Menu
Tax Notes logo

MULTINATIONALS COMMENT ON COST-SHARING RULES.

JUN. 19, 1992

MULTINATIONALS COMMENT ON COST-SHARING RULES.

DATED JUN. 19, 1992
DOCUMENT ATTRIBUTES
  • Institutional Authors
    Price Waterhouse
  • Cross-Reference
    IL-372-88

    IL-401-88
  • Code Sections
  • Subject Area/Tax Topics
  • Index Terms
    related-party allocations
  • Jurisdictions
  • Language
    English
  • Tax Analysts Document Number
    Doc 92-5700 (69 original pages)
  • Tax Analysts Electronic Citation
    92 TNT 135-34

 

=============== SUMMARY ===============

 

Price Waterhouse, Washington, on behalf of 17 multinational companies, has submitted extensive comments on the intercompany pricing and cost-sharing provisions of the proposed regulations under section 482.

The commentators' fundamental objection to the proposed regs "is that they elevate one method (the 'comparable profit interval') to a priority rule for making intercompany pricing determinations." According to the authors, "In many instances, appropriate measures of 'normal' profitability may be useful as a screening method for identifying potential pricing issues and can be used to determine prices, but industry averages are not a substitute for the facts and circumstances of real world cross border transactions when such transactions provide clearer comparables."

The major points covered in the Price Waterhouse comments include the sound-business-judgment test; choice of a tested party; the comparable-profit-interval test; data availability and reliability; foreign exchange gains and losses; the periodic- adjustment requirement; and recommended safe harbors.

 

=============== FULL TEXT ===============

 

June 19, 1992

 

 

Department of the Treasury

 

Internal Revenue Service

 

Office of the Chief Counsel

 

P.O. Box 7604

 

Ben Franklin Station

 

Washington, DC 20224

 

 

ATTN: CC:CORP:T:R(INTL-0372-88) and (INTL-0401-88)

 

 

Enclosed for your consideration are comments submitted by 17 multinational companies (see list enclosed) on the proposed regulations under Section 482 issued in January 1992. The comments relate to the intercompany pricing and cost-sharing provisions of the proposed regulations.

We would be happy to discuss the comments further with you at your convenience. Please call Larry Dildine at 202/822-8591 or Robert Patrick at 202/822-8555 if you have any questions or would like to set up a meeting.

Sincerely,

 

 

Price Waterhouse

 

 

Enclosures

 

 

COMMENTS SUBMITTED

 

TO THE INTERNAL REVENUE SERVICE

 

ON PROPOSED SECTION 482 REGULATIONS

 

Prepared by Price Waterhouse

 

June 19, 1992

 

 

COMMENTS ON PROPOSED REGULATIONS

 

UNDER SECTION 482

 

ISSUED ON JANUARY 24, 1992

 

 

Submitted by: Abitibi-Price Sales Corporation

 

Caterpillar, Inc.

 

Dresser Industries, Inc.

 

E. I. DuPont de Nemours, Inc.

 

Eastman Kodak Company

 

Genentech, Inc.

 

General Motors Corporation

 

Goodyear Tire & Rubber Company

 

Hewlett-Packard Company

 

IBM Corporation

 

Johnson Controls, Inc.

 

Kellogg Company

 

Levy Strauss Co.

 

3-M Company

 

Pier 1 Imports

 

Tenneco, Inc.

 

Time-Warner, Inc.

 

 

Assisted by: Price Waterhouse in consultation with

 

Steve Lainoff, King & Spalding.

 

 

COMMENTS SUBMITTED TO THE INTERNAL REVENUE SERVICE

 

ON

 

PROPOSED SECTION 482 REGULATIONS

 

 

TABLE OF CONTENTS

 

 

I. INTRODUCTION

 

II. SOUND BUSINESS JUDGMENT AND FORECASTS

 

III. TESTED PARTY

 

IV. THE COMPARABLE PROFIT INTERVAL (CPI)

 

V. DATA AVAILABILITY AND RELIABILITY

 

VI. PROFIT LEVEL INDICATORS

 

VII. PROFIT SPLITS AND ALTERNATIVE METHODS

 

VIII. EXCHANGE RATES

 

IX. PERIODIC ADJUSTMENTS

 

X. SAFE HARBORS

 

XI. COST SHARING

 

 

APPENDIX A

 

APPENDIX B

 

 

COMMENTS SUBMITTED TO THE INTERNAL REVENUE SERVICE ON PROPOSED SECTION 482 REGULATIONS

I. INTRODUCTION

These comments are being submitted in response to the request for comments on the proposed section 482 regulations published in the Federal Register on January 30, 1992. The participants submitting these comments represent a broad cross section of U.S. companies engaged in international trade and investment who are affected by significant developments in tax rules governing the transfer of products and intangibles among related parties.

We believe that in a number of important respects the proposed regulations have appropriately adopted an increased degree of flexibility over certain suggestions contained in the IRS/Treasury White Paper of October, 1988. These improvements include the testing of the arm's length standard over a period of years and recognition of a range of acceptable outcomes rather than a single "correct" price. We also support the decision to eliminate the long-standing priorities in the section 482 pricing regulations for sales of tangible property. Nevertheless, we are concerned that the regulations, as proposed, do not appropriately address many of our factual situations and that they significantly increase the risk of double taxation for many U.S. taxpayers.

A. OVERVIEW

Our fundamental objection to the proposed regulations is that they elevate one method (the "comparable profit interval" to a priority rule for mailing intercompany pricing determinations. In many instances, appropriate measures of "normal" profitability may be useful as a screening method for identifying potential pricing issues and can be used to determine prices, but industry averages are not a substitute for the facts and circumstances of real world cross border transactions when such transactions provide clearer comparable.

The comments contained in this document generally follow the order established by the proposed regulations. A summary of the major points included in this submission follows.

SOUND BUSINESS JUDGMENT. The sound business judgment test should be refined to clarify that the employment of sound business judgment is determined as it is in the market -- with knowledge of those facts listing at the time the judgment is made. Further, the use of reasonable business forecasts should be specifically permitted. (Chapter II)

TESTED PARTY. We recommend that a taxpayer be permitted to treat its non-U.S. operations (or the U.S. operations of a foreign taxpayer) in the aggregate as the tested party, with certain exceptions. In addition, where both parties to a related party transaction are performing comply functions, a taxpayer's choice of tested party, where the CPI has been properly applied, should not be challenged by the IRS. (Chapter III)

COMPARABLE PROFIT INTERVAL. We strongly urge that the CPI test, in the absence of a CUP or matching transaction, be placed on the some level of priority as other inexact comparables. The choice between such methods would be based on the quality of the available data, the closeness of the comparables, and the logic of the circumstances. (Chapter IV)

DATA AVAILABILITY AND RELIABILITY. In applying the CPI test, the IRS should: 1) be required to use data that is also available to taxpayers at the same level of detail and 2) clearly provide that tested-party data and third-party data be adjusted, whenever possible, for differences in local accounting standards, market conditions, exchange rates, and other measurable factors. (Chapter V)

PROFIT LEVEL INDICATORS. When a CPI test is applied, the regulations should convey the need for careful definition of PLIs and associated data to fit the specific facts of the tested operation, and should expressly caution against mechanical solutions. (Chapter VI)

PROFIT SPLITS AND OTHER METHODS. Other methods, including inexact comparables (with appropriate adjustments), and under certain circumstances a workable profit split, should be made available without the requirement of a CPI test. (Chapter VII)

FOREIGN EXCHANGE. The tested party's data should be adjusted for differences in economic conditions of the tested party that are not generally faced by the uncontrolled parties used as comparables. In particular, exchange gains and losses should be explicitly allowed in applying the CPI test. (Chapter VIII)

PERIODIC ADJUSTMENTS. We recommend that the periodic adjustment requirement be eliminated, or, at the very least, that it be significantly narrowed in accordance with the approach adopted in the section 6038A regulations. We are concerned that U.S. treaty partners will not accept this methodology, thereby creating double motivation without any realistic possibility of relief. (Chapter IX)

SAFE HARBORS. We recommend the adoption of two safe harbors -- first, a rate of return safe harbor for parties performing simple functions, and second, a safe harbor based on a division of profits between the licensor and licensee. (Chapter X)

COST SHARING. The rebuttable presumption in favor of the cost to operating income ratio should be eliminated. Any allocation method that reasonably reflects anticipated benefits over time should be acceptable and should not lead to after-the-fact adjustments based upon subsequent cost/income ratios. Any utilization of after-the-fact data, such as a cost/income ratio based upon actual experience, should be employed solely as a safe harbor. (Chapter XI)

B. GENERAL COMMENTS

In addition to the specific comments contained in this submission, we believe that it is important to comment on certain positions taken by IRS and Treasury officials since the publication of the proposed regulations. Certain of these comments are discussed in the specific sections below, but two are of a fundamental nature. First, it has been repeatedly argued that the rules contained in the proposed regulations do not place taxpayers in a worse position than they were before. We disagree. For example, intangible cases may no longer be settled on the basis of a comparable transfer unless a comparable profit interval is computed. In the absence of a matching transaction, such cases may not be addressed in any manner other than through a comparable profit interval. Similarly, the resale price or cost plus method may not be applied to tangible sales without also applying a comparable profit interval; fourth methods may not be used without a comparable profit interval. Many (in fact, virtually all) past transfer pricing inquiries have been resolved by taxpayers and the IRS using methods other than a comparable profit method. The data-gathering burden is far greater in many cases than previously imposed, and all transfers must now be re-examined continuously to ensure compliance with the periodic adjustment rules. We submit that these results of the proposed regulations, along with others, would place most parties with respect to the administration of section 482, including the IRS in certain instances, in a worse position than under prior law, both in terms of reasonable transfer pricing planning and the resolution of transfer pricing disputes.

Second, and related to this general point, is the complaint of government officials that taxpayers are reading the proposed regulations as if they were a "cookbook." The point that taxpayers are raising is that the regulations, particularly as related to the primacy of the comparable profit method, do not clearly permit the kind of flexibility that is necessary to prevent the rules from being read as recipes. If it were the intention of the drafters not to have published a cookbook, substantive changes to the regulations are necessary to carry out this intention. For example, in the case of an intangible transfer, both taxpayers and the IRS are directed to compute comparable profit intervals in virtually all cases -- no alternatives are permitted, regardless of how comfortable both sides may be with an alternative inexact comparable, and regardless of how uncomfortable they may be with the applicability of available data for developing profit level indicators. As a result, the regulations practically tell the parties to ignore potentially superior information.

The regulations are, in effect, written with a small number of problem cases in mind. Thousands of intangible and tangible transfers occur each year that do not warrant the type of detailed, time- consuming analysis required under the regulations. Yet the proposed regulations do not permit substitutes to the CPI in these situations. We welcome the public statements of government officials that a cookbook was not intended, but we believe that the regulations need to be redirected if that implication is to be avoided.

II. SOUND BUSINESS JUDGMENT AND FORECASTS

RECOMMENDATION. The sound business judgment test should be refined to clarify that the employment of sound business judgment is determined as it is in the market -- with knowledge of those facts existing at the time the judgment is made. Further, the use of reasonable business forecasts should be specifically permitted.

The proposed regulations introduce a new subjective concept into section 482 determinations by applying a "sound business judgment" standard. Prop. Treas. Reg. section 1.482-1(b)(1).) We have several comments with respect to this new standard.

First, we believe that it is essential to provide further refinement to reflect the circumstances under which business judgment is exercised. For example, judgment may be applied in initiating and negotiating the terms of a transaction, and in deciding whether to complete the transaction, but thereafter business judgment may be severely limited by the terms of the agreement. If a sound business judgment rule is to be retained, it must recognize the limitations and timing in the exercise of judgment that uncontrolled parties face in the business world. The test should specifically provide that the standard's "full knowledge of the relevant facts" refers only to those facts in existence at the time of the decision. If the IRS insists that the sound business judgment test is appropriate, then the test must accurately reflect how sound business judgment is exercised in the real world.

Further, the regulations should specifically provide that the business judgment of the controlled parties is expected to represent a good faith effort to ascertain an arm's length result and not a perfect business decision in all cases. Just as uncontrolled taxpayers frequently enter into transactions that others may question or that fall short of expectations, controlled taxpayers will make good faith efforts that fall short when viewed with the benefit of hindsight. Business decisions are prospective; subsequent events determine whether they were good or bad. The type of perfect judgment contemplated by the objective rules of the proposed regulations is not observed in uncontrolled transactions and should not be required in controlled transactions.

In the absence of such refinements, we seriously question the utility of a new subjective test, particularly since there is no guidance as to how that new test interacts with the other subjective tests retained in the regulations and the objective tests proposed. We understand that the IRS may have wished to surround the more specific language relating to the treatment of related transactions and contractual arrangements with more general language. However, we believe that the appropriateness of those more specific rules should be considered separately from the general sound business judgment language. If the goal of the proposed regulations, as stated in the preamble, is truly to reduce controversy and facilitate determinations, it is difficult to see how the sound business judgment rule, without the above-mentioned refinements, does anything other than place an additional barrier in that path.

The clearest example of how the objective rules of the proposed regulations themselves fail to comport with a sound business judgment standard is the prohibition on the use of forecasts in construction of a comparable profit interval. Although the IRS has recognized that averaging of results over time is appropriate, it has failed to recognize that the use of current and "following year" actual data for planning purposes is impossible. Even preceding year data for third parties will not normally be available when required to set prices under the regulations because of internal lead time requirements and reporting time lags. The requirement to use actual data may be feasible for the IRS to use as an audit tool, but it is simply not practical for a corporation in establishing and administering transfer prices. Attempts to make pricing adjustments in the last months of the three-year averaging period, as has been suggested by IRS and Treasury officials since the publication of the regulations, are virtually certain to provide no relief given that data for uncontrolled transactions will lag so far behind the actual transactions. Thus, the regulations would impose a requirement for compliance that a taxpayer cannot apply in a timely manner.

To the extent that budgeted results vary from actual data, retroactive adjustments would be necessary in order to meet a CPI test under the proposed regulations. However, retroactive pricing adjustments are not desirable for internal management purposes, are not consistent with the forward-looking nature of arm's length behavior and the exercise of sound business judgment, and are not generally allowable by foreign taxing authorities. In particular, the many corporate groups that employ highly decentralized, profit center management styles depend heavily on the certainty of their internal pricing.

As a result, we urge that the final regulations recognize the use of a company's planning budget or forecasted financial data when those projections can be shown to be realistic and made in good faith under the circumstances known at the time of the transactions. The IRS must recognize that budgets, forecasted financial results and historical data for comparables are the only available tools in the business world for exercising sound business judgment.

III. TESTED PARTY

RECOMMENDATION: We offer two specific suggestions regarding the proposed regulations treatment of tested parties.

o Where both parties to a related party transaction are performing complex functions, a taxpayer's choice of tested party, where the CPI has been properly applied, should not be challenged by the IRS.

o We recommend that a taxpayer be permitted to treat its non- U.S. operations (or the U.S. operations of a foreign taxpayer) in the aggregate as the tested party, with certain exceptions.

Given the structure and likely operation of the comparable profit interval, the selection of the tested party is a crucial step that may, in many cases, greatly influence the allocation of profits between the parties under the comparable profit method. In view of the difficulty, discussed elsewhere in these comments, in locating uncontrolled party data that capture the facts and circumstances of the particular tested party, frequently the effect of the comparable profit method will be to assign "routine" or normal profits to the tested party. Any residual profit, or profit shortfall, would then be assigned to the non-tested party, which will generally be the party performing research and development and headquarters activities, or the developer/owner of the relevant intangible property.

The proposed regulations provide several guidelines for selecting the tested party, which may not all point in the same direction. Ordinarily, the tested party is the transferee in the case of the transfer of an intangible; the buyer/reseller where the resale price method is being applied to a sale of goods; and the seller where the cost plus method is being applied to a sale of goods. The logic of applying a CPI to only one side of a transaction would seem to require that the tested party be the side with the simpler functions. In a great many cases, these guidelines will not be consistent with the more general rules that the tested party is the one associated with the most reliable data.

In many situations, intercompany transactions simply do not fit the simple paradigms -- e.g., a U.S. (or foreign) developer/manufacturer selling to a related foreign (or U.S.) distributor or a U.S. developer dealing with a related foreign manufacturer -- that seem to be the focus of the regulations. For example, component goods will often be sold between related parties that are both engaged in manufacturing activities, and intangibles are frequently licensed or goods are sold in situations where both parties to the transaction own valuable intangibles that contribute to profitability. In such cases, it is especially unclear which of the conflicting guidelines for choosing the tested party should be controlling.

In a situation where the general guidelines do not provide a clear answer, there would seem to be no reason for the IRS to second- guess a taxpayer who has picked one of the related parties as the tested party and has managed to perform the calculations required under the CPI with respect to that related party. For example, if a taxpayer chooses his U.S. manufacturing affiliate, rather than his foreign distribution subsidiary as the tested party (where both perform complex functions), and applies the CPI in accordance with that choice, a further inquiry should not be made by the IRS whether the foreign distribution subsidiary would be a "more appropriate" tested party. The regulations should confirm that, in such a case, the IRS will not make any adjustments that would result from selecting the other related party as the tested party and applying the CPI to the operations of that other related party.

More fundamentally, the tested party concept as embodied in the regulations should be modified to alter the focus of the comparisons called for by the CPI methodology. The tested party concept, as currently formulated, fails to take account of the overall structure of the overseas activities of U.S. businesses (or of the U.S. activities of foreign businesses). The proposed regulations recognize, through the use of a three-year computation period in the comparable profit interval, that a taxpayer's operating results for a particular year should not be considered in isolation. Similarly, the results generated from a particular intangible transfer or a particular flow of goods should not be considered in isolation. Components for a finished product can be produced in more than one country, or the same intangible may be used in more than one country, and examining the results of each transaction separately will often be as misleading as the result for a particular year.

Viewing the results of a U.S. taxpayer's foreign operations as a whole would recognize that operating results from different countries are frequently interdependent. Results that in isolation may appear not to be arm's length are offset by results attributable to other countries or products. Determining the tested party on an aggregate basis would also recognize that the IRS generally has little interest in transactions between foreign affiliates, and that offsets by country are just as appropriate as offsets by year. In effect, aggregation provides an approach that is similar to the tests for determining whether a cost-sharing arrangement is bona fide. In the cost-sharing context, the basic inquiry is whether the U.S. participant is receiving income from intangibles developed under the arrangement that is reasonable in light of the costs it incurred. For an intangible that is licensed, the proper inquiry should be whether the U.S. licensor is receiving an arm's length return for the development of the intangible. The precise make-up of the consideration received by the licensor should not be determinative, just as the precise make-up of the income received by the cost- sharing participant is not determinative. (See Prop. Treas. Reg. section 1.482-2(g)(2)(ii)(C)(3).)

There are several ways to address this problem in the structure of the proposed regulations. We would recommend that the IRS permit the CPI methodology to be applied by treating as the tested party a U.S. taxpayer's overall foreign operations (or a foreign taxpayer's overall U.S. operations), at least in the same product area or business classification.

An analogous issue that requires refinement in the proposed regulations is the fact that many corporations have established royalty rates or methods for computing royalties for intangible transfers that apply across the board in their foreign operations. Such a uniformity of worldwide royalty rates or computation methods serves many purposes including the facilitation of the deductibility of the royalty for foreign tax purposes, and the ability to re-source manufacturing, between foreign locations without tax impact. While it may be perfectly appropriate to examine the royalty rate or method on an aggregate basis to determine if the U.S. developer/licensor is receiving an appropriate amount of overall compensation for its intangibles, there is little justification for insisting that the results attributable to each particular transfer or country satisfy the comparable profit interval requirements. A "cherry picking" approach by the IRS can only result in the United States licensor receiving an overall level of compensation that is, in fact, greater than an arm's length amount. We also recommend, therefore, that situations involving uniform royalty rates or uniform royalty computation methods be treated on an aggregate basis.

The case for aggregation of licensees is particularly compelling given the extreme difficulty in developing particularized foreign country data under the comparable profit interval, especially if resort is needed to data from other countries or to United States comparable data. (See, e.g., Prop. Treas. Reg. section 1.482-2(f)(11) Example 8.) In many cases, the comparable data that is available will aggregate the operations of more than one country without any ability to obtain the underlying separate country information. Given the difficulty in data gathering, the regulations should permit the flexibility to use data that aggregates operations in different countries, as well as data than may aggregate different products or business activities as is already contemplated in the regulations. Thus, aggregation should be permitted where comparable aggregate data can be located with respect to comparable operations, or where separate country data can be located and aggregated thereby permitting offsets to be made in the manner previously discussed. As the search for relevant data is expanded to include other operations or other countries, there is no basis to treat separately the tested parties for which that aggregated data is relevant.

Permitting greater aggregation of operating results would also significantly reduce the potential for double taxation and disputes among foreign tax authorities. Aggregation would also help temper the impact of currency fluctuations on the operation of the comparable profit method.

We recognize that the IRS may be particularly concerned about aggregation where the taxpayer conducts operations in low-tax countries such as Ireland or Singapore. However, with one exception discussed below, we see no reason to distinguish among countries for this purpose. If an Irish manufacturing affiliate sells goods to a German sales affiliate for resale in the German market, it is only by combining the results of the Irish and German operations that an accurate picture of the taxpayer's business can be obtained, and the appropriate return to the U.S. parent/licensor be determined. The only situation where aggregation would not be appropriate, we believe, would be a "round-trip" situation, and such situations could be excluded from the aggregation approach.

IV. THE COMPARABLE PROFIT INTERVAL (CPI)

RECOMMENDATION: We strongly urge that the CPI test, in the absence of a CUP or matching transaction, be placed on the same level of priority as other inexact comparables. The choice among such methods would be based on the quality of the available data, the closeness of the comparables, and the logic of the circumstances.

The primacy given for all purposes to the comparable profit interval (CPI) is misplaced. According to the proposed regulations, a CPI would be used not only for testing whether a taxpayer's pricing results differ from those performing similar functions (and might, therefore, invite further inquiry), but is used as the substantive pricing method that takes precedence over all other pricing methods except narrowly defined exact comparables.

There are too many deficiencies in the CPI methodology to justify the emphasis it receives in the proposed regulations. Some of the most important concerns about the CPI method are:

o The CPI method is not equally appropriate across all types of transactions. The use of a CPI alone would seem to be least appropriate where each party has substantial self-developed intangibles. The CPI is most appropriate for contract manufacturers or routine manufacturing operations under license, or for routine distribution arrangements. Affiliates with only these relatively simple functions have become less common among multinational corporations as product development, research, management, finance, and other functions are increasingly internationalized and decentralized.

o The CPI priority dismisses or overly restricts traditional and useful pricing methods such as cost plus, the resale price method and profit splits.

o There may be cases in which the CPI methodology cannot be applied. For example, there may be no uncontrolled parties performing similar functions in a similar market, or such functions may not be separable from other dissimilar functions.

o In establishing the tested party to which a CPI will be applied, even a sensibly identified tested party could be subject to unique market conditions (e.g., product recall) that greatly influence profitability. (For further comments regarding the tested party see Chapter III.)

o In applying the CPI test, the quest is shifted from finding comparable pricing of TRANSACTIONS to finding and debating which companies perform comparable functions, which measures of their profitability are to be used or averaged, and which are to be ignored.

o The proposed regulations assume that "convergence" among "profit level indicators" is easily recognizable when proper analysis of data from similar uncontrolled firms is performed. In practice, however, determination of a meaningfully narrow range of income from actual industry data may be difficult. (To further illustrate this point Appendix A contains excerpts from an article appearing in Tax Notes International, June 22, 1992 which was authored by John M. Simpson, Ph.D. and Garry B. Stone, Ph.D., two economists with Price Waterhouse who specialize in transfer pricing issues.)

o There is a formidable burden in identifying and applying any comparable profit indicators, especially in non-U.S. jurisdictions. While the regulations indicate that one should turn to the next best set of indicators in another jurisdiction or broader line of business, in doing so the comparison becomes increasingly tenuous carrying the role assigned in the proposed regulations.

o As prescribed, the methodology is focused on after-the-fact determinations and does not provide guidance to taxpayers for planning.

o The CPI method is inconsistently applied in the proposed regulations. On the one hand, a tested party is not permitted to earn more than routine or "normal" profits. On the other hand, a tested party that does earn "normal" profits may nevertheless face an adjustment to the most appropriate point within the CPI if the consideration it paid was determined to be "substantially disproportionate" to the value of the intangible. Based on the logic of the CPI, however, the value of the intangible under such circumstances could only be equal to whatever consideration was actually paid.

o There is a basic inconsistency under the CPI, as applied in the proposed regulations, between the recognition that arm's length pricing can result in a range of outcomes and the requirement that adjustments under that method must generally be made to a most appropriate point within such range.

We agree that, under the right circumstances, comparative profitability measures can be probative of whether intercompany pricing arrangements are appropriate. However, intercompany pricing is essentially an inquiry based on facts and circumstances and profitability comparisons may be inappropriate or inapplicable in many cases.

Where the CPI method is appropriate, it should be consistently applied. This would mean that no adjustment be made for "substantially disproportionate" consideration for an intangible if reported operating income fell within a CPI. The determination of a most appropriate point would be dispensed with, and an adjustment would be required only to put the operating income of the tested party at the upper or lower bound of the CPI, i.e., the top (or bottom) of the range of outcomes considered by the CPI method to be arm's length.

We, therefore, urge that the CPI test, in the absence of a CUP or matching transaction, be placed on the same level of priority as other inexact comparables. Resale price, cost plus, comparable adjustable transaction or other methods, if found to be more appropriate under particular facts and circumstances, would not be subject to a CPI test.

V. DATA AVAILABILITY AND RELIABILITY

RECOMMENDATION: In applying the CPI test, the IRS should:

o be required to use data that is also available to taxpayers at the same level of detail; and

o clearly provide that tested-party data and third party data be adjusted, whenever possible, for differences in local accounting standards, market conditions exchange rates, and other measurable factors.

The CPI concept explicitly acknowledges that financial results can be highly variable from one year to the next, and that any set of inexact comparables will, at best, produce a range of results because similar business transactions can have varying outcomes. Except in the case of a true CUP or matching transaction, differences in the quality of management, effectiveness of marketing, product acceptance, market demand and other terms and conditions of unrelated transactions will cause financial results to differ among even the most closely comparable operations.

The essential logic of the CPI test is to "average out" these real-world variations by pooling the results from a number of uncontrolled operations. In general, if these results form an unbiased sample, the greater the number of cases that can be found, the greater the statistical reliability of the comparison. The logic of large samples is defeated, however, when bias is introduced, and sampling bias will be introduced if the sample contains companies with differences in functions, risks and economic conditions that are systematically related to the profit level indicators being used.

The reliability of the CPI as a fair test depends on both the quantity and the quality of the available data. Quantity is better only if the addition of data does not add bias to the sample. Quality requires comparability in the consistency of data, e.g., through preparation according to the same standards for the same time period. Quality also requires comparability in terms of business functions, risks, intangible assets and market conditions. Thus, any sample of comparison cases must be carefully selected to minimize sampling bias by matching the tested party as nearly as possible.

Also, unique characteristics and circumstances of the tested party should be taken into account wherever possible. Because averaging does not eliminate biases, fairness in applying profitability tests depends on the ability to make appropriate adjustments for special circumstances of the tested party such as start-up conditions, unanticipated exchange losses, significant changes in demand or cost conditions and effectiveness of management.

Data from financial reports of U.S. public companies are now readily available through such services as Standard & Poor's Compustat databases. Such data currently represent the best information that is available at the company level -- it is audited, public, and highly automated. At best, however, such reports are one to two years old and provide limited information about the business segments of each company. Other data sources, such as the Statistics of Income (SOI) and credit-rating services, are inferior in terms of level of detail, reliability and accessibility. Regarding SOI data, for example, Assistant Treasury Secretary Gideon testified that "the limitations of using the data for this purpose [to examine the extent to which aggressive transfer pricing . . . practices may have been used] should be recognized." 1 For any specific company, Gideon said, "Such practices can only be determined from a detailed audit examination of tax and other accounting records of the company, including a detailed review of its transactions with other related parties."

In the absence of segment data, conglomerates make poor comparables and those industries that are dominated by highly integrated, multi-product, multinational firms may not include any companies that are suitable as comparisons for any other taxpayer under audit. For example, if the tested operation is a distributor of branded products in the U.S. market, the appropriate comparables would consist of U.S. companies selling similar products, supplied by uncontrolled parties, into the U.S. market. Ideally, any uncontrolled company that has a significant foreign market, non-marketing functions, or dissimilar products, should not be included in the sample unless the segments and functions can be segregated in a way that does not rely on their intercompany pricing. Even for relatively simple operations, however, significant differences in expected profitability of sample companies may still exist because of the existence of valuable, self-developed intangibles.

Tested-party financial data will frequently need to be refined and adjusted to reflect the related-party transaction. The proposed regulations expressly permit adjustments only for variations in the measurement of assets which may be only one of many relevant and measurable differences.

To summarize, we have the following concerns regarding the quantity and quality of available data:

o The IRS may have access to data from tax returns and third- party summonses that are not available to taxpayers either for their planning or for defense of their tax return positions. It is unreasonable to expect taxpayers to conform to profitability data to which they have no access and, therefore, cannot test independently.

o Data segregated by line of business, business functions, and geographic market might improve the comparability and sample sizes required to develop CPIs. Such data are not generally available and, indeed, are less available since FTC line-of- business data were eliminated some years ago. Experience has shown that the collection of such data is very expensive for taxpayers and the government.

o Data for testing foreign operations will be very limited in many countries either because financial reports are not public or because they are not readily accessible. (See Appendix B for an example of the difficulty of obtaining foreign data.) In such cases, the regulations have suggested using available data from other countries, but there may be significant market differences. This problem highlights the importance of relaxing the requirement of a CPI test in every case and the necessity of adjusting all data for differences such as local accounting standards and market conditions.

Both the quantity and quality of data applied to any case should determine the appropriate statistical methods and measures to be used in developing a CPI. In most instances, if the comparison companies are carefully chosen for comparability of functions, lines of business, scale of operations, etc., the number of such companies may be small (e.g., five to 25 cases). In such events, the best approach is to use "order statistics" (e.g., medians, interquartile ranges) and other simple clustering techniques to avoid exaggerating the importance of a few extreme values that often occur in such data. Parametric statistics (e.g., means, standard deviations, etc.) should be used only if the sample is sufficiently large and the profit indicators approximate a "normal" distribution.

In many instances, the luxury to use statistical methods suitable for large samples can be bought only at the expense of a very broad level of comparability, or minimal detail in the data. When comparables are chosen very broadly (e.g., all wholesalers of non-durable goods), certain profit level indicators (e.g., resale margins) will be inappropriate because of their high degree of variability over industry or functional categories.

Any sample of financial results from real companies is likely to include some extreme cases of high and low profitability. Some commentators have suggested that such cases be excluded as prima facie non-comparable. But such logic is circular. Extreme observations must be tested by independent evidence of their functions, risks, or conditions. The final sample must be chosen to fit the facts. Thus, unless the taxpayer has access to all of the underlying cases used in developing a CPI, it would not be possible for the taxpayer to evaluate the quality and relevance of the data on which a pricing adjustment might be made.

VI. PROFIT LEVEL INDICATORS

RECOMMENDATION: When a CPI test is applied, the regulations should convey the need for careful definition of PLIs and associated data to fit the specific facts of the tested operation, and should expressly caution against mechanical solutions.

The proposed regulations do not adequately address the crucial role of profit level indicators in determining the CPI. The proposed regulations specify definitions for four PLIs, offering no explicit guidance or preference, and treating all PLIs as equally applicable in all situations. In fact, the appropriateness of any one indicator, and indeed the appropriate calculation of each, should be defined by the facts and circumstances of the tested transaction.

Return on assets (ROA) is discussed first (Prop. Treas. Reg. section 1.482-2(f)(6)(C)) and was endorsed in the Treasury/IRS White Paper. We believe ROA can be a very useful indicator of a tested party's profits in some cases, noting of course, that an ROA test may not be reliable for a tested party having unusual risks or intangibles which are not fully reflected in book values.

ROA relates income from the tested party's functions to assets employed in these functions. Thus, because ROA relates return to a financial proxy for functions, close comparability of the tested party and uncontrolled taxpayers, as has been traditionally necessary under the resale price and cost plus methods of the existing regulations, is not as necessary under an ROA test. 2 The ROA profit level indicator puts dissimilar companies on a comparable basis by measuring functions in terms of a common value (assets employed).

The proposed regulations define ROA to be the ratio of operating income to book value of assets (OI/A). The IRS should explicitly consider that other measures of ROA may be appropriate.

For example, the logic that supports computing the return before interest expense (so that alternative financing considerations do not influence the calculations) also supports subtracting accounts payable from assets or, alternatively, including an implicit interest expense factor in the numerator. The existence of discounts for immediate payment and interest for extended terms demonstrate that the price of goods contains an implicit interest expense when payment is not due immediately.

We find the discussion of margins to be unsatisfactory because the ratio of operating income to sales (OI/S) seems to be given a mild priority over the gross income to operating expense ratio (GI/OE). The proposed regulations improperly suggest that GI/OE is more sensitive to functional differences among comparable companies than is OI/S when, in fact, the opposite is often true. In the particular case of a distributor, GI/OE (like ROA) relates reward for the distribution function, or value added by that function, to the costs of performing the function. Operating expenses represent a financial measure of the magnitude of the functions performed. We can also view GI/OE to be equivalent to the ratio of operating income to operating expense (OI/OE). OI/OE relates net return to function as measured by expenses incurred performing the functions; it is a "cost plus" mark-up on operating expenses.

When applied to a distributor, OI/S relates net return to the distributor's functions to the gross returns (i.e., sales) of the distributor and manufacturer combined, which can be distorting. We believe that a distributor who must expend $40 in marketing and distribution to sell $100 of goods is entitled, in an arm's length environment, to greater dollar profit than is a distributor who must expend only $10 in marketing and distribution to sell $100 of goods.

OI/S can be useful under certain circumstances, especially when used in conjunction with other PLIs. For example, the proposed regulations point out that different classification of expenses may distort the GI/OE ratio. Because of differences in application of accounting methodologies, the distinction between cost of goods sold (COGS) and operating expense is not always clear. Thus, the OI/S ratio offers consistent valuation of operating income and sales across companies. In general, we feel that for distributors, GI/OE and OI/OE are better measures of appropriate return/effort than OI/S; however, OI/S as one of several PLIs can provide indicators of significant variability in accounting or structural differences. For manufacturers, OI/S is superior to GI/OE and OI/OE conceptually, and is more consistent in measurement.

For the above reasons, selection of appropriate PLIs must depend on the type of transaction being evaluated and the quality of available tested party financial data. No list of PLIs and their variations can be exhaustive, and often no one measure is clearly the best.

VII. PROFIT SPLITS AND ALTERNATIVE METHODS

RECOMMENDATION: Other methods, including inexact comparables (with appropriate adjustments), and under certain circumstances a workable profit split, should be made available as alternative methods to the CPI.

A. PROFIT SPLITS

Under the proposed regulations, comparable profit splits are discussed as a possible profit level indicator (Prop. Treas. Reg. section 1.482-2(f)(6)(iii)(C)(3)). The comparable profit split relies on data from uncontrolled taxpayers who have similar transactions and functions as the group of controlled taxpayers (including the tested party). The allocation of profits among uncontrolled parties is used as an indicator of the allocation of profits among members of the controlled taxpayer's group.

The data required to implement a profit split test under the proposed regulations make it practically impossible that such tests could be made. The proposed regulations require unconsolidated income statement and balance sheet information for each member of the taxpayer's controlled group, as well as equivalent data for uncontrolled parties whose functions and risks parallel that of the taxpayer's group. The prescribed method requires the identification of at least two companies which have elected to structure their operations in essentially the same form, serve the same markets, and perform similar functions as each of the taxpayer's entities, yet ALL transactions must be unrelated. Even where such similarities exist, finding public sources of data, which isolate only the parallel transactions, would be highly unusual.

In addition, meeting the taxpayer's burden of showing comparability between the functions of the controlled group and the unrelated firms essentially requires a functional analysis of both the taxpayer's business and the unrelated firms' businesses which relate to the transactions under examination. It is highly unlikely that a taxpayer, or an independent consultant working on behalf of a taxpayer, would have access to the personnel and information of the unrelated parties' businesses that would be necessary to set or defend a profit split arrangement.

We strongly recommend that a viable profit split option be included in the regulations as an alternative to the CPI for complex entities with significant self-developed intangibles on both sides of the transaction. For transactions between complex entities each of which has significant self-developed intangibles, one possible profit split method may be first to determine the routine manufacturing and distribution profit attributable to each entity using resale price, cost plus, CPI, or other methods. The residual profit or loss remaining after deducting the combined routine manufacturing and distribution profit from the combined reported operating income would then be split between each entity in the same ratio as its share of routine profit, capital employed, value added, relative costs, or any other reasonable basis under the circumstances.

We are cognizant of the IRS's concerns with the use of profit splits in all situations. Therefore, we would also propose exclusion of the profit split method under particular circumstances, specifically when the tested party is located in a low-tax jurisdiction and is not a complex entity. In some other situations, however, the use of an internal profit split may be less arbitrary than the type of comparisons that would otherwise be required by the proposed regulations.

B. ALTERNATIVE PRICING METHODS

We also recommend that inexact comparables (with appropriate adjustments) be permitted without a CPI test for both transactions involving tangible property and intangible property. For example, a royalty rate determination in a patent infringement suit could be quite valuable in determining a royalty rate under section 482 and should not necessarily be viewed as inferior to a CPI analysis. The White Paper acknowledged the usefulness of inexact comparables. Particularly where the licensee is a complex entity, a CPI analysis clearly could be inferior to certain inexact comparables.

For tangible property, prices to third parties could provide the best guidance for transfer pricing even if adjustments for several differences between the two types of transactions are necessary. If the comparable uncontrolled price method is viewed too narrowly, valuable information on third-party transactions could be inappropriately dominated by a CPI analysis under the proposed regulations. The same fate could befall a closely comparable resale price method application.

These failings of the proposed regulations are most severe when the tested party of the CPI analysis is a complex entity. The priority of pricing methods for tangible property under the existing regulations is deficient because it requires rejection of good but imperfect information for higher-priority methods. The result is reliance on a lower-priority method with information that may be even less reliable. The final regulations should not repeat this kind of deficiency. Fundamentally, the final regulations should not systematically exclude information useful for testing transfer pricing. A diverse menu of transfer pricing methods is important because of the broad range of fact patterns.

VIII. EXCHANGE RATES

RECOMMENDATION: The tested party's data should be adjusted for differences in economic conditions of the tested party that are not generally faced by the uncontrolled parties used as comparables. In particular, exchange gains and losses should be explicitly allowed in applying the CPI test.

The proposed regulations have requested comments on the impact of exchange rate changes on transfer prices. This comment focuses on an illustration which traces the exchange rate impact on the combined business of two hypothetical affiliates, a U.S. manufacturer and a French distributor. Because manufacturing costs are in U.S. dollars and third-party revenues are in French francs (ff), exchange rate changes can have significant impact on the combined business. (Accounts receivable/payable risk is not considered in this comment.)

Suppose that the exchange rate is $1 = 6ff at the time transfer prices are set, and planned financial results are shown below, based on a transfer price that has been determined to be arm's length at 40ff and is set for the year.

                           PLAN ($1 = 6ff)

 

 

               U.S.                     French

 

            Manufacturer              Distributor    Combined

 

            ____________              ___________    ________

 

 

Sales       $6.67 = 40ff                 100ff     $16.67 = 100ff

 

COGS        $5.00 = 30ff                  40ff     $ 5.00 = 30ff

 

Marketing

 

  Costs                                   40ff     $ 6.67 = 40ff

 

Profit      $1.67 = 10ff                  20ff     $ 5.00 = 30ff

 

 

Now suppose that actual results equal plan in all dimensions except that the dollar strengthens to an average rate of $1 = 7ff. (Assume there is sufficient foreign competition so that the French affiliate can not raise third-party prices.) The French actual results equal plan because France is not exposed to the exchange rate change. However, U.S. and combined actual results fall short of plan because manufacturing costs denominated in ff have gone up. Stated alternatively, French sales and marketing costs fall when denominated in dollars.

                          ACTUAL ($1 = 7ff)

 

 

                    U.S.                French

 

               Manufacturer         Distributor        Combined

 

               ____________         ___________        ________

 

 

Sales          $5.71 = 40ff             100ff       $14.29 = 100ff

 

COGS           $5.00 = 35ff              40ff       $ 5.00 = 35ff

 

Marketing

 

  Costs                                  40ff       $ 5.71 = 40ff

 

               ____________             _____       _____________

 

 

Profit         $0.71 = 5ff               20ff       $ 3.57 = 25ff

 

 

The unanticipated strengthening of the dollar causes combined profit to drop, a situation which is beyond the realm of transfer pricing. The transfer price is fixed in ff, but falls in dollars. U.S. profitability is down. The strengthening dollar over the year would also hurt the U.S. manufacturer because of exposure to accounts receivable risk. (Conversely, a falling dollar would help combined and U.S. results and would leave French results unchanged when the transfer price is set in ff.)

If the U.S. manufacturer had been the tested party, planned profit of $1.67 could have been in the relevant CPI, but actual profit of $0.71 could be outside the CPI. This could especially be true if the companies used in constructing the CPI had less exposure to international transactions compared to the U.S. manufacturer on its intercompany transactions. By analogy, if the transfer price had been set in the manufacturing company, the distributor would bear the exchange rate risk which could drive the distributor to high or low profits as the currency strengthens or weakens. Also, if the distributor were the tested party, results could easily fall outside a CPI made up of distributors not bearing the same currency risk.

Alternatively, if the transfer price had been set in dollar (at $6.67 rather than 40ff), the French distributor would bear the burden of the exchange rate charges. Then actual results would be as follows:

                          ACTUAL ($1 = 7ff)

 

                          _________________

 

 

                         U.S.                French

 

                    Manufacturer          Distributor      Combined

 

                    ____________          ___________      ________

 

 

Sales               $6.67 = 47ff              100ff     $14.29 = 100ff

 

COGS                $5.00 = 35ff               47ff     $ 5.00 = 35ff

 

Marketing

 

  Costs                                        40ff      $5.71 = 40ff

 

                    ____________              _____     _____________

 

 

Profit              $1.67 = 12ff               13ff     $ 3.57 = 25ff

 

 

If the French distributor were the tested party, it might fare poorly compared to distributors used in the CPI if they were not equally exposed to exchange risk. The tested party's operating income and the CPI should explicitly take into account the impact of foreign exchange fluctuations. Further, the party which bears the risk of foreign exchange movements should reap the benefits or share the losses generated by exchange rate fluctuations.

In general, exchange rates can significantly affect profitability analysis as well as application of other pricing methodologies. Frequently, exchange rate policies vary among related and unrelated companies. Therefore, a taxpayer should not be subjected to adjustments under section 482 for exchange rate fluctuations if he follows a sensible business strategy which includes an exchange rate policy that is not erratic.

IX. PERIODIC ADJUSTMENTS

RECOMMENDATION: We recommend that the periodic adjustment requirement be eliminated, or, at the very least, that it be significantly narrowed in accordance with the approach adopted in the section 6038A regulation. We are concerned that U.S. treaty partners will not accept this methodology, thereby creating double taxation without any realistic possibility of relief.

The proposed regulations require that the determination of arm's length prices be made throughout the period during which the intangible is transferred. (Prop. Treas. Reg. section 1.482-2(d)(6).) While the proposed regulations contain several exceptions, which will be discussed in more detail below, it seems clear that taxpayers will be required to examine repeatedly virtually all of their license or other intangible transfer agreements throughout their corporate group. We believe that this requirement will create an enormous burden and perpetual uncertainty for taxpayers.

We continue to believe that uncontrolled parties in the marketplace do not enter into agreements permitting subsequent adjustments to the basic terms of their license agreements. It must be remembered that the periodic adjustment requirement will apply to essentially all intangible transfers. It is certainly the experience of the companies in the group, and of all other studies that have been made of the issue since the publication of the White Paper, that one party to a license agreement will have the ability to change the terms of that agreement in few, if any, situations. There is simply no way to assert that the periodic adjustment requirement is consistent with what unrelated parties do in the marketplace.

Based on the reaction of foreign governments to the periodic adjustment rules contained in the White Paper, any U.S. adjustments made pursuant to this rule present a significant and real risk of double taxation, or, at best, lengthy competent authority cases. The Treasury and the IRS have had little success in convincing foreign competent authorities that the periodic adjustment requirements are consistent with the international arm's length standard. In the absence of any international consensus, we believe that it is essential that Treasury proceed with extreme caution, since the burden of double taxation that will otherwise result would be dramatic.

These basic problems present, we believe, a compelling case for a complete elimination of the periodic adjustment notion, at least until the risk of double taxation can be reduced. In the event that Treasury and the IRS continue to reject these arguments, however, we have several specific suggestions to the rules as proposed.

As a practical matter, the general rules for intangible transfers require taxpayers to attempt to compute CPIs for all of their intangible transfers. The periodic adjustment requirements mandate that such attempts be undertaken on presumably an annual basis. The three exceptions provided by the proposed regulations require a large amount of data to be collected before it can be determined whether an exception actually applies to a particular intangible transfer. For example, since each of the exceptions normally requires a CPI to be completed, they provide no relief from this burden. In addition, each of the exceptions contains several ambiguous and subjective conditions that are certain to lead to disputes.

To illustrate, the first exception, contained in Prop. Treas. Reg. section 1.482-2(d)(d)(ii)(A), requires a taxpayer to demonstrate that the income of the tested party fell within the CPI for every year since the initial transfer, regardless of the number of intervening years. Further, there must not have been a "major variation" in annual "revenue" attributable to the transferred intangible, but no definition of "major variation" or "revenue" is provided. Since inflation alone may cause revenue to change, it is difficult to see how this exception could ever apply after a few years have passed, assuming that the calculations can actually be performed for all of the intervening years. Further, it will be necessary to determine whether any variation in revenue was attributable to the intangible transferred. While the variation may be from the revenue generated by activities of the transferee (such as improvements, self-developed intangibles or marketing), there is little question that an effort to determine the source of variation will lead to uncertainty and disputes as such issues have in the past.

The 10-year test exception of Prop. Treas. Reg. section 1.482- 2(d)(6)(ii)(B) suffers from the same faults. In general, thirteen years of data must be calculated. The review of such data for a taxable year under examination will be difficult and contentious. Significant definitional ambiguities also exist. For example, when is an intangible "reasonably susceptible of valuation"? One of the major reasons for the enactment of provisions such as the last sentence of section 482 and section 367(d) was the difficulty perceived by Treasury and the IRS in valuing intangibles in the year they are transferred to a foreign subsidiary. Ironically, requiring that the taxpayer convince the IRS that intangibles were reasonably susceptible of valuation in the year of initial transfer strikes directly at the one of the basic premises of the periodic adjustment requirement. Other ambiguities exist as well. For example, what is "significant commercial production"? Is it an annual test? Is "significant" an absolute concept or a relative one, and, if the latter, to what is it relative?

Finally, the unanticipated events exception of Prop. Treas. Reg. section 1.482-2(d)(6)(ii)(C) appears to raise one area of potential dispute after another. What standards will be applied to determine the use of restrictions that are consistent with industry practices? How is consistency with industry practice to be determined where the comparable adjustable transaction involves a different intangible? How can it ever be determined whether operating income moved outside the CPI "solely" because of unanticipated events? Based on our experience, it is likely to be virtually impossible to convince the IRS that a particular factor is the "sole" reason for a particular result. In fact, the example of this exception contained in the proposed regulations illustrates the difficulty taxpayers will have in convincing the IRS that changed economic conditions were beyond the taxpayer's control. In Example 2, the taxpayer's U.S. plant is found to violate a provision of U.S. law, and a decision is made to close that plant and acquire the product from a foreign subsidiary. There is no discussion in the example of what it would take, either in terms of time, cost, lost production, etc., to modify the U.S. plant to comply with U.S. law, thereby indicating that the IRS views such issues as irrelevant in determining whether the change was beyond the control of the taxpayer. Such a view is open to serious questions. One can foresee continuous disputes over what changes are truly beyond the taxpayer's control, and what events are not anticipated or reasonably foreseeable.

It may be possible to tinker with the exceptions to provide some greater certainty, and we are prepared to work with the IRS to refine these concepts. Yet tinkering alone would ignore the critical issue. It is simply an overpowering burden to require continuous review of the effects of every cross-border intangible transfer, particularly given that such review must include a computation of a comparable profit interval in virtually every case. We believe that some reasonable effort must be made to reduce this burden by eliminating from the periodic adjustment requirement the large numbers of intangible transfers that will not reasonably be of concern to the IRS.

We would strongly suggest that the IRS give further consideration to previously proposed exceptions that would rely on such factors as the absolute or relative sales of the product based on the intangible, the absolute or relative profits generated by the product, or similar figures. The same type of burden was recognized by the IRS in the development of the regulations issued under section 6038A as they relate to the requirement to maintain product-based profit and loss data.

The IRS significantly limited the scope of the section 6038A requirements in this regard through tests based on absolute and relative gross revenues (the significant industry segment test of Treas. Reg. section 1.6038A-3(c)(5)) and absolute and relative return on assets (the high profit test of Treas. Reg. section 1.6038A- 3(c)(6)). Similar rules could also be used for purposes of limiting the periodic adjustment requirement.

We realize that such rules may operate in an arbitrary fashion in some cases. The IRS may exempt some transactions from the periodic adjustment requirement that it would like to cover, and taxpayers may be forced to apply the rules to cases that should be exempt. However, the potential adverse impact (on either taxpayers or the IRS) of any such arbitrariness is more than overcome by the reduction in unnecessary burden and uncertainty on both taxpayers and the IRS that would result from reducing the scope of the periodic adjustment requirement. We strongly urge that such steps be taken.

X. SAFE HARBORS

RECOMMENDATION: We recommend the adoption of two safe harbors -- first, a rate of return safe harbor for parties performing simple functions, and second, a safe harbor based on a division of profits between the licensor and licensee.

A. RATE OF RETURN SAFE HARBOR

A safe harbor based on return on assets can provide certainty for taxpayers engaged in transactions with related parties where the related party performs very simple functions, has few intangible assets and bears little or no risk. Specifically, if a low-risk tested party earns a return within a range determined by low-risk returns observed in the market, the tested party could be deemed to have earned an arm's length return.

Returns on low-risk securities are readily available (e.g., published government security returns). The only appropriate adjustments to the tested party's returns should be to reflect the use of assets for which the tested party does not incur an expense (e.g., current liabilities on which interest does not accrue, and interest-free loans from a related party), and to adjust returns so that they reflect the related-party transactions in question (e.g., eliminate assets, and the returns thereon, that are purely financial and not used in the operating business). The lack of risk associated with the tested party means that equity and debt returns should be approximately equal, making adjustments to the tested party to reflect debt-to-equity ratios unnecessary.

We propose that a safe harbor range be constructed consisting of an average (over time) of published low-risk, non-short term securities, plus or minus 25 percent of the published return. Therefore, if government securities averaged 10 percent for the appropriate time period, the safe harbor range would be from 7.5 percent to 12.5 percent. Consequently, if the standards for risk and functions were met and ROA calculations for the tested party produced results within this range, transfer pricing would not be questioned.

B. ROYALTY TO PROFIT SAFE HARBOR

Our second recommendation for a safe harbor is based on the calculation of a ratio of worldwide (i.e., all royalty-paying entities) royalty payments to pre-royalty income. If the results of this calculation follow the IRS's own rule of thumb of dividing profits approximately 25 percent to the licensor and 75 percent to the licensee, the royalty rate would be adopted. 3

There is wide support for the view that a reasonable royalty to pre-royalty net operating income split is 25 percent for the licensor and 75 percent for the licensee. The IRS has recognized and accepted this standard which was most clearly stated in Ciba Geigy, where the IRS "repeatedly emphasized throughout [its] briefs that, as a rule of thumb, a royalty rate generally divides net profits before royalties 25 to 75 percent between the licensor and licensee, respectively." In addition, there are numerous tax and non-tax judicial precedents (both in Tax Court and U.S. District Court) which refer auspiciously to this approach to determining whether a royalty was appropriate. For example, in the Kodak/Polaroid patent infringement case, the Court commented favorably on expert witness testimony that a one- third to one-quarter profit split would be appropriate.

"Crown jewel" (or high profit) intangibles could be excluded from the safe harbor provisions, as could software royalties that can be effectively calculated using a resale margin in a resale (i.e., relicensing) transaction.

We also recommend that, even if these two methodologies are not adopted as safe harbors, they should be permitted to serve as pricing methods in appropriate cases.

XI. COST SHARING

RECOMMENDATION: The rebuttable presumption in favor of the cost to operating income ratio should be eliminated. Any allocation method that reasonably reflects anticipated benefits over time should be acceptable and should not lead to an after-the-fact adjustment based upon subsequent cost/income ratios. Any utilization of after-the-fact data, such as a cost/income ratio based upon actual experience, should be employed solely as a safe harbor. Additional recommendations include:

o A statement that a comparable agreement with a non-controlled party establishes an arm's length arrangement with related parties.

o Clarification and examples of eligible participants.

o Narrowing the definition of U.S. participants.

o Recognition of both single purpose and umbrella cost sharing agreements and basic research funding.

o Non-application of U.S. rules to non-U.S. controlled affiliates not affecting U.S. taxation of U. S. participants.

o Flexibility in measuring contributions.

o Including purchased technology in research costs.

o Expanding buy-in rules to recognize consideration other than cash payments.

o Narrowing the situations in which a buy-out payment is required.

o Clarification of the relationship of the developer/assister rules and existing services regulations.

o Reallocating costs rather than imposing intangible transfer rules.

o Limiting reallocations of cost sharing payments to the year the payments were made.

DISCUSSION

The legislative histories of the 1984 and 1986 Tax Acts affirm the use of cost sharing agreements as a method of intercompany pricing. Cost sharing provides an orderly method for affiliated companies operating in several jurisdictions to match each company's share of risks and current expenses incurred in joint projects to the share of benefits each might reasonably expect to receive. It is a way to share risks of uncertain research among entities who, individually, might not undertake it. One result of this sharing of risk is the encouragement of more research and risk-taking. In this regard, cost sharing is an appropriate method to support U.S. policy encouraging research into new products and technologies.

We believe that the proposed regulations for cost sharing arrangements have moved in a positive manner to recognize that cost sharing is an appropriate alternative methodology for developing and utilizing intangible property. While many arbitrary limitations suggested by the 1988 Treasury/IRS White Paper for cost sharing have been eliminated, cost sharing may be jeopardized by a new after-the- fact rule requiring retroactive adjustments based on actual net profits instead of allowing costs to be shared based on anticipated benefits. The cost/income ratio requires precise symmetry between costs borne by all participants and their income actually received from the developed intangibles (even though the transfer of intangible rules allow for ranges). While initially stated to be an alternative measure: "anticipated benefits may be measured by reference to anticipated units of production . . . anticipated sales . . . anticipated gross or net profit or any other measure that reasonably predicts the benefits to be shared" (Prop. Treas. Reg. section 1.482- 2(g)(2)(ii)(A)), the cost/income ratio test provides the IRS with a basis for adjustments whenever the IRS wishes to invoke it. Example 2 of Prop. Treas. Reg. section 1.482-2(g)(4)(ii)(E) illustrates the overriding application of the cost/operating income rule. In the example, the district director determined that another method reflected a more accurate measure of anticipated benefits but the cost/income ratio test was still applied to determine if the costs were substantially disproportionate.

As discussed below, the rebuttable presumption regarding the cost to operating income ratio should be eliminated. By eliminating the rebuttal presumption, the cost/income ratio method should be treated as another method and should not be the ultimate gauge of testing the allocation of costs. If any special function is to be given to the cost/income ratio based upon actual subsequent experience, it should be as a safe harbor permitting those taxpayers who might be able to track earlier incurred research expenses to subsequent operating income of the participants to demonstrate that the actual cost to operating income ratios were not substantially disproportionate. For the reasons discussed below we do not believe that this could be widely used as a practical matter. In addition, certain other modifications should also be made to the cost sharing regulations.

A. COST/BENEFIT RATIOS

Under the proposed regulations, U.S. participants in a cost sharing arrangement are to be tested by a cost/income ratio. This ratio applies to a moving three-year average of research costs compared to income attributable to intangibles produced by the research. To the extent the U.S. participant's ratio of costs to income is higher than the average of the other participants, the result may be an adjustment of costs, treatment of a portion of technology transferred as requiring an arm's length payment, or, if grossly disproportionate, the disregard of the cost sharing agreement with all transfers to be treated as having been made without adequate arm length consideration. We believe that there are a number of compelling reasons for removing the primacy of this rule.

1. The cost/income ratio assumes the identity of the profitability of intangible property used by every participant in every market (an assumption that is more precise than even the CPI methodology that allows for ranges). This simply does not occur. Suppose that research is conducted at a U.S. facility with participation in cost sharing by U.S., Canadian, French, and U.K. corporations. After developed technology is exploited for a few years, the product becomes a best seller in the U.K. exceeding profitability in other markets. The regulation would appear to require adjustment of the earlier cost shares borne by the U.K. company. The proposed rule does not allow for demonstrating particular facts that led to the surge in profitability in the U.K. market. However, profitability among companies in various countries varies widely due to a number of factors, including impact of trademark names; governmental grants or restrictions; labor costs; accounting methods; regulatory and safety cost differences; management effectiveness; market conditions; currency fluctuations and business cycles.

2. A retroactive adjustment is not what would occur in arm's length arrangements.

3. The regulations recognize that a period of more than three years might be utilized to measure income related to research costs since there can be a significant interval between research (including unsuccessful research) and income generated by developed intangibles. We note that the examples in the proposed regulations do not establish a linkage between the earlier research costs and the income attributed to those costs.

The proposed regulations assume that research produces a net profit without regard to the fact that much research is unsuccessful and that operating companies, for a variety of reasons, incur operating losses.

There is an inherent problem in tracing research conducted over a period of time to subsequent income. The scheme is seriously flawed by a requirement to ascertain what portion of a corporation's income is attributable to specific intangibles. Income attributable to a particular intangible is virtually impossible to segregate in processing operations. Numerous intangibles are installed annually and there are hundreds and thousands of intangibles already in place which contribute to the income of the entire operation. Hence, particularly for process operations, operating income associated with intangibles cannot be determined.

The fact that a profits based requirement should not be the only measure is illustrated by the case of oil industry technology which primarily targets the achievement of operating cost reductions, i.e., making existing refineries or units thereof operate more efficiently. Therefore, the benefit of cost sharing -- developed technology in this context -- is the achievement of efficiencies which anticipate increased long-term profitability. However, improved long-term profitability cannot be guaranteed because factors outside the control of the cost sharing group e.g., acquisition prices, market changes, affect ultimate profitability.

In other industries, such as pharmaceutical research, there may be a decade or more between the beginning of a targeted research objective and any successful product creating a lengthy time gap between costs and possible projects.

4. If the cost/income ratio method is utilized as an alternative method to allocate costs, there is a further uncertainty as to how the profitability calculation is to be made. The cost/income ratio utilizes "operating income" directly or indirectly attributable to the intangible development area. Operating income includes income from the license or sale of intangibles developed under the cost sharing arrangement, and income earned with respect to the sale of products or services incorporating developed intangibles. Prop. Treas. Reg. section 1.482-2(g)(2)(ii)(C)(3). Clarification is required as to how operating income is to be computed when a participant permits related entities (the "subgroup") to use the intangible, e.g., by way of a royalty license. Is the inter-company royalty operating income of the licensor or does a look-through rule apply to look to the operating income of the licensee?

B. RECOGNIZE EQUALITY OF ALTERNATIVE REASONABLE METHODS

The proposed regulations should first recognize that any method currently being used in a cost sharing arrangement should not be disturbed as long as it reflects a reasonable effort to share the costs and risks of development among the participants on an arm's length basis. An allocation method that reasonably reflects anticipated benefits over time (whether based on the most recent operating profits, sales, units of productions, or capital investment) should not be revised by application of an after-the-fact cost/income ratio.

It would be desirable if the regulations provided additional examples of the appropriateness of other reasonable methods. Since the proposed regulations recognize that under appropriate circumstances, anticipated benefits may be measured by reference to anticipated units of productions, anticipated sales, or any other measure that reasonably predicts the benefits to be shared, the inclusion of the following example in the regulations should clarify this intent:

EXAMPLE -- Controlled corporations A, B, C, and D are participants in a qualified cost sharing arrangement and each company operates a process operation. The research costs are allocated among the participants based on prior year units of production or volume in various processes. If volumes are relatively stable from year to year, the participants' agreement to share costs based on prior year results reflects an accurate measure of anticipated benefits. Any variations that may arise in future output will be reflected as an adjustment of costs in that year without the need to reallocate prior year costs.

C. COMPARABLE METHODS

Participants in this submission include companies with longstanding cost sharing arrangements with related parties and some who have undertaken cost sharing projects with unrelated parties. These companies agree that the overriding principle to be followed in defining a qualified cost sharing agreement is the arm's length standard to test whether the provisions of an agreement would be found in agreements between unrelated parties. A cost sharing agreement that allocates costs in a manner that is consistent with methods employed by unrelated parties should not result in a reallocation of costs. We propose, therefore, that just as there are matching transactions or comparable uncontrolled prices, the regulations provide that the existence of a comparable cost sharing agreement with a non-controlled party establishes that a comparable agreement among related parties is arm's length.

D. ELIGIBLE PARTICIPANTS

Eligible participants should include separate entities (members of the controlled group) performing research and research application within a cost sharing arrangement for other participants at cost. While the regulations recognize that a unit of a manufacturing company may perform research, the regulations fail to address existing corporate structures where research is performed by a separate subsidiary that recovers 100 percent of its costs from participants for research and distributes 100 percent of any benefits derived back to the participants. Under the proposed regulations, since the intangible developed will not be used in the active conduct of the subsidiary's trade or business, the participant would be considered an ineligible participant and treated as an assister under the rules of paragraph (d)(8). Under such rules, the district director may make allocations to reflect an arm's length consideration for such assistance. There is no reason for this disparate treatment between taxpayers, where one taxpayer in a manufacturing unit may be allowed to perform research at cost while another taxpayer performing research in a separate entity may be required to charge an arm's length consideration.

The proposed regulations define a specified interest in an intangible as any legally enforceable interest. This should include situations where a separate entity performs the research and holds legal title to the patents while the participants have a royalty- free, nontransferable, nonexclusive license to utilize the technology developed under the cost sharing arrangement. The participants hold an equitable interest in the property developed and impose certain conditions on the separate entity if technology is ever transferred outside the cost sharing arrangement during the term of their license.

The proposed regulations should also expand the definition of an eligible participant to allow non-controlled entities to participate. Otherwise, existing agreements would need to treat non-controlled entities as ineligible participants and reallocate their costs to the other eligible participants. Since the purpose of section 482 is to place a controlled taxpayer on a tax parity with an uncontrolled taxpayer, the inclusion of a non-controlled entity in a cost sharing arrangement should be encouraged in that it reflects that the costs and risks are being shared on an arm's length basis among the participants.

Guidance on several rules would be enhanced by the use of illustrative examples. This could include examples illustrating the eligible participants and related rules such as the designation of a party to act as the participant on behalf of other parties (and the contractual arrangements among them). To recognize the role of a separate research entity, we believe the insertion of the following example in the regulations would be helpful:

EXAMPLE -- Controlled corporations A, B, C, and D enter into a qualified cost sharing arrangement (CSA) for the purpose of developing an intangible. Corporation D does not use the intangibles in the active conduct of its trade or business but it qualifies as an eligible participant because it is strictly a research company that performs all the research and research application work funded by the participants in the CSA. While corporation D holds legal title to the patents developed under the CSA, the technology developed is equitably owned by the CSA participants who funded the technology. The CSA participants have a royalty-free nontransferable, nonexclusive license to utilize the technology developed under the CSA and any benefits (i.e., royalties) received by corporation D are returned to the CSA participants by means of a current cost reduction and are not taxable at the research entity level.

E. DEFINITION OF U.S. PARTICIPANT

We are not certain why a U.S. participant is defined to include a U.S. controlled foreign corporation. Assuming that such inclusion is desired for administration of the rules, we note that the present definition also includes potentially every foreign corporation since only one share of stock owned (or potentially owned) by one U.S. person literally makes the corporation a U.S. participant.

F. SCOPE OF AGREEMENT

We concur with the facts and circumstances approach of the proposed regulations with respect to the subject matter of an agreement. We suggest that additional examples are desirable to illustrate that bona fide arrangements may be either narrow or broad in scope, provided that there is a clear nexus between contributions and possible utilization by the parties. Existing cost sharing agreements used by this group of taxpayers with related parties cover a range of projects and products that have been reviewed and accepted as arm's length. At one end of the range, these agreements focus on single research projects which terminate upon commercialization of the product. Project costs are shared ratably by affiliates, and since the agreements terminate, no adjustments are made. At the other end of the range, all research conducted by affiliates is pooled and shared based upon current sales or units. Such agreements are similar to cost sharing arrangements entered into by unrelated parties.

We are concerned that, while SIC codes are intended to be only one factor to consider, revenue agents may be tempted to routinely challenge cost sharing across SIC code groups. We believe that it would be helpful therefore, to explicitly state that participants in different SIC codes (and involved in different product lines) are permitted to participate in a cost sharing arrangement as long as there is potential use by all participants.

Many research programs are conducted under the umbrella of a cost sharing arrangement where all participants have access to the technology developed under the agreement. The rationale for this is that in the course of operations over the years, there are many research projects which are only of current interest to certain cost sharing members but over time may be of indirect interest to others. Members are interested in pursuing that research at their own cost because they realize that other members will eventually undertake other projects, the results of which will also be available to all members. The regulations should allow barter-exchanges of information or research results under the umbrella of a cost sharing agreement. If every piece of data or the results of every experiment not fully supported by all agreement members had to be tracked and compensation provided, then the administration of the agreement would become unmanageable, especially when such information is of minimal value to the other agreement members. There is no real business or tax revenue purpose served by tracking such de minimis "spillover" of research data.

In addition, many companies perform research that may have no potential use to members within the cost sharing arrangement but which may benefit the overall organization in the distant future. Such research is in the nature of basic or very long range research in areas where it is reasonable to expect that any new products or services developed will not be used in the active conduct of a trade or business of any participant in a commercially significant manner. In order to provide a significant and stable level of basic research each year, the proposed regulations should allow for the funding of such costs within a cost sharing arrangement by the parent of the corporation. Any attempt to require mutualization of such basic research costs may only result in a reduction in the funding of such projects, since many participants will not see the benefits of such research and other governments either restrict or disallow local tax deductions for the payment of such costs. The alternative will be the reduction of U.S. based basic research.

G. EXTENSION OF RULES TO TRANSACTIONS WHOLLY AMONG FOREIGN ENTITIES THAT ARE NOT U.S. CONTROLLED

If any U.S. participant, or U.S. owned participant, has borne its proper share of costs and receives its appropriate share of benefits, we do not believe that U.S. rules need to be mandated for the arrangements applicable among non-U.S. controlled foreign participants. The administrative requirements may be particularly burdensome for these wholly foreign to foreign arrangements.

H. TRANSFERS TO THIRD PARTIES

Assuming that all parties to the cost sharing arrangement have borne their appropriate share of costs and receive their appropriate share of rights to developed intangibles, we fail to see why any participant cannot exploit the rights received by way of a sale or license or other transfer of such intangibles to a third party.

I. MEASURE OF CONTRIBUTIONS

The proposed regulations properly look to the practical means that unrelated parties would employ to fix shares of costs, such as projections of sales or units based upon the most objective information available in the form of current sales, units, or averages of past experience. In the case of capital intensive industries, contributions based on capital employed might be an appropriate basis.

Current units or sales are preferable to projections of future sales or profits. We assume that references to anticipated sales or profits allows projections based on past results. In the case of start-up projects, forecasts can be based upon how comparable products are or have performed in the markets involved. Because costs are incurred in advance of benefits, each of the parties inevitably bears some risk that actual profits or shares will vary from projections.

J. COSTS INCLUDED IN INTANGIBLE DEVELOPMENT AREA -- PURCHASED TECHNOLOGY

The costs of developing intangibles to be shared is defined to include all of the direct and indirect costs of the intangible development area. The definition should be expanded to include the sharing of costs incurred in purchasing technology (e.g., in the form of a lump sum or running royalty license fee). Such license/patent acquisition costs are incurred to provide participants access to relevant technology. The costs of purchasing the third party technology can then be allocated to the participants through their cost sharing payments.

K. BUY-IN/BUY-OUT REQUIREMENTS

The form of consideration for a buy-in or buy-out should also be expanded to recognize other types of consideration in addition to cash payments. Participants in a financially strapped cost sharing arrangement may actively solicit the participation of new members in order to be able to continue research in high risk technology in the hopes of developing a useful product. A new participant's commitment to shoulder a part of these significant future costs should be viewed as consideration in and of itself without the need for a buy-in payment. Similarly, a financially strapped participant who cannot meet its contractual obligation to continue funding significant future costs may be more than willing to give up its interest in the development of a risky intangible in return for being contractually relieved of any future payments.

Moreover, the regulations should be revised to reflect that a buy-out payment is appropriate only where the remaining participants clearly benefit from the departure. Within cost sharing arrangements, participants already have access to information and results of the technology being developed and the abandonment of a participant does not automatically give them any additional rights to the use of the intangible that they did not already possess. Any additional benefit derived by the remaining participants is more than offset by the increased detriment incurred by the participants in shouldering the additional future cost resulting from departing members.

The inequities of a cash buy-out payment is reflected in situations where a participant decides to terminate its funding of a particular research project. Instead of being penalized for breaking their contractual obligations, the proposed regulations would require that a cash payment be made to the departing participant, thereby offering an incentive to participants to terminate high risk research projects in certain years. In other situations, a participant that has been a member of a cost sharing arrangement for many years may be forced to shut down in a particular country because it is unprofitable to continue operations there. The proposed regulations would require that a cash payment be made to the departing participant even though the other participants would not benefit from the departure and would actually suffer a detriment by shouldering the future costs of the departing member.

The examples of buy-out requirements provide for payment to a withdrawing entity of compensation equal to the costs it incurred up to the withdrawal and require additional payments based upon an enhanced market value because the other participants continued with the research and subsequently produced valuable technology. This would not be done in an arm's length situation. A buy-out payment would be appropriate only where the remaining participants clearly benefit from the departure. The amount of the buy-out should be based solely on the additional benefits received by the remaining participants at the time of the departure.

A situation in which the departure of a participant has no effect on a U.S. participant (or upon U.S. taxation interests), occurs where the U.S. participants' contributions are based solely on its future rights in the U.S. The U.S. participant should not be required to compensate a withdrawing non-U.S. controlled foreign participant whose participation was for foreign rights.

L. COORDINATION WITH SERVICES RULE

The proposed regulations have elaborated upon a concept of developer/assister relationships intended to apply to such things as contract research. We believe that it is desirable to further clarify the relationship of these rules to the services regulations. For example, we assume that the tests employed in the services regulations to determine whether payments may be for costs or whether a profit element is required remain unchanged.

M. REALLOCATION OF COSTS SHARED

We believe that where the parties have implemented a written qualified cost sharing agreement, the basic sanction for intercompany pricing adjustments should be an adjustment of costs (with interest) and not deemed transfers of intangibles at less than arm's length value. Recognizing the nature of a cost sharing arrangement, where all activities represent a risk with no assurance of success, reallocation of costs should be the sole recourse where it is determined that the method employed failed to measure the benefits reasonably anticipated. The intangible transfer rules should not apply since a completed intangible has not been transferred. It is not justifiable to disregard the legal relationship created by the cost sharing agreement merely because the cost allocations may be viewed with hindsight as disproportionate. This will likely lead to unwarranted controversies with foreign governments and increased uncertainty during audits. Realigning costs under a cost sharing arrangement to reflect arm's length cost allocations under a cost sharing agreement provides an appropriate remedy.

N. YEAR OF REALIZATION

The regulations require that reallocations of cost sharing payments be included in income in the taxable year under review (with an appropriate interest charged), even if the costs to be allocated were incurred in a prior taxable year. This requirement is of questionable validity and may lead to difficulty in avoiding double taxation if the adjustments relate to years which have become statute barred in the foreign jurisdiction.

 

FOOTNOTES

 

 

1 Ways and Means Subcommittee on Oversight, July 12, 1990, p. 4.

2 Competitive capital markets enhance this comparability because companies as well as individual investment projects must compete for available investment finds. Competitive capital markets tend to force rates of return towards equalization over the long term, thereby enabling meaningful comparison.

3 The safe harbor range for the licensor could be 20 percent to 30 percent, for example.

 

END OF FOOTNOTES

 

 

APPENDIX A

The following is excerpted from . . .

DO YOU SEE "CONVERGENCE"? EXAMPLES OF REAL WORLD DATA AND THE COMPARABLE PROFIT INTERVAL OF PROPOSED REGULATION SECTION 1.482

by John M. Simpson, Ph.D. and Garry B. Stone, Ph.D. 1

I. INTRODUCTION

This article evaluates the practical applicability of "convergence" among "profit level indicators" (PLIs) used to determine acceptable transfer prices in accordance with the newly proposed intercompany pricing regulations. The article compares the hypothetical data of the IRS's primary example as presented in the proposed regulations to real world financial data. Our analysis indicates that the IRS may be overly optimistic regarding the practical implementation of convergence and statistical techniques to determine intercompany prices.

II. OBJECTIVES

The primary objective of this article is to examine actual financial data to determine the extent to which convergence of the PLIs from public companies is discernable. Our second objective is to evaluate appropriate statistical techniques for determining convergence of the PLIs and the "most appropriate point" within a given "comparable profit interval" (CPI).

We focus our analysis on several standard industry classifications (SIC), generally without examining similarities and dissimilarities among the companies in terms of products, functions, intangibles and risks. Such a detailed analysis would likely be necessary to restrict the samples further so that they were only comprised of those companies most similar to the controlled entity in question. We used U.S. data because it is our experience that more data is available for the U.S. than for other countries. In this article, we evaluate the usefulness of statistical techniques in determining convergence to an appropriate CPI. We analyze standard statistical measures of central tendency and dispersion of the PLIs earned by companies in a number of samples. As such, the convergence we seek is a convergence among the sample of companies for each PLI. 2

Prior to analyzing our selected data, it is useful first to examine the hypothetical distributor data from Prop. Treas. Reg. section 1.482-2(f)(11) Example 3 (hereafter referred to as "Example 3") in order to apply the statistical measures to the IRS's hypothetical fact pattern for which convergence occurs.

III. EXAMPLE 3 (PROPOSED REGULATIONS SECTION 1.482-2(f)(11))

Example 3 demonstrates how to construct a CPI and illustrates convergence of PLIs. The following three PLIs are used to test transfer prices of a hypothetical controlled transaction: (i) operating income divided by total book assets (OI/A); (ii) operating income divided by sales (OI/S); and (iii) gross income (i.e., sales less cost of goods sold) divided by operating expenses (i.e., gross income less operating income) (GI/OE). Convergence in this IRS example is illustrated with hypothetical PLIs where the tested party is assumed to be a wholesale distributor of consumer products. The IRS presents a sample of eight hypothetical uncontrolled consumer product distribution companies.

Following the dictates of Prop. Treas. Reg. section 1.482- 2(f)(2), three years (centered on the year under audit) of financial data are averaged before calculating PLIs. It is important to note that the IRS approach eliminates cyclical variability of profits over the three-year period 3 for purposes of the test. While this averaging reduces the effect a single year of unusually high or low profits will have on measures of dispersion, it also reduces the number of available independent PLI observations by two-thirds. Reducing the number of independent observations (i. e., eliminating information) reduces the applicability of statistical tests, especially when the samples are already somewhat limited.

In determining a CPI from the sample of eight companies, the IRS chose the four companies with PLIs that yield the constructed operating incomes (COIs) that "are clustered most closely," (Prop. Treas. Reg. section 1.482-2(f)(11) Example 3 (v)). In effect, the IRS constructed a CPI by selecting the 50 percent of the sample that minimizes the spread between the most- and least- profitable companies in the range. We call this concept "modal range." It is the smallest range within which a given percentage of observations in the sample fall: 50 percent in Example 3 and the rest of this discussion. OI/A ranges from 5.8 percent to 8.0 percent for the four companies in the modal range compared to a range of 2.7 percent to 23.3 percent for the entire sample. The modal range of this CPI is thus relatively narrow in the IRS's example (2.2 percentage points).

The modal range is generally narrower than the interquartile range which makes it seemingly superior for determining convergence. The interquartile range is much larger than the modal range, but narrower than the mean plus or minus one standard deviation. The interquartile range and standard deviation are both common statistical measures of dispersion. Figures 1a, 1b, and 1c compare the three measures of dispersion as applied to data from Example 3. The samples of "real world" data generally exhibit a similar pattern, with the modal range the narrowest measure of dispersion. The concept of modal range suffers, however, from a definitional problem: it does not relate to standard measures of central tendency. In the IRS's Example 3, the sample's mean OI/A of 10.1 percent lies outside the CPI of 5.8 percent to 8.0 percent. The median of 7.3 percent is near the top of this range. Measures of central tendency are to be used to determine the "most appropriate point" within the CPI (Section 1.482-2(f)7)); in this case, the most appropriate point, as determined by one typical measure of central tendency, lies outside the CPI.

Figure omitted

While Example 3 presents some insight into the workings of the CPI, it suffers from two important drawbacks. First, the example suggests using a modal range, which is not a good measure of dispersion because it does not relate well to measures of central tendency. Second, Example 3 could have used real world data. Recent major transfer pricing court decisions (e.g., Sunstrand, Lilly, Bausch & Lomb) or specific industry data could have provided interesting starting points. Below, we explore real world data from several industries.

IV. REAL WORLD DATA -- DISPERSION

The discussion of Example 3 shows the interquartile range to be a narrower measure of dispersion than the mean, plus or minus one standard deviation, and that the modal range is generally narrower still than the interquartile range. The facts that the mean and standard deviation are greatly influenced by "outliers" in a sample and that the modal range (as used as a measure of dispersion by the IRS in Example 3) is not directly tied to any measure of central tendency, lead us to prefer the interquartile range as the most useful measure of dispersion for determining a CPI from a sample of uncontrolled entities. We will, however, discuss dispersion and convergence in terms of modal range because it is generally the narrowest (not the best) of the three measures and is effectively the method used by the IRS in Example 3.

Figures 2a, 2b, and 2c show the modal ranges calculated for Example 3 and the five real world samples for each of the three PLIs. In all cases, the modal range for Example 3 is much narrower (less than half as wide) than that of any of the real world samples. In fact, the modal ranges for real world data are very wide. Consider a tested party under the OI/A having $100 million in assets. The modal range will permit operating income of zero to $5.5 million per year if the tested party were a Clothing Wholesaler or zero to $13.5 million if the tested party were a Computer Manufacturer. Note also that the two manufacturing industries have wider modal ranges than the wholesaling industries. For a tested party under OI/S, a Clothing Wholesaler having $200 million in sales will be permitted an operating loss up to $1.8 million or a profit up to $3.8 million. Similarly, a computer manufacturer will be permitted an income range of $3.6 million to $29.8 million.

[figure omitted]

A high degree of dispersion in the PLI samples indicates that the appropriate profit or narrow range of profits will often require a closer matching of products sold, functions performed, and risks borne between the tested party and a group of uncontrolled comparables. Basically, more information than financial data by SIC code is likely to be necessary to construct an appropriately narrow sample of comparable companies. In general, functional analysis is essential to achieve more narrow convergence. For example, one option might be to eliminate computer software wholesalers contained in SIC 5045: Wholesalers of Computers, Peripherals, and Software, by examining company descriptions from Forms 10-K and annual reports. In Figures 2a, 2b, and 2c comparing Computer Hardware and Software Wholesalers and Computer Hardware (software excluded) Wholesalers, shows that excluding Software Wholesalers establishes a narrower modal range for two of the PLIs. However, this narrower sample still yields a rather wide range. Further analysis of the functions, risks and intangibles of the remaining firms could be used to narrow the sample even further to reach conclusions about arm's length profitability for the tested party. Such analysis relies heavily on the specific facts surrounding the tested party and the related transaction in question.

A complex tested party adds further importance to the functional analysis. The discussion has focused on the financial data of independent companies being loosely comparable to a hypothetical tested party. To the extent that tested parties diverge from typical or routine functions, risks, and intangibles within an industry, more care must be exercised in applying PLIs from other participants in the industry. Complex entities on both sides of intercompany transactions are not uncommon for multinational companies of the 1990s.

V. DO YOU SEE CONVERGENCE?

We find that all of the samples of real world data examined here exhibit greater dispersion than the hypothetical sample of wholesale distributors in IRS Example 3. At what point or at what "spread" between maximum and minimum PLI values have we found convergence in the real world samples that will be useful in determining appropriate intercompany prices? To what extent is convergence in "the eye of the beholder"?

VI. CONCLUSIONS

Our analysis leads us to two conclusions about convergence of PLIs for real world data, of which the second is the most important.

1. CONVERGENCE AMONG MEASURES OF CENTRAL TENDENCY. Most, but not all, of the samples showed the mean or median to be of similar magnitude; the mode often was not close to either the mean or median. In fact, several samples exhibited multiple modes. It would be helpful for setting intercompany prices if (at least) the measures of central tendency "converged." Also, the mean, a measure of central tendency, fell outside of the IRS's CPI in Example 3. This shows that caution must be exercised in selecting measures of central tendency and dispersion that are consistent with each other when determining the "most appropriate point" within a CPI.

2. Convergence of PLI observations within a sample. We have NOT found a narrow dispersion within the selected samples. For example, the modal range of OI/A ratios is 4.0 percent to 12.5 percent for Wholesalers of Computer Hardware and Software (Figure 2a). If this range is used to set transfer prices for an entity with assets of $200 million, the entity's constructed operating income (COI) could range from $8.0 million up to $25 million. This is quite different from the $11.6 million to $16 million suggested in Example 3. If an entity had sales of $200 million, the modal range of 0.8 percent to 4.2 percent for OI/S (Figure 2b) would permit a COI of $1.6 million to $8.4 million. This is also quite different from the $6.2 million to $8.4 million suggested in Example 3.

In past transfer pricing studies, we have frequently used analyses similar to the CPI analysis of the proposed regulations to test the reasonability of transfer pricing. We prefer using simple statistical measures such as the median and interquartile range because real world samples are frequently small (especially samples of foreign data) and extreme "outliers" frequently exist. Our experience indicates that CPI-type analysis works best: (1) when the tested party in a related transaction and the potential comparables are relatively similar (i.e., the tested party and comparables have simple operations, few intangibles, and face minimal risk); and/or (2) when there is a very large difference (in percentages, not just in dollars) of opinion between the taxpayer and the IRS during examination as to the appropriate transfer price (i.e., the CPI serves as a test on the reasonability of vastly differing proposed transfer prices).

The analysis summarized in this article supports our belief that the CPI analysis will be less useful: (1) when both parties to a transaction are complex (i.e., have substantial intangibles and/or risk); (2) when attempting prospectively to establish an exact transfer price; and (3) when trying to distinguish relatively small (again, in percentage terms) differences of opinion between taxpayers and the IRS.

Where CPI is less useful, broader consideration of the facts and circumstances of the related parties and the potentially comparable transactions would be helpful and, in most instances, necessary. To make the best determination of an arm's length nature of a related- party transaction, consideration of inexact comparables (for intangible as well as tangible property transactions) and other financial and nonfinancial information may be necessary. With the myriad fact patterns that occur in intercompany pricing situations, it would be surprising indeed if any single analytical method (e.g., statistical CPI analysis) were most appropriate under all circumstances.

APPENDIX B

PROPOSED SEC. 482 REGULATIONS DIFFICULTIES IN PRODUCING JAPANESE DATA FOR THE CPI TEST

This memorandum discusses three major difficulties in obtaining the necessary data for profit-level indicators of Japanese companies which would be necessary to compute the CPI under the proposed Section 482 regulations. It is assumed that there is a U.S. parent company with a major Japanese operating subsidiary in the business of manufacturing and selling a finished product. The Japanese subsidiary is assumed to be the "tested party."

CONCLUSION

If the Japanese subsidiary is the tested party, there may be substantial difficulties in performing the CPI test due to the lack of necessary data caused by:

o The absence of comparable independent companies engaging in the same business;

o Insufficient public information in general -- even for publicly held companies, the available information may not be useful; and

o Differences in accounting principles which may make a comparison difficult.

ABSENCE OF COMPARABLE COMPANIES

In any industry, the potential comparables may also be multinationals. For example, all of the major computer manufacturers in Japan have substantial intercompany sales to their foreign subsidiaries. Therefore, they are not proper comparables in performing the CPI test. There may be Japanese companies which manufacture components without substantial foreign intercompany transactions. However, the products and functions may not be comparable to the products and functions of a major MNC.

INSUFFICIENT PUBLIC INFORMATION

Publicly available financial information on Japanese companies is limited. Publicly held companies and other designated companies (for example, issuers of debenture bonds traded in one of the Japanese stock exchanges) are required to file Yukashoken-Hokokusho ("YU-HO") with the Ministry of Finance. YU-HO is equivalent to the SEC reports of the United States and is released to the general public.

If the comparable company is publicly held, YU-HO may be used as a source of data. Although YU-HO provides segment information as a supplemental report, the segments in this report differ considerably from those for SEC reporting purposes.

As discussed, it will often not be possible to find a publicly held company which performs comparable functions similar to the U.S. MNC, and which at the same time does not have foreign intercompany transactions.

In addition to YU-HO, large Japanese companies (as defined) under the Commercial Code are required to publish their condensed balance sheet and income statement in a Japanese newspaper or official gazette, as well as registration. The other Japanese companies are required to publish and register their condensed balance sheet. However, this requirement has not been strictly observed due to the lack of penalties. Although the condensed balance sheet and income statement published in the newspaper could be useful for the CPI test, the data do not include any footnotes or segment reports.

In addition, details of how the taxable income is computed are not available to the public. This is not a feasible alternative for the CPI test due to the lack of segment information and other details.

There is a private data service for general corporate information, including financial information for private companies. This database is compiled from the voluntarily submitted information by the participant corporation. Therefore, the accuracy of the data may not be reliable. Also, the reporting standards may not be consistent. Therefore, the application of this private data bank appears limited.

DIFFERENCES IN ACCOUNTING PRINCIPLES

The proposed regulations require that the CPI test be performed on the tested party's income as computed for U.S. tax purposes. However, the available financial data of comparable companies is based on Japanese accounting and tax principles. Therefore, the direct comparison of income data under the different accounting methods may not provide reasonable results. In order to make the comparison useful, the differences in the accounting method used for the compilation of the financial data should be reconciled.

For example, the accounting treatment of the following items is clearly different and the difference could be significant:

o Provision for employees' severance indemnities;

o Provision for doubtful receivables (statutory rate);

o Provision for seasonal employee bonuses;

o Provision for returned sales (guideline formula);

o Depreciation method;

o Foreign currency denominated receivables and payables;

o Statutory reserves;

o Bonuses to directors; and

o Timing of income recognition.

In practice, moreover, there is no information even in YU-HO which is sufficient to reconcile the financial data under the Japanese GAAP to the U.S. rules.

In summary, there are formidable difficulties in developing meaningful profit-level indicators to test the Japanese subsidiary of a U.S. parent under the CPI test.

 

FOOTNOTES TO APPENDIX

 

 

1 Dr. Simpson is a Partner at Price Waterhouse's Washington National Tax Services' Intercompany Pricing Group based in Washington, D.C. and the firm's lead economist in transfer pricing. Dr. Stone is a Senior Manager also specializing in transfer pricing with the same group. The authors wish to express special thanks to Brian Powilatis for his important contributions to the development of this article. We would also like to thank Larry Dildine and Lekha Sivarajan for their input and assistance.

2 We did not test for "convergence" of constructed operating incomes (COI) resulting from the various PLIs. "Convergence" of COIs is a more stringent test than "convergence" for a specific PLI because it requires not only "convergence" within each PLI, but also "convergence" toward a common COI derived from the various PLIs.

3 Some sort of averaging over time is a welcome addition to the proposed regulations as compared with the existing regulations. Three years, of course, may not be sufficient, but if one can demonstrate that the "circumstances indicate that a different period is more appropriate" (Sec 1.482-2(f)(2)) the regulations allow different time periods. We expect that in many situations, three years is not the most appropriate time period. For example, various markets for durable goods could face relatively long business cycles. In addition, it will be impossible for taxpayers to use information about a future year's actual financial results for tax planning purposes.

 

END OF FOOTNOTES
DOCUMENT ATTRIBUTES
  • Institutional Authors
    Price Waterhouse
  • Cross-Reference
    IL-372-88

    IL-401-88
  • Code Sections
  • Subject Area/Tax Topics
  • Index Terms
    related-party allocations
  • Jurisdictions
  • Language
    English
  • Tax Analysts Document Number
    Doc 92-5700 (69 original pages)
  • Tax Analysts Electronic Citation
    92 TNT 135-34
Copy RID