International Taxation: Transfer Pricing and Information on Nonpayment of Tax (Letter Report, 04/13/95, GAO/GGD-95-101). Pursuant to a congressional request, GAO provided information regarding transfer pricing issues and foreign-controlled corporations' (FCC) and U.S.-controlled corporations' (USCC) tax compliance, focusing on: (1) the Internal Revenue Service's (IRS) handling of transfer pricing issues through its examinations, appeals, and litigation functions; and (2) IRS use of available regulatory and procedural tools. GAO found that: (1) recent IRS experiences with transfer pricing cases have been mixed; (2) although there were as many regulatory violations in 1993 and 1994 as in previous years, the value of the 1994 adjustments increased $1.3 billion over 1993 adjustments; (3) a large number of the 1993 and 1994 cases involved pricing methods other than the three methods specifically described in earlier IRS regulations; (4) the outcomes of IRS appeals and legal processes for the 2 years were similar to those in 1987 and 1988, with a sustention rate of about 30 percent of the proposed adjustments' value; (5) IRS has used certain procedural tools, such as simultaneous examinations and arbitration, as effective deterrents to abusive transfer pricing practices; (6) IRS expects to increase its use of advanced pricing agreements; (7) the success of the new transfer pricing regulations remains to be seen; (8) about 75 percent of FCC and 60 percent of USCC paid no U.S. income tax between 1987 and 1991; (9) the corporations that paid U.S. taxes in 1991 held 80 percent of FCC and USCC assets and generated 81 percent of their receipts; (10) the largest nontaxpaying corporations accounted for most FCC and USCC assets and receipts; and (11) factors other than transfer pricing abuse may contribute to the differences in tax amounts paid by FCC and USCC. --------------------------- Indexing Terms ----------------------------- REPORTNUM: GGD-95-101 TITLE: International Taxation: Transfer Pricing and Information on Nonpayment of Tax DATE: 04/13/95 SUBJECT: Tax administration Foreign investments in US Tax nonpayment Tax evasion Foreign corporations Income taxes Price adjustments Litigation Tax law Voluntary compliance ************************************************************************** * This file contains an ASCII representation of the text of a GAO * * report. Delineations within the text indicating chapter titles, * * headings, and bullets are preserved. Major divisions and subdivisions * * of the text, such as Chapters, Sections, and Appendixes, are * * identified by double and single lines. The numbers on the right end * * of these lines indicate the position of each of the subsections in the * * document outline. These numbers do NOT correspond with the page * * numbers of the printed product. * * * * No attempt has been made to display graphic images, although figure * * captions are reproduced. Tables are included, but may not resemble * * those in the printed version. * * * * A printed copy of this report may be obtained from the GAO Document * * Distribution Facility by calling (202) 512-6000, by faxing your * * request to (301) 258-4066, or by writing to P.O. Box 6015, * * Gaithersburg, MD 20884-6015. We are unable to accept electronic orders * * for printed documents at this time. * ************************************************************************** Cover ================================================================ COVER Report to Congressional Requesters April 1995 INTERNATIONAL TAXATION - TRANSFER PRICING AND INFORMATION ON NONPAYMENT OF TAX GAO/GGD-95-101 Transfer Pricing Abbreviations =============================================================== ABBREV AC(I) - Assistant Commissioner (International) APA - Advance pricing agreement FCC - Foreign-controlled corporation IRS - Internal Revenue Service SOI - Statistics of Income USCC - U.S.-controlled corporation Letter =============================================================== LETTER B-260157 April 13, 1995 The Honorable Byron L. Dorgan United States Senate The Honorable Paul E. Kanjorski House of Representatives This report responds to your requests to provide recent data on transfer pricing issues and on tax compliance of foreign- and U.S.-controlled corporations.\1 Specifically, we are providing information and analysis to update our 1993 work on (1) IRS' recent experience in dealing with transfer pricing issues through its examinations, appeals, and litigation functions and (2) IRS' use of available regulatory and procedural tools. We have also used 1990 and 1991 tax data to update our analyses of how many U.S.-controlled corporations (USCC) and foreign- controlled corporations (FCC)\2 , did not pay U.S. income taxes. Our 1993 work in some instances provided information for tax years dating back to 1987. -------------------- \1 We previously testified on these issues in International Taxation: Updated Information on Transfer Pricing (GAO/T-GGD-93-16, Mar. 25, 1993) and provided tax information in International Taxation: Taxes of Foreign- and U.S.- Controlled Corporations (GAO/GGD-93-112FS, June 11, 1993). \2 Foreign-controlled corporations are U.S. corporations of which at least a specified percentage of the voting stock is owned by a foreign individual, partnership, corporation, estate, or trust. U.S.-controlled corporations are other corporations, although certain entities such as subchapter S corporations, which are corporations that are treated for federal income tax purposes like partnerships, are not included in either category. BACKGROUND ------------------------------------------------------------ Letter :1 A transfer price is the price charged by one company for a product or service supplied to a related company, such as the price a parent corporation charges its wholly-owned subsidiary. Any company that has a related company with which it transacts business establishes transfer prices for those intercompany transactions. Although often associated with the pricing of tangible goods, transfer pricing occurs whenever income and expenses are allocated among interrelated companies. For example, the payment of royalties, interest payments for debts, leasing expenses, and fees for other services between interrelated companies are transactions requiring transfer prices. Pricing of intercompany transactions affects the distribution of profits and, therefore, taxable income among the related companies and, sometimes, across tax jurisdictions. Abusive transfer pricing occurs when income and expenses are improperly allocated among interrelated companies for the purpose of reducing taxable income in a high-tax jurisdiction. Underpayment of U.S. income taxes can result from inappropriate transfer pricing between interrelated companies with operations in both the United States and in a country with a lower tax burden. Even when the U.S. corporate tax rate is lower than that of some other country, transfer pricing abuses can occur by shifting income through another related company that operates in a tax haven, that is, a country with low or no taxes. The following is an example of abusive cross-border transfer pricing. A foreign parent corporation with a subsidiary operating in the United States charges the subsidiary excessive prices for goods and services rendered (for example, $1,000 instead of the going rate of $600). This raises the subsidiary's expenses (by $400), lowers its profits (by $400), and effectively shifts that income ($400) outside of the United States. At a 35-percent U.S. corporate income tax rate, the subsidiary will pay $140 less in U.S. taxes than it would if the $400 in profits were attributed to it. Section 482 of the Internal Revenue Code provides IRS authority to allocate income among related parties if IRS determines that the transfer prices used by the taxpayer are inappropriate. To evaluate transfer pricing, the IRS examiner considers what the price would have been if the parties had not been related to each other. Such a price between unrelated parties is called the "arm's length" price. Finding a section 482 violation, that is, a difference between the price a related party charged and the arm's length price, the IRS examiner can propose an adjustment to the taxpayer's income. If the taxpayer does not agree with the proposed adjustment, it can appeal the dispute through IRS' appeals process or take the case to court. In July 1994, IRS issued new regulations under section 482 that differed significantly from previous section 482 regulations. Under the new regulations, taxpayers have great latitude in establishing transfer prices. However, under 1993 legislation, taxpayers are subject to new requirements for documenting their transfer prices, and they face stiff penalties for substantially misstating them. RESULTS IN BRIEF ------------------------------------------------------------ Letter :2 IRS' recent experiences with examinations, appeals, and litigation relating to section 482 issues have been mixed. For instance, in 1993 and 1994, IRS examiners found, as they had in previous years, large section 482 violations. However, the total dollar value of the adjustments in the 2 years differed by $1.3 billion. Examiners also noted that, in a large proportion of cases, transfer pricing activity involved pricing methods other than the three--uncontrolled price, cost plus, and resale price--that were specifically described in section 482 regulations over the years. The outcomes of the appeals and legal processes in 1993 and 1994 were similar to those in 1987 and 1988, with IRS sustaining less than 30 percent of the proposed section 482 adjustment amounts. In 1993 and the first part of 1994, IRS had somewhat better success litigating large transfer pricing cases than in 1990 through 1992. According to IRS officials, certain enforcement tools available to IRS, such as measures to obtain information and stronger penalties, have served mostly as deterrents that altered taxpayer behavior. Alternatives to traditional examinations, appeals, and litigation, such as simultaneous examinations, arbitration, and advance pricing agreements, either have been used infrequently or are expected to grow in number in the future. Similar to examiners' experience, advance pricing agreements have prominently featured pricing methods other than the three specifically described in earlier regulations. How successful the new transfer pricing regulatory regime will be remains to be seen. The flexibility that new regulations allow taxpayers in applying the arm's length standard must be weighed against the flexibility given IRS and the increased documentation required of taxpayers under threat of penalty. A majority of all FCCs and USCCs paid no U.S. income tax in each year from 1987 through 1991, and the percentages of each--nearly three-quarters of FCCs and about 60 percent of USCCs--remained largely unchanged over the 5-year period.\3 Although taxpaying corporations were a minority of all FCCs and USCCs, they owned the majority of corporate assets and generated most of the receipts. Furthermore, the largest nontaxpaying corporations--those with assets of $100 million or more--were relatively few in number but accounted for relatively large proportions of all FCCs' and all USCCs' total assets and receipts. While the differences in tax payment rates between FCCs and USCCs are not convincing evidence of transfer pricing abuse, transfer pricing abuse cannot be ruled out as at least a possible cause for part of the observed differences. Other factors, such as the different types of industries for the nontaxpaying FCCs and USCCs, may also account for some of the observed differences. -------------------- \3 The confidence intervals are plus or minus 4 percent for the FCC data and plus or minus 0.9 percent for the USCC data. (See app. V.) IRS' EXAMINATION, APPEALS, AND LITIGATION EXPERIENCES WITH SECTION 482 WERE MIXED ------------------------------------------------------------ Letter :3 IRS' international examiners continued to propose substantial adjustments to the taxable income of FCCs and USCCs in fiscal years 1993 and 1994, although the total dollar value of the adjustments in 1993 was $1.3 billion less than in 1994. In examinations finished in fiscal year 1993, international examiners proposed adjustments to taxable income of $2.2 billion for 369 corporations--$900 million for 247 FCCs and $1.3 billion for 122 USCCs. For 1994, IRS data showed $3.5 billion in proposed adjustments. Most of the dollar value of these proposed section 482 adjustments was for large cases, those that had total proposed adjustments of $20 million or more. For example, in 1993 examiners proposed $1.8 billion in adjustments for 51 of these large cases--$700 million for 18 FCCs and $1.1 billion for 33 USCCs. Although the 18 large FCCs subject to proposed income adjustments in 1993 represented an increase in number over the previous 4 years, when 11 to 13 FCCs were subject to proposed income adjustments for transfer pricing, they also represented a decrease in dollars. The number of USCCs subject to income adjustments for transfer pricing in 1993, in contrast, was about the same as in previous years, although this group also involved fewer dollars.\4 According to IRS, the 1993 adjustments for transfer pricing might be understated due to data that were lost in implementing a new management information system. Transfer pricing issues for which IRS examiners proposed income adjustments in 1993 occurred about equally in four categories, while pricing of tangible goods represented the largest share of the adjustment dollars proposed. As shown in figure 1, the categories with the most frequent section 482 issues were interest, royalties, pricing of tangible goods, and a nonspecific category of income allocations and deductions, each accounting for between 11 and 14 percent of occurrences found by IRS examiners. Yet, adjustments for pricing of goods accounted for 49 percent of all the section 482 adjustment dollars proposed. Figure 1: Section 482 Issues With Proposed Income Adjustments and Their Related Dollars, As Percentages of All Proposed Section 482 Issues and Dollars, Fiscal Year 1993 (See figure in printed edition.) Note 1: A particular taxpayer may have multiple occurrences--that is, multiple challenges by IRS--for the same section 482 issue. Note 2: Income/deductions is a general category whose full title is allocation of income and adjustments. Source: GAO analysis of IRS data. IRS' recent experience with section 482 issues in the appeals process--the IRS administrative process that attempts to negotiate disputes--and with the Chief Counsel's office, which is also involved in dispute resolution, was in some ways similar to, but also different from, the experience about which we testified in March 1993.\5 The section 482 issues pending resolution in Appeals or in litigation with Counsel involved about the same dollar value, $14 billion in both June 1994 and September 1992, the dates for which information was available for our current report and for our March 1993 testimony. However, fewer taxpayers were involved at the more recent date--114 as opposed to 180 in the earlier time. Foreign-controlled taxpayers accounted for just under 20 percent of the total both times. The sustention rate for section 482 issues, which is the ratio of income adjustments proposed by the examiners to the final income adjustment after the appeals and legal processes ran their courses, peaked at 52 percent in 1990, then in succeeding years fell back to prior levels of less than 30 percent.\6 While the data IRS maintains on examination and appeals staff time do not have sufficient detail to link them to specific tax issues, we determined that IRS examiners, economists, and appeals staff spent about 186 staff years on cases closed in fiscal year 1993 which contained transfer pricing issues, compared to about 227 staff years on cases closed in fiscal year 1992. The 1993 cases, which involved section 482 issues among other IRS findings, represented 148 international examiner years, 13 economist years, and 25 appeals years. According to IRS officials, the number of years associated with closed cases can fluctuate over time because large cases incur time charges over a period of years and close at different times within a year. In calendar year 1993 and the first part of calendar year 1994, IRS was more successful in litigating section 482 issues than it had been in calendar years 1990 through 1992. While IRS did not prevail in any of the four major court cases with section 482 issues that we analyzed, it lost only one and was a partial winner in the other three. -------------------- \4 A few large adjustments significantly affect comparisons of adjustments for FCCs and USCCs and across years because they comprise large percentages of the totals. \5 GAO/T-GGD-93-16. \6 The great majority of section 482 proposed adjustments that were closed in recent years were closed by Appeals as opposed to Counsel. IRS MADE CIRCUMSCRIBED USE OF AVAILABLE PROCEDURAL TOOLS ------------------------------------------------------------ Letter :4 IRS has a number of procedural tools--that is, discretionary measures, including designated summonses and formal document requests, and additional penalties specifically for section 482--that can be invoked to obtain needed documents, to encourage a recalcitrant taxpayer's cooperation, or to otherwise facilitate examinations. IRS, however, actually used these tools infrequently because, according to IRS officials, they had their desired effect as deterrents inducing a positive change in taxpayer behavior, and recordkeeping requirements were also helping. Arbitration, a tool available only if both parties agree, was used only once for section 482 issues--with a favorable outcome for IRS. Simultaneous examinations of related parties by IRS and foreign tax enforcement agencies promote tax compliance and exchange of documents, but, for a number of reasons, including questions of the timing of examinations, IRS and the foreign governments have only agreed to examine five to eight each year. Advance pricing agreements (APA), which allow the taxpayer to detail and IRS to approve in advance the methodology to be used in setting transfer prices, require an upfront investment of resources to negotiate in order to save IRS examination time in future years. APAs increased from the 9 we cited in our 1993 testimony\7 to 26 completed as of January 1995. IRS expected the number of APAs to grow significantly in the immediate future, which will require additional upfront staff time. -------------------- \7 GAO/T-GGD-93-16. SUCCESS OF NEW REGULATIONS IS UNCLEAR ------------------------------------------------------------ Letter :5 Issued in July 1994, the final transfer pricing regulations are too recent for their impact to be known. The new regulations replace the strict priority of pricing allocation methods contained in the prior regulations with the best method rule, which recognizes that the most reliable measure of an arm's length result will vary depending upon the facts and circumstances of the transaction under review. As a result, taxpayers will have flexibility in selecting and justifying transfer prices, and IRS will also be using considerable judgment in applying the arm's length standard. As it was, taxpayers were already using transfer pricing methods other than the three formerly most well-defined methods for a large proportion of their transfer pricing activity reviewed by IRS, including that in APAs. The new regulations also recognize a range of acceptable transfer prices, providing a taxpayer protection for a reasonable and economically justified pricing methodology. The Omnibus Budget Reconciliation Act of 1993 required contemporaneous documentation of the pricing method used in setting transfer prices and established more widely applicable penalties for transfer pricing abuses than existed earlier. The extent to which the new documentation requirements and IRS' use of judgment in applying the arm's length standard will offset the risks of controversy brought on by taxpayer flexibility remains to be seen. THE MAJORITY OF FCCS AND USCCS PAID NO U.S. INCOME TAX ------------------------------------------------------------ Letter :6 In each of the 5 years studied, FCCs were less likely than USCCs to pay U.S. income tax, as shown in figure 2 and appendix V. In 1991, 73 percent of FCCs paid no U.S. income tax compared with 62 percent of USCCs. The relative flatness of the top two lines in the figure demonstrates how little has changed overall during this 5-year period. Yet, the bottom two lines show that between 1987 and 1991 an increasing percentage of large FCCs and USCCs (those with assets of $100 million or more) did not pay U.S. income tax. Figure 2: Percentage of Corporations That Paid No U.S. Income Taxes, 1987-1991 (See figure in printed edition.) Note: Data presented are based on the number of corporate income tax returns filed with IRS in the indicated year. Large FCCs or USCCs are those with assets of $100 million or more in the year the tax return was filed. Source: GAO analysis of IRS data. NONTAXPAYING FCCS AND USCCS HAD SMALLER SHARES OF ASSETS AND RECEIPTS THAN THEIR TAXPAYING COUNTERPARTS ---------------------------------------------------------- Letter :6.1 The nontaxpaying corporations, both FCCs and USCCs, accounted for the majority of all returns filed in 1991, but for much smaller proportions of total corporate assets and receipts. Although 73 percent of FCCs paid no U.S. income tax in 1991, they accounted for only 37 percent of the assets and 31 percent of the gross receipts of all FCCs that year, as shown in figure 3. This means the 27 percent of FCCs that paid U.S. income tax had 63 percent of FCC assets and 69 percent of receipts. Similarly, 62 percent of USCCs paid no U.S. income tax in 1991, but these nontaxpaying corporations accounted for only 20 percent of the assets and 19 percent of receipts of all USCCs. So, the 38 percent of USCCs that paid U.S. income tax in 1991 held 80 percent of the assets and generated 81 percent of all gross receipts. Figure 3: Nontaxpaying Corporations' Proportions of Corporate Returns, Assets, and Receipts--Fiscal Year 1991 (See figure in printed edition.) Note: Percentages are the proportions of nontaxpaying FCCs' or USCCs' numbers of returns filed, total assets, and gross receipts to the returns, assets, and receipts of all FCCs or USCCs. Source: GAO analysis of IRS data. A FEW LARGE NONTAXPAYING CORPORATIONS ACCOUNTED FOR A LARGE SHARE OF GROSS RECEIPTS ---------------------------------------------------------- Letter :6.2 The largest nontaxpaying FCCs and USCCs were relatively few in number but had disproportionately large shares of total assets and gross receipts relative to all FCCs and all USCCs. In 1991, 715 nontaxpaying FCCs and 3,713 nontaxpaying USCCs had assets of $100 million or more. The large nontaxpaying FCCs accounted for only 1.5 percent of all FCCs that filed U.S. corporate income tax returns in 1991 but 31 percent of all FCCs' assets and 22 percent of their total receipts. Similarly, the largest nontaxpaying USCCs represented one-fifth of 1 percent (0.2 percent) of returns but 16 percent of assets and less than 7 percent of receipts generated by all U.S.-controlled corporations in 1991. Conversely, smaller nontaxpaying FCCs (those with assets under $100 million) accounted for 71 percent of the 48,246 FCCs in 1991 but only 6 percent of FCC assets and 9 percent of FCC receipts. For USCCs, the smaller nontaxpaying corporations accounted for 62 percent of all USCC returns but only 4 percent of assets and 12 percent of the gross receipts. REASONS WHY FCCS PAID LESS OR NO U.S. INCOME TAX ARE UNCLEAR ---------------------------------------------------------- Letter :6.3 As already discussed, large FCCs were more likely than large USCCs to pay no U.S. income tax. Furthermore, large FCCs that did pay income tax, on average, paid less in taxes relative to receipts than their taxpaying U.S. counterparts in 1991. Abusive transfer pricing--that is, inflating prices of intercompany transactions to shift income outside the United States and reduce tax liability--is logically suspect as a possible cause of these observations, yet this suspicion is not easily confirmed by analyzing the tax data. For example, costs of goods sold and purchases accounted for larger proportions of receipts for large nontaxpaying FCCs than for USCCs, as would be expected if FCCs were inflating the price of goods more than USCCs were. Yet, it is not true that interest paid as a percentage of receipts is higher for the FCCs than for the USCCs, as would be expected if FCCs were shifting income by inflating interest paid to a related party. So, while abuse of transfer pricing may be occurring and may be a reason why FCCs are more likely to pay no or less tax than USCCs, other factors, such as differences between FCCs' and USCCs' industries, may be at work. Analysis of industry representation found that large nontaxpaying FCCs were more likely to be in manufacturing and wholesale trade, and less likely to be in finance, insurance, and real estate, than the large nontaxpaying USCCs--differences that may explain some of the tax disparities. OBJECTIVES, SCOPE, AND METHODOLOGY ------------------------------------------------------------ Letter :7 As mentioned earlier, our objectives for this report were to provide information and analysis to update our 1993 work on (1) IRS' recent experience in dealing with transfer pricing issues through its examinations, appeals, and litigation functions; (2) IRS' use of available regulatory and procedural tools; and (3) the extent to which USCCs and FCCs did not pay U.S. income taxes. In some cases, the information we were updating dated back to tax year 1987. To meet the objectives relating to IRS' recent experience and use of its tools, we obtained and analyzed the most recent data available from IRS. To update our analyses of nonpayment of U.S. income taxes by FCCs and USCCs, we obtained and analyzed data on U.S. corporate income tax returns for the 1990 and 1991 tax years, again the most recent IRS data available at the time we did our work. See appendix I for a full discussion of our objectives, scope, and methodology. AGENCY COMMENTS ------------------------------------------------------------ Letter :8 We discussed a draft of this report with responsible Treasury and IRS officials on January 27 and 31, 1995. These officials included members of the Office of the Assistant Secretary of the Treasury for Tax Policy, the Deputy Assistant Commissioner (International), Appeals and other representatives of IRS' Office of Chief Counsel, and an employee of IRS' Statistics of Income Division. While generally agreeing with the report's contents, the officials brought to our attention corrected, updated, and clarifying information. We modified the report where appropriate. IRS officials also raised other points that merit special mention. First, they pointed out various initiatives related to transfer pricing that were beyond the scope of our work. These initiatives included recent technological improvements given to IRS international examiners. Second, an IRS official from the Statistics of Income Division noted some limitations of our statistical analysis of the corporations that did and did not pay taxes. By focusing on corporations that paid no tax, we gave little attention to those that paid minimal taxes. By defining large corporations in terms of asset size rather than receipts, we included, for example, a larger number of finance companies such as banks and fewer trade companies than we would have otherwise. Finally, by comparing all FCCs to all USCCs, we did not consider industry differences. Although these limitations are inherent to some degree in our selected methodology, our analyses provide data on corporations that paid minimal taxes (see table V.4) and on industry differences (see table V.6). Further analyses would provide additional insights into the differences between FCCs and USCCs. ---------------------------------------------------------- Letter :8.1 As agreed with your office, unless you publicly announce its contents earlier, we plan no further distribution of this report until 30 days from the date of this letter. At that time, we will send copies to the Commissioner of the Internal Revenue Service, the Secretary of the Treasury, and other interested parties. The major contributors to this report are listed in appendix VI. If you have any questions, please call me at (202)-512-9044. Natwar M. Gandhi Associate Director, Tax Policy and Administration Issues OBJECTIVES, SCOPE, AND METHODOLOGY =========================================================== Appendix I Our objectives directly related to section 482 were to update our 1993 work by providing information and analyzing (1) how much money was involved in transfer pricing cases, and what issues were at hand; (2) what allocation methods were used or proposed; (3) how many resources IRS used in transfer pricing cases; and (4) what regulatory and procedural tools IRS used in transfer pricing cases. Another objective was to update our analyses of FCCs and USCCs that paid no U.S. income taxes by including 1990 and 1991, the latest years for which tax data were available during our review. In some instances, the information we were updating dated back to tax year 1987. To determine how much money and what issues were involved in transfer pricing cases, we used an examination database maintained by IRS' Assistant Commissioner (International) (AC(I)). At the time we used it, this database contained relevant transfer pricing information extracted from examination reports completed in fiscal year 1993 and the first half of fiscal year 1994. We obtained from IRS' Office of Appeals information for a similar period generated from its case database that generally had large issues and issues that met certain tax deficiency or other criteria. Finally, we researched 1993 and early 1994 court cases involving transfer pricing issues. To determine the allocation methods used or proposed in transfer pricing cases, we summarized information from a database that IRS compiled in early 1993 covering fiscal years 1990 through 1992 to help it share examination findings across the country. We also gathered data on the allocation methods appearing in advance pricing agreements as of mid-1994. To determine how many resources IRS used in transfer pricing cases, we again used the AC(I) examination database. From this database, we determined the amount of time international examiners and economists spent on cases that involved section 482 issues and cases that did not. Similarly, the Office of Appeals provided us with the number of hours spent by Appeals personnel on cases containing section 482 issues as well as other issues.\1 To report on the tools used by IRS personnel in transfer pricing cases, we obtained data covering fiscal year 1993 (and 1994 when available) from AC(I) and interviewed IRS and outside officials. We did not audit any of IRS' management information systems from which we obtained section 482 data. An IRS internal audit report pointed out problems with some of the international management systems, which IRS is improving. For instance, due to a programming error, IRS officials told us they could not be sure that the 1993 and 1994 international examination information they were giving us was comprehensive or completely accurate, but they were continuing to refine it. According to IRS officials, problems with the data included the loss of significant information categorizing audit adjustments by specific issue, such as section 482. The management information we used was the best available at the time we did our work. To determine the percentage of foreign-controlled and U.S.-controlled corporations that paid no tax, we obtained from IRS' Statistics of Income (SOI) Division data for 1990 and 1991. As we did in 1993, we broke these statistics out by returns, receipts, and assets. The statistics were based on tax returns as filed and did not reflect IRS audits or net operating loss carrybacks that would result from any losses in future years. The SOI statistics in this report, other than those for corporations with assets of $100 million or more, are based on SOI's probability sample of taxpayer returns and thus are subject to some imprecision due to sampling variability.\2 In this report, we measure the imprecision with the 95-percent confidence intervals that surround estimates of numbers of taxpayers and the assets and receipts for those taxpayers. For example, our finding that 35,138 foreign-controlled corporations did not pay tax in 1991 is surrounded by a 95-percent confidence interval of plus or minus 3,996. This means that the chances are 19 out of 20 that the interval from 31,142 to 39,134 includes the actual number of such corporations. Table I.1 shows this and other confidence intervals for 1991. We worked mainly with IRS' National Office in Washington, D.C. We did our review from April through December 1994 in accordance with generally accepted government auditing standards. Table I.1 Confidence Intervals for Sampled Data, 1991 (Dollars in millions) Lower bound Estimate Upper bound --------------- ------------- ------------- ------------- FCCs not paying tax ------------------------------------------------------------ Number 31,142 35,138 39,134 Assets $675,460 $679,261 $683,062 Receipts $352,812 $358,802 $364,792 USCCs not paying tax ------------------------------------------------------------ Number 1,246,529 1,265,272 1,284,016 Assets $3,249,333 $3,256,986 $3,264,639 Receipts $1,546,707 $1,569,477 $1,592,247 ------------------------------------------------------------ Source: GAO analysis of IRS data. -------------------- \1 Neither AC(I) nor Appeals had a system for capturing staff time used on individual Internal Revenue Code sections. \2 SOI information for the largest corporations is not based on samples and thus is not subject to sampling error. IRS' RECENT EXPERIENCE WITH TRANSFER PRICING IN EXAMINATIONS, APPEALS, AND LITIGATION ========================================================== Appendix II In fiscal years 1993 and 1994, IRS examiners continued to propose substantial section 482 adjustments to income. However, IRS appeals officers, who are charged with resolving tax controversies without litigation, continued in 1993 and 1994 to substantially reduce adjustments proposed by examiners in earlier years. Further, IRS had mixed success in litigating important section 482 cases in the courts, although this was an improvement over its record in earlier years. IRS' EXAMINATION OF SECTION 482 ISSUES -------------------------------------------------------- Appendix II:1 As shown in table II.1, in fiscal years 1993 and 1994, for cases with total proposed adjustments of $20 million or more, IRS proposed section 482 income adjustments for 51 and 64 taxpayers, respectively, of $1.8 billion and $3.5 billion. The number of proposed adjustments was similar to or larger than the numbers in previous years, although the associated dollar amount for 1993 was significantly lower than in several of those years. The dollar amount can fluctuate, however, on the basis of a few large adjustments. Also, proposed adjustments to income may or may not result in increased tax collections, depending on such things as whether a company has offsetting adjustments, offsetting corporate net operating losses carried over from other years, or success in challenging the proposed adjustment. Table II.1 Proposed Section 482 Income Adjustments of Foreign-and U.S.-Controlled Corporations With $20 Million or More of Total Proposed Adjustments (Dollars in billions) Number of Adjustment Number of Adjustment Number of Adjustment Fiscal year corporations amount corporations amount corporations amount ---------------- ------------ ------------ ------------ ------------ ------------ ------------ 1989 12 $0.7 31 $4.1 43 $4.8 1990 11 1.6 26 4.4 37 6.0 1991 12 1.1 23 1.2 35 2.3 1992 13 1.0 37 3.1 50 4.1 1993 18 0.7 33 1.1 51 1.8 1994 31 2.0 33 1.5 64 3.5 ---------------------------------------------------------------------------------------------------- Note 1: A few large adjustments significantly affect comparisons of adjustments for foreign- and U.S.-controlled corporations because they comprise large percentages of the totals. Note 2: We generally used IRS' determinations of whether particular corporations were foreign controlled, but if we were aware that an IRS determination was incorrect, we used our own. Sources: Data for 1989 through 1992 and for 1994 were accumulated by IRS, and data for 1993 were developed by us working with IRS and using information obtained from IRS' International Case Management System. While the cases with proposed adjustments of $20 million or more accounted for the bulk of all the dollars in proposed section 482 adjustments, they did not comprise the majority of all proposed adjustments. For instance, as shown in table II.2, for fiscal year 1993, IRS proposed $2.2 billion in section 482 adjustments to income for 247 FCCs and 122 USCCs.\1 Only about 12 percent of the section 482 issues for any of these taxpayers were for tax years after 1990, the first recent year in which transfer pricing received intensive congressional scrutiny. Tax years ranged from 1975 through 1992. Because we received the 1994 information at the end of our review, we could not similarly analyze it. According to IRS, both 1993 and 1994 adjustments for transfer pricing might be understated due to data that were lost in implementing a new management information system. Table II.2 Total Proposed Section 482 Adjustments for Foreign-and U.S.-Controlled Corporations (Dollars in billions) Number of Adjustment Number of Adjustment Number of Adjustment Fiscal year corporations amount corporations amount corporations amount ---------------- ------------ ------------ ------------ ------------ ------------ ------------ 1993 247 $0.9 122 $1.3 369 $2.2 1994 236 2.0 156 1.5 392 3.5 ---------------------------------------------------------------------------------------------------- Note: Numbers do not always add due to rounding. Source: IRS data on international examinations completed in fiscal years 1993 and 1994. GAO compiled the numbers for 1993, and IRS supplied those for 1994. The nature of the section 482 issues IRS disputed varied from company to company. As shown in table II.3, several different sorts of issues accounted for at least 7 percent each of the section 482 issues IRS raised in fiscal year 1993. These relatively frequently occurring issues were interest, royalties, intercompany pricing of tangible goods, general allocation of income and deductions, and service charges and fees, which accounted for 55 percent of all section 482 challenges by IRS and 71 percent of section 482 dollars involved. Table II.3 Most Frequently Occurring Proposed Section 482 Income Adjustments for Fiscal Year 1993 (Dollars in millions) Percent of all Percent of section all 482 Section section Number of proposed 482 482 occurrence adjustment dollars dollars Issue description s \a s involved involved -------------------------------- ---------- ---------- ---------- ---------- Interest 249 14% $ 32 1% Royalties 216 12 99 5 Pricing of tangible goods 195 11 1,083 49 Allocation of income and 193 11 242 11 deductions (general category) Service charges and fees 120 7 115 5 ================================================================================ Total 973 55% $1,570 71% -------------------------------------------------------------------------------- \a A particular taxpayer may have multiple occurrences, that is, multiple challenges by IRS for the same section 482 issue. Source: IRS data. As shown in figure II.1, for cases closed in fiscal year 1993 or the first half of fiscal year 1994, IRS spent about a third of its total international examiner time, and a much higher percentage of its economist time, on those cases that had a section 482 issue, among other issues.\2 We could not break out the amount of time spent only on section 482 because AC(I) did not have a system for capturing staff time on individual Internal Revenue Code sections. However, using a formula provided by IRS, we estimated that for cases closed in fiscal year 1993, IRS spent about 148 international examiner staff years and about 13 economist staff years on examinations that included a section 482 adjustment. The corresponding numbers for fiscal year 1992 were 164 and 19.\3 According to IRS officials, the time spent on cases closed is expected to fluctuate from year to year because large cases may close that incurred time charges over several years. In addition, most of the time reflected on cases that close early in a fiscal year will likely have been incurred on the cases in previous years. Figure II.1: Percentage of International Examiner and Economist Staff Time Spent on Taxpayers With Section 482 Issues (See figure in printed edition.) Source: GAO analysis of IRS data. -------------------- \1 As we reported in Tax Administration: Information on IRS' International Tax Compliance Activities (GAO/GGD-94-96FS, June 27, 1994), IRS' closed international examinations of FCC tax returns increased 353 percent from fiscal year 1990 to 1993, while examinations of other tax returns declined 31 percent. In its examination of corporations, IRS included in its category of FCCs foreign corporations operating through branches in the United States. \2 According to IRS officials in January 1995, the equivalent percentages for all of fiscal year 1994 were 39 percent of total international examiner time and 75 percent of economist time. \3 In January 1995, IRS indicated that its computations for fiscal year 1994 showed 190 international examiner staff years and 21 economist staff years spent on cases involving section 482 issues. APPEALS CONTINUED TO SUBSTANTIALLY REDUCE EXAMINERS' PROPOSED INCOME ADJUSTMENTS -------------------------------------------------------- Appendix II:2 When adjustments to income are proposed in examinations, the taxpayer can take the dispute to Appeals, which is the administrative body within IRS authorized to settle tax controversies. Also, when litigation is involved, so is IRS' Chief Counsel. In March 1993,\4 we testified that IRS' experience with section 482 issues in Appeals had not improved from the experience we described in our 1992 report.\5 Since that time, while IRS examiners again identified billions of dollars in proposed adjustments to income for section 482 issues, Appeals again substantially reduced previous years' proposals.\6 -------------------- \4 GAO/T-GGD-93-16. \5 International Taxation: Problems Persist in Determining Tax Effects of Intercompany Prices (GAO/GGD-92-89, June 15, 1992). \6 IRS staff do not track every issue related to section 482 but believe they capture the large ones by generally focusing on large issues and issues that meet certain tax deficiency or other criteria. OPEN ISSUES\7 ------------------------------------------------------ Appendix II:2.1 As table II.4 shows, as of June 30, 1994, IRS had 114 large taxpayers with total section 482 proposed adjustments to income of $14.4 billion pending administrative resolution in Appeals or litigation with Counsel as opposed to the 180 taxpayers with proposed section 482 adjustments of $14.4 billion as of September 30, 1992.\8 Thus, although the dollar value of open proposed adjustments stayed the same, the number of taxpayers involved declined.\9 As of June 30, 1994, IRS Appeals officials had spent almost 96,000 hours on the open work units that contained section 482 issues as well as other issues. None of the tax returns involved were for tax periods after 1990. Of the $14.4 billion in proposed adjustments as of June 30, 1994, $1.8 billion was for foreign- controlled issues.\10 Table II.4 Open Section 482 Issues (Dollars in billions) Da Number of Dollar Number of Dollar Number of Dollar te taxpayers amount taxpayers amount taxpayers amount -- ----------- ----------- ----------- ----------- ----------- ----------- Se 34 $2.4 146 $12.0 180 $14.4 p t e m b e r 3 0 , 1 9 9 2 Ju 19 1.8 95 12.6 114 14.4 ne 3 0 , 1 9 9 4 -------------------------------------------------------------------------------- Source: GAO analysis of IRS data. -------------------- \7 "Open" issues are issues referred beyond examination but not yet settled. \8 We compared information for September 30, 1992, and June 30, 1994, because those were the two dates for which data were available for our March 1993 testimony (GAO/T-GGD-93-16) and our current report. \9 IRS officials attributed the decline to more examinations being resolved at the examiner level without being appealed. \10 A small part of the information we received was for taxpayers, such as individuals, that were not corporations. Therefore, in discussing this information, we use the terms "foreign-controlled issues" and "U.S.-controlled issues" rather than FCCs and USCCs. CLOSED ISSUES\11 ------------------------------------------------------ Appendix II:2.2 In March 1993, we testified that from 1987 through 1989, 29 percent of section 482 proposed adjustments to income were sustained by IRS.\12 As shown in table II.5, this number rose to 52 percent in 1990 but declined in the years afterwards to the 21 to 28 percent range. A very large part of the section 482 proposed adjustments were closed by Appeals as opposed to Counsel. Sustention rates for foreign-controlled issues fluctuated more than did those for U.S.-controlled issues. Table II.5 Sustention Rates for Large Section 482 Issues (Dollars in billions) Proposed Sustention Proposed Sustention Proposed Sustention Fiscal year adjustments rate adjustments rate adjustments rate ---------------- ------------ ------------ ------------ ------------ ------------ ------------ 1987 $0.1 12% $0.5 25% $0.6 24% 1988 0.5 34 1.5 25 2.0 27 1989 0.2 9 1.0 43 1.2 37 1990 NA NA NA NA 1.8 52 1991 0.2 24 2.0 28 2.2 28 1992 0.4 5 1.5 30 1.9 24 1993 0.4 20 0.5 32 0.9 27 1994\a NA NA NA NA 1.0 21 ---------------------------------------------------------------------------------------------------- \a We received the 1994 information late in our review and did not have a chance to ascertain the breakout between foreign- and U.S.-controlled issues. Note: The 1991 sustention rate for foreign-controlled issues (24 percent) differs from the 23 percent that we reported in our 1993 testimony because we had more precise information available the second time. Also, the 1992 percentages are estimates derived from less precise information than was available for the other years. Source: GAO analysis of IRS data. In fiscal year 1993, the sustention rate was 27 percent, or $249 million of the $926 million of proposed adjustments to income. This percentage was very close to the 29-percent fiscal year sustention rate for all international recommended tax, credit, and penalty adjustments for large dollar issues that we reported in June 1994.\13 During fiscal year 1993, Appeals and Counsel closed 82 transfer pricing issues (involving 51 taxpayers), and the vast majority of those applied to tax years before 1988. During fiscal year 1994, Appeals and Counsel closed 61 transfer pricing issues and sustained about $208 million of the slightly less than $1 billion proposed by IRS for a sustention rate of 21 percent. Using a formula provided by IRS, we estimated that Appeals personnel spent about 25 staff years on work units that included at least one section 482 issue as well as possibly other issues that were closed during fiscal year 1993. Of these years, 7 percent, or about 2 years, were spent on foreign-controlled issues, and 93 percent, or about 23 years, on U.S.-controlled issues. For cases closed during fiscal year 1992, Appeals personnel spent about 44 staff years. -------------------- \11 "Closed" issues are issues where settlement has been reached. \12 The sustention rate is the ratio of the adjustment amount determined after an issue has been resolved by Appeals, Chief Counsel, or a court to the adjustment amount proposed by examiners. IRS highly qualifies sustention rates because the data collected before 1992 were not gathered to measure sustention rates and/or were not subjected to rigorous accuracy checks. Also, IRS cautions that sustention rates for individual issues like section 482 ignore such distortionary effects as those caused by negative proposed adjustments. Finally, according to IRS officials, the fluctuation in sustention rates from year to year might be due to the resolution of a few cases with large dollar implications. \13 GAO/GGD-94-96FS. REASONS FOR REACHING A SETTLEMENT ------------------------------------------------------ Appendix II:2.3 In recent years, IRS began tracking the reasons why it reached a particular settlement in particular cases. As table II.6 shows, the most prevalent reason why IRS reached a settlement of transfer pricing issues in cases closed during fiscal year 1993 was concern about whether a court would apply the same judgment to the evidence as IRS had. This reason appeared in 29 of the 82 issues closed, accounting for 63 percent, or $426 million, of the $677 million reduction in the original proposed adjustments to income. Table II.6 Reasons for Fiscal Year 1993 Reductions to Proposed Section 482 Adjustments (Dollars in millions) Percent of Number of Amount overall Reason for reductions issues reduced reduction ------------------------ ---------- ---------- ---------- Hazards of litigation 29 $426 63% relating to facts or evidence being open to judgment Hazards of litigation 14 143 21 relating to uncertainty about how the court will apply the law New facts or evidence 14 98 14 was obtained and evaluated by Appeals or Counsel All reasons 82 $677 100% ------------------------------------------------------------ Note: The reasons here are not mutually exclusive; more than one reason may be assigned to the same issue. Also, the closed case information that IRS gave us did not always have IRS reason codes attached. Source: IRS data. Two other reasons appeared in 14 issues each. These were (1) uncertainty about how the court would interpret or apply the law in a particular case and (2) the need to deal with new facts or evidence obtained and evaluated by Appeals or Counsel. IRS' EXPERIENCE IN THE COURTS WITH TRANSFER PRICING ISSUES -------------------------------------------------------- Appendix II:3 As shown in table II.7, IRS' recent experience in the courts with section 482 has been mixed. However, this is an improvement over what we reported in our 1993 testimony--that is, that IRS lost a significant section 482 issue for each of the five cases we examined. Table II.7 Summary of Major Court Cases With Section 482 Issues Decided Between January 1, 1993, and May 20, 1994 Date of Date Amount returns of at Name at issue ruling issue Winner and ruling ------------ -------- ------ ------ -------------------- National 1978- 1994 $122 Mixed--The Tax Court Semiconducto 1982 millio increased the r n in taxpayer's income Corporation income under section 482 by \a $40.6 million to bring the pricing relationships closer to what would have occurred at arm's length. According to the Court, neither party presented the Court with evidence that would satisfy any of the prescribed methods of transfer pricing under section 482 regulations. The Court, however, adopted the IRS expert's analysis with certain modifications as the "least unacceptable methodology" presented. Seagate 1983- 1994 $285 Mixed--The Tax Court Technology, 1987 millio held that IRS' Inc. n in reallocation of income income under section \b 482 was arbitrary and capricious but that the manufacturer failed to prove that the transactions were arm's length. The Court determined the appropriate transfer prices on the basis of its own "best estimate." Perkin- 1975- 1993 $22 Mixed--The Tax Court Elmer 1981 millio noted that in this Corporation n in case, as in other income significant section \c 482 cases, each party spent most of the time attacking the other party's allocation formula rather than establishing the soundness of its own formula. Thus, the Court was left to find a formula without the benefit of sufficient help from the parties. Income allocations to the taxpayer were less than called for by IRS. Exxon 1979- 1993 Over Taxpayer--The Tax Corporation/ 1981 $6.5 Court rejected IRS' Texaco, billio income reallocation, Inc./Aramco n in stating that Advantage taxes\ restrictions from d the Saudi Arabian government made a higher transfer price legally unfeasible. ------------------------------------------------------------ Note: We selected the four cases by reviewing cases decided between January 1, 1993, and May 20, 1994, that involved what we considered to be a major section 482 reallocation issue. IRS officials agreed that we had selected the major cases. \a After an expert's recommendation was taken into account, the maximum total proposed adjustment at trial was less than $77 million. \b As reflected in expert reports introduced at trial, the total proposed adjustment became $171 million. \c IRS' alternative trial position showed income at issue to be $29 million. \d According to commentators, the Aramco Advantage case represents the biggest dollar deficiency in Tax Court history. Sources: Court cases. In this report, we identified four major section 482 cases that had been decided in court between January 1, 1993, and May 20, 1994. These cases took an average of 15 years from the earliest tax year audited until resolution in the courts. As we reported in our 1993 testimony,\14 cases like these illustrate how disputes over section 482 issues can become extremely expensive for taxpayers and the government by requiring the employment of outside experts, resulting in long, drawn out litigation and keeping corporate tax liabilities in an uncertain status for years. In three of the cases, the Tax Court expressed its displeasure with IRS and found its reallocation of income to be arbitrary, capricious, or unreasonable. However, an IRS official noted that under its standard of review, the Tax Court must find IRS "arbitrary and capricious" to differ with adjustments proposed by IRS and then try to find a middle ground between the company and IRS. Thus, even though IRS was found arbitrary, capricious, and unreasonable in the National Semiconductor case, the Tax Court still increased the taxpayer's income by $40.6 million. IRS also noted that in the one case in which the taxpayer was the complete winner--the Aramco Advantage case--the Tax Court limited IRS' authority to propose adjustments under section 482. -------------------- \14 GAO/T-GGD-93-16. IRS' USE OF CERTAIN PROCEDURAL TOOLS ========================================================= Appendix III IRS used certain new and existing procedural tools--designated summonses, formal document requests, additional penalties, simultaneous examinations, and arbitration--sparingly. However, advance pricing agreements (APA) were increasingly used. Gauging the ultimate impact of procedural tools is difficult because, while their direct use is limited, their real effectiveness may be as deterrents that alter behavior and as remedies to uncertainty. Also, while IRS can use some of the tools unilaterally, it must have a partner for others. DESIGNATED SUMMONSES AND FORMAL DOCUMENT REQUESTS ------------------------------------------------------- Appendix III:1 As shown in table III.1, very few designated summonses for transfer pricing cases have been used since they were sanctioned by Congress in 1990. Designated summonses are summonses issued by IRS that could result in suspending the running of the statute of limitations governing the time IRS has for assessing additional taxes against a taxpayer. According to IRS officials, increased taxpayer cooperation has made the use of designated summonses unnecessary, and just the threat that the designated summonses could be imposed is enough to stir taxpayers to action. Table III.1 Number of Designated Summonses and Formal Document Requests Designated Formal document Fiscal year summonses requests -------------------- ------------------ ------------------ 1991 2 28 1992 1 13 1993 0 3 1994 (first half) 0 1 ------------------------------------------------------------ Note: IRS believed that the information for 1993 and 1994 might not have been complete, but that complete information would not have been significantly different. Also, some of the activity listed here may not relate to transfer pricing issues. Source: IRS. A similar reason was given for the drop in formal document requests issued by IRS from 28 in 1991 to 1 in the first half of fiscal year 1994, also included in table III.1. Formal document requests are issued when relevant taxpayer documents are outside the United States. According to one IRS official, the need for formal document requests has been overtaken by the success of section 6038A of the Internal Revenue Code, which outlines recordkeeping requirements for certain foreign-owned corporations. IRS provided us many examples of taxpayers who became more compliant in the face of section 6038A. ADVANCE PRICING AGREEMENTS ------------------------------------------------------- Appendix III:2 Since our 1993 testimony,\1 IRS has continued to make progress with its APAs, which are agreements under a program begun in fiscal year 1991 in which IRS approves ahead of time the methodology a taxpayer volunteering for the program will use in setting transfer prices. According to IRS' APA program director, as of January 1995, 26 APAs were complete, up from the 9 APAs we cited in our 1993 testimony. The director also anticipated substantial additional growth in APAs and growth in APA staff resources in the immediate future as the APA office was entertaining about 100 active matters. He expected the extra resources to result in reduced audit demands and enhanced voluntary compliance. When we asked, IRS did not have information on the number of staff hours it had devoted to APAs. -------------------- \1 GAO/T-GGD-93-16. ADDITIONAL PENALTIES ------------------------------------------------------- Appendix III:3 Section 6662 of the Internal Revenue Code, as modified in 1990 and 1993, imposes substantial penalties on tax underpayments attributable to certain transfer prices that were substantially misstated. According to an IRS official, only one penalty has been proposed since new provisions took effect in 1990 because substantiating penalties was originally difficult on account of the broad exceptions allowed. The difficulty was eased, according to the official, starting in April 1993, when tougher documentation standards went into effect. To head off concerns that penalties might be applied inconsistently or unfairly, future penalties must be reviewed by an IRS Penalty Oversight Committee. Since the latest version of the penalties was enacted for tax years beginning after December 31, 1993, the full force and impact of the penalty cannot be measured at this time. However, IRS officials believe that taxpayers are acting differently than they did before because of the existence of the penalties. SIMULTANEOUS EXAMINATIONS ------------------------------------------------------- Appendix III:4 In simultaneous examinations, the United States and another country examine related parties under their jurisdictions at the same time in an effort to promote international tax compliance and information exchange. As shown in figure III.1, the number of simultaneous examinations proposed in fiscal years 1993 and 1994--18--was substantially lower than the 33 proposed in 1991 and 1992, which in turn was a much higher number than were proposed in previous years. However, from 1991 through 1994, only a moderate number of the proposed simultaneous examinations had been accepted for follow through at the time of our review, although a few were still pending. According to an IRS official, the relatively high number of proposals in 1991 and 1992 was due to a misunderstanding between IRS and foreign countries. Figure III.1: Simultaneous Examinations Proposed and Accepted During Fiscal Years 1991 Through 1994 (See figure in printed edition.) Source: IRS. As we testified in 1993, for years, at least since our 1981 report on transfer pricing, IRS has emphasized that its simultaneous examination program is important in protecting U.S. interests in international tax enforcement.\2 According to an IRS official, however, questions of the timing of U.S. and foreign examinations, examination proposal format, and resources have presented obstacles to doing more simultaneous examinations. However, IRS has developed model procedures to encourage the carrying out of more of these examinations. -------------------- \2 IRS Could Better Protect U.S. Tax Interests in Determining the Income of U.S. Multinational Corporations (GAO/GGD-81-81, Sept. 30, 1981). ARBITRATION ------------------------------------------------------- Appendix III:5 Only one section 482 case has been resolved through arbitration, even though both IRS and the taxpayer considered that arbitration a success. Under Tax Court Rule 124, any time a factual case is at issue and before trial, the parties to the case may move to resolve it through voluntary binding arbitration. In the one case, the arbitration panel ruled in favor of IRS after the agency substantially modified its position after examination. According to the taxpayer's attorneys, arbitration reduced the taxpayer's expenses, improved its settlement opportunities, and produced a binding decision. Although recommendations have been made to improve the arbitration process, a second case has not been forthcoming. Before arbitration can occur in a specific instance, both IRS and the taxpayer have to voluntarily agree to it, which is unlikely if either party is unsure of the strength or desirability of its arbitration position. However, IRS officials indicated their interest in pursuing arbitrations and other means of alternate dispute resolution. IRS was beginning a program that would allow certain cases to be subject to mediation. CHALLENGES OF SECTION 482 WILL CONTINUE ========================================================== Appendix IV Although IRS has issued new regulations on section 482, how successful they will ultimately be in resolving transfer pricing issues is unclear. Subjectivity will still be involved, and thus controversy may still arise. The flexibility that the regulations allow taxpayers for tax planning must be weighed against the flexibility allowed IRS and the more stringent documentation and penalty provisions that have recently been enacted. PROBLEMS THAT HAVE EXISTED WITH ARM'S LENGTH PRICING -------------------------------------------------------- Appendix IV:1 As we testified in 1993, the arm's length standard required that the price charged on a transaction between related corporations be the price that would have been charged if the corporations had been unrelated. To enforce this standard, IRS had to analyze comparable transactions between unrelated corporations to identify the arm's length price that the related corporations had to charge. If IRS found a difference between an arm's length price and the price that the related corporations charged, it could propose an adjustment to the related corporations' income.\1 IRS and taxpayers have used direct and indirect methods for identifying arm's length prices. The comparable uncontrolled price was based on a direct comparison of the prices charged on readily identifiable, comparable transactions between unrelated parties. More indirect methods, such as the resale price, cost plus, and other appropriate methods, based prices on comparisons with unrelated corporations performing similar functions. These methods may require IRS examiners to use considerable judgment and to develop and analyze a great deal of data. A major obstacle in enforcing the arm's length standard has been the difficulty that IRS examiners have had in finding readily identifiable, comparable transactions. The data requirements and the subjective nature of the pricing methods imposed a significant administrative burden on both corporate taxpayers and IRS, and also led to uncertainties for corporations about their ultimate tax liabilities. As shown in figure IV.1, indirect methods have been used for most section 482 cases. For transfer pricing issues completed in fiscal years 1990 through 1992, IRS international examiners reported that the three then most well-defined methods--comparable uncontrolled price, cost plus, and resale price--had been used only about half the time, with other methodologies being used the other half. Similarly, these three methods accounted for only 38 percent of the 75 methods used in APAs that had been started and/or finished without being withdrawn as of July 7, 1994, and that had decided-upon methodologies. Another 36 percent used different profit measures or formulary or allocation apportionment,\2 with 27 percent using miscellaneous other allocation methods. IRS officials indicated that formulary apportionment has been used only in difficult cases and only after obtaining the agreement of affected treaty partners. Because the APA information is more recent than the examination information we have, it is more likely to show recent trends in methodologies. Figure IV.1: Allocation Methods Reported by IRS Examiners and Those Used in Advance Pricing Agreements (See figure in printed edition.) Note 1: Percentages do not add to 100 due to rounding. Note 2: Eighteen of the 430 transfer pricing issues reported by IRS examiners arose in APAs. Source: GAO analysis of IRS data. The difficulty of finding comparables is further illustrated by concerns expressed by the IRS Commissioner's Advisory Group Task Force on Third Party Transfer Price Information. These concerns centered on the confidentiality of price and cost information and the reliability of aggregated data on comparables. The problem of access to information on comparable transactions, and the difficulty of finding such transactions for many intangible properties, make transfer pricing difficult for both taxpayers and IRS. -------------------- \1 The new July 1, 1994, section 482 regulations that will be described later modified the interpretation of the arm's length standard but retained the idea of comparability. The standard is met if the results of a transaction are consistent with the results that would have been realized if uncontrolled taxpayers had engaged in the same transaction under the same circumstances. As a practical matter, whether a transaction produces an arm's length result will be determined by reference to results of comparable transactions under comparable circumstances. \2 Treasury officials told us that they would be surprised if IRS was using formulary apportionment in the sense of applying the same formula to different companies. IMPACT OF NEW SECTION 482 REGULATIONS IS UNCERTAIN -------------------------------------------------------- Appendix IV:2 On July 1, 1994, the Department of the Treasury issued new final regulations on intercompany transfer pricing, replacing temporary and proposed regulations issued on January 21, 1993. According to Treasury, the new regulations emphasize comparability and flexibility. The "best method rule" under the regulations provides that the pricing method the taxpayer chooses must be the one that gives the most reliable measure of the arm's length result under the facts and circumstances. Thus, both taxpayers and IRS will have to use considerable judgment in applying the arm's length standard. To help taxpayers and IRS exercise this judgment, the new regulations have a greatly expanded discussion of the factors to consider in applying the best method rule. As we said in our 1993 testimony,\3 because the best method approach is still based on the facts and circumstances of each case, the task of selecting and justifying transfer prices will remain complex and open to interpretation. Some commentators on the new regulations have echoed this, saying that there is room for aggressive tax maneuvering by taxpayers and warning that examination and litigation controversy between taxpayers and IRS will increase and certainty decrease. According to Treasury officials, however, the general opinion is that the regulations are relatively workable. Further complicating the task of dealing with the regulations are the data problems described earlier. According to the Commissioner's Advisory Group Task Force, the data needed for section 482 regulations are extensive, but the ability to create a single database is inhibited. According to an IRS official, this lack of good data on comparables led IRS to develop profit-oriented methods that do not rely so heavily on information on comparable transactions. Still, the new regulations prefer the use of direct evidence of arm's length prices over profit-related methods. The lack of good data also explains IRS' practice of asking individual companies other than the one being audited to voluntarily submit information on comparables. Without this information, IRS believes that it runs the risk of inappropriately raising or abandoning section 482 issues, settling for far less than the arm's length amount, or failing to sustain litigation. The increased documentation required by new penalty legislation\4 may ease the regulatory risks and IRS' enforcement burden, despite the continuing difficulty in identifying comparables. The requirement for contemporaneous documented support provides IRS with information about the methods used by taxpayers when setting their transfer prices. According to IRS officials, this information provided by taxpayers should enable examiners to determine whether transfer prices are appropriate without, in each case, having to develop alternative methods and data. These IRS officials believed that the documentation requirements combined with the substantial misstatement penalties also provided in the legislation will reduce transfer pricing abuse. The documentation requirements may add to the compliance burden of taxpayers to the extent that transactions will be documented whether or not they are at issue with IRS. However, contemporaneous documentation will be useful to taxpayers in justifying their prices and in avoiding the penalties. Furthermore, taxpayers should benefit from the new section 482 regulations' additional guidance in determining comparable transactions and from the recognition that the arm's length price may belong to a range of prices. Taxpayers with prices within the arm's length range will be protected to some degree from small changes in transfer prices by IRS that result in large increases in tax liabilities. -------------------- \3 GAO/T-GGD-93-16. \4 The Omnibus Budget Reconciliation Act of 1993. A HIGH PERCENTAGE OF FCCS AND USCCS DID NOT PAY U.S. INCOME TAX FROM 1987 THROUGH 1991 =========================================================== Appendix V A higher percentage of FCCs than USCCs paid no U.S. income tax in each year from 1987 to 1991. Larger corporations (both foreign- and U.S.-controlled) were more likely than smaller corporations to pay U.S. income tax. Overall, nontaxpaying corporations accounted for a smaller proportion of total assets and receipts in 1991 than taxpaying corporations did, which means that the relatively smaller proportion of taxpaying corporations generated the majority of receipts. But the very small number of large nontaxpaying corporations accounted for a disproportionately large share of total corporate assets and receipts. Finally, large taxpaying FCCs, on average, paid less U.S. income tax relative to receipts than large taxpaying USCCs. Analyzing foreign- and U.S.-controlled corporations' costs of goods sold, purchases, and other tax data as a percentage of their gross receipts indicated differences but provided no clear evidence of transfer pricing abuses. Comparing the types of industries represented by the largest foreign- and U.S.-controlled corporations may explain some of the differences between them. A HIGHER PERCENTAGE OF FOREIGN-CONTROLLED CORPORATIONS THAN U.S.-CONTROLLED CORPORATIONS PAID NO U.S. INCOME TAX --------------------------------------------------------- Appendix V:1 FCCs were less likely than U.S.-controlled corporations to pay U.S. income tax. In 1991, 73 percent of FCCs versus 62 percent of USCCs paid no U.S. income tax. The trend over 5 years, as shown in table V.1, was relatively constant for FCCs. The change in the percentage of FCCs that did not pay U.S. income tax--from 71 to 73 percent--is small relative to the sampling error in the 1991 estimate of about plus or minus 4 percent.\1 Although the USCC data are also based on statistical samples, the differences for the USCCs that did not pay income tax--from 57 percent in 1987 to 62 percent in 1991--are statistically significant.\2 Table V.1 Number and Percentage of Foreign-and U.S.-Controlled Corporations That Paid No Income Taxes, 1987 Through 1991 Number of Number of Year returns Percent returns Percent -------- ----------- ----------- ----------- ----------- 1987 29,928 71 1,333,470 57 1988 33,636 73 1,310,698 58 1989 32,135 72 1,265,970 59 1990 32,343 73 1,265,462 61 1991 35,138 73 1,265,272 62 ------------------------------------------------------------ Source: GAO analysis of IRS data. The increase from 57 to 62 percent in the USCCs that did not pay income tax from 1987 through 1991 should be interpreted with care. The absolute number of USCCs that did not pay income tax in 1991 was the lowest of the 5 years--1,265,272--as opposed to 1,333,470 in 1987. The explanation can be found in the changing number of USCCs--that is, the number of USCCs overall is decreasing faster than the decrease in nontaxpaying USCCs. -------------------- \1 The estimates based on IRS' statistical samples that in 1989 72 percent and in 1991 73 percent of FCCs did not pay U.S. income tax are plus or minus 4 percent at a 95-percent confidence level. As a result, the 1- and 2-percent changes during these years were small relative to the precision of the estimate for any given year. \2 The confidence interval for the USCC estimates of 59 percent in 1989, 61 percent in 1990, and 62 percent in 1991 is plus or minus 0.9 percent at a 95-percent confidence level. Therefore, the differences of two or more percentage points between years is statistically significant. LARGE FCCS AND USCCS WERE MORE LIKELY TO PAY U.S. INCOME TAX --------------------------------------------------------- Appendix V:2 As table V.2 shows, large FCCs and USCCs were more likely than their smaller corporate counterparts to pay U.S. income tax. In each year from 1987 through 1991, a higher percentage of FCCs and USCCs with assets of $100 million or more paid income tax than did FCCs and USCCs with assets of less than $100 million. Table V.2 Corporations That Paid No Income Taxes, by Ownership and Size, 1987 Through 1991 Number of Percent not Number of Percent not Year returns paying tax returns paying tax -------- ----------- ----------- ----------- ----------- 1987 29,632 71.5 1,330,988 56.9 1988 33,286 73.6 1,307,881 58.4 1989 31,690 72.8 1,262,801 59.0 1990 31,800 74.4 1,261,999 60.8 1991 34,423 73.7 1,261,559 62.0 Assets greater than or equal to $100 million ------------------------------------------------------------ 1987 297 36.0 2,483 29.9 1988 350 31.8 2,817 30.8 1989 445 34.3 3,169 33.7 1990 543 39.1 3,463 35.8 1991 715 46.3 3,713 37.2 ------------------------------------------------------------ Source: GAO analysis of IRS data. While the largest companies were more likely than smaller companies to pay U.S. income tax, an increasing number of the largest companies, both foreign- and U.S.-controlled, paid no tax in these 5 years. The number of large foreign-controlled corporations that did not pay U.S. income tax more than doubled in this period, from 297 in 1987 to 715 in 1991. During the same period, the number of large U.S.-controlled corporations that did not pay U.S. income tax also increased, although not as dramatically, from 2,483 in 1987 to 3,713 in 1991. This trend among the largest corporations is not based on samples and therefore is not subject to sampling error. NONTAXPAYING CORPORATIONS ACCOUNTED FOR A SMALLER SHARE THAN TAXPAYING CORPORATIONS OF TOTAL ASSETS AND RECEIPTS --------------------------------------------------------- Appendix V:3 Nontaxpaying corporations, both foreign- and U.S.-controlled, accounted for the majority of all returns filed, but for much smaller proportions of total assets and receipts. This indicates that many of the nontaxpaying corporations, both foreign- and U.S.-controlled, were smaller in terms of assets and generated fewer receipts than their taxpaying counterparts. While 73 percent of FCCs did not pay U.S. income tax in 1991, these 35,138 nontaxpaying corporations accounted for only 37 percent of the assets and 31 percent of the gross receipts of all FCCs that year, as shown in table V.3. This means the 13,108 taxpaying FCCs accounted for only 27 percent of the returns but 69 percent of receipts. Table V.3 Returns, Assets, and Receipts of Corporations That Did Not Pay Income Taxes, 1991 (Dollars in billions) Di d no t pa y in co me ta x Number Percent Amount Percent Amount Percent -- ----------- ----------- ----------- ----------- ----------- ----------- Al 35,138 72.8 $ 679.3 37.2 $ 358.8 31.4 l F C C s La 715 1.5 568.8 31.1 255.0 22.3 r g e F C C s Al 1,265,272 61.9 3,257.0 20.1 1,569.5 18.7 l U S C C s La 3,713 0.2 2,582.8 16.0 574.1 6.8 r g e U S C C s -------------------------------------------------------------------------------- Note 1: "Large" refers to those USCCs and FCCs that had $100 million or more in assets. Note 2: All percentages are of total FCCs or USCCs, including both those that paid income tax and those that did not. Source: GAO analysis of IRS data. This observation is even more striking for the nontaxpaying USCCs. They numbered 1,265,272 or 62 percent of all U.S.-controlled corporate returns filed, but accounted for only 20 percent of the assets and 19 percent of receipts. So, the 38 percent of USCCs that paid income tax in 1991 had 81 percent of all gross receipts generated by USCCs. A SMALL NUMBER OF LARGE CORPORATIONS PAID NO INCOME TAX BUT HAD LARGE RECEIPTS AND ASSETS --------------------------------------------------------- Appendix V:4 The largest nontaxpaying FCCs and USCCs were relatively few in number but had a disproportionately large share of the total assets and receipts of all FCCs and all USCCs. In 1991, 715 nontaxpaying FCCs had assets of $100 million or more. These 715 corporations represented only 1.5 percent of all FCCs that filed U.S. corporate tax returns in 1991 but, as shown in table V.3, 31 percent of all FCCs' assets and 22 percent of their total receipts. In contrast, the 34,423 nontaxpaying FCCs with assets of less than $100 million were 71 percent of all FCCs but accounted for only 9 percent of total FCC receipts and 6 percent of total FCC assets. The observation is also true for USCCs. The 3,713 nontaxpaying USCCs with assets of $100 million or more accounted for a tiny 0.2 percent (one-fifth of one percent) of returns, but 16 percent of assets and 7 percent of receipts generated by all USCCs in 1991. Thus, the 1,261,559 nontaxpaying USCCs with assets under $100 million represented 62 percent of all USCC returns but only 4 percent of assets and 12 percent of the gross receipts. LARGE FCCS PAID LESS INCOME TAX RELATIVE TO GROSS RECEIPTS THAN LARGE USCCS --------------------------------------------------------- Appendix V:5 As shown in table V.4, the large FCCs that did pay U.S. income tax paid less relative to FCCs' total gross receipts in 1991 than the large USCCs that paid tax. Also, for any level of U.S. income taxes paid, large FCCs paid, on average, fewer taxes and, except for those paying $1 million or more in taxes, had higher gross receipts and assets than their taxpaying U.S. counterparts. Finally, large FCCs that paid no U.S. income tax in 1991, on average, had larger assets and more than double the receipts of the large USCCs that paid no tax. Table V.4 Profile of Large Foreign-and U.S.- Controlled Corporations, by Amount of Income Taxes Paid, 1991 Distribu Income tion by taxes paid income Income per $1,000 taxes Number of Percent of Assets Receipts taxes paid of paid returns returns (millions) (millions) (thous.) receipts -------- ---------- ---------- ---------- ---------- ---------- ---------- No tax 715 46.3 $ 795.5 $ 356.7 $ 0.0 NA paid Less 141 9.1 1,100.7 533.4 29.4 $ 0.05 than $100,00 0 $100,000 258 16.7 633.2 299.4 405.9 1.36 or more but less than $1 million $1 429 27.8 1,749.1 1,175.5 10,567.2 8.99 million or more ================================================================================ Total 1,543 100.0 $1,061.4 $ 590.9 $3,008.6 $ 5.09 Large U.S.-controlled corporations -------------------------------------------------------------------------------- No tax 3,713 37.2 695.6 154.6 0.0 NA paid Less 589 5.9 1,000.9 427.5 31.7 0.07 than $100,00 0 $100,000 2,429 24.4 502.7 141.9 486.7 3.43 or more but less than $1 million $1 3,246 32.5 3,110.7 1,417.0 20,680.4 14.59 million or more ================================================================================ Total 9,977 100.0 $1,452.4 $ 578.3 $6,848.7 $11.84 -------------------------------------------------------------------------------- Source: GAO analysis of IRS data. COST OF GOODS SOLD AND OTHER RATIOS PROVIDE INCONCLUSIVE EVIDENCE OF TRANSFER PRICING ABUSE --------------------------------------------------------- Appendix V:6 Ratios of cost of goods sold and other expense deductions to receipts yield interesting differences between large nontaxpaying FCCs and USCCs but no clear evidence of transfer pricing abuse. We calculated various components of cost of goods sold--including purchases, cost of labor, and inventory--and other items--including interest paid and taxes paid--as percentages of total gross receipts, and then compared the results for the large FCCs and USCCs. These are factors that we considered and reported on in our earlier reports and testimony.\3 Higher costs of goods sold and other items in relation to receipts may indicate, but do not necessarily prove, transfer pricing abuses. Table V. 5 presents the factors that differed most significantly. Table V.5 Ratios of Cost of Goods Sold, Purchases, and Interest to Receipts for Large Foreign-and U.S.-Controlled Corporations That Paid No Income Taxes, 1991 Large foreign- Large U.S.- controlled controlled corporations corporations -------------------- ------------------ ------------------ Cost of goods 65.7% 43.0% sold/receipts Purchases/receipts 47.7 25.7 Interest paid/ 9.5 11.4 receipts ------------------------------------------------------------ Source: GAO analysis of IRS data. For the large nontaxpaying FCCs, costs of goods sold accounted for 65.7 percent of receipts, while purchases accounted for 47.7 percent of receipts. In contrast, costs of goods sold and purchases for large nontaxpaying USCCs accounted for a smaller proportion of receipts--43.0 and 25.7 percent, respectively. This may indicate, but is not clear evidence, that large nontaxpaying FCCs are abusing transfer pricing--that is, shifting taxable income by inflating the cost paid to related parties outside the United States for goods and services. Indeed, since one potential means of shifting income is inflating the interest paid to a related party, we might expect that interest paid as a percentage of receipts would be higher for the nontaxpaying FCCs than for the USCCs.\4 But the opposite appears to be true--that is, interest paid as a percentage of receipts was higher (11.4 percent) for large nontaxpaying USCCs than for their FCC counterparts (9.5 percent). A recent study showed various factors at work in accounting for the depressed earnings of foreign firms in the United States.\5 It attributed these earnings to the firms having bought underperforming U.S. firms at top dollar, borrowed heavily, and spent freely on investment and marketing. The study also found some evidence that would point to profit shifting as a contributor to the low earnings. -------------------- \3 GAO/GGD-92-89, GAO/T-GGD-93-16, and GAO/GGD-93-112FS. \4 Grubert, Goodspeed, and Swenson, Explaining the Low Taxable Income of Foreign-Controlled Companies in the United States, National Bureau of Economic Research (Chicago: University of Chicago Press, 1993) stated that higher debt costs, either to unrelated lenders or to related offshore companies ("earnings stripping"), will lead to low taxable U.S. income because interest expenses are deductible. Yet, they concluded that "debt and earnings stripping do not appear to be major reasons for the low taxable income of foreign-controlled companies." They added that about half of the differential between rates of return for FCCs and USCCs can be explained by factors other than transfer pricing, and that the unexplained half could be due to transfer pricing distortions or other reasons. \5 David S. Laster and Robert N. McCauley, "Making Sense of the Profits of Foreign Firms in the United States," Federal Reserve Bank of New York Quarterly Review (Summer-Fall 1994). DIFFERENCES IN U.S. INCOME TAXES PAID BY FCCS AND USCCS MAY BE RELATED TO DIFFERENCES IN INDUSTRY --------------------------------------------------------- Appendix V:7 The type of business may also help explain the differences observed between FCCs and USCCs in their relative costs of goods sold, purchases, interest paid, and income taxes paid. Specifically, as shown in table V.6, while more than three-quarters of the large nontaxpaying USCCs were in finance, insurance, and real estate, little more than one-third of the FCCs were in these businesses. In contrast, almost one-half of the FCCs were in manufacturing or wholesale trade compared to less than 15 percent of the USCCs, which may explain the large nontaxpaying FCCs' relatively higher costs of goods sold and purchases relative to receipts. Table V.6 Industry Breakdown of Large Corporations That Paid No Income Taxes, 1991 Foreign- controlled U.S.-controlled Industry corporations corporations -------------------- ------------------ ------------------ Mining 3.5% 1.2% Construction 2.1 0.5 Manufacturing 29.9 10.3 Transportation and 2.0 3.3 public utilities Wholesale trade 16.9 3.8 Finance, insurance, 34.5 77.7 and real estate Services 11.2 3.2 Total 100.0% 100.0% ------------------------------------------------------------ Note 1: Because we defined large corporations in terms of asset size rather than receipts, this industry breakdown is skewed in favor of finance companies, which have large assets, and against trade companies, which have large receipts. Note 2: The detail may not sum to the total because of rounding differences. Source: GAO analysis of IRS data. MAJOR CONTRIBUTORS TO THIS REPORT ========================================================== Appendix VI GENERAL GOVERNMENT DIVISION, WASHINGTON, D.C. -------------------------------------------------------- Appendix VI:1 Jose R. Oyola, Assistant Director, Tax Policy and Administration Issues Lawrence M. Korb, Assignment Manager Michael J. Sanchez, Evaluator Nilsa I. Perez, Evaluator Patricia H. McGuire, Senior Computer Specialist James M. Fields, Social Science Analyst BOSTON/NEW YORK FIELD OFFICE -------------------------------------------------------- Appendix VI:2 Claire L. Gambaccini, Evaluator-in-Charge PHILADELPHIA REGIONAL OFFICE -------------------------------------------------------- Appendix VI:3 Carolyn B. Alessi, Evaluator