though many of these studies are carried out dispassionately and by highly competent investigators, the complexity of the issues they study has prevented them from yielding unambiguous answers. By and large the studies have indicated that beneficial effects of employee stock options equal or exceed the dilutive effects, thus indicating that stockholders do benefit from the grant of employee stock options. However, there is at least one unpublished working paper that has reached the contrary conclusion, though, like many of the other studies, it is characterized by serious imperfections.¹ After dealing with some statistical issues in the treatment of the data, we found that the calculated results of the unpublished working paper were consistent with the remaining body of empirical work.² All in all it is our conclusion that this body of investigation has produced no evidence that the issue of stock options generally has detrimental effects upon stockholders -- that is, the empirical analyses show the the beneficial effects of stock options appear at least to equal the dilutive effect on shareholders. Thus, the empirical research to date supports the conclusion that them is no measurable economic cost to shareholders -- or to the firm -- related to the grant of employee stock options.

3. The amount of the section 83 tax deduction taken by a firm related to the spread upon exercise of its employee stock options corresponds to no economic cost to the firm. It is widely accepted by economists that the stock prices generally incorporate all pertinent and publicly known information.³ Under applicable accounting rules, publicly reporting firms must fully disclose information related to their grants of employee stock options. Thus, the stock markets must already anticipate the exercise of the "in the money options" (i.e., options with a strike price less than the current market price of the stock) and the dilutive effect of such exercises must be expected already to be reflected in the stock price. In our judgment the issue of whether there is an economic cost to the firm relating to employee stock options economically must appropriately be measured at the time of option grant. The problems that arise in attempting to separate the dilutive effects from the incentive effects at the time of exercise are particularly acute. Indeed, we believe that the spread on exercise is more likely at least in part to reflect any beneficial incentive effects of the employee stock options than any dilutive effect.

4. The economic evidence clearly supports the conclusion that Xilinx's exclusion of the value or "cost" of employee stock options from its agreement to share development costs with its affiliated entities was entirely consistent with arm's-length dealing by unrelated parties in the marketplace. Economic analysis supports the use of an arm's-length standard for defensible evaluation of any cost and the allocation of its coverage responsibilities among the relevant parties. The unbiased behavior of markets in which transactions are entered into voluntarily distinguishes true economic costs from the accounting fictions that unfortunately often become unavoidable when it is attempted to evaluate costs without reliance upon the guidance of actual market behavior. For the current issue this means that to determine defensibly whether stock options constitute costs that are to be encompassed in cost-sharing agreements between firms should be investigated on the basis of observation of the pertinent arm's- length transactions. Where two firms engage in a cost-sharing agreement at arm's-length it is not to be expected that rational managements will include stock options as an element of the costs that are to be shared. The most obvious reason is that there is no reliable way to measure the value of those options, much less any cost that can defensibly be associated with them.

But there is more than that. The inclusion of stock options as a cost to be shared, if the option's "cost " is to be evaluated on the basis of the value of the issuing company's stock at the date the option is exercised, means that one of the partner enterprises commits itself to an unlimited liability whose future magnitude depends on the stock price of the other party to the agreement. If it were to do so it would in effect be writing a naked call option on the stock of its cost-sharing partner, agreeing to compensate the partner for the value that stock happens to take on some future date, a value and date unknowable at the time of adoption of the cost-sharing agreement. In perhaps simpler terms, this situation is akin to selling short the stock of the firm's business partner. While one would generally hope that one's business partner is successful, under the IRS's contention, the higher the business partner's stock price climbed, the more the other business partner's development "cost" would climb -- even if the development project were a complete failure and the stock price rise had nothing to do with the joint development activities.

Even worse, employee stock options often do not vest for a period of years and can remain outstanding for still many additional years. (Xilinx's employee stock options vested over four to five years and did not expire for ten years.) The IRS is apparently contending that option exercises in the current year -- which necessarily relate to options granted in prior years -- somehow are related to the costs of the jointly shared development activities in the current year. As shown by the empirical studies, there is no general and measurable net cost to a firm related to employee stock options. The IRS's exercise date hypothesis has the further insurmountable logical gap of trying to connect the spread on exercise of options granted in prior years to current year development efforts. There is simply no logical or economic basis for such an assertion.

For all these reasons, economic analysis leads us to conclude that it would not be rational for the cost-sharing firms to include stock options in their agreement. There exist many actual joint development cost-sharing agreements that include no such provision, and we have found none in which arm's-length parties actually include the value or "cost" of employee stock options in practice. Even more significant, we understand that the IRS too has conceded that it is unable to find any such example. In sum, the empirical research shows that there is no measurable economic cost related to employee stock options. Even if it were valid to claim that the value of employee stock options is in some sense an economic cost, the best tools of economics and finance are unable to value employee stock options with any degree of accuracy or economic certainty. Economic analysis leads us to conclude that one would not expect unrelated parties dealing at arm's-length to agree to share the value or any purported "cost" of stock options, and that to do so on the basis of an ex post facto spread upon exercise would be particularly risky and irrational. And in practice, consistent with our economic analysis, not a single example can be found where unrelated parties in the marketplace agree to share the value or purported costs of employee stock options. At the very least, it must be concluded that Xilinx's actions in excluding the value or "cost" of employee stock options from its cost-sharing arrangement were entirely consistent with the theory and practice of arm's-length dealings by unrelated parties in the marketplace.

The Two Purposes of Stock Options

There may be many considerations that lead the management of a firm to undertake an issue of stock options to its employees. However, the literature recognizes two primary objectives of such a step and these must be understood if the relation between this action and cost is to be comprehended. The first of these two purposes is to provide the firm a substitute for some part of the compensation the enterprise would otherwise have to provide to the recipient employees. The second purpose is to solve what economic analysis describes as the principal-agent problem -- the possible divergence between the interests of the management of a corporation and those of its stockholders.

The first of these purposes is straightforward. For example, consider a firm that is strapped for cash and subject to other financial difficulties. Suppose the firm locates an experienced executive with an outstanding track record in dealing with such problems. Such persons are not obtained cheaply, and the cash poor firm may not feel itself in a position to commit itself to providing the compensation needed to induce this individual to join it. Instead, it can offer that person stock options in lieu of a substantial portion of the compensation demanded. An agreement between the company and the individual can then be sought on the quantity of options that will serve as an appropriate equivalent of the foregone compensation.

These options then serve as a substitute for cash payments to the individual in question. But as we will see presently, their status as costs to the firm are very different.

The second of the two primary purposes of the issue of stock options is very different, though such an issue may well be undertaken to serve both objectives. As early as the 1930s, in the classic study of A.A. Berle, Jr. and G.C. Means, The Modern Corporation and Private Property (Macmillan N.Y. 1932), it was recognized that the modem corporation is characterized by separation between ownership and management. Unlike the minuscule enterprise that is overseen by its proprietor, the large corporation's managers are, as it were, hired help who, if the arrangements are inappropriate, may choose to pursue their own agenda rather than those of the true proprietors of the firm. A trivial but obvious example is nepotism in hiring. There are also reasons for believing that management benefits from expansion in the volume of the firm's sales or its market share even if it entails some sacrifice of the profits of the enterprise, that are presumably the stockholders' primary concern.

Here, economists speak of the stockholders as the principals of the firm and the members of management as the agents of those principals. Clearly, without suitable precautionary measures, the principals have good reason for concern about the temptations for the agents, consciously or unconsciously, to give priority to their own interests rather than those of the principals. The recognized way to deal effectively with this dilemma is to modify the nature of the payoffs offered to the agents in such a way that brings their interests more closely into line with those of the principals. That is precisely what stock options are designed to do.

Stock options can achieve this result in a straightforward manner. Because the recipient of the options benefits from them only to the extent that the price of firm's stocks rises above its value at the time the options were issued, the recipient members of management are given the incentive to strive as hard as they can to increase the value of those stocks. But that is precisely what serves the interests of stockholders.

Stock Options, Costs to the Firm and Cost to the Stockholders

Is there a clear-cut cost, or even any net cost to the firm entailed in the issue of stock options to employees of the firm? Before getting to the heart of the matter, it is important to note that the issue of the options for either of the two purposes just described has an inherent offset that is beneficial both to the firm and its stockholders. This is obvious if the options are provided to offer the desired incentives to management -- to deal with the principal-agent problem. If the options induce management to work harder -- to create better products, to cut costs, to promote sales, or otherwise to contribute to profits and to the value of the securities of the corporation -- then they clearly provide a benefit to stockholders. At most, any cost to stockholders that options are said to entail must be lower than that of any equivalent compensation that provides no such incentives.

This is even true of options whose only purpose is as a partial or total substitute for some direct employee compensation payment. The two forms are inherently different from the point of view of the interest of stockholders. The acceptance of options in lieu of direct compensation payment has two offsetting benefits to stockholders. First, it can reduce pressures on the firm's limited supply of cash. Particularly for a firm seriously short of cash or one that can obtain cash only on very disadvantageous terms, this benefit is obvious. But there is a second offsetting benefit to stockholders, whether or not the firm is significantly cash constrained. For an option transfers risk from the firm to the individual who obtains the option. That person has accepted an uncertain future payoff, one dependent on future stock prices, in lieu of a guaranteed stream of payments. The guarantee is a risk that would have been borne by the stockholders if the substitution had not occurred, because so long as the individual continues to be employed he would have to be paid whether the firm prospers or does not. But an option will be exercised and the individual will receive the corresponding payment only if the enterprise prospers. If it fails, it is the option holder who bears the corresponding portion of the consequences and the stockholders are relieved completely of this risk.

But these offsets are only part of the story. There are two other considerations that go much further and undermine straightforward cost interpretation. First, there is the possibility that the incentive and agency effects of stock options may be so substantial and favorable to the stockholder that they generally constitute a net benefit rather than. a cost. As will be shown here later, much of the evidence is indeed consistent with such a conclusion. Many of the available studies indicate that stockholders predominantly are net beneficiaries when firms choose to issue options to their employees. We must admit here, however, that the empirical evidence is neither unambiguous nor conclusive as to whether there is a net benefit to shareholders from the issuance of employee stock options (i.e., that the beneficial effects exceed the negative dilutive effects by a statistically significant amount). While the preponderance of the empirical investigations do reach the conclusion that in general employee stock options offer gains to stockholders, we cannot claim that a statistically significant affirmative net benefit has been shown beyond any reasonable doubt. We are very comfortable in concluding, however, that there is no measurable net economic cost to the firm or its shareholders from the issuance of employee stock options (i.e., to a reasonable degree of economic and statistical certainty, the positive effects of employee stock options are at least equal to the negative dilutive effects to shareholders).

The final consideration here, however, is the most conclusive, though perhaps the least widely recognized. This is the fact that the issuance of employee stock options must be recognized as only constituting a redistribution of benefits between initial stockholders and the new prospective stockholders who have obtained this position by their receipt of the options. It does not result in any reduction in the overall size of the firm's total earnings pie. Rather, it only affects the way in which that pie is sliced and divided up among future shareholders. And that is so even if the options lead to absolutely no change in the performance of management and the firm's future prospects. This is markedly different from the effect of, say, a rise in the cash wages of the company's current employees which, if it does not affect their performance, must result in a net reduction of the total profits of the firm. The latter is a cost to the firm in that, without offsetting benefits, it reduces the size of the earnings pie. The stock option issue, in contrast, leads to no such reduction in the earnings of the firm.

The point in all this is that it would be erroneous to take the cost of a direct expenditure such as a cash wage cost to be equivalent to that of an option. And there is simply no valid empirical evidence showing that the grant or exercise of an employee stock option constitutes a measurable economic cost to the firm. The empirical literature to date shows that the issuance of employee stock options normally either has no measurable cost to the firm or shareholders, or that such an issue actually benefits the firm and its shareholders. It simply cannot defensibly be claimed that the issue of employee stock options is a normal cost to the firm from the empirical research performed to date.

Some Aspects of Option Pricing Models

Since, in the discussion that follows, it will be necessary to refer back to some aspects of the option pricing model, it will be useful here (and in the appendix -- A Primer on Options) to review certain concepts. A call option gives the owner of the contract the right but not the obligation to purchase a share of company stock at a fixed price (the exercise or strike price) on or before a certain date (the expiration date). The buyer of an exchange-traded option pays an amount called the option premium to obtain such a right. The premium (less commission) is given to the option seller (or writer) who takes on the obligation to sell the shares to the option buyer at the exercise price.

Intuitively, we can understand what determines the size of the option premium. Premiums will be larger the longer the time to expiration since more time will be available for an event favorable to the option holder to occur. Premiums will be larger the higher the price of the underlying stock. Obviously an option on a one dollar stock can't be worth more than one dollar (otherwise, you would just buy the stock for one dollar) while a three month option on a hundred dollar stock can be worth five dollars or more. Interest rates also influence option premiums since the option buyer puts up less money than the person who buys the stock outright.

The Crucial Role of Volatility

The most important factor influencing option premiums is the volatility of the underlying shares. Options are worth more if the underlying stock is more volatile. To see why this is so, consider the following example: Suppose we have two stocks currently selling at $30 per share. Suppose that Stock A is very volatile and that in three months time each of five future values is equally likely ranging from a low of $10 to a high of $50. Stock B is less volatile and the equally likely range of future values runs from $20 to $40, Consider now how much a 3 month call option with an exercise price of $30 is worth. At expiration, the option will be worth the difference between the actual stock price and the $30 exercise price. Thus, if the stock sells at $30 or less, the call option expires worthless. But if the stock sells at $40 at the end of the period, the option has an "intrinsic" value of $10 since the holder could simultaneously exercise the option at $30 and sell the stock in the open market at $40. We then can see clearly from the exhibit below that in the case where market prices go up, the high volatility Stock A has larger option payoffs than the less volatile Stock B.

                       The Value of Volatility

The Problem of Estimating Volatility

We present in the appendix a discussion of the principles underlying the Black-Scholes option pricing model. While the mathematics behind the model is advanced and complex, the appendix shows that the model is a natural extension of the "binomial option pricing model." The important point to remember is that the future volatility of the underlying stock plays a crucial role in the model and that estimating future volatility is extremely difficult and becomes increasingly even more difficult the further out in time one attempts to estimate volatility. The Black-Scholes option pricing formula can provide reasonably good measures of the value of exchange-traded, short-term put and call options. Variants of this model produce value estimates for short-term (such as one to three months) options that are not only extremely close to one another, but that also track with considerable precision the actual market prices of these instruments. This is so because recent past volatility tends to be reasonably persistent over the short term. It is important to point out, however, that for longer-term (such as six months to one year) exchange-traded options, the Black-Scholes formula can produce a wide range of estimates, and actual market prices of traded instruments vary substantially from their predicted values. Unfortunately, volatility over the longer term is notoriously difficult to estimate and the longer the time the option has to run the greater the difficulty in arriving at an estimate of its value. This inherent limitation in option pricing models is exacerbated when one moves from so-called "long-term" exchange traded options (i.e., six months to one year) to employee stock options with lives measured in years rather than months. (Xilinx employee stock options were ten year options.)

The problem stemming from the fact that stock volatility is not constant over the longer term has long been recognized by market practitioners. Traders tend to put less reliance on Black-Scholes estimates as the time to expiration increases. The problem is widely recognized and is discussed in texts on option pricing such as the leading text by John Hull:

Complications Arising From the Special Features of Employee Stock Options

When one adds the complications that executive stock options do not vest immediately and are subject both to forfeiture and restrictions on the sale of the option stock, it is virtually impossible to put a precise estimate on the option's value. Each of these factors violates the assumptions underlying the Black Scholes model. Moreover, employee stock options generally have durations of five to ten years (the Minx options have 10 year lives) and, as noted above, the Black-Scholes formula has considerable difficulty even in pricing the longer-term six month to one year exchange-traded options. Finally, unlike exchange-traded options, employee stock options cannot be sold by the employee, violating one of the key assumptions of the Black-Scholes model.

It is widely recognized in the finance literature that the Black-Scholes model is unsuitable for employee stock option valuation, as noted in a recent article by Richard Friedman:

It is, of course, possible to attempt to adjust the Black Scholes model to account for many of the special features of employee stock options. Mark Rubinstein has proposed a rather ingenious model to do this.⁷ The model, however, uses 16 input variables, many of them difficult to estimate, and a wide range of estimates can be derived from the model. It is particularly important, as Rubinstein expressly states in his article, that he is not attempting to take into account incentive effects of the employee stock options, but rather is merely seeking to value the options granted to the employees. Rubinstein points out that the inherent subjectivity of the estimates required can allow firms to report values half or double those for other similarly situated firms. Rubinstein also considers use of "minimum value "accounting the primary method suggested by the Financial Accounting Standards Board for private companies. But even use of this minimum value method can lead to demonstrably inconsistent results for similarly situated companies as the terms of the options can easily alter the features of the employee stock option grant in a way that uses zero as the minimum option value.⁸

We conclude that it is impossible to measure the value of options granted to employees with any degree of precision or economic certainty,

Exercise Date Accounting

There is, of course, one valuation approach that would appear to avoid the ambiguities in valuation that have been discussed above. Under this approach, which we understand to be the approach of the IRS in this case as well as in a January 25, 2002 directive issued by the IRS to its agents, the "cost" would be recorded as the difference between the market and exercise prices on the date at which the executive exercised his or her options. ⁹ This approach obviously has the advantage of avoiding the inherent limitations of the option pricing models and avoiding the need for estimates of long-term future stock price volatility. The problem with this exercise date method is that while one obtains a precise calculation, it is unclear what this calculation measures. What is clear, however, is that this exercise date calculation does not represent an economic cost to the firm.

As discussed above, employee stock options uniquely are associated with both beneficial and detrimental effects upon shareholders. These effects are inextricably intertwined in that the positive incentive effects arise precisely because the employee stock options transfer ownership to the employees (and hence dilute the holdings of existing equity claimants) if stock prices rise and the options are exercised.

Shareholders' willingness to approve employee stock option plans is evidence of an expectation by shareholders that the size of the pie (i.e., the value of the firm) will grow faster than the percentage of the pie owned by current shareholders will shrink. Some researchers have attempted to estimate and separate the expected effects of the stock options, presumably including both the dilative effects and the agency and incentive effects at the date of grant. Many of them include analysts' long-term earnings growth forecasts as variables in the analysis. But such growth is itself apt to be a consequence of the incentive effects of the employee stock options so that the statistical calculations will associate the incentive effect with the growth forecasts, leaving only the dilutive effects to be attributed to the options. In other words, the calculation inadvertently attributes to the options all of their detrimental dilutive consequences, but not all of their benefits. The net result is that the calculation is distorted toward giving the appearance that options benefit initial stockholders less then they really do.

But the problem of separating the dilutive effects of employee stock options from their incentive effects becomes even more acute when considering the recording of costs at the time the options are exercised. The IRS's exercise date theory presupposes that the entire increase in stock value is a measure of the "cost" of the options. The empirical research shows, however, that the shareholders' and the market's expectations at the time of grant -- ex ante -- are that the positive incentive effects will at least equal the dilutive effects. Thus, the increase in stock price is decidedly not the dilutive effect on shareholders -- it represents that increase in stock price that was expected at least to offset the dilutive effects of the option grant. If in fact the positive and negative effects are perfectly in balance, there is no net cost to the firm. If, as some of the studies suggest, the positive incentive effects normally outweigh the dilutive effects, then there would be a net benefit rather than a cost to the options. The point is that simplistic use of the spread on exercise as a measure of the economic cost to the firm ignores the fact that at least part of the change in stock price is likely to result from the positive incentive effects of employee stock options. Further, the stock price can also rise for a host of reasons unrelated to employee stock options.

Moreover, we suggest that the spread upon exercise is far more likely to measure the positive incentive effects of the options as well as their economic cost. When employee stock options are substantially "in the money" (by which we mean that the current stock price is well above the option price), a reasonably efficient stock market will expect that the extra share certainly be issued and so the dilution cost will already be reflected in the stock price. In part, the subsequent increase in stock value will reflect whatever positive incentive and agency effects were created by the option grants themselves. It is certainly not defensible to consider this difference an economic cost to the firm or its shareholders.

Further, there is an obvious break in the logic chain of the IRS's position underlying its exercise date criterion. Employee stock options generally vest over a period of time and can remain outstanding for many years. The Xilinx employee stock options at issue in this case vested over five years and had ten year lives. The IRS contention is that current year spread on exercise -- which necessarily relates to options granted in earlier years -- is somehow related to the costs of jointly shared development projects in the current year. The problems besetting this approach are obvious. As a simple example assume that in years one through five, Firm A grants. stock options to its employees with five year vesting and 10 year life. In years two and three, Firm A's stock price doubles. In year six, Firm A and Firm B enter into a cost-sharing agreement and, as the IRS argues, they agree to share option "costs" on a spread on exercise basis. In year six, Firm A's employees exercise all stock options granted in year one, resulting in a large section 83 deduction for Firm A. Under the IRS position, Firm B would be liable for the section 83 deduction enjoyed by Firm A up to the amount of Firm B's cost-sharing ratio as an additional cost of the joint development project -- even though the grant and exercise of the options had nothing to do with the joint development effort. Firm B would thus be required to underwrite the rise in stock price in Firm A, even though, by definition, the options were granted five years before the development work began and the run up in stock price occurred two years before the development work began.

The point is that the change, in price in Firm A stock may be wholly unrelated to anything connected with the joint development effort with Firm B. For example, Firm A can have an entirely different product, Product A, that is completely outside the scope of the development effort. If suddenly demand for Product A takes off and the firm starts selling huge quantities of the item, then Firm A's stock price may well increase substantially in value for that reason alone. Under the IRS hypothesis, with the voluntary agreement of Firm B, Firm A would charge Firm B for this increase in stock price through the spread on exercise of stock options held by employees working on the joint development effort. Obviously, this rise in stock price would have nothing to do with the joint development effort with Firm B.

Even if Firm A started granting options only at the same time as the joint development project began, the risk to Firm B of entering into a joint development agreement under the IRS exercise date assumption would be sufficient to cause rational firms never to agree to such an arrangement. Specifically, in the IRS exercise date scenario, Firm B essentially writes a naked call option on the stock of Firm A. Writing a naked call option on the stock of Firm A is akin to Firm B selling short the stock of Firm A, its business partner. Firm B has unlimited liability exposure for increases in the stock price of Firm A. In other words, the better Firm A does (regardless of whether such success is based on the joint development effort or occurs for wholly unrelated reasons), the more expensive the joint development effort is for Firm B. On the other hand, the empirical research shows that a representative Firm A will have incurred no cost attributable to the employee stock options. Further, Firm A will receive additional cash when its employees exercise their options in an amount equal to the exercise price of the option which equaled the market price of the stock at the date of grant. No rational firm would enter into such a relationship.

Moreover, we understand that under the law, the option holders' decisions as to date of exercise and date of sale of stock will determine the amount of the firm's section 83 deduction. These choices surely do not affect the firm's revenues or costs in one way or another. Consequently, use of the section 83 deduction amount makes no sense as an indicator of any corresponding cost that would be shared in a cost-sharing agreement.

Conceptually, neither the grant nor exercise of employee stock options is a direct cost to the firm. At grant, the employees theoretically are willing to trade off some cash payment for option grants, thus preserving the firm's cash. At exercise, the employees pay cash to the firm in an amount equal to the fair market value of the stock at the time of grant, again increasing the firm's cash. As discussed above, the issue of the options does not reduce the firm's earnings but rather potentially redistributes a portion of the equity claims on the firm from existing shareholders to the option holders. In theory, the existing shareholders are willing to give up some equity to the employees on the presumption that the beneficial incentive and agency effects stemming from the options will cause the firm's value to grow more quickly.

While the issue of employee stock options has no direct cost effect upon the firm and is expected to improve the firm's performance, there none the less is a possibility that the issue of options can indeed produce an economic cost to the firm. This is so because the firm's shareholders and the market may believe that the dilutive effect of employee stock options is greater than the anticipated benefits from the agency and incentive effects. If the shareholders and the market were to believe the detrimental effects to outweigh the beneficial effects, then the firm's stock price would fall in response to this expected diminution in the value of the firm. If stock prices declined, then the firm's equity cost of capital would be increased. An increase in the firm's equity cost of capital can legitimately be interpreted to constitute a net economic cost to the firm. On the other hand, if the market anticipated that the beneficial effects of options would equal or outweigh the dilutive effects, then the firm's stock price would remain unchanged or increase above that which would otherwise have prevailed. If the stock price remains unchanged or is increased then the firm's equity cost of capital remains unchanged or would decrease, with the issue of the options then having no net economic cost to the firm.

Whether the issue of employee stock options then constitutes such an economic cost to the firm is an empirical question that must be examined by study of the effect of employee stock options on stock price. A number of investigators have attempted to measure empirically whether the net effect of employee stock option grants tend to raise or lower stock price in reality. In this section, we briefly review some highlights of the empirical work. We conclude that while these studies produce different estimates of the effect of option grants on share prices, most find a positive effect on shareholder wealth and none of the studies provides convincing evidence that the act effect on share prices is negative.

Some Methodological Problems

There are some very difficult conceptual, and methodological problems involved in all of the analyses we will review. What we seek to determine is whether the value of options granted has a positive or negative influence on sham prices. Certainly, we know that ordinary expenses tend to depress share prices. For example, if a firm's earnings decline with increased expenses we can expect the stock price to suffer. But we have seen above that the fair value of options granted can only be estimated and the estimates used are far from precise. One method used in the studies is to estimate the value via a Black-Scholes formula as used in the footnotes of the financial statements of the different firms. Unfortunately, since each firm estimates the value of option grants using different assumptions, there can be substantial differences among option expense estimates even for similarly situated firms. Even more fundamentally, the best yardstick available to measure the value of employee stock options -- the Black-Scholes option pricing model -- cannot and does not measure the value of employee stock option grants with any reasonable degree of precision or economic certainty.

There is an even more serious statistical problem to be overcome. Most of the empirical studies attempt to determine the effect of option expense on share price. For this purpose, a number of the empirical studies have used firms' Black-Scholes based option expense estimates from the firms' FAS 123 footnote disclosures. But as noted earlier, the amount of option expense estimated via the Black-Scholes model depends on the price of the shares. As a result, these empirical studies entail a statistical difficulty known as a "simultaneity problem." Option expense may influence share price but share price also influences option expense. Different studies deal with this problem in different ways. In some studies, option expense is estimated in an artificial way and it is hard to know if the empirical results are simply artifacts of the particular method of estimation.

Finally, many of the statistical studies attempt to show the relationship of stock prices to the following explanatory variables: earnings, book value, expected future growth, and the fair value of option grants. If a negative sign is obtained on the option expense variable (i.e., a greater value of options issued is associated with lower stock prices) at least one study has interpreted the result as indicating that options grants depress share prices. But, in reality, all that is being measured is the negative dilutive effect of options. The positive incentive effects are likely to be subsumed by the expected growth variable. Therefore, we can not take any one of these studies as dispositive. Nevertheless, there have been a substantial number of papers written on the subject and fortunately the papers do suggest a tentative conclusion.

A Review of the Empirical Studies

Below we summarize the major conclusions of the empirical studies that attempt to measure the effect of ESOs on stock prices.

Conclusion

Numerous investigators have attempted to measure the net effect of employee stock option grants on the firm and its shareholders. The majority of the studies find that employee stock option programs have a positive net effect on share prices. However, considerable measurement and econometric problems affect all the analyses and it is not surprising that some studies are unable to measure any statistically significant effect at all upon share prices (i.e., the employee stock option programs have no measurable effect on share price). The one unpublished study (Hillegeist and Penalva) that appears to find a significant negative effect on share prices from the value of options granted does not provide robust results. In any event, Hillegeist and Penalva find that firms which increase grants experience better future performance. Thus, even accepting their findings at face value, the net effect of employee stock option grants is a positive one for the firm and its shareholders. Thus, the empirical studies performed so far establish that the issue of employee stock options has either no effect or a positive effect on stock price. Thus, the empirical studies establish, at a minimum, that the issue of employee stock options has no general and measurable economic cost to the firm.

It may be argued that the purely dilutive effect of the issue of a stock option does have a clear opportunity cost because it reduces the price of the firm's shares since it reduces the price below what it otherwise would have been. But the evidence indicates that in general the issue of employee stock options has incentive and agency effects that work in the opposite direction. On average, these incentive and agency effects more than offset the dilutive consequences. Therefore it is clear that any such net opportunity cost must typically be zero or negative. That is, typically there can be no such opportunity cost at all.

An example underscores this point. Assume a firm can either grant an [sic] at the money option for 100 shares to an employee or sell a warrant on equivalent terms into the marketplace.¹¹ If the firm opts to grant the employee stock option, the employee will theoretically take less current cash compensation. As we have discussed, the empirical evidence establishes that the market views the positive incentive and agency effects at least to be equal to the negative dilutive effects. Thus, the stock price does not fall, even though each shareholder's ownership has been diluted. If the firm takes the alternative approach of selling the warrant into the marketplace, the results are not the same. If the warrant is sold, the firm does indeed receive cash. However, this cash receipt will be offset in whole or in part by the current cash compensation foregone by the employee in the other alternative, as the employee will now need to be paid additional cash in lieu of options. The firm's shareholders will also suffer a negative dilutive effect from the sale of the warrant, as they would in the case of the option. However, the sale of the warrant does not generate the positive incentive and agency effects peculiar to the grant of employee stock options. Thus, the sale of the warrant into the marketplace is less valuable to the current shareholders than the grant of the employee stock options. Put slightly differently, because the firm can grant the employee stock option without decreasing the price of its stock, it can sell an additional warrant or share of stock at the same price at which the warrant or share would have sold if the employee stock option had not been granted. The firm can do both. The grant of the employee stock option does not raise an opportunity cost to the firm.

The IRS may argue that there is an opportunity cost of a different sort, that an employee stock option issued when the price was $10 but exercised when the stock price reached $50 entailed an opportunity cost of $40 to the firm. But that is no different than asserting an "opportunity cost" to the firm of issuing a share at a time when its price was $10 rather than postponing the issue to some future distant date when its price might be $50. Clearly, neither of these entails reasonable substitute choices for the firm. For example, for the firm that needs money today it is not an equivalent choice to obtain it, say, four years later. Indeed, this purported opportunity cost calculation is even more severely damaged by the fact that the rise in stock price may itself well be a partial consequence of the issue of the options.

On Arm's-Length Evaluation of Costs

There are many cost elements for which data are not readily knowable or where the information is not known at all. There are even cases in which it is unknowable in principle. As a result, accountants frequently and quite justifiably are driven to adopt simplifying proxies that can be used for calculation purposes, even when they demonstrably have little or no relation to the underlying reality. A prime example is a fully allocated cost that ostensibly purports to specify which portion of some total outlay that inextricably benefits several outputs of a firm is to be considered the responsibility of each of the different benefiting outputs. Since there is no way of assigning the unassignable, the accountant is driven to adopt some arbitrary criterion, such as the values or the weights of the different products, as the basis for the apportionment of the unassignable costs and calculation of the "full costs" of each of the individual products. Similarly, conventions such as straight line depreciation permit easy workability but may have little relationship between the numbers generated by the calculation and the underlying reality. True values, actual costs and relevant practices of reality, however, cannot be determined in this way. Rather, wherever possible, one must took to actual practice -- to the workings of reality to get at such matters. Moreover, it is essential in this search to focus attention on a reality that is not distorted by the interests of the participants. Thus, where assignment of cost responsibility is at issue, and the practices that are followed in that assignment are in question, we are led to took for arm's-length transactions as the most reliable source of the requisite information. Where the parties in question are independent and neither has any interest in biasing matters in favor of the other, one can be confident that the process will not be systematically distorted and that it will not yield distorted information.

The matter here in question is precisely of this variety. The issue is whether the issue of a stock option creates any cost that we can expect to be included in a cost-sharing agreement entered into at arm's-length, and whether in actual practice such possible costs are systematically included or included at all. Reliance on the evidence derived from arm's-length transactions relieves us of the need for recourse to arbitrary valuation and costing procedures, and instead bases our inferences on the firm foundation of market experience. We therefore turn to examination of the pertinent inferences that can be drawn from the examination of arm's-length relationships in the relevant arena.

Stock Options in Arm's-Length Cost-sharing Contracts: the Incentives

There are several reasons why we cannot normally expect cost-sharing contracts entered into at arm's-length to include stock options as generators of costs that are to be stand. This, of course, does not mean that exceptions are impossible. But it does mean that if the empirical evidence is consistent with the analysis, it is illegitimate on the arm's-length standard to claim that such contracts can be deemed implicitly to include costs that are attributable to any issue of stock options by one of the cost-sharing firms. This conclusion, of course, becomes particularly compelling when one cannot even pretend to be able to provide reasonably accurate evaluations of these purported costs, a difficulty the analysis of the preceding section has demonstrated.

The three primary reasons why it is rational for cost- sharing firms to omit stock options from their agreements is easily made clear from our earlier discussion in this statement. The first is the fact that that it is at least arguable that stock options generate no costs at all. As was shown here earlier, the stock options are never a cost to the firm, in the standard sense. At most they constitute a redistribution among old and new stockholders, with no change in the total profits to be allocated among them. In that case, they would constitute a cost to some stockholders with an offsetting benefit to others. But that is not all there is to the story, because the available evidence does not reject the hypothesis that the issue of stock options, presumably through its incentive effects, generally serves to increase the profits of the firm, and thereby becomes beneficial to all of the stockholders, as well as the firm.

The second reason for exclusion of stock option considerations from arm's-length cost-sharing agreements follows from the first. Whatever cost, if any, is to be attributed to an issue of stock options, it is patently very different from a direct financial outlay. The qualitative character of the two types of "costs" raises very cogent questions about the proper way to aggregate them. The old problem raised by any attempt to add apples and oranges clearly re- arises here.

The third reason why cost-sharing agreements cannot be expected to incorporate stock-option components stems from the difficulty of evaluating these options, and the inaccurate results that are obtained even from the most sophisticated methods currently available.

All of this means that even if it were believed that such options entail some cost, their inclusion in a cost-sharing agreement might simply make the agreement complex, ill defined and unworkable. Rather than serving as an effective and easily utilized tool beneficial to the participating firms, such a contract would simply invite contention, increase the costs of operation of the agreement, and very likely make its adoption threaten to be more trouble than it is worth. It can only be concluded that it should not be surprising if it turns out to be difficult to find any real arm's-length contracts that include any cost component associated with the issue of stock options. There is very good reason for those who enter into such contracts to avoid the inclusion of any such element.

But that is not the end of the story because the IRS favors a particular approach to estimation of the purported cost of the options that it takes to be included implicitly in. cost-sharing agreements. It will be shown next that this approach makes it even considerably less likely that any rational firms would enter into an agreement that followed the IRS approach to the sharing of the asserted costs of stock options.

As we understand it, the IRS favors the valuation of stock options on the basis of their price at the future date when they are exercised by their recipients. Under such a procedure the cost to the firm of the granting of the options would be based on the difference in the price at the date the options are exercised and the price at the time the options were issued. As discussed above in detail at pages 21 to 25, this exercise date calculation is not a valid measure of any economic cost to the firm and no rational firm would enter into such a relationship.

The most obvious difficulty raised by this approach is the degree to which it compounds the uncertainties for current evaluation of the value of a stock option at the date it is granted. If the costs are not to be reported on this basis transaction by transaction at the dates the options are exercised, but at the date the options are issued, this does indeed entail a marked aggravation of the problems. There is no way in which one can even be certain in advance whether the options will ever be exercised, as surely they will not be if the firm's stock price falls. Even if one can be confident that they will all be utilized at some future dates, those dates are likely to vary from one option holder to another and even one holder may elect to use different portions of his options at different times. The difficulty of foreseeing the prices that will prevail at those dates is equally clear. The bottom line is that the exercise- date approach to valuation vastly multiplies the uncertainties and computational problems that would beset any cost-sharing firms that contemplated such a course.

But this, too, is not the end of the story. An agreement made on such terms would commit the firm that undertakes to bear part of these costs to a very risky undertaking. Its future cost-sharing payments would fall due at unpredictable dates and at unpredictable prices. In effect, the contracting firm that undertook to do this would have written a call option on the stocks of the other participant in the cost-sharing arrangement. This is akin to selling short the stock of one's business partner. One can hardly imagine that firms would commonly be willing to cater into such agreement voluntarily and at arm's-length.

Stock Options in Arm's-Length Cost-Sharing Contracts: Empirical Evidence,

Even though it should be recognized that inclusion of stock options in arm's-length cost-sharing arrangements is implausible as a matter of economic theory and analysis, it is still conceivable that it does occur in reality. This possibility requires examination of the available evidence. The evidence is necessarily incomplete, because them is a profusion of such contracts and many of there are treated as proprietary and are consequently not open to general inspection. Yet the pertinent information is far from being totally unavailable. As will be shown next, while its incompleteness does not permit us to judge whether or not there are ever any contracts of the sort at issue in which stock options do play a role, we can conclude from the evidence that there is a very substantial number of such arm's-length contracts from which consideration of stock options is precluded. Moreover, neither we nor, apparently, the IRS have been able to find a single example in which stock options are given any role of the sort the IRS proposes to assign to them. The empirical evidence thus confirms the economic theory and analysis. Further, on the basis of the available evidence on pertinent arm's-length transactions, from an economic perspective it is simply illegitimate for the IRS to impute to a cost-sharing agreement in which the issue is not mentioned the presumption that stock options are implicitly included in the agreement.

Our empirical evidence rests on contracts for service between the private sector and the federal government of the United States, which are entered into voluntarily and at arm's-length and in which the government undertakes to reimburse the private firms for the cost incurred in supplying the government with the items provided under the contracts. In particular, we will deal with the billions of dollars of services purchased each year by the government from the private sector and classified as research, development, test and evaluation services ("RDT&E"). We understand that these contracts with the private sector for RDT&E, like all purchases of goods and services by the Executive Agencies of the Federal Government, were governed throughout the years at issue by the Federal Acquisition Regulations System ("FARS").¹²

In purchasing RDT&E or other services from the private sector, the United States Government is precluded from paying for any value of at-the-money options issued by contractors to their employees.¹³ The United States Government takes the position that the grant of employee stock options does not constitute an allowable compensation cost and prohibits private sector contractors from charging the United States for the value of stock options granted to employees working on RDT&E service contracts (or any other service contract).¹⁴

The Directorate of Information, Operations and Reports, of the U.S. Department of Defense ("DOD") publishes information each year regarding the DOD's purchases, including reports on annual DOD purchases of RDT&E in contract amounts of more than $25,000. These reports for the years 1990 through 2000 indicate that each year between two and three thousand different firms have been involved in these transactions. The amounts entailed are well in excess of 16 billion dollars per year, or some 200 billion dollars over the decade of the 1990s. Thus, neither the amounts of money involved nor the number of contracting parties can be considered insignificant.

Moreover, the contracts are evidently voluntary, arm's- length arrangements. None of the contracting firms is obligated or required to enter such a contract and thereby to accept its terms. But, despite the explicit exclusion of stock options and any associated costs from the amounts to be reimbursed, this great multitude of suppliers willingly and consistently has accepted those terms.

Perhaps even more telling is the admittedly complete absence of any contrary evidence that has been found by the IRS. The IRS has admitted that its inclusion of the "cost" of employee stock options in the costs to be shared in the shared-cost contracts is not based on any actual transactions between unrelated parties. Further, the IRS has admitted that it possesses no evidence or written contracts (cost-sharing agreements or otherwise) that demonstrate that any unrelated party at arm's-length actually pays for the "cost" of employee stock options issued to or exercised by the employees of the other contracting party. In particular, the IRS has admitted that it possesses no evidence that unrelated parties base payments for employee stock options on the amount of the tax deduction under section 83.¹⁵

But even if there do exist some such contracts, that surely is hardly enough to show that standard arm's-length practice is consistent with the IRS contentions. And economically, standard arms-length practice, rather than the possibility of exceptional cases, is what is truly the pertinent issue.

Conclusion

The role of employee stock options is complex and continues to be investigated in the economic literature. Much remains to be teamed about the subject. But a good deal is well understood about the topic. We know that their issue can, at least in principle, be beneficial both to the issuing firm and to all of its stockholders. We know, consequently, that they need not entail a cost, as the term is normally and appropriately interpreted. We know that even the value of the employee stock options is not in general accurately and unambiguously determinable. Consequently, a proposal to base the calculation of their purported costs on such a valuation can hardly be expected to provide figures that can pretend to reliability. There is even less logic to a proposal to base evaluation of the purported costs of employee stock options on the spread between the exercise price and the current market price of the stock at the date of exercise, an approach that is wholly indefensible from an economic standpoint. Perhaps most important, the appropriate criterion, the arm's-length standard, has yielded no evidence that supports the IRS position and much evidence that contradicts the contentions that underlie that position. Consistent with our economic analysis, not a single example has been found in which unrelated parties in the marketplace agreed to share the value or purported costs of employee stock options. At the very least, it must be concluded that Xilinx's actions in excluding the value or "cost" of employee stock options from its cost-sharing arrangement were consistent with the theory and practice of arm's-length dealings by unrelated parties in the marketplace.

i) Exchange-traded options

A stock option, just as the name implies, gives the buyer the right (but not the obligation) to buy or sell a common stock (or group of stocks) at a specific price on or before a set date. For example, a call option on IBM might cost the buyer $ 10 a share (the option premium) expiring the third Friday in July (the expiration date) with an exercise price of $100 a share (the exercise or strike price). Thus, for a premium of $10, the buyer of this call option has the right to purchase a share of IBM at $100 at anytime up through the third Friday in July. The seller (or writer) of the option receives the premium and takes on the corresponding potential obligation to sell the share at the contact price. A put option reverses the situation. A put on IBM gives the holder the right to sell IBM shares at a specific price. The seller of the put (called the writer) takes on the potential obligation to buy the shares.

Exchange-traded options exist on the major traded individual stocks as well as on a variety of stock indexes, bonds, and foreign currencies. Options on the S&P 500 index, the NASDAQ 100, and the Dow Jones Industrial Average are traded in Chicago. In addition, options are traded on a variety of smaller capitalization indexes as well as specific industry indexes. The volume of trading in basic options and futures has at times actually exceeded the volume of trading in the underlying assets. Option holders are free to purchase or sell options at any time up until the expiration date.

ii) Employee stock options

These are call options granted to employees of corporations as a form of incentive compensation designed to motivate the employees and to align their interests with those of the shareholders as a group. The employees benefit only to the extent that the share price increases and thus the well being of the employee is tied directly to that of the shareowner.

The typical procedure is to grant the option with an exercise price equal to the current market price. Thus, if IBM were selling at $100 per share today, the employee stock option would have an exercise (strike) price of $100. Thus far, employee stock options (ESOs) closely resemble those traded on the major options exchanges. But there am major differences between the two types of options.

ESOs are typically very long-term instruments with expiration dates often 10 years from the grant date. Few ESOs are exercisable at the time they are granted. ESOs are normally subject to a vesting schedule, typically over a four-year period, before they become exercisable. After the ESOs vest, they can be exercised at any point in time up to the end of the options' life. However, unlike exchange-traded options, ESOs cannot be traded or sold. If the employee leaves the firm or is terminated, the unvested ESOs must be forfeited and the vested ESOs must be exercised (assuming they are in the money) within a short period of time after departure. The tax treatment of ESOs also differs from the treatment of exchange-traded options. Finally, when an employee exercises her ESOs, the firm typically issues new shares in exchange for the employee's cash payment of the exercise price.

b) Valuation of Options: The Role of Volatility

The process by which options are valued is often misunderstood by the general public. The key element making some options far more valuable than others (other than, of course, the time period to expiration) is the characteristic volatility of the underlying stock. A simple example will make the situation clear. Suppose we have two $30 stocks and three-month call options on each, with an exercise (strike) price of $30. But let's assume that one of them tends to be highly volatile (say, a tech stock) and the other far more stable (say, a company selling staple consumer goods). We will assume that future prices in three months will be one of five values -- all equally likely.

Exhibit 1 shows the possible future prices for the two stocks under a high and low volatility assumption. We'll assume that the price in three months is equally likely to take on each of the five possible values listed so that on average the price has an expected future price of $30. No greater expected appreciation is assumed for either stock. But for the high-volatility stock, the highest price is higher and the lowest price lower than is the case for the low-volatility stock. Now, I will show that an option on the high-volatility stock is more valuable than an option on the more stable stock.

c) The Binomial Option Pricing Model

Let's now develop a simple binomial model showing how the value of an option could be determined if its possible returns, both when results turn out to be favorable and when they are unfavorable, are known in advance. Assume a stock currently priced at $80 can take only one of two values at the end of the option period, assumed to be one year. It can either go up to $100 or down to $70, as shown in Exhibit 2. Because there are only two possible outcomes, the model is called a "binomial model." Further, assume there is a call option on the stock expiring in one year with a strike price of $80 (the current market price).

[54] As a step in determining the value of an option, we must first show how it can be used to form a hedge against the risks of stock ownership. This will be done by means of an illustration. We will show that with the payoff possibilities as given here, the following portfolio provides a perfect hedge, i.e., it eliminates risk completely.

Let me now explain how to price the option and how to derive the hedge ratio. We have seen that a portfolio of 2/3 share and one call written is perfectly hedged. It has the same value whatever the price of the stock. In a well-functioning market any investment that has a guaranteed payoff cannot earn more than the risk-free rate of interest (which we normally consider to be the U.S. Treasury Bill rate).

We let C = the price of the call option. What is the amount invested in the hedged portfolio? It is 2/3 (80) - C. What is the payoff? The payoff, if the stock goes up or down, is 46 2/3. The amount invested can only earn the risk-free rate of interest, which we will denote as rF. Therefore, it must be true that

By introducing some simple notation, we can now easily derive the hedge ratio (H). Let CU and CD be the (intrinsic) value of the call option at the upper (U) and lower (D) stock prices at the end of the period (20 and 0 in our example). SU and SD will be, the upper and lower stock prices (100 and 70 in our example). The hedge ratio is easily derived in Exhibit 3 and is shown to be 2/3.

To fine [sic] it, we note that whether the stock attains its upper or lower value, the investor's value will be the value of the call option plus the value of the stock, i.e.,

C_D + HS_D or S_U + HS_U

which must be equal if the hedge is perfect. Therefore, solving for H, we obtain the basic formula of Exhibit 3.

We'll assume that in the "up" state the price increases by 5 percent and in the "down" state it decreases by 3 percent. Suppose that the initial price was 100, so that the two possibilities in the next period are as shown below.

The Black-Scholes Pricing Formula (shown in Exhibit 6) appears to be recondite and forbidding and it does require higher mathematics for its derivation. It can be viewed intuitively, however, as simply the logical extension of the binomial model. While the option value does not depend directly on the expected return from the stock, it does so indirectly by including the current stock price (which depends on the company's risk and return characteristics) and crucially on the stock's volatility -- captured in the standard deviation term.

² In reviewing the unpublished working paper, we noted what we believed to be problems in certain statistical aspects of the analysis. After contacting the authors and obtaining their data set, we asked Dr. Atanu Saha to assist us by making certain alterations to and re-running the various regression equations in the unpublished working paper. Dr. Saha's analysis is summarized below and is set forth in fall in his economic report.

³ This statement reflects who is known as the "weak form" of the Efficient Market Hypothesis ("EMH"). Many leading economists have contended under the "strong form" of the EMH that stock prices anticipate all information, regardless of whether the information is publicly available or widely known.

⁴Both Professors Black and Scholes and Professor Merton cited a paper we wrote with Richard Quandt on the valuation of convertible securities in their Nobel Prize Winning articles. William J. Baumol, Burton G. Malkiel, and Richard E. Quandt, "The Valuation of Convertible Securities," Quarterly Journal of Economics, Vol. 80, February 1966, pp. 48-59.

⁵John C. Hall Introduction to Futures and Options Markets, 3rd Ed., 1999, Practice-Hall Chapter in the Black- Scholes Model, pp. 332-333.

⁶Friedman. R., 2001. "What Are My Options Worth?" Article on the web site of My Stock Options.com.

⁷Rubinstein, M., 1995. "On the Accounting Valuation of Employee Stock Options." The Journal of Derivatives, pp. 8-24.

⁸Rubinstein, op. cite, p. 19.

⁹As described below, the IRS exercise date position is not quite this simple, as it is actually based on the amount of the section 83 deduction recorded by the U.S. participant.

¹⁰We were curious why the Hillegeist & Penalva working paper results were inconsistent with all of the other empirical analyses. Upon inspection of their regression specification and statistical techniques, we noted several statistical techniques that were questionable. We asked Dr. Atanu Saha of the Analysis Group to contact Professors Hillegeist and Penalva and to obtain their data set. We then asked Dr. Saha to re-run their particular Hillegeist and Penalva regressions after correcting the shortcomings we perceived in their particular specification of the regression equations and the statistical techniques. After adjustment for these items, the Hillegeist & Penalva regressions are consistent with the other empirical studies and show that the relationship between estimated option expense and share price is not statistically significant from zero. In other words, the revised Hillegeist & Penalva regressions show that there is no measurable economic cost to the issuance of the options. The details of the work performed by Dr. Saba under our instruction am set forth in his declaration.

¹¹For purposes of this example, we will ignore the problems involved in selling an equivalent warrant into the market, including issues relating to vesting, forfeiture and non- transferability.

¹²48 C.F.R. § 1.101; 48 C.F.R. § 31.103. The FARS governed both commercial and military purchases by the United States. 48 C.F.R. § 1.101; 48 C.F.R. § 31.103.

¹³Section 31.205-6 of the FARS governed compensation for personal services, including RDT&E services, during the years at issue. 48 C.F.R. § 31.205-6(a).

¹⁴48 C.F.R. § 31.205-6(i).

¹⁵Petition ¶¶ 5.a.(37)-(45) & 5.a.(50)-(54) and Answer ¶¶ 5.a.(37)-(45) & 5.a,(50)-(54).