Taxing Multinationals: Transfer Pricing and Corporate Income Taxation in North America 9781442680371

Table of contents :
Contents
Preface
Part I. The Rules of the Game
1. Taxing Multinationals: An Introduction to the Issues
2. The International Tax Transfer Pricing Regime
Part II. Multinationals and Intrafirm Trade
3. The Multinational Enterprise as an Integrated Business
4. Multinationals and Intrafirm Trade in North America
Part III. Transfer Pricing and Taxation
5. The Simple Analytics of Transfer Pricing
6. Taxing Multinationals in Theory
7. Taxing Multinationals in Practice
Part IV. The Rules of the Game in North America
8. The U.S. Tax Transfer Pricing Regulations. Part I: Rules
9. The U.S. Tax Transfer Pricing Regulations. Part II: Procedures
10. The Canadian Tax Transfer Pricing Regulations
11. Transfer Pricing and the Tax Courts
Part V. Reforming the Rules of the Game
12. Reforming the International Tax Transfer Pricing Regime. Part I: Principles and Norms
13. Reforming the Tax Transfer Pricing Regime. Part II: Rules and Procedures
14. Conclusions and Policy Recommendations
Notes
Glossary of Economic and Accounting Terms
Bibliography
Author Index
Subject Index

Citation preview

TAXING MULTINATIONALS

This page intentionally left blank

LORRAINE EDEN

Taxing Multinationals: Transfer Pricing and Corporate Income Taxation in North America

UNIVERSITY OF TORONTO PRESS Toronto Buffalo London

www.utppublishing.com University of Toronto Press Incorporated 1998 Toronto Buffalo London Printed in Canada ISBN 0-8020-0776-7

Printed on acid-free paper

Canadian Cataloguing in Publication Data Eden, Lorraine, 1948Taxing multinationals: transfer pricing and corporate income taxation in North America Includes bibliographical references and index. ISBN 0-8020-0776-7 1. International business enterprises - North America Taxation. 2. Transfer pricing - North America. I. Title. HD2753.A3E32 1998

336.243

C97-931091-1

University of Toronto Press acknowledges the financial assistance to its publishing program of the Canada Council and the Ontario Arts Council. This book has been published with the help of a grant from the Humanities and Social Sciences Federation of Canada, using funds provided by the Social Sciences and Humanities Research Council of Canada.

I would like to dedicate this book to Carl Sumner Shoup, McVickar Professor Emeritus of Political Economy at Columbia University, who introduced me to multinational enterprises and transfer pricing over twenty years ago, encouraged me to take a multidisciplinary and policy-oriented approach to the problem, and has continued to provide me with guidance and the benefits of his knowledge in this complex area of international taxation.

This page intentionally left blank

Contents

Preface xi

PART I: THE RULES OF THE GAME 1 Taxing Multinationals: An Introduction to the Issues 3

Introduction 3 How Should National Governments Tax Multinationals? 4 The Purpose and Organization of This Book 9 Multinationals and Transfer Pricing 12 National Regulation of Transfer Pricing 26 The International Tax Transfer Pricing Regime 32 Explaining the Transfer Pricing Methods 36 The Importance of This Topic 52 Appendix 1.1: Searching For a CUP: The Christmas Tree Case 54 Appendix 1.2: A Typical Business Income Statement 61 Appendix 1.3: Differences in Economic and Accounting Methodologies 62

2 The International Tax Transfer Pricing Regime 63

Introduction 63 The Theory of International Regimes 63 The International Tax Regime 69 The North American Tax Regime 83 An Assessment of the International Tax Regime 95 The International Tax Transfer Pricing Regime 103 The North American Tax Transfer Pricing Regime 112 An Assessment of the International Tax Transfer Pricing Regime 116 Conclusions 116 Appendix 2.1: International Guidelines on Taxing Multinationals 117

viii

Contents

PART II: MULTINATIONALS AND INTRAFIRM TRADE 3 The Multinational Enterprise as an Integrated Business 125 Introduction 125 Why Multinationals Exist 126 Managing the Multinational as an Integrated Business 138 Multinationals and Intrafirm Trade 152 Multinationals and the North American Free Trade Agreement 166 Conclusions 172

4 Multinationals and Intrafirm Trade in North America 174 Introduction 174 North American Trade and Investment Patterns 174 How Large Is Intrafirm Trade in North America? 184 Conclusions 206

PART III: TRANSFER PRICING AND TAXATION 5 The Simple Analytics of Transfer Pricing 211

Introduction 211 Transfer Pricing in a Horizontally Integrated Multinational 213 Transfer Pricing in a Vertically Integrated Multinational 227 Three Problems with the Transfer Pricing Rules for Tangibles 239 Setting Transfer Prices for Intangible Assets 254 Setting Transfer Prices for Support Services 270 Conclusions 277

6 Taxing Multinationals in Theory 279

Introduction 279 Taxes and Transfer Pricing in Theory 280 Transfer Pricing and Trade Taxes 280 Transfer Pricing and a Tax on Pure Profits 287 Some Related Issues 295 Transfer Pricing and the Corporate Income Tax 303 Tax Penalties for Transfer Price Manipulation 308 Transfer Pricing and Unitary Taxation 313 Conclusions 319

Contents 7

ix

Taxing Multinationals in Practice 320

Introduction 320 Transfer Price Manipulation: A Literature Review 320 The IRS versus Foreign Multinationals: Tax Grab or Tax Abuse? 343 Incentives to Manipulate Transfer Prices in North America 356 New Evidence on Taxing Multinationals in North America 367 Conclusions 378

PART IV: THE RULES OF THE GAME IN NORTH AMERICA 8 The U.S. Tax Transfer Pricing Regulations Part I: Rules 383

Introduction 383 The U.S. Approach to Tax Transfer Pricing 384 Section 482: Allocation of Income and Deductions 385 Applying the Arm's Length Standard to Tangibles 390 Applying the Arm's Length Standard to Intangibles 397 Creating One Set of Rules: The New 482 Regulations 419 Conclusions 446 Appendix 8.1: A History of U.S. Tax Transfer Pricing Regulation 448

9 The U.S. Tax Transfer Pricing Regulations Part II: Procedures 454 Introduction 454 General Audit Procedures 454 Enforcement and Penalties 458 Updating the Dispute-Settlement Process 469 Conclusions 481

10 The Canadian Tax Transfer Pricing Regulations 483

Introduction 483 The Canadian Rules for Valuing Intrafirm Transactions 484 The Tax Avoidance Provisions 503 The Canadian Administrative Procedures 515 Conclusions 522 Appendix 10.1: A History of Canadian Tax Transfer Pricing Regulations 523

11 Transfer Pricing and the Tax Courts 525 Introduction 525 Indalex: The Facts of the Case 527

x

Contents The Federal Court, Trial Division, Decision (1986) 535 The Federal Court of Appeal Decision (1988) 540 An Economic Analysis of the Indalex Case 541 Lessons Learned from the Indalex Case 546 Conclusions 546

PART V: REFORMING THE RULES OF THE GAME 12 Reforming the Tax Transfer Pricing Regime Part I: Principles and Norms 549

Introduction 549 Reforming the Principles: International Equity and Neutrality 550 Reforming the Norm: The Arm's Length Standard 556 Replacing the Norm: Unitary Taxation 561 Conclusions 583

13 Reforming the Tax Transfer Pricing Regime Part II: Rules and Procedures 584

Introduction 584 Reforming the Rules: The Rules in Theory 585 Reforming the Rules: The Rules in Practice 611 Reforming the Procedures: Dispute-Resolution Mechanisms 627 Conclusions 633

14 Conclusions and Policy Recommendations 634

Introduction 634 Does Transfer Pricing Matter? 634 The Problem: Multinationals Are Integrated Businesses 635 The Solution: The International Tax Transfer Pricing Regime 637 Current U.S. and Canadian Practice 642 Policy Recommendations 645 Conclusions 652

Notes 653 Glossary 697 Bibliography 705 Author Index 737 Subject Index 743

Preface

In September 1974, I started a Ph.D. in economics, specializing in public finance, at Dalhousie University under the supervision of professors John Head, Carl Shoup, Cliff Walsh, and John Graham. It was an exciting, intense two years. Carl Shoup had offered to supervise my dissertation if I wrote on transfer pricing. (He had just retired from Columbia, was coming to Dalhousie as a visiting professor, and was working, as part of a United Nations Eminent Persons group, on a report on transfer pricing.) So I wrote my dissertation on transfer prices,1 and have been working in the area since that time. On such small things is one's career path determined! After graduation, I spent the next few years writing theoretical journal articles about transfer pricing, taxes, and tariffs, using the standard microeconomic models developed by Tom Horst, Larry Copithorne, myself, and others. In these models, the multinational enterprise (MNE) generally consisted of a parent firm and a wholly owned foreign affiliate. The MNE's goal was to maximize its after-tax global profits. The enterprise was faced with different tax rates, trade taxes, and exchange rates, and had to decide the direction, price, and volume of intrafirm trade. These choices determined how much each firm produced and sold in each country and where its profits were earned. In more sophisticated models, the MNE had an array of additional choices, such as the level of repatriated dividends, royalty payments, and share of head office expenses. These analyses were exercises in comparative statics; that is, in each case, the researcher modelled small changes from existing tax levels and determined the theoretical impact on variables such as output, sales, prices, taxes paid, trade volumes, and so on. Corporate income taxes were generally treated as taxes on the MNE's pure profits, which implicitly meant either that the firm had no equity capital or that all corporate income taxes were shifted, either forward to consumers or onto other factors of production, so that none of the tax was borne

xii

Preface

by shareholders. Models of how the home and host country taxed the MNE's profits populated journals in the 1980s. Tax the parent or foreign affiliate or both? Tax deferral or taxation as accrued? Foreign tax credit or deduction? The theoretical models have become more sophisticated over time (e.g., principalagent models under uncertainty), but the general format has not changed. In 1983,1 helped Alan Rugman organize a conference on transfer pricing that led to our jointly edited book on transfer pricing,2 now unfortunately out of print. The book looked at the theoretical, empirical, and public policy aspects of transfer prices. The messiness of empirical work on transfer pricing (e.g., difficulty in getting data, problems with databases and questionnaires, interpreting results) is evident in the empirical chapters. The last part of the book clearly demonstrates the difficulties in domestic regulation and the need for international cooperation (for example, Carl Shoup proposes international arbitration of transfer pricing disputes). One useful outcome of this exercise was that I learned that while I knew a fair amount about microeconomic modelling of transfer prices, I did not know very much about strategic decision-making processes within MNEs, and that useful work hinged on bringing these two together. My solution was to develop and teach a course on multinationals, starting first as an undergraduate economics course at Brock University in 1985, and then, from 1988 on, as a graduate course in the School of International Affairs at Carleton University. The Reading Economics Department (John Dunning, Mark Casson, Peter Buckley, John Cantwell, Alan Rugman) and the Harvard Business School (Ray Vernon, Michael Porter, Oliver Williamson, David Teece, Charles Kindleberger, Stephen Hymer) dominated the field.3 The Reading school was most closely identified with the OLI paradigm4 developed by John Dunning; the Harvard school followed the business strategy approach5 developed by Ray Vernon and Michael Porter. Only in the past few years have the two schools begun to merge.6 The more I read and learned and taught about multinationals, the less enamoured I became with the existing microeconomic models of transfer pricing. At the time, I saw them as sterile exercises in calculus, with little contribution to make to the real world of regulating MNEs. This realization became even clearer to me when, in 1990, Jack Calderwood and Wayne Voege in the International Audits Division at Revenue Canada Taxation asked me to put together a week-long course on 'Multinationals and Transfer Pricing' for their senior tax auditors and lawyers. The course has now been taught several times under the auspices of Carol Gouin, Jack Calderwood's successor at Revenue Canada. It has been an important building block in this manuscript. I was fortunate that the course participants were tolerant of my ignorance and

Preface

xiii

mistakes. From them I learned of the legal and accounting work that has been done on transfer price regulation, the real difficulties with putting the arm's length standard into practice, and the gap between the academic economists and the practitioners in their understanding of transfer pricing. I have also been fortunate to work with and learn from several well-known transfer pricing experts in these Revenue Canada courses, including Charles Berry, Nathan Boidman, Donald Brean, David Quirin, and Alan Rugman. Each has taught with me at one time or another. I owe a special debt to three internationally renowned individuals who have influenced my own thinking in this book. Carl Shoup, my dissertation supervisor and mentor, has shaped my views on the need for an interdisciplinary approach to this topic. His ability to successfully blend public finance and law, bringing the insights of academic economists together with the needs of realworld policy-makers and multinational managers, sparked my own desire to attempt the same bridging of disciplines and functions. Charles Berry, a Princeton University economist and expert witness in several U.S. transfer pricing court cases, in his own work on transfer pricing shows how economists can play an important and necessary role in this area. For Berry, the economic concepts of opportunity cost and imperfect competition are critical to transfer price regulation, both in writing tax legislation and in solving disputes through the court system. The third individual is Nathan Boidman, an international tax lawyer at Phillips and Vineberg in Montreal and the Canadian representative in the International Fiscal Association. Boidman's writings on the Canadian and OECD regulation of transfer pricing are prolific and insightful. He has been most helpful in providing me with large numbers of transfer pricing documents that I found difficult to obtain elsewhere. Charles Berry and Nathan Boidman also have the unusual gift of being able to make complicated tax court cases vivid and understandable. These three transfer pricing experts have inspired much of my own recent thinking on the theory of tax transfer pricing. However, I am sure there are sections in the text where we have honest disagreements, so I hasten to add that they bear no responsibility for any of the views expressed in Taxing Multinationals. I take full responsibility for the book and for any errors or omissions that may be present. When Allan Maslove, Research Director at the Ontario Fair Tax Commission, first approached me about writing a monograph on MNEs, taxation, and transfer pricing, I eagerly agreed. My goal was to write a more integrated, interdisciplinary study, one that brought the perspectives of the different disciplines together to examine the transfer pricing issue. The monograph, in attempting to do so much, took much longer to write than anticipated, and eventually grew to

xiv

Preface

a full-fledged book. I owe a debt to the Ontario Fair Tax Commission for funding the beginning of the project. Initially I had intended to look at only the Canadian and U.S. transfer pricing regulations. However, as my work progressed, it became obvious that the OECD Committee on Fiscal Affairs has been instrumental in moulding an international consensus around the arm's length standard. The OECD, in my opinion, has successfully established an international regime (in the political science sense of having coherent principles, norms, rules, and procedures) in tax transfer pricing, at least among its 25 member states. The book therefore has been written from the perspective of an international regime approach to taxing multinationals, with detailed discussions of the role of the OECD through its model tax treaties and various transfer pricing reports, the draft transfer pricing guidelines issued in piecemeal fashion over the 1994-6 period. Chapters 2, 5, 12, and 13, in particular, deal with the international regulation of tax transfer pricing, focusing on the OECD's transfer pricing reports. Thus, while Taxing Multinationals deals primarily with the history, economics, and regulation of transfer pricing in Canada and the United States, the analysis is placed more generally in the context of the international situation; this should make the book more useful for a wider audience. Two other key events occurred while the book was in progress. The North American Free Trade Agreement came into force in January 1994. I had also been doing some work for Industry Canada on the organizational and locational decisions of multinationals within the context of regional integration (NAFTA) and technological change (the shift from mass to lean production). These insights turned out to be quite relevant for an analysis of MNE responses to transfer price regulation and corporate income tax changes. They are discussed in some detail in chapters 3 and 4. While I have not specifically devoted a chapter to Mexican tax law in this area, discussions of Mexico in regard to NAFTA can be found throughout the book, especially in chapters 2 and 7. At the same time, the U.S. Treasury was engaged in a major overhaul of the Treasury regulations that accompany section 482 (the transfer pricing section) of the Internal Revenue Code. The final version of the 482 regulations was released in late June 1994, and the final version of the penalty regulations for valuation misstatements (section 6662) in 1996. The book reviews in detail the nearly ten-year development of the new regulations, starting with the 1986 addition of the 'commensurate with income' standard to section 482, and provides a critical analysis of these changes in chapters 8 and 9. Taxing Multinationals was written in and carried between Ottawa, Ontario; Cambridge, Massachusetts; Columbus, Ohio; and College Station, Texas, over a four-year period. During part of the period (July 1992 through August 1993),

Preface

xv

I was on sabbatical leave from Carleton University on a U.S.-Canada Fulbright award as Visiting Professor and Fellow in the Center for Business and Government at the Kennedy School of Government, Harvard University, and as Visiting Professor, Department of Finance, College of Business, The Ohio State University. The Finance Department at Ohio State continued to provide me with office space and departmental support while I was on research leave from Carleton in the fall of 1994. In the summer of 1995,1 left Carleton (remaining on staff as an Adjunct Research Professor in the Paterson School) and moved to join the Department of Management at Texas A&M University, where the final revisions for the book were completed. I would like to thank Carleton, Harvard, Ohio State, and Texas A&M for providing me with the research facilities and colleagues that enabled me to complete this project. I would also like to thank Virgil Duff at the University of Toronto Press for his continuing support and encouragement as the manuscript and I moved from place to place over this period. Helpful comments and assistance on various parts of the manuscript were received from Brian Arnold, Charles Berry, Nathan Boidman, Harry Grubert, Chuck Hermann, Allan Maslove, Christopher Maule, Maureen Molot, Kingsley Olibe, David Quirin, Subi Rangan, Alan Rugman, Carl Shoup, Ray Vernon, Wayne Voege, and from individuals within Revenue Canada, the Department of Justice, and Industry Canada. I am especially indebted to Don Brean for reading and offering detailed comments on the first full draft of the manuscript, and to two anonymous external referees for their detailed comments on the penultimate version of the manuscript. Various parts of the manuscript have been used in several of my Revenue Canada executive training courses on multinationals and transfer pricing, and I thank the course organizer Bill Blair and the participants for their excellent suggestions. Sections of the manuscript have also been given in several university seminars. Cindy Murray has been a superb research assistant, both during the Revenue Canada courses and in helping prepare this manuscript. Usha Viswanathan, Nanette Tello, and Reinhard Hinterreither have been most helpful in finishing the final version. Lastly, I would also like to thank my family, friends, and colleagues - particularly Jessica Eden and Chuck Hermann - for their understanding and support while I was writing and rewriting this book. I am sure they are as grateful as I am that it is finally done. Taxing Multinationals should be useful to a wide variety of individuals: graduate students and academics, legal and accounting professionals, business economists, government policy-makers, and managers of multinational enterprises. Transfer pricing is a subject in many different university and executive training courses: business policy, cost accounting, corporate law, the economics of multinationals, international finance, and international taxation. The reasons are not

xvi

Preface

hard to find. University academics are interested in predicting and measuring the impacts of different taxes and transfer pricing rules applied to multinationals. Legal and accounting professionals and business economists advise multinationals as to acceptable transfer prices, critique government proposals, and are involved in tax court cases. Government policy-makers, tax auditors, and lawyers are responsible for writing and enforcing transfer pricing laws and regulations. Business professionals, particularly in accounting and finance departments, are responsible for developing MNE transfer pricing policies that meet both internal and external objectives. All these groups need to understand transfer pricing regulation. This is the background that has led me to write this book. It is my firm belief that to study transfer pricing and the taxation of intrafirm trade it is essential to bring together the disciplines of economics, law, accounting, political science, and international business. Each in its own way has focused on transfer pricing, but each is seeing only one piece of the puzzle - transfer pricing requires an interdisciplinary set of lenses. LORRAINE EDEN COLLEGE STATION, TEXAS

PART I: THE R U L E S OF THE GAME

This page intentionally left blank

1

Taxing Multinationals: An Introduction to the Issues

Introduction Governments use taxes to finance public spending. Given that the ultimate purpose of taxes is to raise sufficient revenues to finance expenditures, how should these revenues be raised? Most economists would answer this question by arguing that governments should set up their domestic tax systems with two underlying principles of public finance in mind: equity and neutrality. A good tax system should be equitable that is, two taxpayers in similar economic circumstances should pay the same tax (horizontal equity) and taxpayers in different circumstances should pay appropriately different taxes (vertical equity). The system should also be neutral - that is, it should not affect the taxpayer's choice of corporate form, location of the tax base, choice of pricing policy, and so on. This is the so-called 'normative theory of public finance,' which has a long tradition in economic theory.1 Setting up a tax system based on these principles involves choices about the appropriate blend of what Carl Shoup (1991) has called tax architecture, engineering, and administration. Tax architecture (choosing which taxes to include in the tax system), tax engineering (deciding the substantive issues concerning each tax, such as its rate and base), and tax administration (how to implement tax law in practice) all need to be considered simultaneously when setting up or reforming a tax system. This book is about the architecture, engineering, and administration of taxing multinational enterprises. Its specific focus is tax transfer pricing, that is, how governments treat the pricing policies multinational enterprises (MNEs) adopt for intrafirm transfers among their affiliated companies. Multinational enterprises are private organizations that engage in foreign direct investment (FDI) in the form of owning and/or controlling value-adding activities in more than

4

The Rules of the Game

one country. The parent firm and its domestic and foreign affiliates engage in international production, producing and selling products around the world. MNEs can be either horizontally integrated (different affiliates produce, the same product in different markets) or vertically integrated (upstream affiliates produce intermediate products that are further processed by downstream affiliates prior to final sale) or both. Because the MNE is an integrated enterprise, its affiliates engage in substantial amounts of intrafirm transactions. The price of any non-arm's length transaction involving transfers of goods, intangibles, or services between wholly or partly owned affiliates (parent, branch, subsidiary) of a multinational enterprise is called a transfer price. Governments are concerned that multinationals can and do manipulate transfer prices (that is, over- or under-invoice their intrafirm transactions) so as to avoid paying corporate income taxes. Tax authorities, in response, have developed a complex set of rules and procedures at the national and international levels designed to regulate MNE transfer pricing policies. These regulations are the subject of this book. How Should National Governments Tax Multinationals? In making their tax architecture, engineering, and administrative choices, the fiscal authorities generally act as if their power to tax were unbounded, and, under national law, the federal power to tax residents of a country is basically unbounded. However, even if a government's power to tax is legally unbounded, in practice there are limits because jurisdictional reach is restricted by the mobility of individuals and businesses and by the reach of other tax authorities. As Brian Arnold, a well-known Canadian tax law expert, notes: [The] government's power to tax is limited effectively only by the countervailing interests of other governments and the practical difficulties of enforcement and collection. There are no limitations under international law on a nation's power to tax; and in most countries, there are no constitutional limitations. (Arnold 1986, 1)

If there are limits in terms of taxing domestic labour and capital owners, these limitations are even more pronounced in terms of taxing multinational enterprises (MNEs). As the activities of these large, integrated businesses grow and spread out across the globe, so do the interlinkages between national economies. Governments are faced with regulating firms within their borders at the same time as these borders are becoming more permeable. Given the mobility of multinational enterprises and of capital flows in general, countries have to

An Introduction to the Issues

5

accommodate their domestic tax policy choices to the realities of the global economy or see capital flight erode their tax base. The Globalization of the Multinational Enterprise Recent data on multinationals show how the globalization of (what was once) domestic activity is spreading through the activities of large enterprises. Inward and outward investment and crossborder flows of technology, goods, services, and businesspeople have all increased exponentially since 1950, particularly in the Triad economies of the United States, the European Community (now called the European Union), and Japan. For example, the 7993 World Investment Report (UNCTAD 1993, 13) estimates that at least 35,000 parent firms control 170,000 foreign affiliates worldwide. These include both majority-owned foreign affiliates or MOFAs (i.e., the parent firms holds more than 50 per cent equity ownership in the affiliate) and minority-owned foreign affiliates (i.e., the parent holds between 10 and 50 per cent equity ownership). The report estimates that Canada has roughly 1,300 parent firms with over 10,000 affiliates; the corresponding numbers for the United States are at least 3,000 parents with in excess of 15,000 affiliates (UNCTAD 1993, 20).2 The top 100 multinationals, ranked by the size of their foreign assets, in 1990 owned 3.2 trillion dollars in assets, of which approximately 1.2 trillion dollars were held outside the parent firm's home country. These 100 firms account for one-third of the worldwide stock of foreign direct investment (FDI). Five major home countries (United States, United Kingdom, France, Germany, and Japan) are the headquarters for 75 per cent of the top 100 MNEs. The United States is home to 27 firms on this list and accounts for one-third of the top 100 MNEs' foreign assets. Of the top ten MNEs, five are American: Ford, General Motors, Exxon, IBM, and Mobil. Canada is home to three (Thomson, Alcan Aluminium, and Seagram).3 In terms of industries, the petroleum, automotive, chemical, and pharmaceutical industries represent over half the foreign assets of the largest 100 firms (UNCTAD 1993, 22). Thus the largest of the multinationals are concentrated both in terms of geography and industry distribution. Worldwide, the assets of all MNEs have been estimated to total more than nine to ten trillion U.S. dollars; more than $3 trillion are held by foreign affiliates. MNEs worldwide employ more than 55 million workers; foreign affiliates have a labour force in the range of 15 million. Worldwide sales by MNEs total in excess of $13.5 trillion, with $4-4.5 trillion sales made by foreign affiliates. The worldwide sales of the largest firms exceed the gross domestic products of many small countries (Dunning 1993, 16).

6

The Rules of the Game

As these statistics show, large multinational enterprises are the major nonstate actors in the global economy. As economies become more open and more integrated due to the globalizing effects of MNEs, national taxation of multinationals become more problematic. We look below at some of the problems MNEs create for national tax authorities. The Problems Multinationals Create for National Taxation Globalization of MNE activities raises the national salience of the problematique of international taxation: How should national governments tax multinationals in a global economy? Lawrence Summers phrases the issue as follows: Until recently, international taxation has been an arcane subspecies among American tax lawyers, and international considerations have rarely influenced the thrust of tax reform ... Such a provincial approach to tax policy may have been appropriate in an earlier era, but the increasing economic integration of the world requires a more global approach to policy. (Summers 1988, 64)

Multinational enterprises create particular problems for tax authorities that do not occur in taxing domestic firms. The key reason is that the MNE is an integrated or unitary business. The accepted definition of an MNE is two or more firms, located in different countries, but under common control, with a common pool of resources and common goals. The multinational enterprise should be visualized as an interlocking network of activities, working more or less in tandem depending on the control exercised by the parent firm. The enterprise's goals are to survive, make profits, increase its market share, and grow. Its rivals are other large multinationals and its actions are developed as strategic responses to those rivals in an environment characterized by market imperfections, oligopolistic behaviour, and substantial risk and uncertainty. Since, by definition, its activities cross national borders, the MNE has certain characteristics which pose problems for tax authorities: A multinational has affiliates located in several countries. Thus the MNE has a global reach, whereas governments are limited by their geographic boundaries to a national reach. This creates jurisdictional problems for domestic tax authorities and limits the effectiveness of governments in taxing MNEs. All components of a multinational are under the common control of the parent firm. This means the MNE decisions on investment, production, sales, trade, and pricing may be made outside the country.

An Introduction to the Issues

7

All members of the MNE family have common goals such as the maximization of global after-tax profits. This brings the affiliates of the enterprise into conflict with the governments where they are located since each government has its own national goals which most likely will differ from the MNE's goals. A multinational has common overheads and resources. This causes problems for tax authorities in deciding where the tax base is located and how to allocate the income from, and expenses of, MNE activities among jurisdictions. The resources allow the MNE to escape the jurisdiction of national governments (for example, controls over borrowing can be avoided if affiliates can access their parent's funds). Conflicts are inevitable when national governments tax multinationals because domestic tax systems set up for domestic purposes, by definition, are poorly designed to handle the international activities of multinational enterprises. Thus tax authorities and MNEs are likely to disagree about the appropriate tax the enterprise should pay at the national level. Conflicts can also occur between the tax authorities of the countries where the units of the MNE are located as these governments compete for their 'fair' share of an increasingly mobile tax base. Double taxation and/or undertaxation of MNE profits, relative to the taxes that would be paid by a purely domestic firm engaged in comparable activities in comparable circumstances, is highly probable. National Responses to the Problem of Taxing Multinationals Governments have responded in two very different ways to the problems MNEs create for domestic tax systems. The first approach is to use lower taxes to attract MNEs. One method is to set low tax rates at home to attract MNEs to locate inside the country. For example, many governments in developing countries have set themselves up as tax havens (e.g., Bermuda, Bahamas, the Seychelles), trying to attract more inward investment activity through lower tax rates. In addition, many countries offer tax rebates, tax holidays, and other financial incentives for certain locations in their countries; e.g. export processing zones (South Korea, Taiwan, Ireland, Mexico), duty-free zones (the United States) and international banking and financial centres (Canada). A second method is not to tax the income domestic MNEs earn on their foreign activities. For example, some governments tax foreign source income such as dividends only when remitted by a foreign affiliate to its parent firm (the United States). Other tax authorities do not tax repatriated income if the foreign source income is defined as active business income4 (Canada). A third group does not tax foreign source income at all (France).

8 The Rules of the Game The second approach to the problems created by MNEs is the reverse of the first: that is, tighten up tax regulations, eliminate loopholes, broaden the tax base, and/or raise tax rates. This approach is designed to make sure firms that are located within a country pay their 'fair share' of taxes to the domestic government. The United States is the best example of this approach. While all OECD governments have passed tax legislation insisting that multinationals follow the arm's length standard - that is, intrafirm transactions should be priced the same as the prices chosen by unrelated parties engaged in similar transactions under similar circumstances - the U.S. government has developed incredibly complex regulations outlining how this standard is to be followed in practice.5 In addition, the Internal Revenue Service has narrowed the credits given to U.S. MNEs for foreign taxes paid on their foreign source income. The U.S. tax rules require U.S. parent firms to charge their affiliates higher royalty rates and more for headquarters expenses, shifting taxable income to the parents. Penalties for income tax violations have risen, along with the amount of documentation that MNEs are required to file with tax authorities. The amounts spent by the IRS on auditing and enforcement of MNEs, and the numbers of tax auditors allocated to transfer pricing, have also risen dramatically over the past 15 years. Lastly, at the subfederal level, some of the state governments (California) have broadened their tax reach by attempting to tax MNE worldwide income through the unitary tax method. In this second approach, we should also distinguish between those governments that have moved or are moving to a tighter fiscal regulatory system for all firms and governments that have focused mostly on lessening abuses - i.e., on the small percentage of firms that are tax evaders. Of course, while arguing that the purpose of the regulation is to catch abusers, tax authorities may in practice be adopting a confiscatory tax regime that applies to everyone. The distinction is important, as a system designed to penalize abusers should be different from one designed to provide uniform treatment across taxpayers. In the U.S. case, it is clear that the Internal Revenue Service has moved to a tighter regulatory system for all firms, both U.S. multinationals and foreign MNEs located in the United States. At the same time, the U.S. Congress has shown open concern with potential tax abuse in particular sectors (pharmaceuticals), by firms of particular nationalities (Japanese transplants), and in particular categories of transactions (transfer pricing). Problems at the International Level Clearly, these two very different ways of dealing with the global reach of mul-

An Introduction to the Issues 9 tinationals - we can call them the 'low-tax' versus the 'high-tax' approach can and do create international interjurisdictional conflicts among the countries themselves. Disagreements between MNEs and governments, and between governments, over the appropriate 'tax bite' taken by the tax authorities are likely to arise. For example, an enterprise with affiliates in two different tax jurisdictions may find its income double taxed if the definitions and/or methods of taxation are not harmonized between countries. This can happen if one government reassesses the MNE's income and levies a higher tax bill, and the second government is unwilling to provide an offsetting tax adjustment. Differences in tax systems also allow the possibility of tax arbitrage - that is, the shifting of real and/or financial activities from the high taxed to the low taxed location. For example, where one state is a low-tax jurisdiction and another neighbouring state is a high-tax jurisdiction, capital may be attracted into the lower-tax location. Firms may shift revenues to the low-taxed, and deductible expenses to the high-taxed, location. This puts pressure on high-tax states to reduce their taxes and/or to tighten their monitoring and enforcement mechanisms in order to avoid losing mobile firms and employment opportunities. If capital exits, taxes on less mobile actors (e.g., labour) must rise in order to provide the same level of public services. Thus tax differentials can have inequitable and non-neutral effects on multinationals. As a result, the principles of public finance - equity and neutrality - which should underpin a good tax system are unlikely to be satisfied at the international level so that either under- or overtaxation of multinationals is probable. The amount of income to be taxed, and the division of the tax revenues among the countries where the MNE conducts its activities, are unlikely to be seen as fair, either by the MNE or by the revenue authorities; tax neutrality is also problematic. Domestic taxation of MNEs, without harmonization or coordination of national tax systems, in sum, is a recipe for conflict. The international problems caused by multinationals raised above, and the various government responses, fall into two general categories. The first is the general question of tax jurisdiction. Which government has the right to tax what tax base? While we pay some attention to this question (primarily in the next chapter), the focus of this book is on the second question: the issue of income and expense allocation. The Purpose and Organization of This Book Our main interest in this book is the appropriate valuation for tax purposes to attach to intrafirm transactions among various affiliates of the multinational

10 The Rules of the Game enterprise, particularly with respect to multinationals in North America. That is, the purpose of this book is to address two sets of questions. The first set of questions deals with taxing multinationals in general at the national and international levels in terms of the allocation of MNE income and expenses and transfer pricing issues. Here we address questions such as the following: How do multinationals set their transfer pricing policies in theory? In practice? How can MNEs manipulate transfer prices in order to avoid paying corporate income taxes? What are the theoretical benefits and costs of transfer price manipulation? Do multinationals engage in such manipulation of transfer prices? What evidence do we have that MNEs do manipulate transfer prices so as to avoid paying taxes? How should governments regulate MNE transfer pricing policies at the national level so as to ensure that the principles of international equity and neutrality are achieved in terms of taxing the income from multinationals? What are the recommendations of international organizations such as the Organization for Economic Cooperation and Development (OECD) and the United Nations with respect to taxing intrafirm transactions? What principles and norms underlie these recommendations? What transfer pricing methods are recommended, and why? How could these be improved? Is there a regime in place at the international level through which national governments can cooperate to reduce the interjurisdictional problems of taxing multinationals at the national level, a regime specifically focused on the issue of taxing intrafirm transactions, that is, tax transfer pricing? The second set of questions deals specifically with transfer pricing regulations and practice in Canada and the United States. We address these questions: How important are MNEs in the North American economy? How large is intrafirm trade? Is transfer price manipulation a problem in North America? What evidence do we have that MNEs have manipulated transfer prices so as to avoid paying taxes to the U.S. and Canadian governments? How do the U.S. and Canadian governments tax multinationals, and, in particular, what regulations do they have with respect to transfer pricing? Do these regulations satisfy international norms and principles with respect to taxing multinationals? How could these regulations be improved?

An Introduction to the Issues

11

What transfer pricing policies should multinationals in North America adopt in response to these regulations? Taxing Multinationals examines the current tax transfer pricing regime, focusing in particular on the U.S. and Canadian approaches to transfer price regulation under the corporate income tax (CIT). The book deals with the regime in terms of its tax architecture, engineering, and administration at the domestic and international levels. At the national level, we examine how the Canadian and U.S. tax officials have attempted to regulate transfer pricing through the corporate income tax> evaluate the various methods that have been employed in the past, and make proposals for improvements. Policy recommendations are made to improve the overall effectiveness of the U.S. and Canadian approaches to taxing intrafirm trade. At the international level, we look at the role of the OECD as the organization at the heart of the tax transfer pricing regime. We evaluate the current regime in terms of its principles, rules, and procedures, focusing in particular detail on an assessment of the various transfer pricing methods for valuing tangibles, services, and intangibles. We argue that new solutions are necessary, solutions worked out at the multilateral level and not unilaterally imposed by the largest and most powerful governments. We make suggestions for such improvements, and strongly urge policy reform be developed in multilateral forums like the OECD's Committee on Fiscal Affairs. Taxing Multinationals is organized in five parts. Part I, 'The Rules of the Game,' is divided into two chapters: Chapter 1 introduces the book, while Chapter 2 develops the book's framework, the international tax transfer pricing regime. Part II, 'Multinationals and Intrafirm Trade,' consists of chapters 3 and 4, which focus on the multinational enterprise as an integrated business. Chapter 3 develops a theory of the MNE as an integrated business that includes the possible impacts of regional integration schemes (such as the North American Free Trade Agreement) and technological change on intrafirm trade patterns. Chapter 4 provides statistical data on the extent and involvement of multinationals in the North American economy. Part III, 'Transfer Pricing and Taxation,' contains chapters 5, 6, and 7, which deal with taxing multinationals in theory and practice. Chapter 5 develops several theoretical models that explain how MNEs choose transfer prices for goods, services, and intangibles in the absence of external motivations for transfer pricing such as taxes and tariffs. In each case, the theoretical results are compared with the OECD's transfer pricing guidelines. Chapter 6 extends these

12

The Rules of the Game

models to cases in which the MNE faces various types of taxes and government regulations. In Chapter 7 the empirical literature on transfer price manipulation is reviewed, along with an analysis of the recent U.S. and Canadian debates over the tax payments of multinationals. New evidence on taxes paid by multinationals, both domestic and foreign owned, in North America is presented. Part IV, The Rules of the Game in North America,' consists of four chapters that examine Canadian and U.S. transfer pricing regulations. Chapter 8 provides a detailed history of U.S. transfer pricing rules, while Chapter 9 focuses specifically on U.S. tax procedures. The Canadian rules and procedures are reviewed in Chapter 10. Chapter 11 looks at one court case in detail, Indalex versus the Queen, the best known of the Canadian tax court cases in the transfer pricing area. Part V, 'Reforming the Rules of the Game,' contains three chapters that assess the international tax transfer pricing regime, examine possible alternatives and reforms, and make policy recommendations. Chapter 12 evaluates the international tax transfer pricing regime in terms of its principles and norms, with a large section devoted to evaluating the main alternative to the arm's length standard: unitary taxation. Chapter 13 evaluates the regime in terms of its rules and procedures. The last chapter of the book concludes with policy recommendations for Canadian and U.S. taxing authorities. The purpose of this book is to address these questions. We explore the answers to some of the questions very briefly below, and devote the rest of this book to an in-depth study of these topics. Multinationals and Transfer Pricing In this section we look at transfer pricing through the eyes of the multinational enterprise. What is a transfer price, and why do MNEs use transfer prices? What pricing methods do large firms use for their intrafirm transactions? Is one method more commonly used than other methods? What incentives are there to manipulate these prices? What Is a Transfer Price? MNEs supply their affiliates with a package of capital and technology inputs and managerial skills, for which the parent firm receives a stream of dividend and interest payments, royalties, and licence fees. Intrafirm transfers of technology, management services, and financial loans move around within the MNE family. Intermediate goods (parts, components, subassemblies) flow down-

An Introduction to the Issues

13

stream for further processing before final sale to end consumers. Some affiliates provide business services (e.g., legal, accounting, advertising) on behalf of the group. The examples listed in the previous paragraph are all examples of intrafirm trade - that is, trade in goods, services, and intangibles conducted at non-arm's length within the affiliates of the MNE family. The price of any non-arm's length transaction involving goods, technology, or services between wholly or partly owned affiliates (parent, branch, subsidiary) of the MNE is called a transfer price. The multinational may record some intrafirm flows on its books as transactions and formally put a price on these activities. Other flows may not be treated as separate transactions, nor priced internally. Most intrafirm flows of tangibles (raw and semi-finished products, finished goods), are valued by the MNE in one of two ways: on a cost plus basis, that includes direct costs plus some allocation for overhead expenses of the producer, or on a market price basis, where prices charged to nonrelated firms are used to determine transfer prices on related party sales. Within these two general categories lies a wide range for determining the actual transfer price. Which transfer pricing method is chosen will depend on the relative strengths of the various motivations the MNE has for using transfer pricing. The Multinational's Motivations for Transfer Pricing There are both internal and external motivations for transfer pricing. In terms of internal motivations, where different affiliates within the MNE family are treated as stand-alone units called profit centres, transfer prices are needed internally by the MNE to determine profitability of the individual divisions. Transfer prices can also be used for internal measures of performance by individual affiliates and to motivate corporate managers.6 Other units within the MNE, particularly units which provide group services to the MNE family, are likely to be run as cost centres. In such cases, downstream affiliates are generally charged a share of the costs of providing the group service function so that the service provider, in total, covers its costs plus a small mark-up. For example, the price for windshield wiper blades made by a Mexican maquiladora subsidiary of Ford, the North American advertising expenses incurred by the U.S. head office of Toyota, and the tooling charges paid by Ford Canada to its U.S. parent are all examples of transfer prices the MNE is likely to record on its books for internal reasons. On the other hand, affiliates often share in the ongoing goodwill intangibles of the parent, exchange information among themselves, and offer short-

14 The Rules of the Game term assistance when problems arise. These events generally occur without the need for the MNE to price the intrafirm activity. Thus there are likely to be intrafirm transfers where the enterprise has no internal motivation for setting a price. Several external motivations can affect the MNE's choice of transfer prices. Because multinationals operate in two or more jurisdictions, transfer prices must be assigned for intrafirm trade that crosses national borders. Border taxes, such as tariffs and export taxes, are often levied on crossborder trade. Where the tax is levied on an ad valorem (per cent of the value) basis, the higher the transfer price, the larger the tax paid per unit. On the other hand, where border taxes are levied on a per-unit basis (i.e., specific taxes), the transfer price is irrelevant for tax purposes. Another external factor that can affect a multinational's transfer pricing choices is the need to meet the rules of origin that apply to crossborder flows within a free trade area. Since border taxes are eliminated within the area, rules of origin must be used to determine eligibility for duty-free status. Over- or underinvoicing inputs is one way to avoid customs duties levied on products that do not meet the rule-of-origin test. In addition, MNEs must declare profits and pay taxes in the various jurisdictions where they do business. Most governments tax residents on their worldwide income while taxing nonresidents on their domestic source income. The need to declare taxable income means that the enterprise must allocate its expenditures and revenues among its various affiliates, set prices for all intrafirm crossborder transactions, and, at the same time, follow the different (and possibly conflicting) corporate tax rules set down by the various taxing authorities. Thus, the MNE may have to determine and record transfer prices for activities even if there is no internal reason to determine a price. Figure 1.1 shows the various transfer prices that could be involved in crossborder intrafirm transactions between an MNE parent and its foreign affiliate. These include the valuation of goods (where both tariff and tax authorities are involved), services, and intangibles (where tax officials are involved). Where rules of origin must be satisfied in a free trade area, valuation of intrafirm trade (exports and imports) in goods, services, and intangibles is required. All of the transactions identified in Figure 1.1 are examples of intrafirm trade in the sense that they take place between related parties that are not at arm's length with each other. How important are internal and external factors in affecting the actual transfer pricing policies of MNEs? To answer this question, we look at some recent empirical studies on how firms set their transfer prices.

An Introduction to the Issues FIGURE 1.1 External Motivations for Setting Transfer Prices

15

16 The Rules of the Game Transfer Pricing in Practice In practice, the typical MNE values its intrafirm trade flows within the firm, either by having the prices set by headquarters or through bargaining among divisions. The more centralized the MNE, the more likely the transfer price is to be set by headquarters. If an outside market price exists, some reference is often made to that price; however, not all MNEs allow their divisions to buy or sell on the outside market. Several studies have been done, mostly through questionnaires, to find out how MNEs actually set transfer prices for tangibles. We report on four of the most recent studies. The Benvignati (1985) Study Benvignati (1985) analysed data from the U.S. Federal Trade Commission on intracorporate transfers made by U.S. manufacturing MNEs to their foreign and domestic affiliates in 1975. She found that nonmarket pricing was used more frequently (76 per cent of the cases) in transfers to foreign affiliates than to domestic affiliates (49 per cent of cases). (See Table 1.1 below.) The most important nonmarket pricing method was cost plus, used in 57 per cent of all foreign transfers and 29 per cent of domestic transfers. She concluded that foreign transfers were potentially more problematic for tax authorities because such transfers typically were not based on market prices. Benvignati then used regression analysis to explain the greater use of nonmarket prices in foreign transfers. The dependent variable, the percentage of foreign affiliate transfers priced at market prices, was regressed against a large number of industry- and firm-related variables. The significant variables were the firm's advertising intensity (negative sign), the dollar value of total MNE transfers to foreign affiliates (negative), foreign branch activity (positive), domestic company size (positive), and number of foreign affiliates (positive). On the basis of these regressions, she reached several conclusions. First, most of the variation in pricing behaviour came from firm-to-firm differences and not from industry variations. Second, U.S. MNEs were more likely to use nonmarket prices if products were heterogeneous, total transfers were large, and the MNE was small and did not have branches or many affiliates. Third, the U.S. tax treatment of branches on an accrual basis made it less profitable to engage in transfer price manipulation. And, fourth, large-size MNEs with many subsidiaries were also more likely to use market-based pricing perhaps because of a higher likelihood of tax audits, higher propensity for conflicts with management objectives, or greater administrative costs associated with keeping 'two sets of books' (Benvignati 1985, 209).7

An Introduction to the Issues

17

TABLE 1.1 The Benvignati (1985) Survey of Transfer Pricing Methods

Per cent of transfers that used a market-based pricing method Per cent of transfers that used nonmarket-based pricing methods: Cost plus method Other cost-based methods Other methods Total per cent using nonmarket-based pricing methods

Domestic transfers

Foreign transfers

49.44

24.04

29.27 18.83 2.46 50.56

57.24 14.48 4.25 75.96

SOURCE: Benvignati (1985, 197)

The Al-Eryani, Alam, and Akhter (1990) Study Al-Eryani, Alam, and Akhter (1990) surveyed 164 U.S. multinationals in 1987 concerning their transfer pricing policies. Large MNEs were selected and then broken into two groups depending on whether the majority of their foreign affiliate activity occurred in developed or developing countries. Their results are summarized in Table 1.2. The authors found that 50 per cent of the sample MNEs with affiliates primarily in developed countries used cost-based transfer pricing methods; another 34 per cent used methods based on market prices. For MNEs mainly in developing countries the percentages were 41 and 38 respectively. Overall, 47 per cent of the sample MNEs used cost-based pricing while 35 per cent used market-based pricing. Two-thirds of the 'actual cost' cases included a fixed markup on top of actual full cost per unit of output; one-third had no markup. A markup also characterized most of the 'standard cost' cases. Marginal or opportunity cost was rarely adopted. Therefore about two-thirds of the cost-based cases used a cost plus transfer pricing methodology. The second most frequent transfer pricing policy was market price with the cases evenly split between the prevailing price and an adjusted market price. What Table 1.2 does not say is who sets the transfer price - head office, a centralized purchasing division, or the firms themselves. The only case that suggests complete autonomy of the divisions is the negotiated price approach. Approximately 15 per cent of respondents negotiated prices - i.e., the related firms would set the transfer price through bargaining between themselves. The IBFD( 1991) Study A third survey was conducted by the International Bureau of Fiscal Documentation (IBFD) in 1991 (see Hamaekers 1992, 605). This voluntary survey,

18

The Rules of the Game

TABLE 1.2 The Al-Eryani et al. (1990) Survey of Transfer Pricing Methods U.S. MNEs with affiliates primarily in developed countries

Number

Per cent of total

15 16

20 21

34 45

20 26

2

0

0

2

1

50

50

31

41

81

47

13 2

13 2

12 4

16 5

25 6

15 3

100

66

47

62

112

65

17 15

18 16

12 16

16 22

29 31

17 18

32

34

28

38

60

35

97

100

75

100

172

100

Per cent of total

Actual cost Standard cost Marginal cost or opportunity cost Total cost-based methods

19 29

19 29

2

Negotiated price Other Total nonmarket (cost + other) methods Market price Adjusted market price Total market pricebased methods Total number of responses

U.S. MNEs with affiliates in all host countries

Per cent of total

Number

Transfer pricing method

U.S. MNEs with affiliates primarily in developed countries

Number

SOURCE: Al-Eryani et al. (1990, 420)

answered by 67 MNEs, 50 in manufacturing and trade and 17 in services, from 13 countries, inquired about their use of the CUP, RP, C+, and other methods. Over three-quarters of the firms used one pricing method, but another 7 per cent used two methods, and 16 per cent used three methods in combination. The most used method was cost plus - 42 per cent of the respondents used it as the only method, while as many as 65 per cent used it in conjunction with the resale price and CUP methods. None of the MNEs that responded to the questionnaire used profit comparisons to establish their transfer pricing policies, but six firms used profit splits. The results of this survey are summarized in Table 1.3.

An Introduction to the Issues

19

TABLE 1.3 The IBFD (1991) Survey of Transfer Pricing Methods

Transfer pricing method or combination of methods

Percentage of respondents using method 20 8 42 7 16 7

Comparable uncontrolled price (CUP) Resale price (RP) Cost plus (C+) Resale price and cost plus CUP, resale price, and cost plus Other methods Total

100

SOURCE: Hamaekers (1992, 605)

The Tang (1993) Study Roger Tang (1993, 271) performed a similar analysis of transfer pricing policies for domestic and international transfers, using a sample of 143 firms from the Fortune 500. His results are summarized in Table 1.4. In evaluating the methods used by respondents (noting that many of these firms use more than one domestic or international transfer price), Tang found that, for domestic transfers, 46.1 per cent used cost-based methods, 36.7 per cent used market-based methods, and 17 per cent utilized other methods. By contrast, for international transfers, 41.4 per cent used cost-based methods for pricing, 45.9 per cent used marketbased methods, and 12.7 per cent made use of other methods. Note that Tang also found that 13-17 per cent used a negotiated-price approach where the affiliates bargained to set the transfer price, a percentage similar to that of the previous study. From these four surveys, it is clear that MNEs use a variety of transfer pricing methods, based primarily on cost plus or market price with cost plus dominating as the preferred method. In about 15 per cent of the remaining cases, the two affiliates negotiated the transfer price. Since methods based on cost plus and/or market price appear to make economic sense, why the considerable attention paid by national tax authorities to transfer pricing? The Problem of Transfer Price Manipulation (TPM) Governments have developed sophisticated, complicated rules for valuing intrafirm transactions for tax purposes. The reason for these rules is not transfer pricing per se, but the fear of transfer price manipulation.

20 The Rules of the Game TABLE 1.4 The Tang (1993) Survey of Transfer Pricing Methods For domestic transfers

Pricing methods Variable cost Full cost Variable cost plus lump sum subsidy Full cost plus markup Other Subtotal % of total

Market price Other Subtotal % of total

Negotiated price Other methods % of total Total all methods

Number of firms

8 54 2 37 2

103

Per cent of total

82 36.7

37 1 17.0 223*

Number of firms

Per cent of total

Methods based on cost 3.6 2 24.2 17

10.8

0.9 16.6 0.9 46.2

1.3 26.8 1.3 41.4

46.1

56 26

For international transfers

2 42 2 65 41.4

Methods based on market price 25.1 41 11.6 31 36.7 72 45.9

16.6 0.5 100.0

Other methods 20 0 12.7 157*

1.2

26.1 19.8 45.9

12.7 0 100.0

*Many firms use more than one domestic or international transfer price. SOURCE: Tang (1993, 71)

It is important to distinguish between the terms 'transfer price' and 'transfer price manipulation.' Transfer pricing is a normal, legitimate, and, in fact, required activity. Firms set prices on intrafirm transactions for a variety of perfectly legal and rational internal reasons, and, even where pricing is not required for internal reasons, governments require it in order to determine how much tax revenues and customs duties are owed by the MNE. The image of giant MNEs manipulating millions of dollars of crossborder flows in order to evade or avoid payment of taxes and tariffs, on the other hand, is an image of transfer price manipulation. Transfer price manipulation is the deliberate setting of the price paid by one company to a corporate affiliate located in another taxing jurisdiction for the purpose of reducing the aggregate 'tax' burden of the company and its affiliates, where 'tax' is broadly defined as

An Introduction to the Issues

21

FIGURE 1.2 Intrafirm Transactions within the Multinational Enterprise

any external constraints on the MNE - e.g., taxes, tariffs, compulsory minority shareholders, quota regulations, and so on. Figure 1.2 illustrates some complexities that multinationals create for national tax authorities in terms of transfer pricing. For example, even a small MNE (call it USCO) with five foreign affiliates (e.g., in Canada, Latin America, Europe, Asia, Australia) has the possibility of 30 (six times five) different international, one-way channels through which intermediate and final products of the MNE can pass. If we classify products as goods, services, or intangibles, that brings the total number of channels by broad category up to 90 (30 times three). Most firms have dozens of product lines and each product requires hundreds or perhaps thousands of parts. We could similarly divide service and intangible flows into dozens of categories. International transactions can also occur with frequencies as low as once a year (e.g., dividend payments) to as high as several times an hour (e.g., foreign exchange transactions). Therefore the number of individual transactions within one MNE group can number in the millions per year. Regulating these transactions in any meaningful way at the national level is impossible because each government sees, and has control over, only part of the whole integrated business. No government has

22 The Rules of the Game BOX 1.1 External Motivations for Transfer Price Manipulation

Intrafirm trade

Type of transfer price manipulation

Motivation: The corporate income tax rate on affiliate A's income is higher than the tax on other MNE affiliates. A's imports and exports of goods

Overinvoice A's imports and underinvoice A's exports to shift profits to other MNE affiliates where tax rates are lower. Repatriated dividends If A is parent firm and A's government taxes only upon repatriation, delay dividend parent from affiliates to firm payments to A to avoid additional tax. If A is affiliate, deferring dividends can avoid the dividend-withholding tax. Head office services If A is parent, undercharge for head office to affiliates to services. If A is affiliate, overcharge for services as long as head office fees are deductible against host country's income tax. Technology transfers If A is parent, undercharge for transfers and defer royalty payments to parent. If A is to affiliates subsidiary, overcharge and speed up royalties paid to parent as long as royalties are deductible against host income tax. Motivation: A's government levies a tariff on imports. Imports

If ad valorem tariff, underinvoice imports.

Motivation: A's government subsidizes (taxes) exports. Exports

Over-invoice (underinvoice) exports if subsidy (tax) is ad valorem.

Motivation: Transactions in volatile currencies are subject to exchange rate risks. Payments for intrafirm Overprice and lead payments to shift profits transactions into a strong currency; underinvoice and lag payments to shift out of a weak currency. Motivation: The government forces the MNE to take on minority shareholders in A. All transactions

Minority shareholding acts as an effective tax on MNE profits earned in the affiliate because each dollar of profits must be shared. Use TPM to reduce the profits declared in A (see above methods for CIT).

An Introduction to the Issues

23

the funds to oversee all these transactions, nor would a benefit/cost analysis justify such a close examination. Many exogenous factors might cause the MNE to adjust its transfer prices. Some external reasons why MNEs might be motivated to engage in transfer price manipulation are outlined in Box 1.1, along with the type of manipulation (under- or overinvoicing, leading or lagging payments) that we might expect these firms to use. The external motivations include customs duties, export taxes/subsidies, differences in corporate income tax rates, and foreign exchange restrictions. Most of the strategies identified in Box 1.1 are methods to (1) shift taxable income into low-tax jurisdictions, and (2) shift tax-deductible costs into high-tax jurisdictions. Both income tax and customs duty officials need to be concerned about potential transfer price manipulation (see Figure 1.1). Customs officials worry that MNEs underinvoice inbound transfers so as to minimize the duties they pay on imported parts and finished goods. Tax authorities worry that MNEs overinvoice tax-deductible items (e.g., cost of inputs, service charges) and underinvoice income receipts (e.g., from sales of goods, service fees, royalties, and dividend income) so as to avoid paying corporate income taxes. Governments worry about transfer price manipulation because they are concerned with the loss of revenues through tax avoidance and/or evasion, and they dislike the loss of control this implies. Overall MNE profits after taxes may be raised by either under- or overinvoicing the transfer price; such manipulation for tax purposes, however, comes at the expense of distorting other goals of the firm, in particular, evaluating management performance. Thus taxes and tariffs are only some of the variables that influence the transfer pricing policies of MNEs; the MNE must also pay attention to its internal constraints. An example of how MNEs can use transfer price manipulation to reduce corporate income tax payments is provided in Box 1.2 and illustrated in Figure 1.3. Assume USCO, a U.S. multinational, makes 100 widgets at a per-unit cost of $2.00. The manufacturer charges cost plus 65 per cent, or $3.30 per unit, as the transfer price on intrafirm sales to its affiliate MEXCO. MEXCO markets and distributes the widgets in the Mexican domestic market, incurring costs of $1.00 per widget. Total manufacturing and distribution costs per widget are $4.30. (We ignore customs, insurance, and freight costs for simplicity.) MEXCO sells the widgets to consumers for $5.00 each. USCO makes a pre-tax profit of $130; MEXCO of $70. Assuming the effective tax rate on USCO's profits is 50 per cent, and on MEXCO's profits is 30 per cent, we calculate the total tax paid as $65 by USCO and $21 by MEXCO, for total MNE taxes of $86. USCO is left with $65 in after-tax profits, MEXCO with $49, for total after-tax MNE profits of $ 114.

24 The Rules of the Game BOX 1.2 Transfer Price Manipulation Reduces Overall Taxes The initial situation Assume USCO manufactures widgets and charges standard cost plus a 65 per cent markup on its sales of widgets to its affiliate MEXCO. MEXCO markets and distributes the widgets for final sale at a price of $5.00 per unit. USCO's profits are taxed at 50 per cent; MEXCO's profits at 30 per cent. (We ignore tariffs and transport costs.) Output Per-unit cost Unit cost of imports

USCO 100 2.00

MEXCO

100 1.00 3.30

The MNE 100 3.00

Total cost Price Total sales

200.00 3.30 330.00

430.00 5.00 500.00

300.00 5.00 500.00

Profit before tax Tax paid

130.00 65.00

70.00 21.00

200.00 86.00

65.00

49.00

114.00

Profit after tax

The MNE reduces the transfer price Assume the price is $2.20 (i.e., a 10 per cent markup). All other conditions are unchanged. USCO 100 2.00

MEXCO 100 1.00 2.20

THE MNE 100 3.00

200.00 2.20 220.00

320.00 5.00 500.00

300.00 5.00 500.00

Profit before tax Tax paid

20.00 10.00

180.00 54.00

200.00 64.00

Profit after tax

10.00

126.00

136.00

Output Per unit cost Unit cost of imports Total cost Price Total sales

Conclusion: Underinvoicing raises MNE after-tax profits Where the tax rate on USCO is higher than on MEXCO, the MNE receives larger global profits, net of tax, by underinvoicing USCO's exports to MEXCO. Shifting profits to the lower-taxed firm causes a drop in profits for exporter, but the MNE as a whole is better off.

An Introduction to the Issues

25

FIGURE 1.3 Transfer Price Manipulation Raises After-Tax Profits for the Multinational

Now assume the multinational can underinvoice the product USCO sells to MEXCO, setting a mark-up of 10 per cent, making the new price $2.20 instead of $3.30. Working through the example again, we find that pre-tax profits of the manufacturer fall (to $20 from $130) while those of the distributor rise (to $180 from $70). USCO pays less tax ($10 versus $65) while MEXCO's tax payments

26

The Rules of the Game

rise (to $54 from $21). On the other hand, for the multinational as a whole, pretax profits are unchanged (at $200);8 tax payments are lower ($64 versus $86) and after-tax total MNE profits are higher ($136 instead of $114). The pre-tax and post-tax profits of the two firms and the MNE as a whole are illustrated in Figure 1.3. Note that the difference between the two circles represents taxes paid. By understating profits in the high-tax country and overstating them in the low-tax country, the MNE realizes greater overall after-tax profits. Thus transfer price manipulation can be an effective way to avoid paying taxes where tax differentials exist between jurisdictions. Clearly, the USCOMEXCO case outlined above is a very simple example. In practice, large MNEs engage in thousands of transactions in goods, services, and intangibles with affiliates in dozens of countries each year. The potential avenues for tax avoidance are therefore myriad. In order to reduce the likelihood of transfer price manipulation, governments have responded to this possible source of revenue drain through ever-increasing regulations. These regulations have centred on a concept known as 'the arm's length standard.' In the next section, we turn from transfer pricing as seen by the multinational, to transfer pricing as regulated by the tax authorities. National Regulation of Transfer Pricing In this section, we look at government attempts to regulate MNE transfer prices and to resolve transfer pricing disputes at the national and international levels. Underpinning these tax regulations is the concept of the arm's length standard, so we first explain this standard. We then provide short outlines of transfer pricing regulation in the United States and in Canada. The U.S. regulations on tax transfer pricing are the oldest and most detailed regulations in the world. They are also the most complicated. The Canadian rules are based on the U.S. ones but are shorter and simpler. We provide a brief summary of the historical development of these rules below. The Arm's Length Standard The most common solution that tax authorities have adopted to reduce the probability of the transfer price manipulation is to develop particular transfer pricing regulations as part of the corporate income tax code.9 These regulations (e.g., U.S. Internal Revenue Code section 482, the Canadian Income Tax Code section 69, the OECD transfer pricing reports) are generally based on the concept of the arm's length standard, which says that all MNE intracorporate activities should be priced as if they took place between unrelated parties acting at arm's

An Introduction to the Issues

27

FIGURE 1.4 The Arm's Length Standard METHOD # 1 Use the price negotiated between two unrelated parties C and D to proxy for the transfer price beteen A and B.

METHOD # 2 Use the price at which A sells to unrelated party C to proxy for the transfer price between A and B.

Both methods require either (1) the same product be transferred under the same circumstances or (2) adjustments to be made to quantify these differences.

length in competitive markets. The arm's length price is the price two unrelated parties would reach through bargaining in a competitive market. The 1979 OECD Report defines the arm's length standard (ALS) as prices which would have been agreed upon between unrelated parties engaged in the same or similar transactions under the same or similar conditions in the open market. (OECD 1979, 7) The arm's length standard asks the following question: What would the parties have done had they been unrelated? What price would they have negotiated? Since the firms are related in the transaction under scrutiny by the tax authorities, any answer to this question must be hypothetical. The best answer is a proxy, done in one of two ways (see Figure 1.4). In the first method, the price negotiated by two other unrelated parties which were engaged in a comparable transaction under comparable circumstances is a

28

The Rules of the Game

proxy for the arm's length price in the transaction in question. In this case, the regulator looks for two other firms, unrelated and engaged in similar activities as the related parties in question, and then uses the price negotiated by the unrelated firms, adjusted if necessary for differences in product and functional characteristics, as the arm's length price. As Figure 1.4 shows, the arm's length price negotiated between firms C and D is used to proxy for the transfer price between the related firms A and B. In the second method, the price set by one of the related parties in a comparable transaction under comparable circumstances with an unrelated party could be used as an estimate. Where the MNE either buys outside or sells outside, in comparable circumstances (e.g., product characteristics, functional level, time horizon, risks taken), the price negotiated with unrelated parties can be used as the arm's length price. That is, in Figure 1.4, the arm's length price that A charges unrelated firm C is used to proxy for the transfer price that A charges related firm B. In practice, the method used will depend on the available data. Are there unrelated parties engaged in the same, or nearly the same, transactions under the same, or nearly the same, circumstances? Does one of the related parties also engage in the same, or nearly the same, transactions with an unrelated party under the same, or nearly the same, circumstances? Where there are differences, are they quantifiable? Do the results seem reasonable in the circumstances? If the answers to these questions is yes, then the arm's length standard will yield a reasonable result. If the answer is no, then alternative methods must be used. We provide an illustration of the arm's length standard as it was used in an actual court case: J. Hofert vs. the Minister of National Revenue (DTC 1962) in Appendix 1.1 at the end of this chapter. The concept of the arm's length standard was first developed in, and then refined by, the U.S. Treasury. The Canadian government also follows this standard in its transfer pricing rules. In the two sections below, we briefly look at how these two governments have adopted and used the ALS in pricing MNE transactions for purposes of determining corporate income taxes. The U.S. Approach to Transfer Pricing Regulation As the principal source of outward-bound foreign direct investment (FDI), and also, since the early 1980s, the key destination for inward foreign direct investment, the United States has well-developed policies for taxing multinationals. The biggest of the national tax authorities is, of course, the U.S. Internal Revenue Service (IRS). The IRS has the world's largest staff of specially trained international examiners and economists located in an International Enforcement

An Introduction to the Issues

29

Division for the auditing of MNEs. This staff is equipped with a range of policies designed to reduce and penalize MNE that engage in transfer pricing manipulation. In the United States, transfer pricing law is developed in the U.S. Treasury, passed by the U.S. Congress, interpreted and applied by the Internal Revenue Service (IRS, or the Service), and interpreted by the U.S. tax courts. The keystone of the U.S. approach to tax transfer pricing is section 482 of the U.S. Internal Revenue Code (IRC), first passed in 1917 and broadened in 1928, which applies to all intracorporate transfers, both tangible and intangible. In the legislation, the IRS Commissioner has the right to reallocate income and deductions between related parties in order to prevent tax avoidance and to determine the true taxable income of each party. Section 482 is responsible for ensuring that the income earned on transactions between related parties is determined on an arm's length standard. In 1968, the U.S. Treasury developed its first set of regulations on section 482. IRS auditors were to evaluate intrafirm transactions using these regulations, and multinationals were encouraged to follow them in pricing their own transactions. The regulations specify various types of transactions: loans, rentals or sales of tangible property (i.e., goods); transfer or use of intangible property (e.g., patents, copyrights); and performance of services (e.g., managerial, technical). Sales of tangible property are tested against an arm's length standard based on one of four methods (in order of priority): comparable uncontrolled price (CUP), resale price (RP), cost plus (C+), and so-called fourth or other, methods. The CUP, RP, and C+ methods are transactions-based methods that look for comparable transactions between unrelated parties in order to proxy for the related party transaction. The most serious problem associated with section 482 has been the lack of comparables, making the CUP, RP, and C+ methods difficult to use in practice, and necessitating the use of fourth methods. This problem was accentuated when non-U.S. MNEs were involved since information was often less readily available than for U.S. multinationals. In addition, U.S. law historically encouraged the offshore, below-cost transfer of intangibles by U.S. parents to their foreign affiliates, even though such underinvoicing was directly in conflict with the spirit of section 482. As a result, since the early 1960s, transfer pricing regulation has been an acrimonious area of U.S. tax law with dozens of tax court cases, many dragging on for up to a decade and more through the court process. In order to address these problems, starting in the early 1980s, the U.S. Treasury has engaged in major, and frequent, revisions to its transfer pricing regulations. In 1986, the U.S. Congress passed a law requiring that transfers of

30 The Rules of the Game intangibles be priced commensurate with the income (CWI) from the intangibles. Since then, the U.S. Treasury has worked on integrating the CWI standard into the 482 regulations. Over the 1992-4 period three versions of new transfer pricing regulations were introduced; the final ones were approved in June 1994. In the final 482 regulations, the number of specified methods have increased by two - the comparable profits method (CPM) and the profit split (PS) method', the hierarchy of methods has been eliminated, and instead taxpayers are supposed to select the best method in terms of the facts and circumstances of the case; and periodic adjustments (i.e. re-evaluations) of intangible prices will be made, with certain exceptions, to ensure that the CWI standard is satisfied. Taxpayers are expected to use functional analysis (an economic evaluation of the activities, responsibilities, resources, and risks of each of the related parties) to explain their transfer pricing policy. The procedures used by the IRS to handle transfer pricing disputes are also changing. A new Advance Pricing Agreement procedure was introduced in 1991 whereby a taxpayer and the IRS negotiate an agreed transfer pricing methodology that is binding on both parties for a specified time period, generally three years. In 1994, the Service and Apple Computer first used binding arbitration to settle their transfer pricing dispute rather than going to the tax courts. Both parties were happy with the outcome and the method is likely to be used by other MNEs. In addition, new penalty regulations for misvaluations (section 6662) were added to the Internal Revenue Code in order to ensure MNE compliance with the new section 482 rules. The traditional bilateral approach has been through competent authority provisions of bilateral tax treaties that bring the two tax authorities together to settle transfer pricing disputes. Thus, over the past ten years (1986-96), the United States has been engaged in reforming its tax transfer pricing regulations. The new rules are finally in place and the U.S. Treasury is unlikely to engage in such major reform again for quite some time. Therefore it is a good time to stand back and look at these changes and assess the new regime in place in the United States. We provide a critical, historical review of the U.S. approach to tax transfer pricing in Chapter 8 (the U.S. rules), Chapter 9 (the U.S. procedures), a more general assessment in chapters 12 and 13 (reforming the tax transfer pricing regime), and some specific suggestions for reform in Chapter 14 (conclusions and policy recommendations). The Canadian Approach to Transfer Price Regulation In Canada, tax transfer pricing law is written by the Department of Finance, passed by Parliament, interpreted and implemented by Revenue Canada, and

An Introduction to the Issues

31

interpreted by the Canadian tax courts. The Canadian transfer pricing legislation, section 69 of the Income Tax Act, was first passed in 1972. The section is in three parts. Section 69(1) applies a fair market value criterion to the arm's length criterion for intrafirm domestic transactions. This section is designed to prevent related firms within Canada from artificially shifting income and/or deductions among their divisions. Sections 69(2) and 69(3) apply to intrafirm international transactions and use the 'reasonable under the circumstances' approach as the criterion for ensuring arm's length transactions. Section 69(2) insists that intracorporate crossborder payments not exceed a reasonable amount, whereas section 69(3) insists that such receipts be not less than a reasonable amount. In 1987, Revenue Canada adopted Information Circular 87-2 designed to clarify the Canadian approach to tax transfer pricing. The Canadian regulations also apply the same four methods as the pre-1994 U.S. regulations: CUP, resale price, cost plus, and fourth methods, with CUP having priority. Rules are developed for transfers of tangibles, business services, and intangibles that roughly follow the U.S. approach. The key test in the Canadian rules is whether the MNE's transfer price for a particular transaction was reasonable given all the facts and circumstances. Unlike the United States, very few court cases in Canada have focused specifically on transfer pricing issues; most cases have involved tax havens and pricing of tangibles where the paperwork was shunted through a tax haven so as to move the profits offshore to a low-tax jurisdiction. We examine some of these court cases in Chapter 11. While the U.S. Treasury engaged in the overhaul of its transfer pricing regulations between 1986 and 1994, Revenue Canada and the Department of Finance have watched the U.S. upheaval and done little to modify Canadian rules. The only significant pronouncement has been a January 1994 news release, clarifying for Canadian taxpayers that they should follow Canadian tax law and use the competent-authority process under the Canada-U.S. bilateral tax treaty in which differences in Canadian and U.S. law lead to double taxation. In addition, Canada has developed its own Advance Pricing Agreement procedure, the final version should be available in early 1997. Now that the United States has completed the overhaul of rules and procedures, it is perhaps a good time to evaluate the Canadian approach to tax transfer pricing and to suggest where the rules might need changing. Chapter 10 reviews the history of transfer price regulation in Canada, while Chapter 14 offers some proposals for change. We turn now to international attempts to develop a common regulatory framework for taxing multinationals.

32 The Rules of the Game The International Tax Transfer Pricing Regime In this section we provide a brief history of government efforts at the international level to develop a set of rules and procedures to guide tax authorities and MNEs. We argue that an international regime has developed whereby tax authorities have attempted to establish certain principles and norms centred around the arm's length standard in order to reduce international taxation disputes. We outline the structure of this regime, and compare it briefly with an alternative approach: unitary taxation. Multilateral Solutions: A Brief History Historically, the most common international solution has been the bilateral tax treaty. In a tax treaty, two governments spell out which one has jurisdiction over what tax base and how the tax base is to be measured and allocated. By signing tax treaties with close trading and investment partners, two countries could better regulate their crossborder transactions and provide a more secure legal environment for crossborder investments. Beginning in the 1920s, tax authorities started to develop a set of international principles for tax treaties which were designed to reduce the probability of interjurisdictional conflict. The key idea behind the principles was the need to prevent both undertaxation and double taxation of MNE income. First through the League of Nations and then through the United Nations and the OECD, groups of tax experts have developed a set of principles in the form of model tax conventions to guide national taxation of multinationals and the bilateral tax treaty process. Three principles underpin these conventions: inter-nation equity (tax revenues should be allocated fairly between jurisdictions), international neutrality (taxes should not interfere with private decisions), and international taxpayer equity (taxpayers in the same jurisdiction should be treated equally regardless of the source of their income). In the transfer pricing area, these principles are embodied in the international norm, the arm's length standard. Thus the ALS, first developed in the United States, has become the benchmark for pricing intrafirm transactions. The first model tax conventions incorporating these principles and norms appeared in the late 1940s. Since the late 1960s, when the United States first set out its regulations on applying IRC section 482, the international tax community - tax authorities, lawyers, and public finance economists - has been involved in developing a set of rules and procedures designed to specify how different international intrafirm transactions should be priced so as to satisfy the arm's

An Introduction to the Issues

33

length standard. These guidelines have given us the comparable uncontrolled price (CUP), cost plus, and resale price methods. Guidelines have appeared in the form of OECD and UN model tax conventions, general guidelines on MNE-state relations, and transfer pricing guidelines, all centred on the principles of international equity and neutrality and on the norm of the arm's length standard. We argue in this book that the national approach to tax transfer pricing has expanded into an international regulatory network. Led by the OECD, this regulatory network has become much more sophisticated in its approach to taxing multinationals, and can now be described as an international regime, the international tax regime. Nested within this regime is another: the international tax transfer pricing regime - the true subject of our book. What Are International Regimes? Problems of interdependence at the international level can perhaps best be handled through international regimes, a form of international governance structure (Krasner 1983; Preston and Windsor 1992). International regimes are institutions, sets of functional and behavioural relationships among national governments. These relationships embody the principles underlying the regime, the expected behaviour patterns of regime members, and the formal arrangements that implement the international agreements and understandings that form the regime. Regimes are useful as a way to reduce international transactions costs in an interdependent world. When a clear legal framework establishing property rights and liability is missing, markets for information are imperfect, and/or incentives exist for actors to behave opportunistically, regimes can improve the functioning of international markets. International regimes can increase the predictability of behaviour, provide generalized sets of rules, and improve the information available to participants. Thus regimes are ways to manage interdependencies among nations. The Current Tax Transfer Pricing Regime The problems created for governments by the global reach of MNEs make it impossible to regulate these large firms effectively at the domestic level. Either MNE income goes untaxed or double taxation occurs. Both events cause conflict: conflict between MNEs and nation-states over who pays what tax to whom and how much, or conflict between tax authorities in different states over their fair share of MNE income. Therefore making regulations on transfer pricing at the national level to meet national goals is an unsatisfactory approach to

34 The Rules of the Game an international problem, and the need for multilateral approach to taxing multinationals is clear. The domestic reach of national jurisdictions is ill-suited to regulate the global reach of the multinational enterprise, and the number of interjurisdictional tax conflicts is growing along with the increase in MNE crossborder activity. We contend that there is an international tax regime with principles, norms, rules, and procedures designed to facilitate cooperation between national tax authorities in order to better regulate crossborder taxable activities of multinationals. The goals of the regime are the avoidance of double taxation of income and the prevention of tax avoidance and evasion. These goals are to be achieved through coordination and harmonization of national tax systems. Examples of government cooperation in the tax area that form components of the tax regime include a variety of national tax policies, bilateral tax treaties (BTTs), and model treaties and guidelines developed by institutions such as the OECD and United Nations. The international tax regime deals with both jurisdictional issues (who has the right to tax what) and allocational issues (how should costs and revenues be allocated and priced). Within the international tax regime is nested another regime dealing with the taxation of intrafirm trade. The international tax transfer pricing (TTP) regime focuses on the international allocation of MNE income and expenses, specifically on the pricing of intrafirm trade flows within the various affiliates of the multinational enterprise. Government cooperation in the transfer pricing area is based on a variety of national tax policies, BTTs, and model treaties and guidelines developed by institutions such as the OECD and the United Nations. International bodies of experts such as the OECD's Committee on Fiscal Affairs and the International Fiscal Association (IFA) have played important roles in developing international policies and norms. We argue that the combination of these behaviours and functional relations can be seen as constituting an international tax transfer pricing regime. At the core of the regime is the OECD's Committee on Fiscal Affairs and the OECD Model Tax Treaty. The treaty incorporates an arm's length standard for allocating income between firms and their subsidiaries, parents, or sister enterprises. Each unit of the MNE is expected to declare, for tax purposes, the profits which it would have made had it been a distinct and separate enterprise operating at arm's length from its parent and sister affiliates.10 We view U.S. tax law and the OECD Model Tax Treaty rules with respect to transfer pricing as central components of the TTP regime. Most members of the OECD adhere to the arm's length standard and have developed regulations loosely based on either the U.S. regulations or the OECD Model Tax Treaty

An Introduction to the Issues

35

sections. We argue that this regime has its own norm (the arm's length standard), principles (international equity and neutrality), rules (various methods for valuing intrafirm trade), and procedures (e.g., competent authority rules, advance pricing arrangements, appeals, and arbitration). In Chapter 2 we develop the concept of the tax transfer pricing regime and document the ways in which the OECD, United Nations, and the U.S. Treasury have influenced the development of the regime. We focus in particular on the model tax treaty and 1979 and 1984 transfer pricing reports, as developed by the OECD's Committee on Fiscal Affairs. The appendix to Chapter 2 summarizes the various international publications on tax transfer pricing. In 1992, the OECD released a new model tax treaty. In addition, the Committee on Fiscal Affairs has engaged in a major overhaul of the OECD's earlier transfer pricing reports (OECD 1994b, 1995a,b, 1996). In chapters 5 and 13 we examine these revisions and make suggestions for improving their effectiveness as part of the international tax transfer pricing regime. An Alternative Approach: Unitary Taxation As part of our analysis of the tax transfer pricing regime it is important to look at alternatives to the arm's length standard that underpins the regime. In particular, there is one major alternative that has been proposed: to replace the arm's length standard with a formulary apportionment approach commonly referred to as 'unitary taxation.' The arm's length standard is based on the separate accounting or separate entity approach. The borders of a firm are defined according to national boundaries; this is known as the 'water's edge.' Domestic affiliates and foreign branches are consolidated with the parent firm for tax purposes, but foreign subsidiaries and other affiliates of the MNE are treated as separate firms. Income of the multinational is measured using separate accounting for the domestic and international units of the MNE. Since the parent's tax return is consolidated with its domestic affiliates and foreign branches, transfer prices for intrafirm transactions among these affiliated parts of the MNE are not required for tax purposes. However, intrafirm transactions between the parent and its foreign affiliates must be measured and accounted for. These transfer prices must be valued as if they occurred at arm's length - i.e., using the arm's length standard. Transfer price rules (CUP, C+, RP) are used to ensure that such transactions approximate the prices unrelated firms would choose in comparable circumstances. This is the standard that lies at the heart of the international tax transfer pricing regime, a standard endorsed by the OECD and followed by the Canadian and U.S. tax authorities.

36 The Rules of the Game A quite different approach is that of unitary taxation. Unitary taxation is taxation of the worldwide income of a unitary business - that is, the tax authority measures the income of all the related affiliates of a multinational enterprise that do business within the taxing jurisdiction, and then assesses tax according to the share of the worldwide business of the MNE that occurs within that particular jurisdiction. Unitary taxation is normally based on & formula apportionment method whereby a firm's share of certain factors (e.g., sales, labour costs, and capital costs), as a percentage of the worldwide MNE amount of these factors however weighted, is multiplied by the MNE's total worldwide income to compute the tax to be paid in that jurisdiction. Unitary taxation has been little used in practice. The U.S. states and the Canadian provinces use formulary methods to allocate domestic subfederal corporate tax revenues among themselves. In addition, a few of the U.S. states, in particular California, have taxed firms located in their jurisdiction on a. pro rata share of the worldwide income generated by the MNE corporate group. Most recently, the IRS has signed several advance pricing agreements with international banks, using a formulary approach to allocate the income from global trading. The OECD dislikes and has actively discouraged the use of global formulary methods on the grounds that they are arbitrary and do not satisfy the norm of the arm's length standard. While California and some other states use the method, the U.S. government has also discouraged its spread. The Canadian government is also on record as opposing to unitary taxation. We discuss unitary taxation in several places in this book. Chapter 2 discusses the OECD's views on unitary taxation. In Chapter 6 we explain the economic effects of unitary taxation in a situation in which one government uses, the method and another does not, and one in which all governments follow this approach. In Chapter 12 we discuss the pros and cons of unitary taxation and separate accounting, and evaluate two U.S. experiences with unitary taxation: the recent Barclays Bank case, and the global trading APAs. Finally, in Chapter 12 we suggest that formulary apportionment could be used for tax purposes for North American multinationals that derive most of their income within the North American Free Trade Area. It is clear from our study that the approach, long vilified by both multinationals and tax authorities, is slowly spreading throughout the OECD community; we expect it to spread more quickly within regional blocs such as NAFTA and the European Community in the coming years. Explaining the Transfer Pricing Methods Having briefly reviewed the history of U.S., Canadian, and international regula-

An Introduction to the Issues

37

tions on how to price intrafirm transactions, in this section we look at the five major methods now in place in the U.S. and Canadian regulations and the OECD's new guidelines. Our purpose is explain how each of the major methods works, apply it to a numerical example, and point out some of the strengths and weaknesses. In subsequent chapters we examine and assess these methods in much greater detail in their international and national contexts. Here we simply want to set the stage for what follows. In the past, the OECD rules, and Canadian and U.S. practice, recommended that tax authorities use one of three methods to price intrafirm transactions: comparable uncontrolled price (CUP), resale price (RP), or cost plus (C+). Where none of these specified methods applied, the income tax auditor turned to fourth (other) methods. The new U.S. transfer pricing regulations detail two alternatives: the profit split (PS) method and the comparable profits method (CPM). In the new rules, the hierarchy of methods has been eliminated and the tax authority is supposed to apply the best method. In practice, we suspect that the de facto ranking is likely to remain, given the OECD and IRS commitment to using transactions-based (CUP, C+, RP) methods over profit-based (CPM, PS) methods. The Canadian and U.S. regulations spell out how the tax auditor is supposed to apply these methods to estimate the arm's length price. The regulations are also meant as a guide for taxpayers - i.e., the multinational is supposed to also set its transfer pricing policy for different transactions using one of these methods. Thus, the tax authorities have attempted to make this trio of methods (CUP, RP, C+) not only the regulator's chosen transfer pricing methods but also the multinational's transfer pricing methods. We examine each of these methods briefly below. For those readers who may not be familiar with the accounting terms that will be used in these examples a glossary of terms can be found at the end of the book. In addition, a simple income statement for a business can be found at the end of this chapter in Appendix 1.2, together with a list of some of the differences between economic and accounting terms in Appendix 1.3. The Comparable Uncontrolled Price (CUP) Method The CUP method looks for a comparable product to the transaction in question, either in terms of the same product being bought or sold by the MNE in a comparable transaction with an unrelated party, or the same or similar product being traded between two unrelated parties under the same or similar circumstances. The product so identified is called a product comparable. All the facts and circumstances that could materially affect the price must be considered - e.g., the

38

The Rules of the Game

BOX 1.3 The Comparable Uncontrolled Price (CUP) Method

CANCO sells television sets directly to its U.S. subsidiary USCO. CANCO and other Canadian firms also sell TV sets in the United States to unrelated parties through commission sales agents. By custom, the product is sold FOB (free on board; i.e., without freight or insurance added) from the purchaser's plant. An average U.S. transaction price, based on sales by commission agents, is available from these agents. The transfer price per television set is calculated as follows: Average retail price in the United States

$ 500.00

MINUS Adjustment for saving the agent's commission (5 per cent of the transaction price)

25.00

Freight adjustment (amount reflected in average daily transaction price less actual cost)

30.00

Total deductions Transfer price using the CUP method

55.00 $ 445.00

characteristics of the product, the market location, the trade level of the firms, and the risks involved. Adjustments are made to the external price to more closely estimate the arm's length price. Box 1.3 provides a numerical example of the CUP method, which is illustrated in Figure 1.5. In this example, CANCO, a Canadian manufacturer of television sets, sells TVs both inside the MNE (to its U.S. subsidiary) and outside the MNE (to unrelated firms in the United States). The average external market price ($500), adjusted for the trade level (the agent's commission of five per cent) and for transport costs (the freight adjustment, $30), is used to calculate an FOB transfer price of $445 per unit. This is an example of the second method for determining an arm's length price (see Figure 1.4). Where the MNE sells the same product under the same

An Introduction to the Issues

39

FIGURE 1.5 The Comparable Uncontrolled Price (CUP) Method

circumstances both inside and outside the enterprise, the outside price can proxy for the transfer price. In this case, the products are the same but the circumstances are slightly different (i.e., there is a sales agent and freight costs are incurred in the outside sales), so some adjustment is required to find the correct price. Tax authorities prefer the CUP method over all other pricing methods for at least two reasons. First, it incorporates more information about the specific transaction than does any other method; i.e. it is transaction and product specific. Since the arm's length standard is a transact!onal approach to valuing the MNE, the best method is the one that focuses most closely on the product and the transaction under consideration. Second, CUP takes both the interests of the buyer and seller into account since it looks at the price as determined by the intersection of demand and supply. The method assumes two firms are willing to bargain and that the comparable uncontrolled price is the outcome of that bargaining.

40 The Rules of the Game

BOX 1.4 The Resale Price Method

CANCO is the Canadian distributor for its British parent's established line of automobiles. Comparable independent distributors in Canada earn profit margins of eight per cent. CANCO performs extra advertising and warranty services not normally provided by these distributors. The transfer price to CANCO for a particular automobile is calculated as follows: Final retail price in Canada

$ 20,000.00

MINUS Margin earned by comparable Canadian distributors (8 per cent off the retail price) Allowance for expenses borne by CANCO not normally borne by comparable independent distributors -advertising - warranty work Total deductions Transfer price using the resale price method

1,600.00

100.00 600.00 2,300.00 $ 17,700.00

The Resale Price (RP) Method Where a product comparable is not available, so that the CUP method cannot be used, an alternative method is to focus on one side of the transaction, either the manufacturer or the distributor, and to estimate the transfer price using a functional approach. Under the resale price method, the tax auditor looks for firms at similar trade levels that perform similar distribution functions (i.e., a functional comparable). The RP method is best used when the distributor adds relatively little value to the product so that the value of its functions is easier to estimate. The assumption behind the RP method is that competition among distributors means that similar margins (returns) on sales are earned for similar functions. 11 A distributor is likely to charge the same or a similar sales margin for carrying TV sets as for carrying washing machines or other white goods.

An Introduction to the Issues

41

FIGURE 1.6 The Resale Price (RP) Method

Given a large number of distributors, averaging over these unrelated firms can be used to proxy for the margin that the distribution affiliate would have earned in an arm's length transaction. Subtracting this margin from the retail price (the price to the consumer, which is known), one can estimate the transfer price. In Box 1.4 we give an example of the resale price method for the case of a Canadian distributor of British-made cars. The example is illustrated in Figure 1.6. We assume the U.K. parent sells directly to its Canadian subsidiary. CANCO has the sole distribution rights in Canada for these autos, which retail for $20,000. The question is the transfer price that UKCO charges CANCO. The tax authority knows that the profit margins earned by independent Canadian distributors of automobiles average about eight per cent; however, the Canadian affiliate incurs advertising and warranty costs of $700 that are not normally borne by independent distributors. Subtracting the eight per cent discount from the retail price, and then adding in an adjustment for additional costs incurred by CANCO, yields a transfer price of $17,700, using the RP method.12

42 The Rules of the Game Thus the resale price method 'backs into' the transfer price by subtracting a profit margin, derived from margins earned by comparable distributors engaged in comparable functions, from the known retail price to determine the transfer price. As a result, the RP method evaluates the transaction only in terms of the buyer. The method ensures that the buyer receives an arm's length return consistent with returns earned by similar firms engaged in similar transactions. Since the resale margin is determined in an arm's length manner, but nothing is done to ensure that the manufacturer's profit margin is consistent with margins earned by other manufacturers, the adjustment is one-sided. Under the RP method, having determined the buyer's arm's length margin, all excess profit on the transaction is assigned to the seller. Thus the resale price method tends to overestimate the transfer price since it gives all unallocated profits on the transaction to the upstream manufacturer. We can call this the contract distributor case since, effectively, the manufacturer is contracting out the distribution stage to the lowest bidder. The Cost Plus (C+) Method In the cost plus method, the tax auditor looks at the other side of the transaction: the manufacturer. The method starts with the costs of production, measured using recognized accounting principles, and then adds an appropriate mark-up over costs. The appropriate mark-up is estimated from those earned by similar manufacturers. The assumption is that in a competitive market the percentage mark-ups over cost that could be earned by other arm's length manufacturers would be roughly the same.13 Thus, this method is also a functional comparable like the RP method. The cost plus method works best when the producer is a simple manufacturer without complicated activities so that its costs and returns can be more easily estimated. Box 1.5 gives an example of a perfume manufacturer, CANCO, that manufactures perfume for itself and three sister affiliates at a standard cost of $4.40 per ounce. This is illustrated in Figure 1.7. The formulations for the foreign affiliates are customized for tastes in each market; customizing normally adds an additional five per cent over standard cost. Since other perfume manufacturers in Canada prepare bulk formulations for a mark-up over standard cost of 20 per cent, this mark-up is taken as an estimate of the arm's length mark-up that should be earned by the manufacturer. Adding the mark-up to standard cost, together with the cost of the ingredients and the additional cost of customizing, gives a transfer price of $7.50 using the cost plus method.14 In order to use the cost plus method, the tax authority or MNE must know the accounting approach adopted by the unrelated parties. For example, what costs

An Introduction to the Issues

43

BOX 1.5 The Cost Plus Method

CANCO, a wholly owned subsidiary of a French perfume multinational, produces an expensive perfume for sale in Canada using active ingredients purchased at arm's length. The active ingredients cost $2.00 per ounce of perfume; CANCO's standard manufacturing cost is $4.40. The firm also does custom formulations for other affiliates of its French parent. The industry average mark-up for bulk formulations performed by other perfume manufacturers in Canada is 20 per cent above standard cost. Custom formulations normally add an additional 5 per cent over standard cost. The transfer price per ounce of perfume for a particular shipment by CANCO to one of the foreign affiliates is calculated as follows: CANCO standard cost per ounce (excluding active ingredient costs) ADD Cost of active ingredients

$ 4.40 2.00

Mark-up received by functionally comparable manufacturers in Canada 20 per cent of standard cost

0.88

Additional cost of preparing custom formulation for the affiliates 5 per cent of standard cost

0.22

Total additions Transfer price using the cost plus method

3.10 $ 7.50

are included in the cost base before the mark-up over costs is calculated? Is it actual cost or standard cost (costs which have been standardized for cyclical fluctuations in production as in the example in Box 1.5)? Are only manufacturing costs (cost of goods sold, which includes labour, overhead costs, including depreciation, and material input costs) included or is the cost base the sum of manufacturing costs plus some portion of operating costs (i.e., selling, general, and administrative (SG&A) expenses and R&D costs)? The larger the cost base (i.e., the more items put below the line and thus into the cost base), the smaller should be the profit mark-up, or gross margin, over costs.

44 The Rules of the Game FIGURE 1.7 The Cost Plus (C+) Method

As a one-sided method, the cost plus method focuses only on the profit markup of the seller and insists that the seller should earn only what arm's length sellers engaging in similar transactions would earn in a competitive market. Therefore the C+ method tends to underestimate the transfer price because it gives all unallocated profits from the transaction to the buyer. This argument is generally known as the contract manufacturer case where the transfer price is set such that the manufacturer earns only costs plus a small mark-up with the majority of profits going to the downstream firm. In sum, the product comparables method (CUP) is the preferred transactional method for determining the transfer price. Where it cannot be used, functional comparables methods (RP, C+) are the second choice. Historically, the U.S. regulations used this hierarchical approach: CUP first, RP second, and C+ third. In the 1994 final section 482 regulations, the hierarchy of methods was abandoned in favour of the best method rule (i.e., use the 'best method' for the facts and circumstances of the case). In Canada, as in the OECD transfer pricing reports, the resale price and cost plus methods are given the same (second place) priority after CUP. The disadvantage of both the RP and the C+ methods, vis-a-vis

An Introduction to the Issues

45

CUP, as we have shown, is that they only focus on one side of the transaction, either that of the seller or the buyer. The Profit Split (PS) Method Where none of the three basic transfer pricing methods can be applied, either because there are no suitable product comparables (the CUP method) or functional comparables (the RP and C+ methods), generally the regulations suggest the use of fourth/other methods. The most common other method in practice has been the profit split (PS) method, whereby the profits on a transaction earned by two related parties are split between the parties. The profit split method allocates the consolidated profit from a transaction, or group of transactions, between the related parties. Where there are no comparables that can be used to estimate the transfer price, this method provides an alternative way to calculate or 'back into' the transfer price. Various ratios can be used to split the profits on the transaction between the related parties; the most commonly recommended one is return on operating assets (the ratio of operating profits to operating assets). An example of the PS method, using return on operating assets to divide the profits, is provided in Box 1.6; see also Figure 1.8. In the example, the financial statements of two related firms are shown individually and on a consolidated basis (so that intrafirm transactions cancel out). Firm A, the manufacturer, produces and sells 100 lamps each time period to firm B at a transfer price of $1.50 per lamp, for total revenue of $150. A incurs cost of goods sold (COGS) or manufacturing cost of $120; this amount represents the costs of material inputs, labour, and overhead costs. After subtracting COGS from total sales, the firm earns gross profit of $30. In addition, the firm incurs operating expenses (SG&A and R&D costs) of $10, leaving it with $20 in operating profit. After subtracting cost of goods sold and operating expenses, the manufacturer makes an operating profit of 20 cents per lamp, which is a four per cent return on A's operating assets. Lastly, A has net interest expense on its debt of $3, for a final net profit or income of $17. The purchaser, firm B, sells the lamps for $2.00 per unit, giving it an operating profit after costs of 30 cents per lamp, for a return of two per cent on its operating assets. The consolidated operating profit of the MNE as a whole is 50 cents per lamp, with an average return on assets of 2.5 per cent. Note that firm A has only one-third the operating assets of firm B, yet it receives a higher return than B. If return on operating assets is used to divide the overall profit of $50, then one-quarter of the operating profits should go to A and three-quarters to B. Thus A's profit should be $12.50 and B's should be

46 The Rules of the Game BOX 1.6 The Profit Split Method (Based on Return on Operating Assets) Assume firms A and B are related. Each period, A makes and sells 100 lamps at a transfer price of $1.50 to B, and B distributes and sells the lamps to consumers at a price of $2.00. A's and B's financial statements are reproduced below: Financial statement

Firm A

Firm B

Consolidated

Quantity of lamps Selling price

100 $1.50

100 $2.00

100 $2.00

Total sales revenue Cost of goods sold

$ 150 120

$ 200 150

$ 200 120

Gross profit Operating expenses

30 10

50 20

80 30

Operating profit N e t income exp

$ 20 3

$ 30 2

$ 50 5

Net income

$ 17

$ 28

$ 45

Operating assets

$ 500

$ 1,500

$ 2,000

Rate of return on assets (%) (ROA)

4.0%

2.0%

2.5%

Note that A's sales revenue of $150 is the value of intrafirm trade; it equals, and cancels, B's cost of goods sold when the accounts are consolidated. The operating profit earned by the two firms is $50. A's return on operating assets (the ratio of operating profit to operating assets) is 4 per cent, higher than B's return of 2 per cent. However, A has only one-third the assets of B. Under the profit split method, the transfer price is set so that each party shares in operating profit in proportion to the party's share of MNE operating assets. The ratio of A's operating profits to B should therefore be one-to-three, the same as their ratio of operating assets. Another way of saying this is that each firm should earn the average ROA across both firms or 2.5 per cent. Thus the transfer price should be set such that (1) total profits = $50, and (2) RORA = RORB = 2.5 per cent. This means A's profit should be $12.50 and B's should be $37.50. Working back in the financial statements, for A's operating profit to be $12.50, the transfer price must be ($12.50 + $10 +$120)7100 = $1.425. This lower transfer price gives B operating profits of $200 - $142.50 $20 = $37.50. The ratio of operating profit of A to B is $12.50/$37.50 = 1/3, and the rate of return of both affiliates is 2.5 per cent.

An Introduction to the Issues

47

FIGURE 1.8 The Profit Split (PS) Method

Using the profit split method A's profit / B's profit 20/30 A's assets / B's assets 500/1,500 Since ratio of assets is 'A, profit ratio should be Vs. A's profit is therefore 12.50, B's profit is 37.50. Use profit = sales - costs to "back into" the transfer price.

$37.50, implying a transfer price of $1.425 instead of $1.50. Thus the profit split method 'backs into' the transfer price through the allocation of profits between the related parties. The method is intuitively simple - split the profits - but the key questions are not simply answered, that is: (1) Which profit measure? (2) How should the profit be split? (3) On what activities? Answering these questions proves that the devil is in the details; i.e., the method is deceptively simple, and can give rise to results that are inconsistent with the arm's length standard. We come back to a discussion of the types, benefits, and costs of profit splits in Chapter 8 (the U.S. rules) and Chapter 13 (reforming the transfer pricing regime: rules and procedures). The Comparable Profits Method (CPM) Starting with the proposed section 482 regulations in 1992, the U.S. Treasury has advocated the use of the comparable profits method (CPM). The method

48 The Rules of the Game was widely criticized by tax practitioners, multinationals, other governments, and the OECD on the grounds that it was not compatible with the arm's length standard because (1) it was not a transactions-based method, (2) it did not take the contractual obligations of the parties into account, (3) did not reflect the facts and circumstances of the case, and (4) it could lead to substantial double taxation of income if other governments did not accept the method. The U.S. Treasury modified the method in 1993 and again in 1994, each time simplifying the method, and reducing its priority vis-a-vis the other methods. In the final 482 regulations, the CPM method is one of several possible methods, must be tested against the best method rule, and is generally considered a method of last result when transactional approaches (CUP, C+, RP) fail.15 In the U.S. regulations, there are nine steps to applying the CPM method: 1. The tax auditor or the MNE chooses the tested party - that is, one of the two related parties, preferably the one with the simplest functions and for which the best data are available. 2. The line of business activity to which CPM is to apply is determined; it may be one product or a product line or even broader. 3. Unrelated firms are selected as comparables; comparability is a question of facts and circumstances. Adjustments are made for differences in responsibilities, risks assumed, resource capabilities, and other material differences. 4. A profit level indicator is selected as the benchmark for determining the tested party's estimated profit; a common indicator would be the ratio of operating profit to operating assets. 5. The profit level indicator is calculated for each of the uncontrolled firms, and their ratios are applied to the tested party to determine a range of operating incomes (the arm's length range). 6. The firm's operating income is compared with the arm's length range. If it falls within the range estimated using the profit level indicators of the unrelated firms, the transfer price that generated this ratio is accepted by the tax authority. 7. However, if the firm's operating income lies outside the range, the government can set the firm's income equal to any point in the range, generally the median or mean of the range. 8. Given the final constructed income of the party (the arm's length result) the tax authority or MNE then 'backs into' the transfer price that would generate this arm's length result. 9. All remaining profits are allocated to the other related party, as determined by the final transfer price.

An Introduction to the Issues

BOX 1.7 The Comparable Profits Method Assume the example in Box 1.6 is continued - i.e., that A makes lamps and sells them to B at a transfer price of $1.50. A's financial statement, unchanged from Box 1.6, is reproduced on the left-hand side below. The right-hand side shows A's income after the comparable profits method has been applied to re-estimate A's operating income according to the return on assets earned by comparable uncontrolled manufacturing firms. The changes are highlighted in bold. We explain them below. A's statement before CPM Quantity of lamps Selling price

A's statement after CPM 100 $1.50

100 $1.625

Total sales revenue Cost of goods sold

$ 150.00 120.00

$ 162.50 120.00

Gross profit Operating expenses

30.00 10.00

$ 42.50 10.00

Operating profit Net income expense

$ 20.00 3.00

$ 32.50 3.00

Net income Operating assets ROR on assets (%)

$17.00

$29.50

$ 500.00 4.0%

$ 500.00 6.5%

Assume the selected profit level indicator is the rate of return on assets as measured by the ratio of operating income to operating assets. The rate of return on comparable uncontrolled firms, as estimated by the tax authority, is as follows: minimum return = 5.0%, maximum = 8%, mean = 6.5%. Given these returns, the arm's length range for A's operating income can be calculated as follows: Minimum constructed income = 5% ($500) = $25 Maximum constructed income = 8% ($500) = $40 Arm's length range = $25 to $40 Midpoint of arm's length range = 6.5% ($500) = $32.50 Since A's actual operating profit of $20 lies outside the arm's length range, the tax authority adjusts the profit to the midpoint of the range, i.e., to an arm's length resulted $32.50. This implies a transfer price of ($32.50 + $10 + $120)/100 = $1.625. This price is shown above in the right-hand side of the financial statement. Note that a $1.625 transfer price leaves firm B with an operating profit of ($200 - $162.50 - $20 = $17.50, and a return on $1,500 worth of operating assets of only 0.117 per cent.

49

50 The Rules of the Game FIGURE 1.9 The Comparable Profits Method (CPM)

The Comparable Profits Method Assume A is the tested party and ROA is the profit level indicator. Industry ROAs vary between 5 and 8% with a mean of 8%. Therefore As profit should be in the 25-40 range. A's profit is 20, so it is outside the range. Set A's profit at the midpoint of the range (32.50). Then use profit = sales - costs to "back into" the transfer price.

It is not clear how CPM will be used by the Internal Revenue Service in practice. The comparability requirements, as outlined in step 3, could be quite daunting. On the other hand, if comparability is loosely defined and industrywide statistics accepted, CPM can be calculated quite simply. All one needs to

An Introduction to the Issues

51

do is look up industry rates of returns on assets, as available on the Compustat database for example, for firms performing similar functions. This means the tested party could be defined as a simple distributor or contract manufacturer, industry returns on assets for distribution or contract manufacturing calculated and applied to the tested party, and all remaining profits allocated to the other related firm. For example, let us take the case presented in Box 1.6 and apply the comparable profits method. This is shown in Box 1.7 and illustrated in Figure 1.9. Suppose the tested party was firm A, the manufacturer: the profit level indicator was the ratio of operating profit to operating assets; and a sample of contract manufacturers earned a rate of return on assets that varied between five and eight per cent. Applying this range of returns to A's operating profit implies an arm's length range of constructed operating incomes that varies between $25 and $40. Since A's operating income is only $20, it lies outside the range. The mean (average) of the range is $32.50. If the arm's length result is set at the mean of the range, the required transfer price to give A an operating profit of $32.50 is ($32.50 + $10 + $120)7100 = $1.625. This leaves firm B with an operating profit of ($200 - $162.50 - $20 = $17.50. Alternatively, since the median (50 per cent of the observations above, 50 per cent below) of the range could be either higher or lower than the mean, the final transfer price is dependent on the choice of mean versus median. However, as this simple example illustrates, it also depends on several other choices: the tested party, the comparable unrelated parties, the profit level indicator, and the allocation mechanism. Each choice affects the final determination of the transfer price in ways that can only become clear with experimentation. Note also that the outcome is quite different from that of a profit split. The PS method ensures that both related parties earn the same return on assets; the CPM, on the other hand, ensures that one of the two parties earns the average or median of returns earned by comparable uncontrolled parties. CPM is therefore somewhat like the cost plus and resale price methods in that it focuses on only one side of total profits - that generated by the tested party - whereas the PS method looks to both sides. We will come back to this discussion in Chapter 8 (the U.S. rules) and Chapter 13 (reforming the tax transfer pricing regime: rules and procedures). Summary These simple examples serve only to outline the basics behind the variety of methods that have dominated the transfer price regulations of most governments. The CUP-RP-C+ trio was first adopted by the U.S. Treasury in 1968, then spread to other countries including Canada, and now forms the core of the OECD's recommended transfer pricing methods. The profit split method has

52

The Rules of the Game

been a fourth method on paper for some years, and in practice the method used by the U.S. tax courts to allocate taxable income in transfer pricing disputes. The comparable profits method is the newest of the five rules and the method which enjoys the least support outside of the U.S. Treasury. We will return to the methods in much greater detail in Chapter 5 (the simple analytics of transfer pricing), chapters 8 and 10 (the U.S. and Canadian rules, respectively), and Chapter 13 (reforming the transfer pricing methods). This concludes our discussion of transfer pricing from the viewpoint of the multinational and from the regulator's perspective, both at the national and international levels. In the last part of this chapter we address the question of the importance of this topic, and provide a brief outline of the remaining chapters in the book. The Importance of This Topic Taxing Multinationals deals with the treatment of intrafirm transactions under the corporate income tax, focusing on the transfer pricing choices of the multinational, the U.S. and Canadian transfer pricing regulations, and the international tax transfer pricing regime. Transfer pricing and tax policy is an important area of research for several reasons. First, the globalization of markets, the Canada-U.S. Free Trade Agreement (FTA) and its 1994 successor the North American Free Trade Agreement (NAFTA), and the growing importance of technology and services in international trade, are all issues dominated by the presence of multinationals. These are large integrated businesses, designed to maximize net-of-tax global profits, and engaged in strategic manoeuvres with their rival firms. How we tax their transfer prices can and does affect their output, sales, and intracorporate trade decisions. In a country as heavily populated with multinationals as is Canada or the United States, understanding the effects of the domestic tax system on multinationals, both domestic and foreign, can have important positive benefits for the economy.16 Second, MNE intrafirm trade in tangibles and intangibles has risen rapidly as a share of total Canadian and U.S. trade. About two-thirds of Canada's trade and investment flows are conducted with the United States. Manufacturing accounts for much of this crossborder activity. Roughly 30 to 40 per cent of shipments, value added, investments, and assets in Canadian manufacturing are generated by U.S.-controlled subsidiaries, and over half of their trade is intrafirm (i.e., between affiliates of the same multinational).17 As we show in Chapter 4 approximately half of Canada-U.S. trade in goods is conducted within MNEs, and up to 70 per cent of trade in business services is in-house.

An Introduction to the Issues

53

According to Revenue Canada (RC) officials, over 12,900 Canadian-based corporations were engaged in intrafirm, crossborder trade in 1995. The total dollar amount exceeded Can$318 billion, 69 per cent in tangible property. Three countries (U.S. 73.8%, U.K. 4.9%, Japan 4.4%) dominated this trade. Since most multinationals are headquartered in Ontario, a sizeable proportion (perhaps over 50 per cent) of these flows occur in one province. The dollar amounts are therefore huge and the tax implications clearly important. Third, this study is a useful complement to studies in other areas. For example, differences in tax bases and tax rates are one of the factors influencing MNE output, pricing, and locational decisions. Therefore this book has implications for interjurisdictional tax comparisons and corporate investment decisions. Statistics on differences in marginal and average tax rates, as calculated by other researchers, can be combined with this study to examine the effects of tax revenue avoidance through transfer price manipulations, both at the federal and subfederal levels. As another example, since transfer pricing falls in the tax avoidance/evasion area, and large MNEs are best placed of all businesses to engage in such manipulations, this book can usefully complement other studies on tax administration, compliance, and enforcement costs. Fourth, this book is innovative in that it explores transfer pricing from the viewpoint of several different disciplines, attempting to bring them together in a more holistic approach. Economists examine transfer pricing using complicated mathematical models of vertically and horizontally integrated firms to predict the effects of small changes in government policy (e.g., changes in tax rates or bases) on firm behaviour and performance. Lawyers examine transfer pricing from the viewpoint of redefining the laws in a constantly changing and difficult area of international taxation. Business professors look at transfer pricing as a management tool, while cost accountants worry about allocating revenues and expenses among units of the MNE. This book attempts to bring all these areas together, to lessen the 'dialogue of the deaf that exists among the different disciplines in the transfer pricing area. And, finally, the timing is right. The U.S. Treasury has finished its decadelong overhaul of its tax transfer pricing regulations. The final version of the section 482 rules are in place and the section 6662 penalty regulations are finalized. The Advance Pricing Agreement (APA) process is well underway and the first summary of the APAs (as applied to global trading) has been released. The U.S. Supreme Court has finally decided that the state of California can legally apply unitary taxation to foreign multinationals doing business in the state. So, inside the United States, the tax transfer pricing regime has finally clarified and settled. At the international level, the OECD's Committee on Fiscal Affairs has released its overhaul of the 1979 and 1984 transfer pricing reports in pieces as

54 The Rules of the Game OECD (1994b, 1995a, 1995b, 1996). The final, complete version will be OECD (forthcoming). If the first period of the historical development of the international tax transfer pricing regime was from the early 1920s until the mid-1960s, and the second period began with the U.S. transfer pricing regulations in 1968, the third period clearly began with the U.S. Congress enacting the Commensurate with Income standard in 1986. That event unleased an enormous flood of legislative changes over the 1986-96 period. The 'baby has been birthed' in the United States; a new set of transfer pricing rules, developed over ten years (1986-96), is now in place. We may be at the beginning of a new phase of consolidation in transfer pricing regulation. Where does Canada fit in this picture? The United States is Canada's largest trading and investment partner. Canada is an active participant in developing, and is committed to abiding by, the rules of the OECD. As the OECD and the United States change their tax transfer pricing rules, should Canada not look at its own policies? The Canadian regulations have been in place since 1987 when Revenue Canada issued Information Circular 87-2. The timing is right for examining the changing international regime at the OECD and U.S. levels to see whether or not Canada's rules and procedures should be overhauled in the light of events elsewhere. One purpose of this book is to provide such an examination. APPENDIX 1.1 SEARCHING FOR A CUP: THE CHRISTMAS TREE CASE This appendix provides an example of the information needed to find a comparable uncontrolled price for an apparently simple case - the pricing of Christmas trees.1 In the 1950s, J. Hofert Corporation was a U.S. parent firm in the Christmas tree business, with a Canadian subsidiary that harvested and shipped Christmas trees to its parent. The dispute with Revenue Canada arose over the transfer price for the Christmas trees. The Facts of the Case The Christmas tree industry is a natural resource industry, based on the harvesting, processing, and shipping of evergreen trees for sale during the December holiday season. Trees are grown year-round, cut in the late fall, checked for size and quality, and shipped in bulk by truck or rail boxcar to urban centres, where they are distributed for sale primarily in small lots adjacent to shopping malls. It

An Introduction to the Issues

55

appears to be a reasonably simple business, with a readily ascertainable price for the tree, depending on its height, type, and quality. Therefore a comparable uncontrolled price should be readily available; the Hofert case demonstrates that there can be many slips between the CUP and the transfer price. J. Hofert Company was a U.S. corporation, headquartered in Los Angeles. In 1946, the firm (hereinafter referred to as Hofert USA) set up a Canadian subsidiary, J. Hofert Limited (hereinafter referred to as Hofert Canada), located in British Columbia. The subsidiary's purpose was to harvest and ship Christmas trees under long-term contract to its parent. (In terms of Box 3.3 in Chapter 3 the subsidiary was a category 2 affiliate, a processor.) In the early 1960s, Hofert USA was the largest dealer in Christmas trees in the United States. The contract between Hofert USA and its Canadian subsidiary, originally written in 1946 and still in force in 1962 at the time of the court case, obliged Hofert Canada to sell and deliver each November as many Christmas trees to Hofert USA as the parent required. The subsidiary was to deliver trees of 'merchantable quality ... free from disease' and 'subject to inspection by Buyer, which shall have the right, prior to shipment thereof, to reject any trees not in conformity with the specifications' (62 DTC, 50-1). The parent firm paid its subsidiary for the costs of 'buying, hauling, inspecting, grading, tagging, tying and loading' the trees onto railroad cars plus a mark-up of eight per cent over cost (62 DTC, 51). Where Hofert Canada cut down trees from its own lands, Hofert USA paid its subsidiary for the costs of cutting down the trees plus ten cents for each delivered bale2 of trees. In addition, Hofert USA supplied Hofert Canada with twine, labels, and staples free of charge 3and advanced $5,000 to the subsidiary as part payment for the trees; final payments were made in May of the following year. The tax issue arose because Hofert Canada not only sold Christmas trees to its U.S. parent, but also sold them to unrelated buyers in Western Canada. Table A 1.1 provides data on these sales for the tax years 1954-6, the period audited by the Department of National Revenue (what we now call Revenue Canada). The department argued that Hofert Canada's sales to its U.S. parent had not been negotiated at arm's length, and that the price charged the parent firm (between $2.00 and $2.04 per bale) was far lower than the price charged to unrelated Canadian customers (between $2.75 and $3.19 per bale). As the table shows, the average difference over the three-year period between the two prices was 90 cents, making the average U.S. price approximately 31 per cent below the average Canadian price. Citing section 17(2), the predecessor of section 69(1), the department argued that the taxpayer had sold the trees to a related buyer at a price less than fair market value. Hofert Canada appealed the assessment.

56 The Rules of the Game TABLE A 1.1 Hofert Canada's Total Sales, 1954-1956

1954 Number of bales to Hofert USA 124,824.00 $2.00 Price to Hofert USA 249,648.00 Total sales to Hofert USA 18,309.00 Number of bales sold in Canada $2.75 Price to Canadian buyers 50,349.75 Total sales in Canada 83.22 Related sales as a % of total sales 0.75 Canadian price minus U.S. price Difference in prices due to trade level 0.55 27.27 Price difference as % of the Canadian price 1.85 Basic cost of production 0.35 Estimated unit profit on Canadian sales Estimated markup over basic cost 18.92 on Canadian sales (in %)

1955

1956

Total

138,491.00 $2.00 276,982.00 20,246.00 $2.87 58,106.02 82.66 0.87 0.71 30.31 1.85 0.31

139,462.00 $2.04 284,502.48 14,048.00 $3.19 44,813.12 86.39 1.15 0.98 36.05 1.89 0.32

402,777.00 ** 2.01 811,132.48 52,603.00 ** 2.91 153,268.89 84.11 0.90 ##0.70 30.93 1.86 0.35

16.76

16.93

18.82

average price charged (total sales divided by total quantity sold). calculated as the sum of the products of the price difference multiplied by the U.S. quantity, all divided by the total U.S. quantity. SOURCE: Based on data in J. Hofert Ltd. v. Minister of National Revenue 62 DTC, pages 51-53

The Tax Appeal Board Decision (1962) Judge R.S.W. Fordham, Q.C., heard the case at the Tax Appeal Board. The judge's decision began by defining fair market value as a 'commercial and not a legal term ... [that] involved a question of fact into which many considerations might enter' (62 DTC, 52). He then asked for the facts. 'What was the fair market value of Christmas trees in Western Canada in 1954, 1955 and 1956 and how was it determined?' (62 DTC, 52). Unfortunately, he said, the department had not provided any facts, other than the prices at which Hofert Canada sold trees to unrelated buyers in Canada. However, these prices, he argued, were not fair market value because the circumstances were 'entirely different from those that prevailed where the American purchaser was concerned' (62 DTC, 52). Judge Fordham then proceeded to outline the differences between the two sets of prices. First, he noted, it cost Hofert Canada more to sell trees in Canada than to its U.S. parent because the trade levels were different. Hofert USA was a middleman, buying the trees and reselling them to distributors, wholesalers, and retailers in the U.S. market. About 60 per cent of the parent's sales were directly to

An Introduction to the Issues 57 retailers. Canadian sales, however, were all to retailers. Therefore the subsidiary's expenses were higher on its Canadian sales because the firm was responsible for distribution and wholesale costs, for which it was not responsible on sales to its U.S. parent. These additional costs included 'the payment of wages and other expense incurred between roadside and delivery points ... of 55 cents, 71 cents and 98 cents per bale' respectively (62 DTC, 52). As Table Al.l shows, these additional costs, averaging 70 cents per bale, account for almost 80 per cent of the average price difference (that is, $0.70/$0.90). Second, the volume of sales differed significantly. Over 80 per cent of Hofert Canada's shipments went to its parent firm. Since bulk buying offers certain economies of scale, the U.S. price should be lower, reflecting these economies. Third, in spite of the contract which allowed Hofert USA to reject any trees that did not conform with its specifications, in practice, all bales purchased by the parent firm were paid for by the parent even if some of the trees were unsatisfactory. The subsidiary's Canadian customers, on the other hand, did not pay for unsatisfactory trees. In addition, if trees were not sold by the end of the Christmas season, Hofert Canada had to take back the Canadian ones but not those sold to its parent. Therefore the U.S. price should be somewhat lower than the Canadian price, reflecting these differences. Fourth, the parent firm provided twine, labels, and staples free of charge to the subsidiary, and also advanced it $5,000 in funds to carry out the agreement, again justifying a lower price. Lastly, the judge noted that the yearly net profit on Hofert Canada's local sales was no higher than on sales to its U.S. parent. Since the profit-to-sales ratios were similar, Judge Fordham concluded it was difficult to argue that preferential treatment had been given to the parent firm. He therefore concluded that there was no essential relationship between the Canadian and U.S. prices and found in favour of the taxpayer. An Economic Analysis of the Hofert Case The Hofert case demonstrates that an appropriate transfer price depends on the facts and circumstances of the case. The price Hofert Canada charged its parent for Christmas trees was clearly less than the price the subsidiary charged its Canadian customers. That, however, did not automatically mean that Hofert Canada was undercharging its parent. The key issues, both identified by Judge Fordham, were (1) whether the two prices were comparable, and, if not, (2) what was a comparable uncontrolled price (CUP)? Box Al. 1 outlines some of the factors that must be considered in finding product comparables. For both homogeneous and differentiated products, factors that

58

The Rules of the Game

BOXA1.1 Searching for a CUP - Guidelines for Comparables General factors that must be considered in finding comparables: Volume or lot size: How many are you buying? Is there a discount for larger sizes? Contract duration and price protection: How long is the contract for? Product form: Is the commodity purchased in bulk or packaged? Delivery point: What is the method of transportation and what are the transport costs? Quality: Is the good technical grade, 99 per cent pure or 99.99 per cent pure? Trade level: What are the franchise's rights and obligations? Warranty provisions and credit terms Additional factors to be considered in finding comparables for differentiated products: Product features: factors such as size, materials, weight Price differentials that may be related to the product features Intangibles: Are intangibles tied up with the product, e.g. the name and reputation of the company? Place in product line: Is the product part of a line of differentiated products?

must be considered include volume, contract duration, product form, delivery schedule, quality, trade level, warranty and credit terms. In addition, differentiated products require additional consideration. Factors include product features, price differentials related to these features, where the product fits in the MNE's product line, and the uniqueness of any intangibles included in the package. On the first issue, the court rejected the Canadian price as the fair market value on the grounds that the products were not comparables. The volume of shipments and the functions the taxpayer performed varied so much that the prices were not comparable. This is illustrated in Figure A 1.1, which shows prices, quantities, and sales by Hofert Canada to its U.S. parent and to unrelated Canadian customers. On the second issue, the judge noted that no one had presented evidence for

An Introduction to the Issues

59

FIGURE Al.l Explaining the Price Differential in the Hofert Case, 1954-1956

another suitable CUP. There was evidence that the differences in terms and conditions justified the U.S. price being lower than the Canadian one, and some evidence presented that perhaps 80 per cent of the difference (an average of 70 cents) could be explained simply by differences in trade levels. Adding volume discounts, credit advances, quality of the trees, and provision of free inputs, the judge concluded that the differences in the terms and circumstances fully explained the price differential (see Figure Al.l). As a check against his calculations, the judge also used the rate of return Hofert Canada earned on sales, arguing that the net profit rates it earned on Canadian and U.S. sales were similar. Although the calculations are not in the case, we can interpret the judge's argument as follows. Hofert USA paid its subsidiary for basic costs plus eight per cent; thus the

60

The Rules of the Game

subsidiary was a cost centre for the parent firm, and paid on a cost plus basis. The final price was only known when the total amount paid by Hofert Canada was calculated on a per-bale basis; that is, the U.S. price was (1 + 0.08) times the average cost of production. Since the U.S. price was almost constant over the three-year period, this meant Hofert Canada's average costs were also constant. An average price of $2.01 implies an average cost of $1.86 and a mark-up over costs of 15 cents. Since Hofert Canada incurred these basic costs on both its Canadian and U.S. sales, if we add in the costs due to the difference in trade level, and subtract that total from the Canadian price, we can calculate an upper bound to the firm's per-unit profit on its Canadian sales; that is, $2.91 - ($1.86 + $0.70) = $0.35 per bale of trees. (Annual estimates are shown in Table ALL) This mark-up is more than double the average mark-up over costs for U.S. sales of $0.15. As a per cent of the basic cost level of $1.86, the mark-up on Canadian sales averages 18 per cent compared with 8 per cent for U.S. sales. Clearly, some portion of the difference in mark-ups - perhaps as high as ten points - is really a discount to Hofert USA for its volume purchases, as was argued by Judge Fordham. A closer look at the data suggests this, in fact, is the case. As Table A 1.1 shows, the price in Canada rose 16 per cent over the threeyear period, from $2.75 to $3.19, while the price to Hofert USA hardly moved from its initial level of $2.00. Since the basic cost of production also hardly moved, this meant either that some other costs associated only with Canadian sales increased between 1954 and 1956, or that Hofert Canada raised its price as the Canadian market tightened. Since Hofert USA's share of the subsidiary's total sales varied little over this period (between 83 and 86 per cent), we can rule out changes in purchasing economies as responsible for the increasing gap between the two prices. In addition, other terms of the long-term contract (e.g., the financial advance, provision of free inputs) did not change. In fact, as the data show, one cost that did substantially increase over the period was that associated with the difference in trade levels; this cost increased by more than 50 per cent, from 75 cents to $1.15 over the three years. If we calculate the profit mark-up (after the difference in trade levels) over basic costs, the mark-up on Canadian sales stays roughly constant at 17-19 per cent of basic costs (see A 1.1 Table). The mark-up on U.S. sales is eight per cent each year, according to the long-run contract. Therefore, even though the two price series diverge over the period, the mark-ups over cost do not change. From this we conclude that Hofert Canada was earning approximately the same per-unit net profit on its intrafirm sales to its U.S. parent as the subsidiary earned on unrelated sales to its Canadian customers, once we allow a discount for differences in the volume of purchases.

An Introduction to the Issues 61 Lessons Learned from the Hofert Case The Hofert case is a nice example of a dispute over the pricing of a tangible. Christmas trees vary by height, quality, and volume; the producer can sell at different trade levels in markets that vary by distance; the contract terms can vary in length and financial conditions. Each of these factors is part of the facts and circumstances of the case - facts and circumstances that can turn an apparent CUP into an irrelevant comparison, as Judge Fordham properly concluded. NOTES 1 The summary of the Hofert case is based on J. Hofert Limited, v. Minister of National Revenue (62 DTC, pages 50-3). See also Nathan Boidman and Gary Gartner (1992). 2 A bale contained between one and eight trees depending on the size of the tree; the shorter the tree, the larger the number in a bale. 3 An extra wrinkle not discussed in any detail in the case is the tariff issue. Hofert Canada had to declare a customs value, and pay customs duty, on these imports of twine, labels, and staples, even if the parent firm furnished them free of charge. Hofert Canada then had to apply for duty drawbacks once the imported inputs were used on the trees and the trees subsequently exported to the U.S. parent.

APPENDIX 1.2 A TYPICAL BUSINESS INCOME STATEMENT Income Statement minus

Net sales Cost of goods sold (COGS) Gross profit

minus

Operating expenses Operating profit

minus

Other income/(expenses) Earnings before interest and taxes (EBIT)

minus

Net interest expense Net income before taxes

minus minus

Provision for taxes Extraordinary items Net income after taxes

SOURCE: Based on Chandler and Plotkin (1993, 45)

62 The Rules of the Game APPENDIX 1.3 DIFFERENCES IN ECONOMIC AND ACCOUNTING METHODOLOGIES! Unit of analysis

Economic concepts

Accounting concepts

The firm

single proprietorship with one entrepreneur-cum-manager running the business and receiving all profits

legal entity

Profit

normal profit (opportunity cost of entrepreneur) and economic profit (any return over and above normal profit, competed away in perfect competition)

accounting profit

Costs

economic costs

accounting costs

How costs are measured

opportunity cost (next best alternative use, so that sunk costs are sunk)

contractual outlays

Gains

gains on an accrual basis

gains on a realization basis

Model of firm behaviour

profit maximization

market share plus profit floors; dividends to shareholders

Model of competition

perfect competition, all firms are price takers

oligopoly, firms are price makers but compete primarily on the basis of nonprice competition

How assets are valued

value assets at current replacement cost

value assets at historical cost

2

The International Tax Transfer Pricing Regime

Introduction The purpose of this chapter is to explain how and why tax authorities regulate the transfer pricing policies of multinational enterprises. We see these regulations as having a coherent structure and focus, such that they may be characterized as part of an international tax transfer pricing regime (the TTP regime). International regimes are sets of functional and behavioural relationships among national governments in particular issue areas of the international political economy. We argue that there exists an international TTP regime in which national tax authorities have cooperated to develop certain principles, norms, rules, and procedures designed to facilitate state regulation of multinationals and to reduce conflicts between MNEs and nation-states in the corporate income tax area. In this chapter, we first outline the general theory of international regimes and provide one example. We next develop the concept of an international tax regime, and examine its characteristics (purpose and scope, principles and norms, and rules and regulations). We argue that nested within the international tax regime is an international tax transfer pricing regime and then we explore the characteristics of this regime. Appendix 2.1, at the end of the chapter, outlines a variety of approaches to international taxation of multinationals that have been recommended by the United Nations, the Organization for Economic Co-operation and Development (OECD), and the Harvard University Model Tax Code. The Theory of International Regimes1 Here we outline the theory of international regimes and then illustrate the theory with one well-known application, the international trade regime based on

64 The Rules of the Game the General Agreement on Tariffs and Trade (GATT), before turning to apply this theory to international taxation. What Is an International Regime? Problems of interdependence at the international level can be handled through international regimes. A regime is an international governance structure, a way to reduce international transactions costs in an interdependent world.2 Regimes can be seen as sets of functional and behavioural relationships among national governments in a particular issue area of the international political economy. These relationships embody the principles underlying the regime, the expected behaviour patterns associated with the regime, and the formal arrangements that implement the international agreements and understandings that form the regime (Preston and Windsor 1992, 7). Thus regimes are a way to manage interdependencies among nations. The generally accepted definition of a regime is: Regimes can be defined as of sets of implicit or explicit principles, norms, rules, and decision-making procedures around which actors' expectations converge in a given area of international relations. Principles are beliefs of fact, causation and rectitude. Norms are standards of behaviour defined in terms of rights and obligations. Rules are specific prescriptions or proscriptions for action. Decision-making procedures are prevailing practices for making and implementing collective choice. (Krasner 1983, 2; italics added)

When a clear legal framework establishing property rights and liability is missing, markets for information are imperfect, and/or incentives exist for actors to behave opportunistically, regimes can improve the functioning of international markets. International regimes can increase the predictability of behaviour, provide generalized sets of rules, and improve the information available to participants. A formal international organization may be either involved or absent from an international regime. For example, there are international regimes in international finance (centred around the International Monetary Fund and the Bank for International Settlements), debt (based on the World Bank), and security (based around NATO and the Nuclear Nonproliferation Treaty). On the other hand, the gold standard was an important international monetary regime in the early twentieth century, a regime that did not have an international organization at its centre. Similarly, the international trade regime is based on the General Agreement on Tariffs and Trade (GATT), where GATT is an international

The International Tax Transfer Pricing Regime

65

treaty, not an organization. One can also hypothesize that there exists a foreign investment regime, based on the UN Code of Conduct for Transnational Corporations (TNCs), the Freedom of Commerce and Navigation clauses, and bilateral investment treaties, although there is no international organization at its centre. International regimes vary in their characteristics. We can distinguish three general categories: purpose and scope, principles and norms, and rules and procedures (Haggard and Simmons 1987; Keeley 1990; Krasner 1983; Preston and Windsor 1992, 90-1). The strength of the regime depends on the extent to which the members conform to the characteristics of the regime. The purpose of the regime refers to the objectives the regime is supposed to accomplish, as seen by the participants. The purpose is normally defined by the problems to be managed. The scope of the regime refers to both the issue area and the geographic area covered by the regime. Regimes may be narrowly or broadly defined by subject matter, and cover many or few countries. The principles of the regime are the beliefs that underlie the regime. These may include principles such as equity, efficiency, neutrality, and/or nondiscrimination. The countries participating in an international regime commit themselves to certain norms, or standards of behaviour, designed to achieve the principles of the regime. These norms are expressed in terms of the rights and obligations of the parties. Norms can be descriptive or prescriptive. A descriptive norm is a course of conduct everyone follows in practice, whereas a prescriptive norm is a course of conduct individuals should follow (Langbein 1986, 628). The rules and procedures refer to the organizational structure of the regime. This includes such factors as membership requirements, presence or absence of an international organization, method of sharing the benefits and costs, and the specific, detailed procedures involved in the regime. Through international regimes, nation-states cooperate to regulate crossborder activities. As an example of an international regime let us look at the international trade regime. An Example: The International Trade Regime The international trade regime, or GATT regime as it is known, is perhaps the regime most familiar to the general public and certainly the one that occupies the most space in the newspapers. In this section, we outline its purpose and scope, principles and norms, and rules and procedures. These are summarized in Box 2.1. The purpose of the postwar international trade regime has been to reduce tariff barriers on international trade in goods (Jackson 1989, Zacher and Finlayson

BOX 2.1 Comparing the International Trade and the International Tax Regimes

Definition

International trade regime

International tax regime

Purpose

Goals to be achieved by the regime, problems to be managed.

To reduce tariffs and prevent tariff warfare. Seen as problems that lower world welfare by preventing gains from specialization and exchange.

To reduce double or undertaxation caused by overlapping tax jurisdictions. Seen as problems that cause distortions in capital markets and inequities among taxpayers.

Scope

Breadth of regime in terms of issue areas covered, number of members, geographic spread.

Trade in goods, broadened to include nontariff barriers. New areas include intellectual property and trade-related investment measures. About 140 contracting parties around the world.

Anything involving crossborder transactions that is subject to national taxation. OECD member countries and their tax treaty partners.

Principles

Beliefs of fact, causation, and rectitude that underlie the regime.

(1) most favoured nation treatment (MFN) and (2) nondiscrimination between domestic and members' products (National Treatment).

Three principles: inter-nation equity, international neutrality, and international taxpayer equity.

BOX 2.1 (concluded) International tax regime

Definition

International trade regime

Norms

Standards of behaviour, rights, and obligations of members, as identified in the general patterns in international agreements. Norms can be either prescriptive, prescriptive, or descriptive.

The GATT provides a general code of conduct on trade. Members commit themselves to the nondiscrimination principle and to GATT articles. Central obligation is to limit tariffs levied on contracting parties. Are protected from arbitrary impositions of tariffs on domestic products.

Tax conventions establish general norms as to which country has the right to tax, what the tax base and rates should be - e.g., the source country has the primary right to tax business income, the residence country is obligated to eliminate double taxation.

Rules

Specific prescriptions or proscriptions for action as identified in the provisions of international agreements.

Rules on voting, admission of new members. Cannot raise tariffs above bound levels. Rules on legal exceptions to GATT.

Rules defining nexus. Specific rules on the use of corporate income and withholding taxes. Tax deferral, exemption or credit rules for residence countries.

Procedures

Prevailing practices for making and implementing collective choice.

Antidumping duty and countervailing duty procedures. Dispute settlement procedures. Practice of trade negotiation rounds and reciprocal tariff cuts in these GATT rounds.

Auditing and dispute settlement procedures, both domestic and international. Includes the mutual agreement procedure under bilateral tax treaties (competent authority), penalties, advance pricing agreements, and arbitration of transfer pricing disputes

68

The Rules of the Game

1981). Tariffs are barriers to the efficient flow of international trade; when tariffs are reduced, the volume of international commerce can be increased and countries can reap the benefits of international specialization and the division of labour in terms of higher incomes and welfare levels. The issue scope of the international trade regime historically has been quite narrow: the GATT was set up after the Second World War to deal with tariffs levied by the developed market economies of Europe and North America on their imports of goods. However, the range of issues has broadened significantly since the 1979 Tokyo Round and the recently concluded Uruguay Round. The issue scope now includes new markets (services, intellectual property, agriculture) and new policies (nontariff barriers, trade-related investment measures [TRIMs], and trade-related intellectual property measures [TRIPs]). The geographic scope of the regime has also broadened as the number of signatories has increased from the original 23 contracting parties to almost 140 countries today. The basic principle that underlies the GATT trade regime is nondiscrimination (ND)? Nondiscrimination has two parts: the most-favoured nation (MFN) principle (ND between the products of the different contracting parties) and the national treatment principle (ND between domestic and member country products). The MFN principle means a country must treat the activities of any particular foreign country at least as favourably as it treats the activities of any other foreign country. In terms of GATT, this means that each contracting party must grant every other contracting party the most favourable treatment which it grants to any country in terms of exports and imports. National treatment means that a country treats foreign activities performed within its borders the same as it treats domestic activities; both are provided with the same treatment. In GATT terms, this means that foreign goods must be treated the same as domestic goods, once they have cleared customs and become part of a country's internal market. The norms of the trade regime are standards of behaviour defined in terms of rights and obligations. Contracting parties commit themselves to certain GATT obligations. The primary obligation is to limit the level of tariffs imposed on other contracting parties; in addition, countries commit themselves to the nondiscrimination principle and to the various GATT articles (e.g., on antidumping and countervailing duties, the customs valuation code, a subsidies code, various procedures).4 The rules and procedures - the organizational structure of the international trade regime - are centred around the GATT treaty to which countries sign on as contracting parties, not as members since there has been, until perhaps in 1995, no central international organization.5 Each contracting party agrees to specific rules (e.g., on voting, admission of new members) designed to support the principles and norms of the regime (Jackson 1989). The procedures or activities of

The International Tax Transfer Pricing Regime

69

the GATT regime include: multilateral tariff-round negotiations; publication by the GATT Secretariat of studies on national trade barriers;6 and the provision of GATT dispute mechanisms through which countries can bring their bilateral trade disputes for settlement. In successive GATT tariff-cutting rounds, one common procedure has been the practice of reciprocity - that is, the removal of tariff barriers by the major negotiating parties on a mutual and equivalent basis (Jackson 1989, 123-5).7 In sum, the theory of regimes is a useful and illuminating exercise to conceptualize certain problem areas in international political economy as being organized through an international governance structure. The international trade regime, organized around the GATT, provides a well-known example. Let us now see if these concepts and characteristics also can be used to describe an international tax regime. The International Tax Regime8 We describe the problems countries face in taxing multinationals, and then examine the existing sets of functional and behavioural relationships that have developed among national tax authorities to manage these interjurisdictional interdependencies. These relationships embody the principles, expected behaviour patterns, and formal arrangements that implement the agreements and understandings that form the international tax regime. The Problem: Overlapping Tax Jurisdictions Tax authorities have had to deal with entities doing business across international borders for a long time. Multinational enterprises, in industries such as automobiles and petroleum, have been with us since the late 1880s. However, the rapid growth in MNEs in the post-World War II period has significantly increased the degree of interdependence between national economies and reduced the sovereignty of national tax authorities. To quote the preface to the 1979 OECD report on transfer pricing: While taxation problems arising from international investment are not new, they have become more important in recent years as a consequence of the growing internationalization of economic activity. One characteristic of this process is the development of socalled 'multinational enterprises' ... This increasingly common phenomenon of related companies operating in a group with some degree of centralized management, yet with individual members of the group operating under different national law, has given rise to important problems regarding the taxation of corporate profits. (OECD 1979, 7)

70 The Rules of the Game Some examples of important problems in taxation raised by MNEs are the following: How should we define the MNE's tax base - its income - for purposes of calculating the corporate income tax? What if the tax base arises in more than one country? Which government should have the right to tax this income base? If two governments both claim the same right to tax, should one government's claim have priority over the other's? Should tax relief be given to prevent double taxation of the income? These are all questions of overlapping tax jurisdictions, caused by the integrated nature of the multinational enterprise. Multinationals are integrated businesses. A working definition of an MNE is that it consists of two or more firms located in different countries, where the firms are under common control and share common goals and a common pool of resources. By definition, the MNE's activities cross national borders and thus bring it under the jurisdiction of more than one tax authority. The enterprise therefore poses certain problems for tax authorities:9 More than one country: The MNE has transactions, income, and assets in more than one country. This creates the key problem with respect to taxing MNEs: that of jurisdictional allocation. Which nation has the right to tax the income base, and how can double taxation and conflicts between tax jurisdictions be avoided? Common control: The sine qua non of the MNE is intrafirm trade, the enterprise's way of integrating affiliates across national borders. As the per cent of intrafirm trade rises, open economies find that MNEs are setting trade, output, sales, and pricing policies as an integrated business. Domestic firms that decide to reduce their tax payments have ample opportunities to do so; global corporations have many more opportunities to hide profits and reduce taxes on a worldwide basis. Common goals: The MNE maximizes global after-tax profits. This brings it into conflict with national jurisdictions which focus on national variables. Governments are defined by their borders, MNEs by their share of world markets. The global reach of MNEs gives them the ability to avoid the national reach of government regulations. By shifting activities outside the reach of a national government, the MNE can avoid paying taxes. Thus underpayment of taxes is a problem for governments. Common pool of resources: The affiliates of the MNE share common overhead and resources. How should these resources be allocated among jurisdictions? Common resources are a source of competitive advantage for the members of the MNE family. They also are a source of interdependencies that make it difficult to disentangle the MNE for tax purposes. The problems created for governments by the integrated nature of the multi-

The International Tax Transfer Pricing Regime

71

national enterprise make it difficult to regulate MNEs at the domestic level alone. Governments are aware of these conflicts and inefficiencies, and, as a result, international taxation is one of the few areas where governments and MNEs have sat down to develop rules and procedures to manage these interdependencies. As Ray Vernon argues: I can find only one functional area in which governments have made a serious effort to reduce the conflicts or resolve the ambiguities that go with the operations of multinational enterprises. The industrial countries have managed to develop a rather extraordinary web of bilateral agreements among themselves that deal with conflicts in the application of national tax laws. Where such laws seemed to be biting twice into the same morsel of profit, governments have agreed on a division of the fare. Why governments have moved to solve the jurisdictional conflict in this field but not in others is an interesting question. Perhaps it was because, in the case of taxation, the multinational enterprises themselves had a major stake in seeing to the consummation of the necessary agreements. (Vernon 1985, 256) The results of government cooperation in the tax area include a variety of national tax policies, bilateral tax treaties (BTTs), and model treaties and guidelines. The latter have been developed by institutions such as the Organization for Economic Co-operation and Development (OECD), founded in Paris in 1960 to facilitate cooperation among developed market economies, and the United Nations. International bodies of experts such as the Committee on Fiscal Affairs at the OECD and the International Fiscal Association (IFA) have played important roles in developing international policies and norms. In Appendix 2.1 to this chapter, we review the guidelines and model income tax conventions that represent tangible evidence of intergovernmental cooperation to deal with the problems of overlapping tax jurisdictions. We argue that the combination of these government behaviours and functional relations can be seen as constituting an international tax regime. The regime reduces transactions costs associated with international capital and trade flows; resolves conflicts between tax authorities and multinationals, and between home and host governments; and reduces the possibilities for opportunistic behaviour by MNEs and nation-states. We now turn to outlining the characteristics of the regime: its purpose and scope, principles and norms, and rules and procedures. These are also summarized in Box 2.1 (see above). Purpose and Scope The purpose of the international tax regime is clearly outlined by Stanley Surrey:

72

The Rules of the Game

We live in one world with many national tax jurisdictions and with taxpayers whose activities cross national boundaries. As a consequence, countries have in various ways sought harmonization and coordination of national assertions of jurisdiction in order to reduce undue overlapping of tax burdens on investment and trade. (Surrey 1978,410) The OECD expands on Surrey's point about overlapping tax burdens: Since 1956 the OECD has sought to build-up a set of internationally accepted 'rules of the game' which govern the ways in which Member countries tax profits arising from international transactions. The main instrument used to achieve an internationally consistent approach to the taxation relating to such international transactions has been the development of an OECD Model Tax Convention ... [Its] purpose is the avoidance of international double taxation and to assist tax authorities in counteracting tax evasion and avoidance. (OECD 1993b, 1) The goals of the international tax regime, following the OECD, are therefore (1) the avoidance of double taxation of income and (2) the prevention of tax avoidance and evasion. These goals, according to Surrey, are to be achieved through coordination and harmonization of national tax systems. The principal method for encouraging such harmonization is through bilateral tax treaties. The OECD has played an important role in the development of the tax treaty network through its own model tax convention. Indeed, the purposes of the international tax regime are clearly visible in the title suggested by the OECD (1963, 1977) for bilateral tax treaties: 'Convention Between (State A) and (State B) for the Avoidance of Double Taxation and Prevention of Fiscal Evasion with Respect to Taxes on Income and on Capital.' As we saw in the general section on international regimes, the scope of a regime is a function of its issue coverage and geographic reach. Since the purpose of the international tax regime is the avoidance of over- or undertaxation of income, anything that involves crossborder transactions is potentially subject to tax and therefore potentially covered by the tax regime. Any income earned in or received from another location is potentially subject to taxation by two jurisdictions: where the income arose and where it was paid. The types of taxes involved are also numerous: corporate and personal income taxes, withholding taxes, value-added taxes, and mining taxes, for example. The issue scope of the regime is therefore potentially very broad. The geographic scope of the regime is also extensive. Its progress has been strongly influenced by the OECD. The OECD has encouraged the growth of tax treaties through its own Model Tax Convention. The United Nations has performed a similar function for developing countries. (See Appendix 2.1 for more

The International Tax Transfer Pricing Regime

73

details.) The network of these intergovernmental agreements is now vast. In 1995, Tax Analysts catalogued over 2,800 tax treaties, agreements, protocols, and similar documents; for example, the United States has signed tax treaties with 82 countries (Tax Analysts 1995, 5, 333-6). Principles and Norms Principles: Equity and Neutrality The principles of an international regime are the 'beliefs of fact, causation and rectitude' that underlie the regime (Krasner 1983, 1). We argue that three key principles underlie the international tax regime: (1) the inter-nation equity principle, determining which jurisdiction has the right to tax; (2) the international neutrality principle, ensuring that the international tax system does not distort private decisions; and (3) the international taxpayer equity principle, ensuring that taxpayers are treated fairly by the tax authorities.10 We examine each in turn. Inter-Nation Equity. The inter-nation equity (or, as it is sometimes called, the jurisdictional allocation) principle requires that tax shares be allocated fairly among countries. The problem is how to define 'fairness' since it involves several issues: which jurisdiction has the right to tax, the selection of a tax base, and the method for providing relief from double taxation (Arnold 1986, 54). In terms of which country has the right to tax, there are two possibilities under generally accepted international law, the source and residence principles.11 In the source principle, the country that is the source of the business income (called the source or host country) has the right to tax business income earned within its borders regardless of to whom that income is paid. Thus, income arising within a country is taxed whether or not the recipient is a resident of that country. The key criterion for source taxation is a nexus between the economic activities producing the income and the taxing jurisdiction. In the residence principle, the country where the owners reside (called the residence or home country) has the right to tax the owners of the business that creates the income. The residence principle looks at the relationship between the taxpayer and the taxing jurisdiction; if the taxpayer is a resident of the jurisdiction, his/her income can be subject to tax. The source of the income does not affect the right to tax; the key criterion is residency. The source and residence principles, by definition, generate conflict between tax authorities in different countries. The classic type of conflict arises between the home and host countries. A foreign subsidiary earns income and remits it in the form of dividends to its parent. At least three possible taxes are involved here: the subsidiary's income can be taxed in the source country (the host coun-

74

The Rules of the Game

try's corporate income tax or CIT); the dividends, when repatriated, can be taxed by the source country (the dividend withholding tax); and lastly, the dividends, when received, can be taxed by the residence country as income of the parent (the home country's CIT). The potential for double taxation of foreign source income is therefore real. Conflicts can also arise between two residence countries or two source countries. For example, even if all countries were to tax on a source (territorial) basis, the question of defining where an activity takes place (nexus) remains.12 Tax authorities are concerned about the conflicts inherent in the source and residence principles because they interfere with the goals of avoiding the double taxation of income and preventing tax avoidance and evasion. Double taxation and undertaxation are seen both as distorting the international allocation of capital and as being inequitable - that is, as interfering with the economic principles underlying a good tax system. This brings us to the second principle underlying the international tax regime. International Neutrality. A fundamental principle guiding national tax systems is tax neutrality. A neutral tax system for business income would leave business decisions unaffected by the tax. This means that governments should levy taxes in a manner that does not affect the taxpayer's choice of corporate form, location of the tax base, debt-equity level, choice of pricing policy, and so on, within domestic borders. The principle of tax neutrality is somewhat different from the principle of economic efficiency. A neutral tax means that the tax does not affect private decisions, whether or not these decisions are efficient (i.e., whether or not market price equals marginal social cost).13 The principle of economic efficiency, on the other hand, requires intervention by the government so as to ensure that private decisions equate market price with marginal social cost. Tax neutrality, therefore, is a weaker condition than economic efficiency. It implies a less interventionist government, one that desires only to avoid distorting private decisions and does not correct for private inefficiencies. The domestic neutrality tax principle has its equivalent at the international level. An internationally neutral tax system would neither encourage nor discourage choices such as whether to invest at home or abroad, work at home or abroad, or consume foreign or domestic goods. The decisions of individual decision makers (investors, workers, consumers) in terms of location would not be affected by the international tax system (Musgrave 1983, 280). While tax neutrality should apply to all types of taxpayer decisions, within the public finance literature neutrality is generally defined in terms of business investment decisions. We will follow that practice here.

The International Tax Transfer Pricing Regime

75

Where tax rates and/or tax bases differ between countries, we can conceptualize of international neutrality from the perspective of either the source country or the residence country.14 Therefore we must distinguish between capital export neutrality and an alternative view, capital import neutrality. The capital export neutrality (CEN) perspective sees the home investor as choosing between a domestic investment and a foreign investment (FDI). At the margin, the investor, in the absence of tax considerations, would attempt to balance the returns from the two investments. For CEN, this choice should not be affected by the tax system. An investor should be indifferent between domestic and foreign investments with the same pre-tax returns. Thus tax rates should be the same for investments with the same pre-tax return. Under capital import neutrality (CIN), who owns the investment should not affect the taxes paid on the investment. The relevant criterion is that domestic and foreign investors located in the same country should receive equal after-tax returns from identical pre-tax investments; that is, all capital within a jurisdiction should be treated similarly by the tax system regardless of ownership. Capital import neutrality is therefore equivalent to the nondiscrimination principle in the international trade regime, since nondiscrimination under GATT requires that imported goods be treated the same as domestic goods. CEN focuses on investments from the perspective of the home country (i.e., the choice between domestic investment and FDI), whereas CIN focuses on nondiscrimination among capital owners with investments in the same host country. The first gives primacy to the residence principle, the second to the source principle. An internationally neutral tax system, however, would not necessarily be a fair one. Thus the third fundamental tax principle is international equity. International Taxpayer Equity. Domestic tax systems also have the principle of fairness or equity in terms of the tax treatment of residents. Equity or fairness in taxation has several dimensions. First, fairness means that two taxpayers in similar economic circumstances should pay the same tax - this is, 'equal treatment of equals' or horizontal equity. Second, vertical equity, the appropriate treatment of unequals, must also be addressed by the tax. Generally, economists argue for progressive income taxation on the grounds that richer taxpayers have the ability to pay more than poorer ones; in addition, they may receive more benefits and thus should pay more using a benefit-cost approach.15 At the international level, the tax system should also be equitable. International taxpayer equity requires that all taxpayers resident in the same jurisdiction receive equal tax treatment regardless of the source of their income. This means that if the pre-tax returns from foreign source income and domestic

76

The Rules of the Game

income are the same, so should be the after-tax returns. Equity is even more difficult to define at the international level than at the domestic level. Which persons should be treated equitably: only residents, or should nonresidents also be included? Should equity be defined in terms of domestic taxes only, or in terms of the total burden of domestic plus foreign taxes? Should we distinguish between individuals and corporations, or between branches and subsidiaries, on equity grounds? These issues historically have been left to the residence country on the grounds that only the home country can tax global measures of income (Musgrave 1983, 281). In sum, three principles should guide the architecture (Shoup 1991) of a good international taxation structure: inter-nation equity, international neutrality (defined as capital export or import neutrality), and international taxpayer equity. These principles are explained in mathematical terms in Box 2.2. Norms: Separate Entity, Water's Edge, and First Crack The norms of the international tax regime represent standards of behaviour, defined in terms of rights and obligations of the national tax authorities, which are designed so as to achieve the principles of the regime. Norms are captured in double tax conventions. As Langbein (1986, 629) notes: While the particular provisions of double taxation conventions constitute rules, the general patterns of the conventions constitute norms. Moreover, there exist international model conventions which embody, and indeed direct, the general patterns, and thus explicitly constitute prevailing prescriptive norms. The norms of the international tax regime must therefore satisfy the three principles of the regime: inter-nation equity, international neutrality, and international taxpayer equity. To satisfy the principle of inter-nation equity, tax conventions must agree: (a) to establish a generally acceptable entitlement rule which spells out the source country's right to tax, (b) out of that entitlement rule to establish the base which may be taxed, (c) to lay down common definitional rules to ensure that there are no overlaps or gaps in the tax base which is divided among the countries of source, and (c) to set mutually agreed rates of tax which may be applied to that base. (Musgrave 1983, 282) Since the 1963 convention, the OECD has endorsed the concept of the separate entity as the underlying basis for allocating taxing rights between countries. Permanent establishments within a country are treated as separate entities. Each taxing authority has jurisdiction over the income and assets of this sepa-

The International Tax Transfer Pricing Regime 77

BOX 2.2 International Neutrality and International Equity

Let H be the residence country and F be the source country. Capital export neutrality (CEN) Residents in H must be indifferent, after tax, between identical pre-tax investments at home or abroad. That is, if rh(H) and rh(F) are before-tax returns to investments in H and F made by residents of H, and rh(H) = rh(F), CEN requires that:

\ ' ~~ 'h(H)Kh(H) = ( I — th(F))''h(F) where th(H) is the tax rate on investments in H and th(F) is the rate on investments in F, made by residents of H. Capital import neutrality (CIN) Identical pre-tax investments in the source country, F, must be treated the same by the tax system, regardless of ownership of the investments. That is, if rf(F) is the pre-tax return to investments in F made by residents of F, and rh(F) the pre-tax return to investments in F made by residents of H, and rf(F) = rh(F), then CIN requires that:

(' ~ tf(F))rt(F) = (i - th(F))rn(F) where tf(F) is the tax on investments in F by residents of F and th(F) is the tax on investments in F by residents of H. International taxpayer equity All residents in H should be treated equally in terms of taxation of their investments, regardless of the location of their investments. This principle, in practice, is the same as CEN since residence is the basis for comparison:

(1-th(H))rh(H) = (1- th(f))rh(F) Inter-nation equity The international allocation of income among countries should be fair. This principle is harder to specify since definitions of fairness vary. One possibility is that the tax rates on equal-yield foreign investments, rh(F) = rf(H), should be the same in both countries. That is, F's investments in H should be taxed by H at the same rate as H's investments in F are taxed by F, so that host countries treat foreign investors reciprocally. The reciprocity principle would be:

(' ~ th(F))rh(F) = (1 - tf(H)rf(H) Reciprocity is currently followed for withholding taxes negotiated under bilateral tax treaties, but could be defined more generally as 'eciprocally equal total taxes on capital income accruing to non-residents' (Musgrave 1983,284).

78

The Rules of the Game

rate entity, earned or received within the country up to its water's edge. Where MNEs are involved, affiliates are treated as separate legal entities and income is apportioned between them, assuming intrafirm transactions take place at arm's length prices. The right to tax depends on the existence of a connection or nexus between the taxing jurisdiction and the business enterprise. The nexus differs under the source and residence principles. For a taxable nexus to be established under the source principle, the business must have a 'permanent establishment' in the taxing jurisdiction.16 Once the MNE's income is effectively connected to a country, the source country can tax all items of income that arise within its borders. Under the residence principle, the definition of residency can vary between countries. In some countries (e.g., the United States), a business is resident in the jurisdiction where it is incorporated; in others (e.g., Canada, the United Kingdom, Australia), location of the 'seat of management' determines residency. In the latter case, de facto control matters more than de jure control (Arnold 1986, 10). The jurisdiction of residence has the right to tax both the domestic and foreign source income (i.e., the 'worldwide income') of its residents. Some countries follow an exemption system and exempt income earned abroad from taxation; others tax worldwide income. Under the jurisdictional norms, 'first crack,' or the primary (but not exclusive) right to tax business profits, is given to the country of source. The residence country has the primary right to tax most other categories of income (Arnold 1986, 174; Langbein 1986, 630). Examples of different types of income and which country normally has the primary right to tax are illustrated in Box 2.3. The principles of international neutrality and taxpayer equity are recognized through the obligation placed on the residence country to eliminate double taxation. Since the source country has the prior right to tax, the residence country is expected to modify its rules to take account of source country taxation. The tax boundaries established in most developed countries are therefore roughly the same: the fiscal authority taxes the worldwide income of its residents and the domestic source income of its nonresidents (Arnold 1986, 3). Many countries - for example the United States - tax worldwide income of their residents, but defer tax on foreign source income until it is repatriated. In calculating the home country tax, a foreign tax credit is granted for the corporate income taxes and withholding taxes paid in the host country, up to the level of the home country tax. In certain cases, the residence country exempts all foreign source income from tax and taxes only on a territorial basis. In still others, certain categories of foreign source income are exempt while others are taxable as earned. For example, in Canada active business income earned abroad is not

BOX 2.3 Source Rules for Taxation of Business Income

Type of business income

Jurisdiction for tax purposes

Dividend income paid by X to Y

where X, the paying corporation, is resident

Interest income paid by X on a loan borrowed from Y

where X, the debtor, resides

The income from personal services provided by X to Y in location Z

in Z where the services are physically performed

Rents received by X from personal property owned by Y in location Z

in Z where the property is put to use

Royalties received by X from the licensing of technology to Y for use in location Z

in Z where the licensed property is put to use

Income received by X from real property owned by X in location Z

in Z where the property is situated

Gain/loss received by X on the sale of real property in location Z

in Z where property is located

Gain/loss received by X on sale of personal property in location Z

passage-of-title test (if sold in X, location in X)

SOURCE: Based on Hufbauer (1992, 203-7)

80 The Rules of the Game TABLE 2.1 Statutory Marginal Federal Corporate Income Tax Rates, Selected OECD Countries, 1981-1992

Australia Canada France Germany Ireland Japan Netherlands Sweden Switzerland United Kingdom United States

1981

1985

1989

1992

0.46 0.483 0.50 0.56 0.45 0.42 0.48 0.58 0.098 0.52 0.46

0.46 0.483 0.50 0.56 0.50 0.43 0.43 0.52 0.098 0.40 0.46

0.39 0.391 0.39 0.56 0.43 0.40 0.35 0.52 0.098 0.35 0.34

0.39 0.391 0.34 0.519 0.40 0.384 0.35 0.30 0.098 0.33 0.34

SOURCE: Data from Cummins et al. (1995, 195)

taxable in Canada, while income from passive investments is taxable as earned (Brean 1993). Rules and Procedures Rules: Corporate Income and Withholding Taxes The international tax regime has specific rules, embodied in double tax conventions, which are designed to allocate the tax base to either the residence or source country. The source country, for example, normally levies a corporate income tax (CIT), allowing the enterprise to deduct expenses incurred in the production of the income.17 Up to the mid-1980s, tax incentives were widely used by OECD governments to encourage investment. As a result, statutory CIT rates (i.e., posted rates) could vary significantly from marginal effective rates (i.e., rates taking account of incentives, credits, and deductions). Led by the United Kingdom in 1984, which announced the elimination of tax incentives and reduction in statutory tax rates, most governments have been flattening their rates and broadening their tax bases over the past ten years (OECD 1990, 152). Table 2.1 shows this trend to reducing statutory federal CIT rates for selected OECD countries over the 1981-92 period. Most countries grant tax incentives for particular types of activities. For example, it is common to differentiate by asset type (machinery versus buildings), industry (manufacturing versus commerce), and source of finance (debt

The International Tax Transfer Pricing Regime

81

TABLE 2.2 Marginal Effective Corporate Tax Rates, Selected Countries, 1990

Asset type: Machinery Buildings Inventories Industry type: Manufacturing Other industry Commerce Sources of finance: Debt New share issues Retained earnings Overall corporate tax rate

Canada

U.S.

U.K.

Japan

15.5 35.9 30.7

18.5 25.3 26.3

8.0 49.7 39.8

8.8 2.5 7.0

24.5 29.1 25.0

34.0 11.7 21.8

24.8 21.2 37.8

6.7 5.9 5.2

-6.3 47.2 47.3 25,9

-14.7 44.1 43.7 24.0

-15.9

4.1 40.5 28.0

-74.6 70.9 62.8

6.1

The marginal effective corporate tax rate (MECTR) is calculated for the corporate income tax (CIT) at the corporate level only, ignoring the personal income tax. The MECTR is the difference between the before-CIT rate of return to the after-CIT rate of return (this difference is called the 'tax wedge') divided by the before-CIT rate of return (also called the 'marginal product of capital.' Thus the MECTR is the ratio of the tax wedge to the marginal product of capital. SOURCE: Data from Jorgenson (1993, 984)

versus equity). As a result, the marginal effective tax rate varies widely across assets, industries, and sources of finance. Some evidence on source country corporate income tax rates from Jorgenson (1993) is provided in tables 2.2 (for the CIT) and 2.3 (for the CIT and personal income tax combined). Table 2.2 focuses on the corporate level, with data on 1990 marginal effective CIT rates for Canada, the United States, the United Kingdom, and Japan in terms of taxation for various assets, industries, and sources of finance. Table 2.3 focuses on the shareholder, looking at the combined corporate and personal income tax rates on corporate source income. Canada's overall corporate income tax rate is 25.9 per cent for corporations (higher than the U.S. and Japan but lower than the U.K.) and 40.2 per cent on corporate source income (highest of all four countries). Economic theory predicts that marginal, not statutory, tax rates affect real investment decisions (see Chapter 6 for details). Differences in marginal CIT rates provide opportunities for multinationals to arbitrage these imperfections, shifting activities to lower-taxed locations. For example, since Canada's marginal effective tax rates are at the high end of the range, the tax differential may

82 The Rules of the Game TABLE 2.3 Marginal Effective Tax Rates on Corporate Source Income, Selected Countries, 1990

Asset type: Machinery Buildings Inventories Industry type: Manufacturing Other industry Commerce Sources of finance: Debt New share issues Retained earnings Overall corporate tax rate

Canada

U.S.

U.K.

Japan

32.9 47.2 44.0

34.1 39.6 40.3

23.0 55.2 47.2

29.4 25.3 28.3

38.5 42.1 41.1

46.4 28.8 36.8

35.5 32.7 45.6

27.8 28.2 27.3

33.7 60.3 49.6 40.2

8.8 64.1 53.3 38.5

35.1 25.5 43.8 37.9

-8.3 76.7 66.2 27.7

The marginal effective tax rate on corporate source income (METRcsi) is calculated for the corporate income tax (CIT) and personal income tax (PIT) combined. The METRcsi is the difference between the before-CIT-and-PIT rate of return to the afterCIT-and-PIT rate of return (we can call this difference the 'joint tax wedge') divided by the before-CIT-and-PIT rate of return. Thus METRcsi is the ratio of the joint tax wedge to the marginal product of capital. SOURCE: Data from Jorgenson (1993, 989)

significantly affect multinational behaviour in Canada, particularly longer-run investment responses by footloose MNEs. One way to deter mobile financial flows is through withholding taxes. Withholding taxes are generally levied on income paid to nonresidents that arises from passive investments or casual, nonrecurring activities in the source country. Interest, dividends, rents, royalties, and management fees are examples of types of income remittances that normally attract a withholding tax. Tax rates are normally in the 10-25 per cent range, but are generally reduced through bilateral tax treaties to zero to 10 per cent. The practice of cutting withholding taxes through bilateral tax treaties provides an example of using reciprocity on the grounds of inter-nation equity.18 The residence country normally levies a corporate income tax on the enterprise's income, however defined, allowing the enterprise to deduct expenses incurred in the production of the income. Generally, the net income from all the units of the enterprise are consolidated for tax purposes. Procedures: Dispute Settlement Both domestic and international procedures are part of the international tax

The International Tax Transfer Pricing Regime

83

regime. For example, at the domestic level, national tax authorities publish regulations and have auditing and dispute-settlement procedures. At the international level, a network of bilateral tax treaties is used to settle interjurisdictional disputes. Most tax treaties are based on the OECD Model Tax Convention (see Appendix 2,1). The basic purpose of a tax treaty between two countries is to clarify their respective tax jurisdictions - that is, the nature of the transactions to be taxed and the per cent of the tax base each country has the right to tax (Kwatra 1988). Where disagreements occur, tax treaties contain a Mutual Agreement Procedure for cooperation where the representatives of each government (called the Competent Authorities) get together to resolve disputes (Skaar 1992). Where two countries do not have a tax treaty between them, there is no easy method at present for resolving interjurisdictional taxation disputes. Within the international tax regime we can see at least two regional groupings emerging, the first in North America and the second in the European Union. The North American tax regime consists of the domestic tax rules and procedures for taxing MNEs in each of the three countries (Canada, the United States, Mexico) and the bilateral tax treaties that they use to determine tax jurisdictions, define tax bases, and settle crossborder disputes. In the following section, we outline the principal components of this regional tax regime. The North American Tax Regime The North American tax regime consists of three national tax systems for taxing multinationals - that is, the foreign-source income of domestic MNEs, and the domestic income of foreign MNEs, together with three bilateral tax treaties. We examine each country's approach to taxing multinationals below, and then review their BTTs. Table 2.4 provides some information on U.S. and Canadian corporate income tax rates on MNEs, and the withholding taxes negotiated under the Canada-U.S. tax treaty, as of 1995. The U.S. Approach to Taxing Multinationals19 In the previous section, we outlined the principles, norms, rules, and procedures of the international tax regime. Now let us look at the U.S. approach to taxing multinational income. Taxing the Foreign Source Income of U.S. Multinationals In terms of the residence principle, the United States taxes its residents - persons and corporations - on their worldwide income. U.S. rules distinguish between a foreign branch (an entity, owned 100 per cent by its parent, which

84 The Rules of the Game TABLE 2.4 U.S. and Canadian Corporate Income Tax (CIT) and Withholding Tax Rates, 1995 (in percentage terms) Type of tax

U.S.

Federal CIT rate - general Federal CIT rate - manufacturing

34

Subfederal adjusted CIT rate - general* Subfederal adjusted CIT rate - manufacturing*

12

Federal plus subfederal CIT - general Federal plus subfederal CIT - manufacturing

46

29

I 15

I 44 36

10(5) 15 (10) 10 (zero) zero zero

Withholding tax on investment dividends# Withholding tax on interest payments#t Withholding taxes on royalties# Withholding tax on management fees# Withholding tax on copyright income# Total CIT plus dividend withholding tax - general** Total CIT plus dividend withholding tax - manufacturing**

Canada

51

5

° 42

Subfederal U.S example is New York State (state CIT rate varies from 0-12 across all U.S. states). State CITs are deductible against federal tax, deductibility already factored in for ease of comparison. Subfederal Canadian example is Ontario (varies from 8-17 percent across all Canadian provinces). Provincial CITs are additional to the federal CIT. Calculated as w (1 - CIT) where w is the withholding tax rate. #As in the Canada-U.S. tax treaty; the 1995 Canada-U.S. tax protocol rates are in parentheses. tSince 1984, portfolio interest paid by U.S. borrowers to unrelated foreign lenders (other than banks lending in the normal course of business) has been exempt from the U.S. withholding tax. SOURCE: Author's calculations using Boidman (1995b, A8-A11), Boidman and Gartner (1992, 30, 34), and the bilateral tax treaties.

does not have an independent legal existence separate from the parent) and a controlled foreign corporation (CFC) (an entity, more than 50 per cent owned by U.S. shareholders, incorporated in the foreign country and thus considered an independent entity). Branch profits are taxed as earned, but U.S. firms are permitted to defer U.S. taxes on income earned by their foreign subsidiaries until the CFC income is repatriated. The U.S. CIT applies to domestic income of U.S. MNEs plus accrued foreign branch profits, head office fees, and interest payments remitted from foreign affiliates. Dividends are grossed up by the amount of foreign CIT and brought into taxable income. A foreign tax credit (FTC) is provided for (1) withholding taxes on remitted interest, head-office payments and dividends, (2) foreign

The International Tax Transfer Pricing Regime

85

branch taxes, and (3) foreign CITs on dividends.20 The credit cannot exceed the U.S. rate of tax. Since 1986, the U.S. rules 'look-through' or characterize types of incomes and place them in separate baskets with separate F TC calculations, in order to reduce tax avoidance.21 Foreign earnings must be pooled and the FTC calculated using cumulative rather than annual foreign-source income and taxes. The rules also require U.S. parents to allocate a percentage of overhead to their foreign subsidiaries; creditable expenses are calculated, however, on a consolidated basis rather than by affiliate. In addition, since the Subpart F rules were passed in 1962, passive income earned by foreign subsidiaries with U.S. parents, in situations considered abusive, has been taxable as earned. These situations primarily involve income in tax haven countries. For example, dividends, interest, rents, and royalties received by a U.S. citizen from a closely held company in Bermuda or the Cayman Islands would be taxable as accrued.22 Taxing the U.S. Income of Foreign Multinationals In terms of the source principle, the United States taxes the income of permanent establishments and any income effectively connected to the United States at the basic federal CIT rate. (In 1995 this was 34 per cent; see Table 2.4 for details.) A U.S. corporation owned by nonresidents is subject to the CIT on its profits. American states and some cities also tax corporate income. The rates vary from zero to 12 per cent, and are deductible against the federal CIT (Boidman and Gartner 1992, 30-1). In addition, when the corporation remits dividends to its parents, a withholding tax of 30 per cent is levied on the dividends. The dividend withholding tax can be reduced to as low as 5 per cent through BTTs; under the Canada-U.S. tax treaty the rate is 10 per cent.23 Before 1986, no tax comparable to the dividend withholding tax was levied on U.S. branches with foreign parents. In the 1986 Tax Reform Act, the U.S. government introduced several reforms that reduced tax rates and widened the income tax base.24 A 30 per cent branch profits tax was introduced as a de facto withholding tax on profit remittances by U.S. branches to their foreign parents. The U.S. government has become increasingly concerned with the (apparent) underpayment of U.S. taxes by foreign-owned firms in the United States. The U.S. Department of Commerce refers to these firms as 'foreign affiliates' or 'foreign-controlled corporations' (FCCs). We discuss this issue in Chapter 7, (Taxing MNEs in Practice), and provide new evidence on the tax payments of FCCs and of U.S. majority-owned foreign affiliates (MOFAs) located abroad. In chapters 8 and 9, on U.S. tax transfer pricing regulations, we review and assess the ways the U.S. Treasury and the Internal Revenue Service have

86 The Rules of the Game attempted to increase the surveillance of, and taxes paid by, MNEs in the United States. The U.S. Tax Treaty Network As of 1992, the United States had signed 42 bilateral income tax treaties with other countries (Hufbauer 1992, 216-19). In 1995, the United States had bilateral treaties, of various types (information sharing, social security, income, defence spending, estate and gift, income from shipping and aircraft), with over 80 countries and principalities (Tax Analysts 1995,136-43). Generally, the treaties define residency, taxation of business profits, attribution of profits, allocation of expenses, and treatment of sources of income. As a home country, a major goal of U.S. treaty policy has been to convince treaty partners to reduce their withholding taxes on remitted dividends, royalties, and head-office fees. The U.S. competent authority is responsible for administering and implementing tax treaties. Given the enormous size of Canada-U.S. intrafirm trade and financial flows, probably the most important of these tax treaties is with Canada. The Canadian Approach to Taxing Multinationals15 Taxing the Foreign Source Income of Canadian MNEs Canada taxes residents - individuals and corporations - on their worldwide income; nonresidents are taxed only on their Canadian source income.26 Taxation of worldwide income means residents pay Canadian taxes on their foreignsource income. However, Canada distinguishes between two types of foreignsource income. Income earned through a foreign corporation (called a 'foreign affiliate') owned by a Canadian firm is not taxed as earned. The tax treatment depends upon whether the foreign affiliate earns exempt surplus or taxable surplus. Exempt surplus can be defined as 'active business income earned in certain listed countries (generally countries with which Canada has or is negotiating a tax treaty)' (Arnold 1986, 154-5). Dividends paid from foreign direct investments are exempt from Canadian taxation if they are paid out of exempt surplus in the foreign affiliate. Foreign-affiliate active business income losses cannot be deducted from the parent's income. Taxable surplus basically consists of passive income and active business income earned in unlisted countries. Under the Foreign Accrual Property Income (FAPI) rules, dividends paid out of taxable surplus (e.g., foreign portfolio dividends) are included in the Canadian parent's income and subject to tax at the basic Canadian CIT rate; foreign withholding taxes are creditable, or may be deducted, against the Canadian tax (Arnold 1986, 153-4). Passive income is

The International Tax Transfer Pricing Regime

87

taxed only when repatriated because the foreign firm is considered to be a separate entity and not resident in Canada, for tax purposes. The definition of what is and what is not FAPI income is difficult. The FAPI rules, which apply only to controlled foreign affiliates, operate such that if the income is classified as active business income then it is not subject to the FAPI rules.27 Therefore FAPI income is basically passive investment income - for example, income from property, inactive businesses, and certain types of service income and capital gains. Thus, Canada has a mixed exemption-credit system: exemption for active business income while dividends out of other income are taxed when repatriated with a foreign tax credit given for host country taxes. This regime was instituted in 1972; up to that point all foreign-source income was exempt from Canadian tax.28 Now dividends out of ABI in listed countries can be repatriated tax free; dividends out of other income are taxed when repatriated. In 1992, the Auditor General of Canada, in his annual report to Parliament, questioned the amount of taxes paid by Canadian MNEs on their foreign-source income, arguing that hundreds of millions of dollars was not being paid due to weaknesses in the Canadian regulations. Specifically, the report questioned: (1) the tax-exempt status of certain dividends received by Canadian MNEs from their foreign affiliates; (2) problems with the anti-avoidance FAPI rules; and (3) the tax-deductible status of interest on borrowed money used by Canadian MNEs to earn foreign-source income. The Department of Finance, however, disagreed with the Auditor General, arguing that Canadian tax regulations had to conform to international norms as established by the OECD; that the rules were designed not only to raise revenues but also to facilitate the competitiveness of Canadian MNEs on their foreign activities; and that little tax revenues would be gained by tightening the rules. We discuss this dispute, and the hearings held by the House of Commons Committee on Public Accounts, in Chapter 10 (Canadian rules and procedures). Taxing the Canadian Income of Foreign Multinationals In Canada, in terms of the source principle, foreign-controlled permanent establishments are taxed at the federal plus provincial statutory corporate income rate, with most tax deductions and credits available to domestic firms also available to foreign establishments. Profits earned from manufacturing and processing activities are taxed at a lower rate. In 1995, the federal rate was 29 per cent on general profits, reduced to 22 per cent for manufacturing profits (see Table 2.4); additional provincial CITs were payable, varying from 8-17 per cent depending on the province. Foreign-owned branches pay an additional 25 per cent of taxable income, from which the CIT is deductible, as a branch tax. Withholding taxes on remittances (except interest payments) are levied when these

88 The Rules of the Game funds are repatriated. The general withholding tax rate on dividends is 15 per cent; since 1980 under the Canada-U.S. tax treaty the rate has been 10 per cent. The 1995 Canada-U.S. tax protocol reduces this rate to 5 per cent (see Table 2.4). The Canadian Tax Treaty Network Canada has an extensive network of bilateral treaties of various kinds (income, social security, capital gains, income from shipping and aircraft) with 59 countries (Tax Analysts 1995, 17-21). Canada's first income tax treaties were signed, not surprisingly, with the United Kingdom (1935) and the United States (1936). Based on Canada's long-standing concern with the inward FDI, the tax structure is set up so as to not directly discriminate against nonresidents, but tax preferences are generally restricted to residents. Canadian tax treaties only include a narrow definition of nondiscrimination, compared with the definition in article 24, of the OECD Model Tax Treaty (Eden 1988a,b). Canada typically precludes discrimination against resident nationals of the treaty partner and against residents of the treaty partner with permanent establishments in Canada. These groups receive treaty protection equivalent to Canadian nationals in Canada. Canadian tax treaties, however, do not include the second paragraph of article 24, which extends nondiscrimination protection to nationals of a treaty country that are not residents of either of the treaty countries. Nonresidents of Canada generally receive only most-favoured-nation (MFN) treatment, not full nondiscrimination. This means that Canada is committed to treating foreignowned enterprises the same as other foreign-owned firms, but not equivalently to resident-owned enterprises. Nondiscrimination applies only to the taxes specified in the treaty, not to all forms of taxation. The Mexican Approach to Taxing Multinationals29 Taxing the Foreign-Source Income of Mexican MNEs Mexico taxes individuals and businesses on their worldwide income. In 1988, the rate was 42 per cent, which was subsequently reduced to 35 per cent, and then, in 1994, to 34 per cent for both individuals and corporations (Perez de Acha 1994, 623). A foreign tax credit is offered up to 34 per cent of taxable foreign-source income. Table 2.5 provides some information on Mexican, U.S., and Canadian CIT and withholding tax rates, as of 1995. While the Mexican income tax at first looks similar to the U.S. and Canadian systems, there are some striking differences (del Castillo et al. 1995; McLees 1992, 1994). First, the definition of income in Mexico is different. Most income bases and deductions, unlike those in the United States and Canada, are indexed

The International Tax Transfer Pricing Regime

89

TABLE 2.5 Statutory Corporate Income Tax Rates within North America, 1995 (in percentage terms) Canada T Initprl

Federal corporate income tax rate Subfederal CIT rate (see note below) Profit sharing tax Combined federal/state CIT rate Withholding tax on direct dividends CIT plus dividend withholding tax

States

Mfg.

Non-Mfg.

Mexico

34 12 0 46

22 14

29 15

34 0 10 44 0

Canada Mexico Canada Mexico

0 44

36 10 5 51 49

10 42

50

44

NOTES: 1 The lowest tax rate in each category is in bold. 2 Subfederal U.S. example is New York State (state CIT rate varies from 0-12 across all U.S. states). State CITs are deductible against federal tax; deductibility already factored in for ease of comparison. Subfederal Canadian example is Ontario (varies from 8-17 percent across all Canadian provinces). Provincial CITs are additional to the federal CIT. There are no state income taxes in Mexico.

for inflation (for example, real not nominal interest costs are deductible expenses). Thus the tax base can differ even if rates are roughly equivalent. Second, net income is generally subject to a mandatory 10 per cent profitsharing payment, which is not deductible against the CIT to the extent that the firm's Mexican employees receive nontaxable fringe benefits. Third, Mexico also levies a 2 per cent business assets tax as a minimum income tax, unless the taxpayer declares at least a 5.7 per cent taxable return on the assets tax base.30 Lastly, Mexico has no state income taxes. Assuming a corporation pays more than the minimum income tax, the effective statutory CIT rate in Mexico is therefore 44 per cent. Taxing the Mexican Income of Foreign Multinationals Nonresident individuals working in Mexico pay a statutory withholding tax rate of 30 per cent on gross income, regardless of where it is paid or the form it takes, even if the company does not have a permanent establishment in Mexico. Reduced rates apply to tax treaty countries and to the maquiladoras (McLees and Reyes 1992). Dividends paid either to residents or nonresidents from previously taxed income are not taxable, so there is no withholding tax on dividends.

90 The Rules of the Game Thus the statutory rate of tax on remitted profits of permanent establishments is 44 per cent. The business assets tax applies to permanent establishments and can be a source of double taxation.31 The Mexican Tax Treaty Network As of 1995, Mexico has a limited, and very recent, group of bilateral tax treaties (BTTs) with 12 countries (Tax Analysts 1995, 82). Its first treaty was an exchange of notes in 1964 with the United States on taxing income from shipping and aircraft. The first income tax treaties (with Canada, France, and Italy) were signed in 1991; by 1993, Mexico had 12 BTTs in place, including treaties with the United States and United Kingdom. Clearly, the prospect of Mexico becoming a member of the OECD and setting up the NAF TA were strong impetuses to formalizing a BTT network with the other OECD and NAFTA countries. Bilateral Tax Treaties in North America The history of bilateral tax treaties in North America is only a long one in the case of Canada and the United States, reflecting their long history of crossborder tax and FDI flows. Mexico's BTTs with Canada and the United States are both quite recent. Table 2.6 outlines the key features (withholding tax rates, national treatment norms, other provisions) of the current BTTs among the NAFTA countries. The 1980 Canada-U.S. Tax Treaty and 1995 Tax Protocol Canada and the United States have had a bilateral tax treaty since 1936 (Tax Analysts 1995, 137). The 1980 Canada-U.S. tax treaty, as amended by protocols in 1983 and 1984, was ratified and came into effect on 1 January 1985. Starting in 1990, the two governments began negotiations on a new protocol, which was first signed in August 1994, revised and re-signed in March 1995, and came into effect on 1 January 1996. The main components of the new protocol, as related to multinationals, are (Arnold 1994; Boidman 1994a, 1996a,b): Nondiscrimination: The nondiscrimination clause, which previously had applied only to the CIT in both countries, is extended to all taxes imposed by Canada and the United States. Thus the principle of nondiscrimination is extended to all taxes as they apply to Canadian and U.S. multinationals. Withholding taxes: The protocol substantially reduces withholding taxes on crossborder financial flows. Withholding taxes on investment dividends fall from 10 to 5 per cent, on interest payments from 15 to 10 per cent, and are

The International Tax Transfer Pricing Regime 91 TABLE 2.6 Bilateral Tax Treaties in North America

1991 Canada-Mexico tax treaty

1992 U.S.-Mexico tax treaty and 1994 tax protocols

1995 Canada-U.S. tax protocol

Withholding tax rates Direct dividends1

10%

Portfolio dividends

15%

2

5% 10% (15% for 5 years)

5% (7%, 6%, then 5%) 10%

Interest

15%

4.9% (bank, 10% for 5 years) 15% (other)

Royalties

15%

10%

National treatment

(1) The tax on non-nationals shall be no more burdensome than that levied on nationals in the same circumstances. (2) The tax on a permanent establishment shall be no less favourable than that levied on residents carrying on the same activities.

10%

0

National treatment and most-favoured-nation norms

A resident of a state that is a NAFTA party may qualify for treaty benefits in certain circumstances.

National treatment for NAFTA investors Most-favoured nation (MFN)

The tax on a company owned or controlled by residents of the treaty partner shall be no more burdensome than that levied on companies owned or controlled by residents of a third country.

Restricted MFN

If Mexico signs a treaty with an OECD state setting a withholding tax on interest or royalties below 15%, Mexico grants Canada the lower rate, but not below 10% (1994 protocol).

If the U.S. signs a treaty with a third country that provides a lower withholding rate on direct dividends, both parties shall apply the lower rate (protocol).

92 The Rules of the Game TABLE 2.6 (concluded)

1991 Canada-Mexico tax treaty

1992 U.S.-Mexico tax treaty and 1994 tax protocols

1995 Canada-U.S. tax protocol

Other clauses in the bilateral tax treaty Exchange of information

The parties agree to exchange information about taxes covered by the Convention.

The parties agree to exchange information about all taxes (protocol).

The parties agree to help collect each other's taxes.

Assistance in tax collection Arbitration panels

The parties agree to exchange information about all taxes.

After 3 years the parties shall consult about exchanging notes to establish a binding-arbitration procedure for disputes that cannot be resolved by competent authority (protocol).

After 3 years the parties shall consult about exchanging notes to establish a binding-arbitration procedure for disputes that cannot be be resolved by competent authority.

NOTES: 1 Mexico does not levy a withholding tax on direct dividends. 2 The United States does not levy a withholding tax on interest payments.

eliminated on royalties.32 See Table 2.4 for a comparison with existing withholding tax rates. Exchange of information: The Canada-U.S. tax treaty allows the exchange of information on income, estate, and gift taxes between the two federal taxing authorities. The new protocol expands the exchange to cover all taxes imposed by two countries, and to allow the disclosure of information related to income or capital taxes to provincial and state tax authorities. Tax collection assistance: The protocol adds a new article, XXVIA, dealing with mutual assistance in tax collection; each country undertakes, but is not obliged, to collect the other's 'finally determined' taxes as if they were its own taxes. Secondary adjustments: Under article IX of the Canada-U.S. Treaty, if a tax authority (e.g., the IRS) adjusts the transfer prices, and thus the taxes payable, of a taxpayer, the other authority (Revenue Canada) must make a corresponding adjustment if it agrees with the adjustment and if it is notified

The International Tax Transfer Pricing Regime

93

within six years. If the requisite notice is not given to the competent authority and the taxpayer, the first authority (i.e., the IRS) cannot adjust the transfer prices to the extent that such adjustment causes double taxation. This last obligation (on the IRS) to not adjust the transfer price is removed from the protocol; the taxing authority may grant tax relief but is not obliged to do so. Arbitration: A voluntary arbitration procedure may be added to the mutual agreement article if the two parties agree; this decision is to be made three years after the protocol enters into force. Starting in January 1996, the statutory withholding taxes on intracorporate financial flows between Canada and the United States began, with the exception of interest payments, to fall to negligible levels and will soon cease to have any real impact on MNE financial decisions. Statutory CIT rates in the two countries, as Table 2.4 shows, are already similar, although there are some differences at the subfederal level and in terms of specific activities.33 Recently, both Canada and the United States have also signed bilateral tax treaties with Mexico. In addition, the North American Free Trade Agreement (NAFTA) will, by 2003, lead to the removal of all tariffs on Canada-U.S.Mexico intracontinental trade, and to the removal, reduction, or harmonization of most nontariff barriers (e.g., quotas, subsidies, preferential procurement policies). NAFTA also puts into place strong investment legislation protecting the rights of North American investors and investments in terms of the principles of national treatment and most-favoured nation. The net impact of these changes is to substantially eliminate or harmonize the national treatment of North American firms and investors in the three countries. The prospect of NAFTA was one of the factors which led Mexico to seek bilateral tax treaties with Canada and the United States, as one way to encourage inward FDI (i.e., by providing a more secure and similar tax regime for foreign investors). The 1991 Canada-Mexico Tax Treaty In 1990, Canada and Mexico signed their first information-exchange agreement, which was followed in 1991 by their first bilateral tax treaty, effective 1 January 1992. The treaty reduced Canadian withholding taxes on direct dividends paid to Mexican investors from 30 to 10 per cent; Mexico does not levy a withholding tax on direct dividends. Withholding taxes on portfolio dividends, interest income, and royalties were reduced to 15 per cent. The treaty provides several examples of a move towards harmonization at the tax level. First, national treatment is provided both in terms of taxes being no more burdensome on non-nationals than on nationals, and in terms of being no

94

The Rules of the Game

less generous to nonresidents than to residents. Second, two types of MFN clauses are introduced. The first is a general commitment to ensuring that the tax on a nonresident company be no more burdensome than that afforded to residents of a third country. For example, if the Canada-U.S. tax treaty offered U.S.-controlled permanent establishments in Canada a better tax rate than Mexican-controlled affiliates received under the Canada-Mexico tax treaty, this article would ensure Mexican affiliates received MFN treatment. Subsequently in the 1994 protocol to the treaty, an additional article was added providing Canada with partial MFN treatment for Mexican withholding taxes on interest and royalties. The 1992 U.S.-Mexico Tax Treaty and 1994 Tax Protocols The United States and Mexico signed their first bilateral tax treaty34 in September 1992; it took effect in January 1994, followed quickly by two protocols.35 As Table 2.6 shows, the withholding tax rates are generally lower than those negotiated under the Canada-Mexico tax treaty. As these rates are phased in, the MFN clauses in the second treaty should provide additional treaty benefits to Canadian investors in Mexico. One of the interesting components of the U.S.-Mexico tax treaty is Mexico's 4.9 per cent withholding tax on interest payments. Almost all interest payments flow north from Mexico to the United States; therefore any reduction in the withholding tax reduces Mexican tax revenues per dollar of interest outflows. U.S. banks, however, want a low withholding tax rate so their income falls in the general financial services basket rather than in the high withholding tax basket. Five per cent was the rate at which interest payments would have to move into the U.S. high withholding tax basket, so a 4.9 per cent rate was the maximum Mexico was able to negotiate (Morrison 1994). The U.S.-Mexico treaty incorporates the same national treatment article as the Canada-Mexico treaty. In addition, it provides a (so far unique) form of national treatment for NAF TA investors. The definition of subsidiaries eligible for benefits under the U.S.-Mexican tax treaty is any subsidiary that is wholly owned, directly or indirectly, by publicly traded companies in any of the three NAF TA countries, with a minimum 50 per cent ownership in either the United States or Mexico. Thus a 51/49 per cent U.S.-Canadian joint venture in Mexico is eligible for U.S.-Mexico tax treaty benefits (Morrison 1994, 829-31). While no general MFN clause exists, there is a restricted clause whereby the United States agrees that if it should negotiate lower withholding taxes on direct dividends with a third country, both parties will adopt that lower rate. Two interesting extensions appear in the 1994 protocols to the treaty. In anticipation of NAFTA, U.S.-Mexico crossborder flows have significantly

The International Tax Transfer Pricing Regime

95

increased and are expected to do so in the future. Therefore tax authorities on both sides of the border have become more interested in data collection for tax purposes. In one protocol the two governments agree to exchange information on all taxes, not just those listed in the Convention (which is the standard article; see the Canada-Mexico treaty for an example). Second, the two governments have agreed to discuss in three years' time the establishment of a binding-arbitration procedure for resolving bilateral tax disputes. The 1994 protocol also details how such a procedure would work.36 Summary: The North American Tax Regime The evolution of corporate income taxation in North America is interesting to study, and offers opportunities for speculation about future directions. All three countries tax foreign-source income of domestic MNEs and domestic income earned and repatriated by foreign MNEs. Excluding preferentially treated sectors (e.g., manufacturing in Canada, the maquiladoras in Mexico, possessions corporations and tax incentive zones in the United States), statutory CIT rates are roughly similar (see Table 2.5), at 44-6 per cent of taxable income. Withholding tax rates are being harmonized through BTTs and the most-favourednation clauses in the Mexico treaties. However, the three approaches do differ in significant ways. Both Canada and the United States have federal as well as subfederal CIT rates; Mexico does not. Mexico has a profit-sharing tax; the other two countries do not. Canada and the United States levy withholding taxes on dividend repatriations; Mexico does not. Canada exempts foreign-source income, if it is active business income, from taxation; the United States and Mexico tax MNEs on their worldwide income and offer a tax credit for foreign income taxes. As long as tax bases and rates differ, the potential for undertaxation or double taxation of MNE income remains. Thus, the North American tax regime is only a partial one: it is in place, it is deepening (particularly since the MexicoCanada and Mexico-U.S. treaties were signed), but it is not complete. The same can be said for the international tax regime, as we conclude below. An Assessment of the International Tax Regime We have outlined the goals, scope, principles and norms, and rules and procedures of the international tax regime. Developed on a piecemeal basis by national tax authorities to deal with the problem of overlapping tax jurisdictions caused by multinational enterprises, the goals of the international tax regime are to avoid the under- or overtaxation of corporate income. The three principles

96 The Rules of the Game underlying the regime are inter-nation equity, international neutrality, and international taxpayer equity. Achieving these principles has led to a complicated system based on the rules for nexus and the methods for coordinating source and residence taxation. International regimes are generally assessed on the basis of their strength. A regime's strength depends upon answers to questions like the following: Are the rules and procedures of the regime an accurate reflection of the underlying principles and norms? Do the actors abide by the regime's principles, norms, rules, and procedures? Is adherence to the regime in terms of numbers of states and geographic areas increasing or decreasing over time? Can the regime effectively monitor and punish offenders? Are there defectors/renegade states that do not abide by, or act so as to deliberately undermine, the regime? Are the regime's norms more honoured in the breach than in practice? Is there a hegemonic state with the willingness and capacity to underwrite the costs of the regime? Space precludes a full answer to this question in terms of the international tax regime. Suffice it to say that the evidence is mixed. The OECD and the U.S. government, in particular, have been the basic forces behind the development, support, and extension of the tax regime. The OECD model tax treaty is widely copied as a model for bilateral tax treaties. Tax treaty states commit themselves to international equity and neutrality principles, expressed in terms of the avoidance of double taxation and the prevention of tax evasion and abuse. A degree of uniformity does exist in terms of the source and residence principles as applied to the corporate income and withholding taxes; that is, the source country has 'first crack' at taxing MNE profits while the residence country is supposed to prevent double taxation. On the other hand, tax havens do exist, and there have been strong criticisms of the international tax system as it currently functions. We address each of this weaknesses below. Tax Havens: Renegades in the International Tax Regime*1 Tax havens38 are countries that enable MNEs to escape the consequences of the international tax regime. Most, with the exception of Switzerland and the Netherlands, are outside the OECD and thus may be considered outside the tax regime as it currently exists. Tax havens generally have (1) zero or low rates of income tax39 (e.g., a low tax on foreign investment or sales income and a low dividend withholding tax on dividends paid to the parent firm); (2) an absence of foreign currency controls (minimal regulation); (3) strong bank, commercial secrecy laws or admin-

The International Tax Transfer Pricing Regime

97

istrative practices which the country is unwilling to breach; (4) modern communications facilities to support financial services; (5) a stable currency; and (6) aggressive self-promotion as an offshore financial centre (IRS 1981, 14-19). Tax havens often have a disproportionately large banking and finance sector measured as a per cent of gross domestic product, or measured in terms of the size of foreign assets of deposit banks relative to foreign trade.40 A 'smell' or reputation test is also one way to recognize a haven: if it looks like a haven and taxpayers treat it as a haven, it probably is a haven. MNEs engage in transactions through tax havens for very different reasons (IRS 1981, 60-1). First, there are transactions with no tax motivation per se in that total tax payments are unaffected by the transaction through the haven. Second, a transaction may have a tax effect, but be completely within the 'letter and the spirit of the law" (IRS 1981, 61); this is tax planning. Putting the headquarters of a shipping firm in a flag-of-convenience location would be an example. Third, tax avoidance is aggressive tax planning that takes advantage of loopholes in the domestic tax system to shelter income from taxation - for example, shifting the ownership of high-profit intangible assets to tax havens, and allocating R&D expenses to high-tax locations. Tax avoidance is legal, at least on the surface. Fourth, tax evasion is the escape of legal obligations through illegal means that is, the activity is a 'sham' (an artificial transaction that unduly reduces taxable income) or it involves the illegal hiding of taxable income.41 Tax evasion can be of two kinds: evasion of taxes on legally earned income and evasion of taxes on illegal income (e.g., from narcotics). It is this fourth category, tax evasion, which distinguishes renegade states from simple free riders in the international tax regime. MNEs and private individuals use tax havens, first, because of their low tax rates, but also for other reasons such as confidentiality, freedom from currency controls, freedom from banking controls (such as domestic reserve requirements which restrict bank loans), and the attractiveness of high interest rates on bank deposits and/or lower rates on loans. In fact, tax havens may not provide much of a tax advantage to MNEs in high-tax locations. The advantage only occurs if the home country does not tax income earned in havens on an accrual (earned) basis but either exempts such income from home country tax or permits deferral of the tax until the income is repatriated. In spite of the reduced tax advantages, secrecy laws, high rates of return on capital due to minimal regulation, and low lending rates continue to be powerful magnets. Not surprisingly, tax havens only afford this special status to nonresidents. This extreme need for secrecy has led many tax havens to become subjects of controversy. Tax havens are often home to laundered and criminal

98

The Rules of the Game

money or to flight capital - 'hot money' (Friman 1994; Gilmore 1992; Naylor 1987; Palan 1994; Strange 1986). Many MNEs have set up 'letterbox" companies to collect patent royalties, licensing fees, and loan interest tax-free (Johns 1983, 64). Due to the unwillingness of tax havens to provide information on banking activities to third countries, clients can keep their financial activities hidden (whether legal or illegal). Let us define a renegade state in the international tax regime as a state that does not comply with the practices of the majority of members of the regime; that is, a tax renegade is an 'abusive tax haven' (Eden and Hermann 1995). It has some or all of the following characteristics: (1) the state has a zero or very low tax rate on business income in general; (2) domestic secrecy laws are strong and the state refuses to exchange information with other tax authorities; (3) the state actively promotes itself as a tax haven where tax avoidance and evasion practices are allowed - for example, money laundering and tax evasion are not illegal; (4) the state is known as a drug conduit state; and (5) the state does not have a network of bilateral tax treaties. We argue that the key characteristic is tight domestic secrecy laws, including the refusal to exchange information with other tax authorities. This encourages the movement of illegal activities into these havens.42 We need to distinguish free riders from renegade states. Free riders abide by the general practices of regime members but cannot or do not contribute their proportionate share to the costs of regime maintenance (the collective costs). For many developing countries, building the tax infrastructure needed to comply with the OECD's practices would be an onerous burden. Such small states are likely to be free riders. Renegade states, on the other hand, do not comply with the regime's practices. Some renegade states are inside the international tax regime because they are members of the OECD. These states would include Switzerland43 and the Netherlands (through the Netherlands Antilles). Others are outsiders (non-OECD members). These would include most Caribbean tax havens (e.g., Bahamas)44 together with countries such as Hong Kong and Liberia. The reasons why some states become renegades in the tax arena, engaging in abusive tax behaviour, may be due to simple economic motivation. Small, poor states lacking natural resources or other obvious attractions to foreign direct investment may turn to tax haven status in order to induce inflows of foreign banking and commercial activities. Historical ties with rich countries that included preferential status for their investments in the poorer partners also encourage low tax rates since the home country effectively engages in 'tax sparing'; the effective tax rate is therefore the host country rate (e.g., Puerto Rico and the United States). These motivations, however, suggest reasons for tax

The International Tax Transfer Pricing Regime

99

haven status per se, and not for abusive status. We hypothesize that the more important are criminal elements (e.g., proximity to narcotics-producing countries so that this state becomes a transit state), the more likely is the tax haven to become a renegade state. Tight secrecy laws and unwillingness to exchange information also encourage abusive haven activities. In terms of dealing with renegade states in the international tax regime, it is clear that reducing tax evasion and avoidance on a global basis cannot be accomplished by individual states. The Gordon report (IRS 1981) notes in this regard: The United States alone cannot deal with tax havens. The policy must be an international one by the countries that are not tax havens to isolate the abusive tax havens. The United States should take the lead in encouraging tax havens to provide information to enable other countries to enforce their laws ... However, such steps taken unilaterally would place United States businesses at a competitive disadvantage as against businesses based in other OECD countries. Accordingly, a multilateral approach to deal with tax havens is needed. (IRS 1981,10)

On the other hand, most of the legislation so far has been domestic. Many OECD countries have enacted domestic tax rules designed to lessen the attractiveness of tax avoidance and evasion through tax havens. Eliminating tax deferral for foreign branches and subsidiaries in 'blacklisted' countries, or for certain types of passive income earned in these locations, is one common approach (e.g., the U.S. Subpart F rules,45 the Canadian FAPI rules). Transfer pricing regulations (e.g., section 482) that ensure intrafirm prices must be based on the arm's length standard are another method of reducing the possibility of tax avoidance. Doctrines of sham (artificial transactions designed so as to unduly reduce taxes) and general anti-avoidance legislation are also used, together with tax regulations that shift the burden of proof to the taxpayer and/ or require substantial amounts of information about the transactions from the taxpayer. Refusing to sign tax treaties with haven states unless they commit to information exchange and anti-abusive rules is another common response. Many tax havens have historically had a special relationship with an onshore economy, often as remnants of the colonial period. The Netherlands Antilles, for example, has taken advantage of its privileged relations with the Netherlands; as a Dutch protectorate, the 1948 U.S.-Netherlands tax treaty was applicable on its territory. A similar arrangement existed between Britain and numerous Caribbean islands. When many British colonies achieved nation status in the 1960s, the 1945 U.S.-British tax treaty was simply extended to these new nations (IRS 1981, 149).

100 The Rules of the Game TABLE 2.7 U.S. Bilateral Tax Treaties with Tax Havens, 1995 U.S. tax treaties with this country [date signed in brackets]

Exchange of information Bahamas Barbados Bermuda Costa Rica Hong Kong Isle of Man Jamaica Liberia Libya Liechtenstein Luxembourg Netherlands*

Taxation of shipping and/or air transport income

Income tax treaty-notesprotocol

X [1987] X [1984, 1991] insurance income only [1986]

X [1984] X [1988] X [1989-91] X [1989] X [1989] X [1988]

X [1980-81] X [1982, 1987]

X [1989, 1993] X [1926]

Panama

X [1941, 1987]

Singapore Switzerland Trinidad & Tobago

X [1985, 1988] X [1989-90]

X [1962] X [1948, 1955, 1963, 1965, 1986, 1987, 1992-3] withholding tax only [1963] X[1951] X [1966-70]

Number of countries signing treaties with this country

1 8 1 1 1 2 9 3 5 1 28 65 1 32 62 11

*U.S.-Netherlands income tax treaty extended to Netherlands Antilles [1955, 1963]; benefits terminated in 1987. Benefits to Aruba terminated in 1995. SOURCE: Calculated from data in Tax Analysts (1995)

As a result of the 1981 Gordon report, the U.S. government terminated several tax treaties with Caribbean havens. New treaties are only negotiated if there is a strong exchange-of-information clause attached that overrides foreign bank secrecy laws in the tax havens.46 U.S. bilateral treaties are to be restricted to residents of a treaty country, so that 'treaty shopping' cannot be used by nonresidents to gain the benefits of the tax treaty (IRS 1981, 12-13).47 In the absence of a tax treaty, firms located in these states do not get treaty benefits - in particular, access to bilateral dispute-settlement procedures is denied. MNEs in these countries face much higher withholding tax rates and lose other tax privileges. Table 2.7 provides some evidence on U.S. bilateral tax treaties with haven

The International Tax Transfer Pricing Regime

101

countries, as of 1995. It is clear from the table that the United States has primarily shipping and air transport treaties with tax havens, other than countries that are OECD members (Switzerland, Netherlands), and that these treaties have mostly been signed in the late 1980s. In general, few countries have signed income tax treaties with havens (see the last column of Table 2.7). Non-tax haven countries also find themselves in an extremely hypocritical position, for it is their citizens that make the most use of tax havens. In the majority of instances, it is OECD companies and/or elites that carry out businesses and maintain their private savings in tax havens. Tax havens thrive precisely because of the existence of foreign banks and service companies, largely from non-tax haven countries. The so-called 'high tax' countries are, therefore, under pressure from their own financial sectors to ensure a certain regulatory laxity (Palan 1994, 11). It is as though tax haven and high-tax countries have implicitly agreed to allow for a certain amount of deviant behaviour within the global financial system.48 In summary, tax haven countries are a 'hole' in the international tax regime areas where the principles, norms, rules, and procedures of the regime do not apply. There are strong political economy motivations on the parts of haven governments and multinationals for these holes to exist. However, they represent a clear weakness in the regime. Is There an International Tax Regime ? There have been strong criticisms made of the international tax system as it currently functions. One long-time critic, Richard Bird, a well-known Canadian economist and international finance expert, argues that strictly speaking, there is no such thing as an international tax system. No law limits national jurisdiction. Each country may adopt whatever taxing rules it sees fit (whether it can enforce the rules it adopts is quite another matter)... Often ... the key features affecting international income have been accidental ... [or] they have been additions patched onto the system to cope with specific problems as they became apparent... In yet other instances, countries seem simply to have copied such complex rules as those on transfer pricing and controlled foreign corporations from the United States, which has been the dominant exporter of both capital and tax policy notions throughout most of the postwar period. The taxation of international income ... has thus developed more by chance than by design. The present international tax order as a whole is a patchwork structure that makes little sense in terms of its purported objectives. (Bird 1988, 293)

Bird goes on to argue that bilateral tax treaties are principally designed to

102

The Rules of the Game

reduce withholding tax rates on payments to nonresidents because capital exporting countries like the United States have a vested interest in negotiating lower withholding taxes on payments from foreign affiliates to their parent firms. At the same time, tax competition among host countries to attract inward foreign investment also persuades these countries to reduce their withholding taxes; once a few countries reduce or eliminate withholding taxes, others follow in order to prevent capital flight. Interest payments, for example, are now generally free from withholding tax by source countries for this reason. And, given the fungibility of different categories of financial payments and the freedom of the MNE to choose the way it remits income to the parent firm, the net effect may be to reduce the effective withholding tax on foreign portfolio income to near zero. The net result is to encourage outward investment at the expense of domestic investment. From the perspective of the home country, the taxing authority must make the following choices: between exempting or taxing foreign-source income, between taxing on an accrual or deferral basis, between a foreign tax deduction or credit. Where foreign-source income is tax exempt (as it is in Canada), the residence country simply cedes jurisdiction to the host country. This however encourages outward investment in low-taxed countries and favours foreign capital over domestic investment. Exemption is likely to encourage the growth of tax havens since the effective tax rate for local investment is not the home rate (as it would be under an accrual with foreign tax credit scheme) but the host rate, and competition among host countries for investment will most likely drive down tax rates. Tax deferral, Bird argues, is equivalent to tax exemption with a penalty added for repatriation, so it makes even less sense than exemption. His conclusion therefore is that 'the present treatment of international capital flows is inefficient and inequitable" (Bird 1988, 295). He goes on to propose several radical solutions: (1) replacing the corporate income tax with a consumption-based business tax; (2) eliminating tax deferral; or (3) moving to a global formulary apportionment system (unitary taxation) for allocating MNE worldwide profits. His preferred solution is the third because of the integrated nature of the multinational enterprise. Is the Glass Half Empty or Half Full? Our view in this book is not as pessimistic as Bird's. We do see a structure to the international tax system that has the characteristics of an international regime, with regime supporters, underlying principles, an international organization at its centre, and so on. Clearly, there are problems with the way the regime functions in practice, problems that the OECD's Committee on Fiscal

The International Tax Transfer Pricing Regime

103

Affairs, the International Fiscal Association (IFA), and the competent authorities of various countries debate regularly. In the absence of this structure, the system would most likely rapidly degenerate into wholesale international tax competition, with significantly higher non-neutralities and inequities compared with the current international tax regime. In part, the issue is the base comparison with which the current structure is compared: an ideal world with complete international neutrality and equity (however defined) or no regime at all. Transfer pricing issues are clearly part of this regime. MNEs through their intrafirm trade in intermediate goods, technology, and services create opportunities for overlapping tax jurisdictions. Governments have developed specific principles, norms, rules, and procedures with regards to the taxation of transactions among related parties. These form the basis of the tax transfer pricing regime, to which we now turn. The International Tax Transfer Pricing Regime Nested within the international tax regime is the international tax transfer pricing (TTP) regime, centred around the international norm of the arm's length standard. Government cooperation in the transfer pricing area is based on a variety of national corporate income tax regulations, BTTs, and model tax treaties. The OECD's Committee on Fiscal Affairs and the International Fiscal Association have also played important roles in the TTP regime.49 The characteristics of the tax transfer pricing regime are outlined in Box 2.4 and discussed below. Box 2.4 is set up in the same manner as Box 2.1 in order to facilitate comparisons between the international trade, tax, and tax transfer pricing regimes. The Problem: Allocation of Income and Expenses Because their activities cross national borders, multinational enterprises fall under the jurisdiction of more than one tax authority. MNEs therefore create problems in regulation; in particular, the integrated nature of the enterprise makes it difficult for governments to devise tax rules to allocate income and expenses among the units of the MNE. As Jill Pagan and Christopher Wilkie, two well-known international tax lawyers, state: Transactions where transfer pricing is relevant are increasing rapidly and the tax authorities are having to devote more and more resources to dealing with transfer pricing inquiries. The first stage towards finding a solution is to recognize the difference between tax authority thinking and commercial thinking. Tax authority thinking is national, not

104 The Rules of the Game BOX 2.4 The International Tax Transfer Pricing Regime

Purpose

Scope

Principles Norms

Rules

Procedures

To avoid double taxation of MNE income and prevent tax avoidance and evasion caused by overlapping tax jurisdictions. Seen as problems that lead to distortions and inequities. Issue area: the appropriate allocation of income and expenses among members of a related group of businesses located in different jurisdictions; the valuation of all crossborder transactions among related parties. Geographic area: OECD members and their bilateral tax treaty partners. Three principles: Inter-nation equity, international tax neutrality, and international taxpayer equity. The arm's length standard: each unit of the MNE is expected to declare, for tax purposes, the profits it would have made had it been a distinct and separate enterprise operating at arm's length from its parent and sister affiliates. Governments adopt a transactions-based, water's-edge approach to allocating the MNE's income and expenses among jurisdictions. Different rules or methods apply for valuing goods, services and intangibles, but all follow the arm's length standard. Acceptable methods include comparable uncontrolled price (CUP), cost plus (C+), resale price (RP), and fourth methods. Domestic procedures: Auditing process with appeals through the domestic courts Publication of rules, procedures, acceptable methods Functional analysis to evaluate the functions of the MNE's entities Advance Pricing Agreements Penalties and documentation regulations to ensure compliance. International procedures: Bilateral tax treaties based on OECD Model Tax Treaty defining jurisdiction Mutual agreement procedure using competent authorities for interjurisdictional disputes International arbitration Exchange of information among tax authorities Simultaneous examination procedures.

The International Tax Transfer Pricing Regime

105

global, and principally uses methodology for establishing individual transaction profit. By contrast, the commercial thinking of the MNE is global, not national, and the emphasis is on consolidated accounts or results. (Pagan and Wilkie 1993, 26-7) The characteristics of the MNE that we identified earlier - common control, common goals, and common resources - all complicate international valuation of the MNE revenues and expenses, and thus the taxation of its worldwide profits. Transfer prices, as we saw in Chapter 1 are unlikely to be the same prices arm's length parties would negotiate. The prices of traded tangibles, intangibles, and services within the various units of the enterprise are basically accounting or bookkeeping prices set for internal reasons. However, since MNE activities cross national borders, transfer prices must be provided to tax authorities and used to calculate both border taxes (tariffs, export taxes) and corporate income taxes. Therefore internal and external factors will influence the MNE's choice of transfer prices. The fear of tax authorities is that external factors will dominate and the MNE will set its transfer prices so as to avoid or evade taxes. However, tax avoidance is not the only reason national authorities regulate transfer pricing. As Jeffrey Owens, Head of Fiscal Affairs at the OECD, admits, the OECD and national tax authorities have never regarded transfer pricing issues as being 'mainly about tax avoidance.' ... Neither the 1979 report nor our current work assumes that all multinational enterprises (MNEs) are manipulating transfer prices to minimize their tax liability, although it would be naive to believe that the temptation is not there and that none will succumb. (Owens 1994, 877)

Even if MNE transfer prices are set for reasons other than taxes, genuine disputes between tax authorities and multinationals can occur as to the proper valuation of the revenue and expenditures incurred by the various affiliates of the MNE around the world. We expand on this below. Purpose and Scope The TTP regime was developed in order to deal with the complexities of determining the appropriate allocation of MNE revenues and expenses, at the national level, in a global economy.50 The concern is that MNE transfer pricing policies might distort these allocations and thus not accurately reflect an 'appropriate' amount of taxable profits to individual tax authorities, and, in particular, that profits might be too low so that undertaxation occurs. The purposes of the TTP regime are the prevention of tax abuse (i.e., the underpayment of taxes) and double taxation of income (i.e., the overpayment of

106 The Rules of the Game taxes). The 1979 OECD report on transfer pricing clearly enunciates these objectives. The first statement below focuses on undertaxation of income, the second on overtaxation, as rationales for government intervention. Thus the purpose of the TTP regime is the 'proper' taxation of income. On undertaxation: Multinational enterprises may adopt transfer prices which are not arm's length prices in order to minimize tax ... or they may adopt them for other reasons, but whatever the reason, whenever intra-group transfers are not carried out at arm's length prices, the result is likely to be that profits are shifted from one company to another in the group and the tax liability of the relevant companies distorted in consequence. Since national tax authorities need to determine the proper allocation of taxable profits of the affiliated enterprises operating within their respective jurisdictions, the transfer pricing policies of MNEs are of great importance to them. (OECD 1979, 8)

Therefore, if MNEs choose transfer prices to minimize tax, the authorities have the right to intervene; that is, the first goal of the tax transfer pricing regime is to eliminate undertaxation through transfer price manipulation. And, even if the MNE's choice of a transfer price was not chosen to minimize taxes, the tax authorities can also intervene to impose the arm's length price. Thus transfer prices set even for legitimate internal or external reasons can be overturned for tax purposes; motive is not a relevant criterion (see also Pagan and Wilkie 1993, 52-3). The second rationale is avoidance of double taxation. On overtaxation, the 1979 OECD report states: In the organization of their intra-group relations MNEs are necessarily confronted with transfer pricing problems - essentially a price has to be charged for every transaction ... One of these problems could be the danger of double taxation, if national authorities differ in their approach for tax purposes. It is therefore of importance to multinational enterprises also that common approaches to the resolution of transfer pricing problems should be developed. (OECD 1979, 9)

The overall impact is that the tax authorities have the right to adjust the MNE's income and expenses whenever the authorities believe the firm has not chosen a 'proper' allocation of income for tax purposes. Clearly, disputes are likely to arise with such a broad definition of transfer pricing problems. This leads us directly into the question of the scope of the TTP regime. The scope of the TTP regime is broad in both issue area and geographic area. The issue area covers the valuation of all activities of multinational enterprises that affect their

The International Tax Transfer Pricing Regime

107

global income, and thus their tax base. Thus the valuation of tangibles, intangibles, and services is included within the scope of the regime. As we have seen above, any valuation that might result in an 'improper' allocation of taxable income can be questioned by the tax authorities. In terms of geography, the scope of the TTP regime is less broad. All members of the OECD have adopted the 1979 OECD report on transfer pricing, although they may apply the report in varying ways and to varying degrees (see Pagan and Wilkie 1993).51 Langbein (1986), in particular, argues that the arm's length principle is more honoured in the breach than in practice. Outside the OECD, there is little legislation in this area, although many governments historically have been concerned that MNEs might be using transfer prices to avoid taxes and other regulatory barriers such as foreign exchange controls (UNCTAD 1978). In 1994, the OECD began a dialogue with the East Asian governments and with the newly reforming countries of the ex-Soviet Union, to help these countries develop transfer pricing guidelines based on the OECD standard. Thus we anticipate that the geographic scope of the regime will spread. Principles and Norms Principles: Equity and Neutrality The principles of the TTP regime are the same ones that characterize the international regime - that is, inter-nation equity, international neutrality, and international taxpayer equity. National governments are concerned with transfer pricing for two reasons. As explained above, transfer pricing can be used to avoid or evade taxes. Where tax havens exist or MNEs can find tax loopholes to reduce their tax burden through under- or overinvoicing intrafirm trade flows, both international taxpayer equity and international neutrality are compromised. A domestic firm dealing at arm's length with another party, even if the party is located in a tax haven, cannot arrange its transactions in this manner. Thus taxpayers in the home country are not being treated equally and taxpayer equity is not achieved. The choice of investment location is also affected, as more investment is directed into low-tax activities, so the tax neutrality principle is also violated. Second, where government regulations of transfer pricing differ, double taxation is also possible. In such cases, international neutrality and equity are also violated. As Jill Pagan recently noted: No two countries have exactly the same (or same combinations of) tax rates, tax bases, levels of withholding taxes, and tax treaty networks and provisions. As long as this state of affairs exists, there are tax arbitrage possibilities. (Pagan 1994b, 1391)

108 The Rules of the Game Therefore national regulation of transfer pricing, by itself, cannot satisfy the basic principles of public finance: tax neutrality and equity. The need for intergovernmental cooperation is clear, and the tax transfer pricing regime was developed for this purpose. Norms: The Arm's Length Standard The fundamental norm or standard behind the current TTP regime is the arm's length standard.52 Every member of the OECD has adopted arm's length as the basic standard for valuing MNE income and expenses. As Stanley Langbein (1986) shows in great historical detail, the arm's length standard as an international norm developed through two distinct historical episodes.53 We would update his analysis by arguing that a third period began in 1994 with the finalization of the U.S. section 482 regulations and the first draft of a new OECD report on transfer pricing. The first period, the 1920s through the early 1960s, was characterized by meetings of experts held first under the auspices of the Financial Committee of the League of Nations, and subsequently under the OECD's Committee on Fiscal Affairs and the United Nations. During this period, the separate entity or independent enterprise standard and the first tax convention models were developed by the League of Nations in 1928. The models were integrated and revised at a Mexico meeting in 1943 (becoming the precursor of the United Nations model tax convention) and in London in 1946 (the precursor of the OECD model). The second period, tentatively dated 1968 to 1993, began when the U.S. Treasury under Assistant Secretary Stanley Surrey developed the methodology of the arm's length standard (published in 1968 as the Internal Revenue Code's Section 482 Regulations) and persuaded the OECD Committee on Fiscal Affairs to adopt the transactions-based approach to this standard. Under the transactions approach, an integrated group of business enterprises (a multinational) is separated according to transactions between related parties, with each of the transactions being valued according to the arm's length standard. The 1977 OECD Model Tax Treaty and its 1992 update incorporate an arm's length standard for allocating income between firms and their subsidiaries, parents, or sister enterprises.54 Article 9 of the 1977 Model Tax Convention defines the arm's length standard. The full quote is in Appendix 2.1, but we can paraphrase it here as: Each unit of the MNE is a separate legal entity for tax purposes. The entity is expected to declare, for tax purposes, the profits it would have made had it been a distinct and separate enterprise operating at arm's length from its parent and sister affiliates. Where the affiliate has not done so, a taxing authority may reallocate the profits and tax the entity accordingly. In 1979, the OECD's Committee on Fiscal Affairs issued its first transfer

The International Tax Transfer Pricing Regime

109

pricing guidelines on the allocation of income and expenses between related enterprises (OECD 1979). The guidelines, developed over five years, apply generally to all MNE transactions. The model tax conventions (see Appendix 2.1) are designed to deal primarily with international double taxation, while the concerns of the transfer pricing guidelines focus not only on double taxation but also on tax abuse and, more generally, on the 'proper allocation' of income among countries. The transfer pricing guidelines, which are meant to apply to all MNE transactions, endorse the arm's length standard for allocations among related parties, although, somewhat surprisingly, the report does not contain a formal definition of the arm's length principle. Four regulatory methods are set out in the report: comparable uncontrolled price, resale price, cost plus, and other methods. The report argues strongly against the use of formulary apportionment (unitary taxation). These guidelines have been widely adopted as the basis for transfer pricing regulations by OECD member countries. In 1984, the OECD published another report dealing with three specific issues in transfer pricing: (1) corresponding adjustments and Mutual Agreement Procedures, (2) the taxation of multinational banking enterprises, and (3) the allocation of central management and service costs. Due to its specific nature, this report has had much less circulation, or impact, compared with its earlier sister document. In addition, parts of the document were controversial (e.g., the allocation of management fees) and were not adopted by certain countries such as Canada. See Chapter 5 (the simple analytics of transfer pricing) and Chapter 10 (the Canadian regulations) for more details. The third period is perhaps now just beginning. A six-country group within the Committee on Fiscal Affairs has been working on new transfer pricing guidelines for MNEs and tax administrations, updating the OECD's 1979 and 1984 transfer pricing reports. Preliminary versions of the report were published in 1994 and 1995, consisting of three documents: Part I: Principles and Methods; Part II: Applications; and Part III: Special Topics (OECD 1994b, 1995a,b, 1996, forthcoming; Hay et al. 1994). The draft report is heavily influenced by the new 1994 section 482 regulations (see Chapter 13 for details). It endorses the arm's length principle as the international transfer pricing standard for tax purposes (OECD 1994b, 160), provides several justifications for this standard, and recommends against the use of formulary apportionment. Although there was no formal definition of the standard in the OECD 1979 report, the 1994 transfer pricing guidelines endorse the definition in article 9 of the 1992 OECD Model Tax Convention. We discuss the OECD's new proposals in chapters 5 (simple analytics of transfer pricing) and 13 (rules and procedures). The arm's length standard is not the only norm that could be used to guide the international tax transfer pricing regime. There is an alternative to the arm's

110

The Rules of the Game

length standard: the global or unitary method of taxing the MNE. Under this norm, what we could call the integrated enterprise standard or global formulary approach, the MNE's worldwide income would be taxed and allocated among countries according to a formulary approach. Respected international public finance economists such as Richard Bird and Charles McLure and international tax lawyers such as Stanley Langbein have been strongly supportive of unitary taxation (see Chapter 12 for details). However, both the OECD and the United Nations have been quite hostile to global methods of taxing MNE income, preferring to rely on the transactions-based, arm's length standard. The OECD argues that global methods are incompatible with articles 7 and 9 of the Model Tax Treaty. Moreover: Proposals for radical reformulations of the approach to intragroup transfer pricing would move away from the arm's length standard towards so-called global or direct methods of profit allocation ... are not endorsed in this report ... Such methods would necessarily be arbitrary, tending to disregard market conditions as well as the particular circumstances of the individual enterprises and tending to ignore the management's own allocation of resources, thus producing an allocation of profits which may bear no sound relationship to the economic facts. (OECD 1979,14)

We will come back to the issue of global methods for taxing multinationals in Chapter 12.55 Rules and Procedures Rules: CUP, Resale Price, and Cost Plus The rules of the tax transfer pricing regime have to do with putting the arm's length standard into practice. In general, the arm's length standard has been interpreted to mean arm's length pricing of individual transactions.56 Under the corporate income tax, governments normally tax the net income of firms located in their jurisdictions, minus any tax deductions or credits. Net income is defined as gross revenues (product sales to households and other firms, royalty income, licence fees, etc.) minus cost of goods sold (factor costs, materials purchased from other firms), general expenses, and other allowable expenses. Where firms are unrelated, governments take the firm's prices as market or arm's length prices and accept the transactions as being determined in the marketplace. However, where the firms are related, the MNE must prove that its transfer prices are equivalent to those that would have been negotiated by unrelated parties engaged in comparable transactions or the tax authorities will substitute their calculation of arm's length prices.

The International Tax Transfer Pricing Regime

111

Many different pricing methods are consistent with the arm's length standard, but four methods are the most widely adopted by tax authorities: the comparable uncontrolled price, resale price, cost plus, and 'fourth or other' methods. These are transactions based methods in that they are based on pricing individual transactions in accordance with the prices that unrelated parties in similar circumstances would have negotiated. These four methods were the formulas specified for intrafirm trade in tangibles in the 1968 U.S. Treasury regulations. The 1979 OECD transfer pricing report recommends that its members adopt these U.S. methods not just for tangibles but also for transactions in services and intangibles. The traditional, transactions-based methods (CUP, resale price, cost plus) are also recommended in the draft OECD transfer pricing guidelines (1994b, 1995a,b, 1996), but the guidelines suggest occasions when moving beyond these methods may be necessary (e.g., global trading). The profit split method and the transactional net margin method (which is similar to the comparable profits method) are specified as alternatives, to be used as a last resort when the traditional methods do not work. The guidelines focus on comparability of transactions - that is, the arm's length price is determined through comparisons with the pricing of transactions between unrelated parties. Transactions should be comparable in terms of: the characteristics of the property or service, functions performed by the parties, contractual terms, economic circumstances, and business strategies of the firms. Procedures: Dispute Settlement As far as procedures are concerned, several methods are available at both at the national and international levels to facilitate common standards and rules across tax authorities (see Box 2.4). These include, at the domestic level, auditing and appeals processes, the use of functional analysis by the tax authorities to evaluate MNE activities, the introduction of Advance Pricing Agreements, and documentation requirements and penalties. The OECD report also recommends the use of functional analysis - a direct survey of the contributions an enterprise makes to the overall MNE - as important to determining the facts and circumstances of the case. At the international level, procedures include bilateral tax treaties that include the Mutual Agreement Procedure (MAP) to settle inter]urisdictional disputes, possible international arbitration of disputes, and information exchanges among tax authorities. BTTs are probably the most important of these international procedures. Tax treaties can affect transfer price regulation in one of four ways. First, they define a particular basis for allocating income and help establish interna-

112 The Rules of the Game tional standards for the treaty network as a whole. For example, the OECD Model Tax Treaty defines associated parties, outlines the arm's length pricing and the various pricing methods, and recommends corresponding adjustments to eliminate double taxation (see Appendix 2.1). Second, tax treaties identify transactions to which the basis will apply, such as trade between related parties in goods, services, and intangibles. Third, treaties provide for resolution of disputes, and fourth, they provide for mutual assistance between the tax authorities. This latter involves the mutual agreement procedure and exchanges of information. The MAP, as Kwatra (1988) notes, has been primarily used to settle transfer pricing disputes. The North American Tax Transfer Pricing Regime Just as we argued above that a North American tax regime was forming at a regional level within the international tax regime, so too can we argue that a North American tax transfer pricing regime is forming at a regional level within the international TTP regime. We outline the main features of this North American TTP regime below. The U.S. Approach to Transfer Price Regulation51 The most important part of the U.S. corporate income tax law in the transfer pricing area is Internal Revenue Code section 482, which applies to all intracorporate transactions, tangible and intangible. The U.S. regulations, first developed in 1968, identify five types of intrafirm transactions: loans, rentals, sales of tangible property (i.e., goods), transfer or use of intangible property (e.g., patents, copyrights), and performance of services (e.g., managerial, technical). Section 482 requires that the income earned on transactions between related parties be determined on an arm's length basis. Sales of tangible property are tested against an arm's length standard based on one of four methods (in order of priority): comparable uncontrolled price (CUP), resale price (RP), cost plus (C+), and the so-called 'fourth methods.'58 The most difficult problems associated with section 482 arise in the pricing of intangibles, particularly where non-U.S. MNEs are involved since information is less readily available. The IRS and the U.S. courts often use the fourth method where a functional analysis is used to split profits on the transaction between the related parties. A recent tax change is the introduction of section 1059, which requires transfer prices on import transactions between related parties to not exceed those prices used for U.S. customs valuation purposes. The U.S. customs value therefore becomes a quasi-fifth method (IRS 1988, 520).

The International Tax Transfer Pricing Regime

113

Since the early 1980s the United States has engaged in major - and frequent - revisions to its transfer pricing regulations. In 1986 Congress added the 'commensurate with income' standard to section 482, making it applicable to valuation of intangibles.59 As a result, the revised section 482 now requires the tax authority to allocate the actual profit from the intangible to the related parties in proportion to their contributions to that income. In 1988 the Treasury White Paper suggested that a functional analysis based on arm's length rates of return (the basic arm's length return method, or BALRM) should be used to satisfy the commensurate with income standard for intangibles (U.S. Department of the Treasury, 1988, Chapter 11). Where both marketing and manufacturing intangibles are involved, they should be separated and the residual income after the arm's length rate of return analysis should be split between these categories. After the White Paper, the Internal Revenue Service, in 1992, issued proposals for reforming section 482 based on the comparable profits method. The proposals were roundly condemned by domestic and international experts, including the OECD, as inconsistent with the arm's length principle (OECD 1993a,b). In February 1993, the IRS issued temporary regulations reaffirming the U.S. government's commitment to the arm's length standard, revising and clarifying its proposals but keeping a modified form of the computed profit method. In June 1994 the Treasury issued final regulations similar to the temporary ones (see Chapter 8 for more details). A recent procedure introduced by the IRS is the Advance Pricing Agreement (APA). This policy allows a multinational to sit down with the IRS and to negotiate an acceptable transfer pricing policy in advance of the actual transactions. This pricing policy remains in force for up to three years, after which the IRS and MNE can renegotiate the APA. In effect, the IRS is setting up a 'safe harbour' for specific transactions with particular multinationals. The first APAs are being negotiated primarily by MNEs, which are already subject to repeated audits by IRS investigators (e.g., Japanese auto and consumer electronics MNEs). The Canadian Approach to Transfer Price Regulation^ The Canadian regulations, section 69 of the Income Tax Act together with Information Circular 87-2, are much less developed than the corresponding U.S. regulations. Section 69(1) of the act is designed to prevent related domestic firms from artificially shifting income and/or deductions among their divisions. Sections 69(2) and 69(3) apply to international transactions; 'reasonable under the circumstances' is the criterion for ensuring arm's length transactions.

114

The Rules of the Game

Section 69(2) insists that intracorporate crossborder payments do not exceed a reasonable amount, whereas 69(3) insists that such receipts are not less than a reasonable amount. Revenue Canada basically follows the approach outlined in the OECD 1979 report on MNEs and transfer pricing (OECD 1979, Messere 1979). Although the 1987 Canadian tax reform did not directly involve transfer pricing, Revenue Canada issued Information Circular 87-2, which was designed to set out its approach to applying section 69.61 The circular defines 'fair market value' in 69(1) and 'reasonable under the circumstances' in 69(2,3) as the same and equivalent to the arm's length price. The circular states that the primary method for calculating arm's length prices is the comparable uncontrolled price. Other methods include resale price and cost plus. A functional analysis is recommended when exact comparables do not exist. Revenue Canada has also recently introduced its own Advance Pricing Agreement based on the U.S. model. Thus the Canadian rules and regulations, while not as detailed as their U.S. counterparts, do follow the same path. The Mexican Approach to Tax Transfer Price Regulation Mexico, until 1991, did not have a transfer pricing standard in its tax code, although the Mexican tax authority has generally recognized the OECD standards since 1976 (del Castillo et al. 1995). The 1991 amendments, under article 64A of the corporate income tax code, now grant specific authority for applying the arm's length standard to transactions involving the use of funds; rendering of services; the use, enjoyment, or disposal of tangible assets; and the use of intangible property (McLees 1992, 999, n. 19). In 1994, the transfer pricing provisions were amended and four methods introduced: CUP, RP, C+, and profit splits (del Castillo et al. 1995, All 1-15; Perez de Acha 1994, 624-5). Profit splits are to be applied solely to permanent establishments and fixed bases in Mexico of residents abroad. The method allocates the total profits of these residents in proportion to the income or the assets these establishments represent in Mexico to the total. Mexico began transfer pricing audits in 1994, after hiring and training tax auditors, with the assistance of the IRS (Matthews 1993a, 233). The first transfer pricing adjustment was collected by the government in October 1994. The Mexican CIT and transfer pricing rules were initially not intended to apply to the maquiladoras because the Mexican government did not treat them as typical permanent establishments. They were treated as cost centres for purposes of calculating the Mexican corporate income tax and were also exempt from the business assets tax.62 This situation has now changed.

The International Tax Transfer Pricing Regime

115

The Mexican tax authority, Secretaria de Hacienda y Credito Public (Hacienda, the Mexican counterpart to the IRS), began to apply the new transfer pricing rules to the maquiladora plants effective January 1995, requiring that prices for maquiladora exports be set according to the arm's length standard. In effect, this meant that profits on maquila operations would be shared between the U.S. and Mexican tax authorities. Hacienda, however, agreed not to audit a maquila for compliance if its taxable income meets a five per cent of asset value test.63 In March 1995, the government changed the maquilas' exemption from the Mexican business assets tax. The maquilas could continue to avoid the business assets tax but only if (1) they paid taxes amounting to at least five per cent of the value of assets employed in the maquila, or (2) the firm obtained a ruling from Hacienda that a lower transfer pricing would satisfy Mexico's transfer pricing regulations. Just over half of the 2,200 maquiladoras opted to elect one of the two options instead of paying the tax; of those, only 20 per cent chose to ask for a ruling from Hacienda (Fernandez 1995, 1276-7). In 1994, Mexico initiated an Advance Pricing Agreement process and has encouraged firms to apply for rulings. Formal APA regulations were issued in July 1995. The new rules recommend the 'return on capital employed' method to establish an arm's length price - in effect, treating the maquila like a contract manufacturer for its U.S. parent.64 The first APA was issued in November 1995, and used a cost plus approach with comparables from a group of U.S. service providers, on the grounds that the maquila was in the business of providing labour services (Fernandez 1995, 1276). It is clear from the above that Mexican transfer pricing regulation is coming more and more to resemble the U.S. regulations.65 Since Canada's rules are also based, like the U.S. ones, on the arm's length standard, does this mean that the potential for tax transfer pricing disputes should lessen, so that the concerns expressed by MNEs in the Ernst & Young survey are temporary? In Chapter 7 we return to this issue and provide estimates of the incentives for tax transfer price manipulation; these estimates suggest in fact that the reverse may be the case. Summary: The North American Tax Transfer Pricing Regime The United States, as the headquarters for one-third of the world's multinationals and simultaneously the major attraction of inward foreign direct investment in the 1980s, has been the leader in developing new rules and procedures to deal with these large, integrated businesses in a globalized economy. However, the U.S. rules and procedures in the transfer pricing area change constantly and are often driven by domestic political processes, rather than by the need to develop

116 The Rules of the Game internationally acceptable standards. As a result, the United States has also been the most destabilizing member of the international tax transfer pricing regime. Canada, on the other hand, has played quite a different role in the TTP regime. Canada has been active as a major host country to U.S. multinationals and as a middle power within the regime, supporting the development of international tax norms and principles, working to strengthen its own rules and procedures in line with the arm's length principle, and adopting defensive mechanisms to deal with the capricious and rapidly changing policies of the U.S. tax authorities. The third country, Mexico, is new to the game, and, not surprisingly, is taking its rules and procedures from the OECD and U.S. practice. The government has clearly adopted the arm's length standard and is moving to establish detailed policies and dispute-settlement mechanisms (through its BTTs) to deal with transfer pricing issues. An Assessment of the International Tax Transfer Pricing Regime We have argued above that there exists an international tax transfer pricing regime nested within the international tax regime. Section 482, the OECD Model Tax Treaty, and the OECD reports on transfer pricing have been important in the development of this regime. Most members of the OECD adhere to the arm's length standard and have developed regulations loosely based on the U.S. regulations or the 1979 OECD report. The TTP regime has its own norm (the arm's length standard), principles (international equity and neutrality), rules (the four methods), and procedures (e.g., competent authority rules, advance pricing arrangements, appeals and arbitration). The rest of this book is devoted to an analysis of the international TTP regime, focusing in particular on the roles played by the OECD, the U.S. Treasury, and Revenue Canada. Our assessment of the overall regime - its strengths and weaknesses - can be found in chapters 12 (principles and norms of the regime), 13 (rules and procedures of the regime), and 14 (conclusions). Conclusions The main goal of Taxing Multinationals is to assess international regulation of transfer prices, particularly as the regulations are applied in Canada and the United States. In this chapter we have attempted to show that international regime theory can be usefully applied to the issue areas of international taxation and, in particular, to regulation of transfer pricing. We have discussed each regime in terms of its purpose and scope, principles and norms, and rules and

The International Tax Transfer Pricing Regime

117

procedures. In subsequent chapters we go into more detail on each of these topics, returning in chapters 12 through 14 to the question of problem areas with the tax transfer pricing regime, and possible solutions for strengthening it. This concludes Part I of Taxing Multinationals. We return to the U.S. and Canadian tax treatment of transfer pricing in Part IV, and to an assessment of the international tax transfer pricing regime in Part V. We move first, however, to look at the theory of the multinational enterprise as an integrated business and to examine statistical evidence on the size and characteristics of intrafirm trade in North America in Part II, 'Multinationals and Intrafirm Trade.'

APPENDIX 2.1 INTERNATIONAL GUIDELINES ON TAXING MULTINATIONALS There are at least three sources of international guidelines on taxing multinationals. First, both the OECD and the United Nations have developed codes of conduct or guidelines for MNEs and nation-states that deal with taxation issues. In addition, the OECD, United Nations, and Harvard University have model tax conventions which can be used as guides when individual countries negotiate bilateral tax treaties. The third source are guidelines on transfer pricing issued by the Committee on Fiscal Affairs of the OECD (OECD 1979, 1984, 1994b, 1995a,b, 1996, forthcoming). We outline the first two below; the third group is not reviewed here since the reports are discussed throughout this book, particularly in chapters 2, 5, 12, and 13. The OECD Guidelines and Model Tax Conventions The OECD's involvement with taxation of MNEs goes back to the first guidelines it issued in 1976. The most recent is the ongoing work on new guidelines on transfer pricing. The 1976 OECD Guidelines on Multinationals In 1976, the OECD countries adopted a Declaration on International Investment and Multinational Enterprises, to which was appended Guidelines for Multinational Enterprises. The purpose of the OECD guidelines is to inform MNEs about matters host countries see as sensitive, and to encourage MNEs to behave in appropriate ways vis-a-vis host countries. Host governments are encouraged to treat domestic and foreign firms in a similar fashion (national

118

The Rules of the Game

treatment) and to abide by their contractual, international obligations (e.g., full and fair compensation for expropriation). Dispute-settlement procedures are encouraged. The guidelines are not binding on MNEs, but states can publicize the names of persistent offenders. In terms of transfer pricing, the 1976 OECD guidelines direct MNEs to provide information necessary to correctly determine taxes and to refrain from using transfer prices that do not conform to an arm's length standard, as below: Upon request of the taxation authorities of the countries in which they operate, provide, in accordance with the safeguards and relevant procedures of the national laws of these countries, the information necessary to determine correctly the taxes to be assessed in connection with their operations, including relevant information concerning their operations in other countries; and refrain from making use of the particular facilities available to them, such as transfer pricing which does not conform to any arm's length standard, for modifying in ways contrary to national laws the tax base on which members of the group are assessed. (Dunning 1993, 525)

The 1963, 1977, and 1992 OECD Model Tax Conventions1 The OECD's Committee on Fiscal Affairs has also developed an income tax convention which serves as the model for bilateral tax treaties between OECD member countries and their treaty partners. Thus, the OECD model is primarily used by developed market countries. There have been three model conventions: a 1963 draft convention, the first formal convention in 1977, and a new model issued late in 1992. In between these periods, the Committee collected proposed amendments to the treaty before issuing the new model. The OECD calls the new loose-leaf version an 'ambulatory model tax convention' (Turro 1994d, 211) because the Committee intends to provide periodic updates rather than wait for several years to issue another model. The model convention incorporates an arm's length standard in two contexts: for allocating income both between home firms in one state and their branches in another state, and between firms and their subsidiaries, parents, or sister enterprises. The OECD has, from the first model convention in 1963, endorsed the concept of the separate entity as the fundamental basis for allocating taxing rights between countries. The MNE's tax base is to be allocated internationally according to the concept of a permanent establishment, with affiliates treated as separate legal entities and income apportioned between them assuming intrafirm transactions take place at arm's length prices. Article 7 (business profits), paragraph 2, provides that where an enterprise of one state carries on a business through a permanent establishment in another

The International Tax Transfer Pricing Regime

119

state, the establishment has attributed to it 'the profits which it might be expected to make if it were a distinct and separate enterprise engaged in the same or similar activities under the same or similar conditions and dealing wholly independently with the enterprise of which it is a permanent establishment.' The 1992 commentary to this article notes that the allocation is to be based on the arm's length principle, and applies this to the allocation profits on transactions between permanent establishments in the same MNE group. Article 9 (associated enterprises), paragraph 1, defines the arm's length standard. The paragraph provides that where an enterprise of one state 'participates directly or indirectly in the management, control or capital' of another enterprise, or where the same persons do so, in another contracting state, and where conditions are made or imposed between the two enterprises in their commercial or financial relations that differ from those that would have been made between independent enterprises, then any profits which would, but for those conditions, have accrued to one of the enterprises, but, by reason of those conditions, have not so accrued, may be included in the profits of that enterprise and taxed accordingly.

Article 7, but not article 9, explicitly qualifies the commitment of the model to pure 'arm's length' theory. Article 7(4) says that insofar as it has been customary in one state to determine profits of a branch by apportioning the total profits to the various parts of the enterprise, the state can continue to use apportionment. Where one government reallocates taxable income according to the arm's length standard, the question immediately arises as to how the other government will respond. If no change in assessment is made, increased taxation by the first government leads to double taxation of the MNE's profit. Therefore the OECD model treaty spells out the situation when the second government should provide a corresponding adjustment. Paragraph 2 of article 9 says: Where a Contracting State includes in the profits of an enterprise of that State - and taxes accordingly - profits on which an enterprise of the other Contracting State has been charged to tax in that other State and the profits so included are profits which would have accrued to the enterprise of the first-mentioned State if the conditions made between the two enterprises had been those which would have been made between independent enterprises, then that other State shall make an appropriate adjustment to the amount of the tax charged therein on those profits. In determining such adjustment, due regard shall be had to the other provisions of this Convention and the competent authorities of the Contracting State shall if necessary consult each other.

The commentary on article 24 in the 1992 OECD model income tax conven-

120 The Rules of the Game tion recognizes that tax problems often arise in triangular situations, not just bilateral ones between two tax jurisdictions. The typical triangular case deals with dividends, royalties, or interest income paid by a firm in A (the country of source) to a permanent establishment in B owned by an enterprise in C (the country of residence). Who should credit the withholding tax levied by A's government, the government in B or C? The commentary proposes that state B should credit the withholding tax, but only up to the level allowed by the tax convention between A and C (Matthews 1993b, 254). Article 25 (mutual agreement procedure, or MAP) permits a resident taxpayer to petition his or her tax authority for relief from double taxation caused by the unilateral actions of one of the two authorities. MAP requests are generally made for issues such as determining whether or not a permanent establishment exists or what is the residence of a taxpayer; the most common MAP request is to settle transfer pricing disputes (Kwatra 1988). The commentary, written by the OECD Committee on Fiscal Affairs, to article 25 in the 1992 model treaty states that the MAP can be used to determine (1) whether a transfer pricing adjustment is well founded and (2) whether the size of the adjustment is appropriate. The amount of the adjustment must be justified under the arm's length principle before a corresponding adjustment would be required of the other competent authority. This proposal was suggested in response to the U.S. proposals for changing its section 482 regulations, in ways that the OECD considered to be moving away from the arm's length standard. With respect to corresponding adjustments of profits after a transfer pricing adjustment, the commentary encourages communication between the competent authorities and the taxpayers. The MAP should be kept flexible, with few formalities. The United Nations Guidelines and Model Tax Treaty Here we have the 1977 UN Code of Conduct on transnational corporations (TNCs), which has not formally been adopted, the 1978 UN Model Tax Convention, and the 1993 World Bank guidelines on TNCs (UNCTAD 1993). The 1977 UN Draft Code of Conduct on TNCs The United Nations Economic and Social Council set out a proposed text for a Draft Code of Conduct on Transnational Corporations in 1977.2 The code is designed particularly for MNEs with activities in developing countries. It generally proscribes MNE activities while giving rights to host countries. For example, host governments are given the right to regulate the entry and estab-

The International Tax Transfer Pricing Regime

121

lishment of MNEs, and the right to nationalize or expropriate MNE assets in return for adequate compensation. As a result, the Code of Conduct has not be adopted. In terms of transfer pricing, the UN code says: Transfer Pricing 33. In respect of their intra-corporate transactions, transnational corporations should not use pricing policies that are not based on relevant market prices, or in the absence of such prices, the arm's length principle, which have the effect of adversely affecting the tax revenues, the foreign exchange resources or other aspects of the economy of the countries in which they operate. Taxation 34. Transnational corporations shall not, contrary to the laws and regulations of the countries in which they operate, use their corporate structure and modes of operation, such as the use of intra-corporate pricing which is not based on the arm's length principle, or other means, to modify the tax base on which their entities are assessed. (Preston and Windsor 1992, 257-8)

Clause 44(e), on disclosure of information, also commits firms to providing information on their transfer pricing policies to national governments. The 1978 UN Model Tax Treaty In 1967, the United Nations established the UN Group of Experts on Tax Treaties between Developed and Developing Countries. Stanley Surrey wrote the introduction to the UN Group of Experts report. The commentary is explicit in its requirement to use arm's length prices (unlike the OECD Model Tax Treaty). The group produced a model tax convention in 1978 for use between developed countries and developing countries. Articles 7(2) and 9(1) of the UN model are identical to the corresponding articles in the OECD model. Unlike the OECD model, the UN model includes provisions regarding transfer pricing in the commentary in article 25, Mutual Agreement Procedure. The 1992 World Bank Guidelines on TNCs The World Bank guidelines contain prescriptions to host-country governments on how to treat foreign investors; the guidelines do not discuss how MNEs should treat host countries. Thus they are written from the opposite perspective to the United Nations Draft Code of Conduct (see UNCTAD 1993, 29). The guidelines encourage governments to admit foreign investors, while preserving

122 The Rules of the Game the state's right to regulate FDI. National treatment, and fair and equitable treatment, are to be accorded foreign investors after entry. Specific guidelines on expropriation are outlined; the state can expropriate but only in return for prompt and adequate compensation. Lastly, dispute settlement through MNEhost country negotiations or United Nations-sponsored dispute-resolutions mechanisms is recommended. The 1992 Harvard University Model World Tax Code More recently, Harvard University's International Tax Program has developed a new model tax code for use in developing countries or economies in transition to a market economy. A preliminary edition of The Basic World Tax Code (Hussey and Lubick 1992) contains only one reference to transfer pricing section 77, which is very similar to IRC section 482: Section 77: Allocation of Income and Deductions Among Taxpayers: In the case of two or more organizations or businesses (whether or not incorporated, whether or not organized in Progress, and whether or not affiliated) owned or controlled directly or indirectly by the same interests, the designated officer may distribute, apportion, or allocate gross income, deductions, credits, or allowances between or among such organizations or businesses if the designated officer determines that such distribution, apportionment, or allocation is necessary in order to prevent evasion of taxes or to clearly reflect the income of any such organization or business.

Brian Arnold has criticized section 77 as being 'incredibly broad' with 'a number of technical difficulties' (1993b, 265). For example, Arnold argues that transactions between almost any two firms, however related, are covered since no equity ownership threshold is specified; also, whether only tax evasion, or avoidance and evasion, is to be included is not clear. Arnold's criticisms are typical of the ones OECD member countries have levied against the wording of U.S. Internal Revenue Code section 482; in fact, the wording of section 77 above is almost identical to the wording of section 482. We discuss this in more detail in Chapter 8. NOTES 1 See Pagan and Wilkie (1993, ch. 9), Turro (1994d). 2 The full text, as of 1992, can be found in Dunning (1993, 588-96) and in Preston and Windsor (1992,249-67).

PART II: MULTINATIONALS AND INTRAFIRM TRADE

This page intentionally left blank

3

The Multinational Enterprise as an Integrated Business

Introduction In this chapter we briefly explain what multinational enterprises are, why they exist, and how they organize their activities. The chapter is divided as follows. We first outline the basic explanation for why MNEs exist and why they are successful - that is, the OLI or eclectic paradigm. We then outline the theory of global strategic management of MNEs, and the coordination and configuration decisions that such firms face. We discuss the types of activities in which integrated businesses are engaged, and the factors that influence MNE intrafirm trade patterns. Lastly, we focus on one locational factor, regional integration, and its impacts on MNE strategies, concluding with a case study of the impacts of regional integration on MNEs in North America. This chapter is an important building block in Taxing Multinationals for several reasons. First, the chapter explains how multinationals are integrated businesses engaged in formulating and implementing strategies at the global level. Understanding the integrated nature of the MNE helps us understand the difficulties national tax authorities, such as the Internal Revenue Service and Revenue Canada, face in taxing these enterprises, and is necessary in order to assess the effectiveness of the international tax transfer pricing regime. Second, transfer price regulations based on the arm's length standard are derived from economic analysis of the multinational enterprise. Functional analysis, the method now most commonly used by U.S. tax authorities to provide support for transfer pricing methodologies, requires an economic analysis of the MNE's organizational structure, production and sales patterns, FDI and intrafirm trade flows, and pricing policies. This chapter provides the necessary background for understanding, and preparing, economic analyses of transfer pricing.

126 Multinationals and Intrafirm Trade We turn first to the general theory of why multinationals exist and are successful. Why Multinationals Exist Why do we have MNEs? Why do firms go abroad, set up foreign affiliates, and engage in international production and intrafirm trade? The key concept in terms of understanding what multinational enterprises are and why they exist is that the MNE is an integrated or unitary business. By definition, the MNE consists of two or more firms under common control, with a common pool of resources and common goals, where the units of the enterprise are located in more than one country (Eden 1994b, 193-4). The generally accepted explanation for the MNE and its foreign investment decisions is that provided by the OLI paradigm. The OLI Paradigm Multinational enterprises are the most successful form of business organization because of their ownership, locational, and internalization (OLI) advantages. The OLI paradigm, developed by John Dunning, provides a general explanation as to why firms engage in, and are successful at, international production.1 The model is an amalgamation of three basic approaches to the question of why MNEs exist and why they are relatively more successful than domestic firms: the location, ownership, and internaiization approaches. The location approach explains FDI in terms of differences in country endowments and characteristics. First, international trade theorists hypothesized that international capital movements were due to differences in interest rates (e.g., Macdougall 1968). However, this confused foreign portfolio investment (FPI) with foreign direct investment (FDI). Portfolio investments are passive investments in which the owner of the bond or stock does not exercise any control over the firm. Foreign portfolio investment is generally thought to be influenced by differences in national interest rates. High interest rates attract FPI inflows and vice versa. Foreign direct investment, on the other hand, involves some degree of control. The actual definition of FDI varies from country to country; the general minimum is an ownership level of 10 per cent of equity assets in the firm, with some exercise of control. FDI is generally seen as a package of capital, technology, and management skills; whereas FPI involves only capital flows. When economists realized that direct investment was driven by different motives than portfolio investment, the model was reworked to argue that differences in country locational factors (e.g., resource endowments,

The Multinational Enterprise as an Integrated Business

127

labour costs, energy, transport and communications costs, market size, income levels, tariff barriers) provide an explanation for international direct capital movements (e.g., Koijima 1978). Vernon (1966, 1971, 1979) used the product life cycle, together with differences in resource endowments, market size, and income levels, to explain the pattern and timing of FDI. In the 1960s and 1970s, industrial organization theorists shifted the focus from a macro perspective on FDI to a micro perspective on the firm and its activities (the ownership approach). Industrial organization theorists argued that MNEs were generally found in oligopolistic markets, so that monopolistic advantages provided an explanation for MNEs (e.g., Hymer 1976). Another theory was that MNEs owned intangible assets on which they could earn additional profits if the assets were deployed in other countries (e.g., Caves 1971, Johnson 1970, Magee 1977). A third approach to explaining FDI was to assume that external markets were characterized by high transactions costs. Internalizing the market through a wholly owned foreign affiliate allowed the firm to reduce transactions costs by creating an internal market to replace the imperfect external one. The internalization approach (e.g., Buckley and Casson 1976; Casson 1982; Rugman 1980, 1981, 1982, 1986) dominated much of the FDI literature throughout the 1980s. The OLI paradigm builds on these various approaches. It argues that three types of advantages affect how and why MNEs go abroad and are successful: ownership (O), locational (L), and internalization (I) advantages. The components of the paradigm are outlined in Box 3.1. Let us explore each in some detail. Firm-Specific Advantages (The O Factor) An MNE operating a plant in a foreign country is faced with additional costs compared with a local competitor. The additional costs could be due to (1) cultural, legal, institutional, and language differences; (2) a lack of knowledge about local market conditions; and/or (3) the increased expense of communicating and operating at a distance. So if the MNE is to be profitable abroad it must have some advantages not shared by its competitors. These advantages must be (at least partly) specific to the firm and readily transferable within the firm and between countries. These advantages are called ownership or firm-specific advantages (FSAs) or core competencies. The firm owns this advantage: it has a monopoly over its FSAs and can exploit them abroad, resulting in a higher marginal return or lower marginal cost than its competitors, and thus in more profit. These advantages are internal to a specific firm. They may be location-bound advantages (i.e., related to the home country, such as monopoly control over a local resource) or non-location-bound (e.g., technology, economies of scale and scope from simply being of large size).

128 Multinationals and Intrafirm Trade

BOX 3.1 The OLI Paradigm

Ownership/firm specific advantages (FSAs): The O factor Knowledge/technology: new products, processes, marketing and management skills, innovatory capacity, the noncodifiable knowledge base of firm. Economies of large size: economies of scale and scope, product diversity and learning, access to capital, international diversification of assets and risks. Monopolistic advantages: privileged or exclusive access to markets due to patent rights, brand names, interfirm relationships, ownership of scarce natural resources. Location/country-specific advantages (CSAs): The L factor Economic advantages: spatial distribution of factor endowments, costs and productivity of inputs, size of market and income levels, international transportation and communication costs. Social/cultural advantages: cross-country differences such as psychic distance, language barriers, social and cultural factors. Political advantages: political stability, general public attitude and government policies towards MNEs, specific policies that affect MNEs (e.g., trade barriers, taxes and FDI regulations, investment incentives). Internalization advantages: The I factor Natural or endemic market failure (natural imperfections) Difficulties in pricing knowledge: information impactedness, opportunism, uncertainty, public goods characteristic of knowledge, failure to account for all costs and benefits. Transactions costs of making markets under conditions of risk and uncertainty: search and negotiation costs, problems of moral hazard and adverse selection, lack of futures markets and insurance, risk of broken contracts. Structural market failure (imperfections created by the MNE) Exertion of monopoly power: using oligopolistic methods (e.g., predatory pricing, cross-subsidization, cartelization, market segmentation), creating barriers to entry that distort external markets and cause structural market failures. Arbitraging government regulations: exploiting international differences in government regulations such as tariffs, taxes, price controls, and other nontariff barriers.

The Multinational Enterprise as an Integrated Business 129 Box 3.1 provides a list of the various types of FSAs which the MNE can possess. We identify three basic types of ownership advantages for a multinational enterprise. These include: knowledge/technology, broadly defined to include all forms of innovatory activity; economies of large size (advantages of common governance) such as economies of scale and scope, economies of learning, broader access to financial capital throughout the MNE organization, and advantages from international diversification of assets and risks; and monopolistic advantages that accrue to the MNE in the form of privileged access to input and output markets through patent rights, ownership of scarce natural resources, and the like. As Dunning (1993, ch. 4) notes, some of these O advantages can be found with de novo firms (i.e., first-time overseas investments); others come from being an established affiliate in a large, far-flung multinational enterprise. Economies of common governance clearly belong to the latter category. Therefore FSAs can change over time and will vary with the age and experience of the multinational. Country-Specific Advantages (The L Factor) The firm must use some foreign factors in connection with its domestic FSAs in order to earn full rents on these FSAs. Therefore the locational advantages of various countries are key in determining which will become host countries for the MNE. Clearly the relative attractiveness of different locations can change over time so that a host country can to some extent engineer its competitive advantage as a location for FDI. The country-specific advantages (CSAs) that influence where an MNE will invest can be broken into three categories: E, S, and P (economic, social, and political). Economic advantages include the quantities and qualities of the factors of production, size, and scope of the market; transport and telecommunications costs; and so on. Social/cultural advantages include psychic distance between the home and host country, general attitude towards foreigners, language and cultural differences, and the overall stance towards free enterprise. Political CSAs include the general and specific government policies that affect inward FDI flows, international production, and intrafirm trade. An attractive CSA package for a multinational enterprise would include a large, growing, high-income market, low production costs, a large endowment of factors scarce in the home country, and an economy that is politically stable, welcomes FDI, and is culturally and geographically close to the home country.

130 Multinationals and Intrafirm Trade Internalization Advantages (The I Factor) The existence of a special know-how or core skill is an asset that can generate economic rents for the firm. These rents can be earned by licensing the FSA to another firm, exporting products using this FSA as an input, or setting up subsidiaries abroad. The ownership advantages of MNEs thus explain why they go abroad, while the locational advantages of countries explain where MNEs set up foreign plants. How they go abroad is another issue. The OLI model argues that external, arm's length markets are either imperfect or in some cases nonexistent. As a result, the MNE can substitute its own internal market and reap some efficiency savings. For example, a firm can go abroad by simply exporting its products to foreign markets; however, uncertainty, search costs, and tariff barriers are additional costs that will deter such trade. Similarly, the firm could license a foreigner to distribute the product but the firm must worry about opportunistic behaviour by the licensee. The OLI model predicts that the hierarchy (the vertically or horizontally integrated firm based on internal markets) is a superior method of organizing transactions than the market (trade between unrelated firms) whenever external markets are nonexistent or imperfect. The theory predicts that internalization advantages will lead the MNE to prefer wholly owned subsidiaries over minority ownership or arm's length transactions.2 It is therefore the internalization advantages part of the OLI paradigm that explains why MNEs are integrated businesses, are producing in several countries, and are using intrafirm trade to ship goods, services, and intangibles among their affiliates. Internalization within the MNE is designed to reduce market failures by replacing missing or imperfect external markets with the hierarchy of the multinational organization. These market imperfections are of two basic types: natural and structural (Dunning 1991). Natural market imperfections are caused by failures in, or the lack of, private markets; these failures arise naturally in the course of market making. Structural market failures are due to MNE oligopolistic behaviour arising from general exploitation of markets, or from arbitraging differences in government regulations between countries. We examine each type below. Several general types of market imperfections arise naturally in, and are endemic to, external markets. Two of the most important are imperfections in, or the lack of, a market for knowledge, and the existence of transactions costs in external markets.3 Other important market failures occur because of risk and uncertainty, and interdependence of demand and supply. First, the external market for knowledge may fail due to three inherent characteristics: (1) transactions in knowledge suffer from impactedness and

The Multinational Enterprise as an Integrated Business

131

opportunism; (2) uncertainty plagues this market; and, most importantly, (3) knowledge is an intermediate good with strong elements of publicness. Because technology is intangible and firm-specific, it is difficult for either the owner or the potential buyer to assess its value. The seller must explain to the buyer how it can be used without telling enough that the buyer could replicate the knowledge; hence, knowledge is impacted. This can cause opportunistic behaviour as each party attempts to shift the terms in his or her favour. Impactedness and opportunism are worsened by uncertainty, leading the buyer to underestimate the benefits. If both parties are risk-averse, the private market underproduces knowledge. Most markets in intangibles are faced with the above problems. Knowledge is not unique in this respect; however, its publicness characteristic is rarer and more serious. Knowledge, once created, is easily disseminated and jointly consumed; thus the marginal cost of provision to an additional consumer is minimal. Once produced, individuals cannot be excluded from consumption via the price mechanism since no one has property rights to this knowledge. Thus consumers free ride and cannot be forced to pay. Publicness, the combination of jointness and nonexcludability, implies that firms cannot profitably produce knowledge, and the private market fails.4 Where markets are missing, integration may create a market. The standard example here is a nonroutine or high profit intangible which the firm is unwilling to license to unrelated firms, for fear of dissipation of the rents from the intangible. The firm may be willing to transfer the intangible to a wholly owned subsidiary but not to a potential competitor. Therefore horizontal integration allows the MNE to exploit the rents from nonroutine intangibles in different markets, opportunities that would not be exploited if the MNE had to engage in arm's length contracting to establish a licence. A second source of natural market failure are the transactions costs which are incurred in overcoming market imperfections or obstacles to trade in all external markets. The higher the costs, the smaller the volume of trade. All markets are faced with the costs of search, communication, specification of details, negotiation, monitoring of quality, transport, payment of taxes, and enforcement of contracts (Casson 1982). Transactions costs may be reduced if the two parties are jointly owned. For example, it may be difficult to conclude a long-run, fixed-price contract if comparable, external prices are not readily available since future price fluctuations will benefit one party at the expense of the other. If the two firms merge, the probability of making a market increases. Therefore vertical integration can reduce transactions costs and increase supply (Casson et al. 1986). In addition, quality control can be improved through backwards integration. Vertical inte-

132 Multinationals and Intrafirm Trade gration to ensure quality control, for example, can be found in the high-quality, high-priced end of the market (e.g., name-brand perishable produce such as Dole bananas or Blue Bell ice cream). A third type of natural market failure arises because external markets fail to deal adequately with risk and uncertainty. Risk is the possibility of loss; risk aversion can be a motive for foreign direct investment (Vernon 1983). Under uncertainty, individuals can only make rational decisions within an area bounded by what they know. As a result, individuals with information not available to the other party may use this information to behave opportunistically, in order to improve their bargaining position vis-a-vis the other party. Internalization lessens the incentives for opportunistic behaviour by buyers and sellers. It can also compensate for the lack of futures markets since individual units of an MNE are less concerned about future price changes within the MNE than are independent entities. Internalization therefore can provide a form of insurance against unexpected price changes, particularly in the long run and where futures markets do not exist to provide such a hedging cushion. A fourth type of natural market failure occurs where interdependence of supply exists due to economies of scale and/or scope; as a result, vertical integration may follow naturally. Suppose large differences in minimum efficient scale (MES) exist such that a single plant can supply the entire needs of several downstream plants; the upstream firm is in effect a natural monopoly. Vertical integration of the firms avoids the problem of determining individual prices for each of the buyer firms. Other examples of firms drawing on a common pool of resources are inventory holdings, cash balances, accounting and legal departments, and research laboratories. Purchasing departments may, through their larger volume of bulk buying, be able to get volume discounts not available to individual buyers.5 Economies of scope can also lead to horizontal integration if, once an input (capital, technology) is acquired for one purpose, it is usable for other projects. Similarly, where interdependence of demand exists, in terms of either substitutes or complements, integration may follow naturally (Casson 1982). For example, firms producing complementary products may benefit from economies of scope in advertising and technology development. This will lead MNEs to diversify into related markets, generating horizontal integration if foreign affiliates are set up to supply these markets and benefit from the economies of scope. Interdependence in demand is also present where internationally mobile buyers are looking for the same standard of service from the same suppliers (e.g. from hotels and banks) when the buyers go abroad. Lastly, consumers that want to be able to order products in one country for delivery in another generate a demand for international distribution and transportation networks which are linked on the demand side. In each of these cases, firms may benefit from internalization.6

The Multinational Enterprise as an Integrated Business

133

Note, however, that even though there are benefits to internalization, there are also costs involved in being an integrated business. One of the most important of these is governance costs - that is, the costs of administering a large, vertically and horizontally integrated enterprise with its complicated internal markets for goods, services, and intangibles. Secondly, integrated businesses, in order to compete on a global scale, also require enormous financial resources that may not be available to the firm or only available at a cost that is higher than that available through other forms of organizational structure - for example, through more loosely related structures such as business networks and strategic alliances (Rugman and D'Cruz 1994). Thirdly, new lines of business may require core competencies or co-specialized assets not possessed by the MNE; rather than either forgo entering these areas or incur the costs of entry, the firm may choose a looser contractual arrangement. The combination of high governance costs, inadequate financial resources, and missing FSAs or co-specialized assets may rule out vertical integration as a mode of entry or expansion, even where the wholly owned subsidiary route is the most preferred route for the firm. Structural market failures are created by the multinational enterprise as it exploits its monopoly power in domestic and international markets. First, because multinationals, especially the largest ones, are powerful and mobile non-state actors in the global economy, their ability to move assets and incomes has been a constant bone of contention with host-country governments, especially developing countries (whose GDP may often be smaller than the global sales revenues of the biggest MNEs; see Dunning 1993). Nationstates fear that multinationals can and do abuse their relative bargaining power in ways that benefit the MNEs at the expense of host-country citizens, businesses, and governments. For example, as members of an international oligopoly, MNEs can raise global profits by segmenting domestic markets and price discriminating, erecting entry barriers to limit competition from domestic firms, restricting the decision-making and R&D activities of its subsidiaries by centralization within the parent firm, using transfer pricing to shift rents out of host countries, and so on (Hymer 1970; Murray 1981). These are endogenous market imperfections, caused by the international oligopolistic nature of the MNE (Rugman and Eden 1985, ch. 1). Second, when governments levy taxes, tariffs, and other forms of trade barriers, these regulations create additional costs for firms that reduce profits. Although the regulations generally have a legitimate economic purpose (e.g., raising government revenue), from the firm's point of view these are exogenous factors distorting international markets. Unrelated firms trading across international borders must pay these taxes; however, MNEs can, through transfer pricing and other financial manoeuvres, at least partly arbitrage these exogenous

134 Multinationals and Intrafirm Trade imperfections. These are exogenous, government-imposed market imperfections (Rugman and Eden 1985, ch. 1). Where government regulations exist, integration can therefore reduce the regulatory burden on firms. MNEs can arbitrage government regulations such as tariffs or differences in tax rates. Ad valorem tariffs can be avoided by underinvoicing imports. If the profit tax rate is higher in one country than another, tax payments can be reduced by overinvoicing intrafirm imports to, and underinvoicing exports from, that country. The amounts and timing of head-office fees, dividends, and royalties can all be manipulated to reduce tax payments. Thus, the MNE through internalization can arbitrage exogenous market imperfections, in the process earning higher after-tax and tariff profits than can unrelated firms engaging in similar transactions.7 We explore the impacts of taxes and tariffs on the MNE's incentives to engage in transfer price manipulation in chapters 6 (taxing MNEs in theory) and 7 (taxing MNEs in practice). In summary, the internalization part of the OLI paradigm therefore answers the 'how?' question about EDI - that is, how does the MNE go abroad? The model predicts that the MNE will internalize markets in order to reduce natural market imperfections; on the other hand, such internalization creates structural market failures. Thus internalization has both an efficiency-enhancing effect (because natural market failures are reduced) and an efficiency-reducing impact (because structural imperfections are increased) on global welfare. No wonder nation-states (and economists) have a love-hate relationship with the integrated enterprise we call the multinational! In practice, however, the choice between the market and the hierarchy is not so simple. There are many different modes of engaging in international production, ranging from simple exporting on the one hand, through subcontracting, licences, and joint ventures, to the polar extreme of a wholly owned subsidiary or branch. Each has its own benefits and costs to the MNE, and these vary depending on the home and host countries, potential partners, the market for the product, government and nongovernmental barriers to trade, and so on. The MNE compares the advantages and disadvantages of these various contractual arrangements. Generally, we expect that the MNE prefers the wholly owned subsidiary route to other contractual arrangements, unless the costs of governance (running the hierarchy) exceed the benefits of internalization (in terms of internalizing natural and structural market failures). We return to the topic of modes of entry into the foreign market later in this chapter. Summary: The OLI Paradigm This section has described the OLI or eclectic paradigm, which explains the

The Multinational Enterprise as an Integrated Business

135

existence of multinationals. The O factor answers the 'why?' question; that is, why the firm goes abroad. The reason is to exploit its firm-specific advantages in other markets and countries; these FSAs allow the firm to overcome the costs of transacting and producing in a foreign location. The L factor answers the 'where?' question of location. Since international production requires the use of foreign factors in conjunction with the firm's FSAs, the MNE chooses its 'where' to locate its foreign operations by comparing each country's locational attractiveness in terms of country-specific economic, social/cultural, and political factors. The I factor answers the 'how?' question as to what mode of entry the firm uses to penetrate the foreign location. The MNE has a variety of alternative contractual arrangements, ranging from arm's length international trade through the wholly owned foreign subsidiary, and weighs their relative benefits and costs to determine how the enterprise enters the foreign market and expands its operations over time. The successful MNE simultaneously combines these ownership, location, and internalization advantages to design its network of activities and affiliates in ways that maximize its market shares and growth. Now let us look at one example of how the OLI model can be applied to a particular industry in order to provide some concreteness to this theory. An Example: The Pharmaceutical Industry in Canada In this section, we use the pharmaceutical industry in Canada to illustrate the OLI paradigm (Eden 1989). In the model, foreign direct investment depends on three factors: firm-specific advantages (FSAs, the O variable), country-specific or locational advantages (CSAs, the L variable), and internalization advantages (the I variable). Ownership Advantages in Pharmaceuticals FSAs are unique, intangible, wholly owned advantages that are readily transferable within the affiliates of a multinational. FSAs allow the MNE to compete in foreign markets where domestic firms have the advantage of better knowledge of Jocal conditions, lower communication costs, and no cultural or language impediments. Examples of FSAs are technological advantages, economies of scale, product differentiation, access to cheaper capital, international diversification of risk, and access to raw materials. The most important FSA possessed by MNEs in the pharmaceutical industry is technological knowledge. The industry is among the most R&D-intensive of all manufacturing industries. It has been estimated that a new drug takes

136 Multinationals and Intrafirm Trade upwards of $100 million and ten years to develop. Knowledge in this industry includes the discovery of new drugs and production processes, superior marketing and distribution networks, and expertise in product differentiation via advertising. Other FSAs include brand names, patents, and the ability to price discriminate between markets. Patents are of particular importance, as is access to raw materials (e.g., chemicals). Firm-level economies of scale (e.g., in R&D and other management functions) lead to centralization at the head office; little R&D is done outside the parent firm; this is typical of most MNEs.8 Locational Advantages in Pharmaceuticals The second component of the OLI-internalization model is location. Once a firm has decided to internalize its FSAs through wholly owned subsidiaries, the decision to go abroad depends on the locational attractions of potential host countries. These country specific advantages (CSAs) are of three types: economic, noneconomic, and governmental advantages. Canada offers many locational advantages and, as a result, has attracted large investments by pharmaceutical MNEs. The labour force is well educated, income levels are high, and labour costs are similar to those in the United States. The economy is politically stable and close to the United States (home to over half the drug MNEs). As a result, foreign MNEs dominate over 90 per cent of the Canadian drug market. Foreign-owned subsidiaries in Canada (FOSCs) in this industry, however, are truncated subsidiaries; investments are concentrated at the preparation/packaging and marketing stages. A small domestic market, a small chemical sector, and low levels of indigenous R&D have inhibited development of the R&D and chemical stages. The third CSA affecting location of pharmaceutical subsidiaries is government policy. The chief policy influences in the Canadian case historically have been tariffs, corporate taxes and R&D incentives, the Foreign Investment Review Agency (FIRA), compulsory licensing,9 provincial formularies, and government involvement in insurance. Canadian policy has been traditionally welcoming of FDI in the pharmaceutical industry. In the early 1900s, U.S. drug firms jumped the Canadian tariff wall to set up branch plants. Tax and patent policies were similar to U.S. levels. No screening of FDI existed. In MNE eyes, Canada was regarded as a good host country. Government policies in the 1970s, when compulsory licensing and FIRA were introduced, were interpreted as unfriendly and may have discouraged pharmaceutical FDI. More recently, the Canadian government has eliminated the compulsory licensing of brand-name pharmaceuticals (long demanded by the drug MNEs and the U.S. government) and established a drug prices review board, a 'watch-

The Multinational Enterprise as an Integrated Business

137

dog' agency over consumer drug prices, in return for a commitment to expand R&D activities in Canada. The Canada-U.S. Free Trade Agreement and the North American Free Trade Agreement now commit the Canadian government to national treatment for North American investors in Canada. Thus the policy environment for foreign pharmaceutical MNEs is more hospitable than at any time since the early 1960s. Internalization Advantages in Pharmaceuticals The third component of the OLI model is internalization advantages, which can be decomposed in the MNE's ability to internalize natural and structural market imperfections. The existence of special know-how is an asset that can generate economic rents for the firm, either through licensing the FSA to another firm, exporting products using this FSA as an input, or setting up subsidiaries abroad. Natural market failure can be reduced if the MNE chooses the internal market over exporting and/or licensing. Internalization problems associated with knowledge, for example, characterize the R&D-intensive pharmaceutical industry. Appropriation of rents is essential due to the uncertainty involved in R&D projects. Few investments produce saleable products that pass health regulations, are patentable, and earn large profits. Thus these products must generate sufficient returns to recoup earlier investments and fund ongoing research projects. Hence, average cost is higher than marginal cost and estimates of monopoly rents are likely to be deceptive. Secrecy, lead time, and investments in marketing are alternatives to patents as methods of appropriating the rents from knowledge. Secrecy is particularly important for drug processes, due to difficulties in obtaining patents for many processes and fear that competitors will 'invent around' the patent. Secrecy is much harder to maintain for products. Because confidentiality is better protected through wholly owned subsidiaries, it is not surprising that internalization is the preferred investment form. In the Canadian case, the requirement for compulsory licensing of brandname drugs after an initial patent period probably discouraged inward FDI because the drug MNEs did not receive compensation levels equivalent to those available in other markets. On the other hand, total MNE investment in technology development was probably unaffected because the Canadian market was too small as a percentage of the global market to have an impact on overall MNE returns. In terms of transactions costs that afflict external markets and that would induce pharmaceutical MNEs to opt for internalization, several types of transactions costs affect the global drug market. Costs of ensuring quality are high since most governments have strict regulations about quality. Economies of

138 Multinationals and Intrafirm Trade scale in fine chemicals, the upstream stage, are significant. Drug MNEs also have a strong international advantage in market-making skills due to brand images, patents, and product differentiation. Controls over the safety and efficacy of drugs often encourage local backwards integration into the clinical testing stage. The international drug industry has long been accused of creating structural market failures through its oligopolistic practices, including excessive pricing of prescription drugs, and activities to keep out generic drug companies. In addition, there are documented cases of transfer price manipulation in pharmaceuticals, particularly in host developing countries (see Lall 1979 and Chapter 7 of this text). In Canada, three Canadian drug policies - compulsory licensing, insurance, and provincial formularies - all offer potential for arbitrage. For example, MNEs use advertising to link brand name with high quality, in order to offset the price advantage of generic drugs. The inelastic nature of demand due to high insurance coverage encourages price discrimination (although compulsory licensing works against this). Offering bulk discounts to pharmacists and encouraging no-substitution prescriptions lessens the price-dampening effects of provincial formularies. Overinvoicing of chemicals and drugs in bulk form imported from MNE affiliates has two arbitrage effects: it reduces tax payments and increases the profits shifted out of Canada through trade flows (thus partly offsetting the lower rents yielded by compulsory licensing). In summary, we conclude that the OLI model can be a critical 'lens' through which one can view, and analyse, multinational enterprises in a particular industry. We now turn to the issue of managing the multinational enterprise. How do integrated businesses manage their global activities? In the next section, we outline the strategic management process within the enterprise, which will then lead us, in the following two sections, to address the questions of how and with whom an MNE coordinates its international activities, and how and why these activities are configured (geographically allocated) by the MNE. Managing the Multinational as an Integrated Business As we have seen above, there are several reasons why an MNE would want to internalize product and/or factor markets - that is, integrate vertically or horizontally to form a unitary business. The general theory of the MNE argues that markets are imperfect and that integration can be used to reduce these imperfections. The hierarchy is an alternative to the market that, under certain circumstances, can be more efficient than the market as a way of offsetting naturally arising or government-imposed imperfections. Even given these costs, most

The Multinational Enterprise as an Integrated Business

139

firms prefer the wholly or majority-owned subsidiary over the external market, particularly where intangible assets are involved. Given that there are likely to be net benefits to the multinational enterprise from being integrated, what do these integrated businesses look like? What do they do? How are they managed? In this section we outline the theory of global strategic management - that is, how managers of an integrated business formulate and implement strategies for the international economy. We first address the question: what does strategy mean? Choosing an International Business Strategy International strategic management is a comprehensive, ongoing management planning process aimed at formulating and implementing strategies that enable a firm to compete internationally (Griffin and Pustay 1995, ch. 10; Hitt et al. 1995). To develop an international strategy the firm must answer the following questions: What products/services should we sell? Where and how should we make these products and/or deliver these services? Where and how should we sell them? Where and how will we acquire the necessary resources? How do we expect to outperform our competitors? The strategies implied by answers to these questions fall into three categories: (1) corporate strategy, (2) business strategy, and (3) functional strategies. These strategies can be visualized as nested within one another: corporate strategy at the overall MNE level, business strategy at the level of the MNE's business units and/or major product lines, and functional strategies at the level of the MNE's individual functions or activities (e.g., marketing, finance, production). Corporate strategy defines the domain of businesses in which the firm intends to operate, that is, the overall organization of the MNE. The enterprise can be organized as a single business, in which all revenues come from one business, product, or activity. Alternatively, the MNE could operate in several related businesses, industries, or markets at the same time. This is called related diversification. The third possibility is unrelated diversification, in which the MNE operates in different unrelated industries and markets. The most common corporate strategy is related diversification because it offers economies of scale and scope, and allows the MNE to achieve synergies across related product lines and markets (Griffin and Pustay 1995, ch. 10).

140 Multinationals and Intrafirm Trade

Business strategy is the process by which managers of the MNE study the external environment surrounding the firm, evaluate the strengths and weaknesses of the MNE's firm-specific advantages (core competencies), and create a three-to-five-year set of action plans designed to achieve the firm's goals, given its resources and its constraints (Shrank and Govindarajan 1993, ch. 6). The goal of business strategy is to create and improve the long-run competitiveness of the firm vis-a-vis its competitors. Business strategy focuses on the specific businesses, subsidiaries, or operating units within the firm. How should the MNE compete in each market it chooses to enter? Business strategy therefore deals with how the MNE should compete within a product line or strategic business unit. There are three generic business strategies a firm can follow in order to achieve a lasting and sustainable competitive advantage (Porter 1985). The first is a low-cost strategy, whereby the firm achieves market leadership through being the lowest cost competitor. A cost leadership strategy requires a close attention to firm-specific factors such as economies of scale, learning curve effects, and tight control of overhead costs. Firms are likely to outsource footloose activities, locating these stages in low-labour-cost locations. The second is a product differentiation strategy, whereby the firm creates a unique and successful product (e.g., Calvin Klein jeans, Rolex watches). A product differentiation strategy means the MNE must concentrate on providing superior customer service, better dealer networks, achieving brand loyalty, and constantly improve product technology and product quality, as methods of establishing and maintaining a real and/or perceived image of the distinctiveness of the MNE's products in the eyes of consumers. A third strategy is focus, whereby the firm concentrates on a particular market niche, usually either a product-line niche (e.g., mountain bikes, undergraduate textbooks) or a geographic niche (e.g., the North American market). Smaller MNEs are more likely to use a focus strategy. Such a strategy can include either a differentiation or low cost component. For example, a focused low-cost strategy would concentrate on the smallest buyers or lowest-entry products in the market, whereas a focused differentiation strategy would concentrate on customized or specialized products in a narrow product line or for a single customer (Hitt et al. 1995, 115-16). The third type of overall MNE strategy is at the functional level. Functional strategies deal with how the firm manages its functions (e.g., finance, marketing, operations, human resources, R&D) in ways that are consistent with the MNE's international corporate and business strategies. Having chosen its overall corporate and business strategies, the MNE is then confronted with developing a strategic plan. Such a plan must include underly-

The Multinational Enterprise as an Integrated Business

141

ing structural choices such as: (1) scale - which activities of the value chain to engage in (vertical integration) and how big an investment to make in each; (2) scope - how many product lines to develop using the MNE's intangible assets (horizontal integration); (3) experience - how often to continue with a certain activity; and (4) technology - what product and process technologies to use. Given our focus on integration and intrafirm trade, we look at two key strategic management decisions: how to coordinate the activities performed within the MNE (who does what?), and how to configure these activities (where does it get done?). The concept of the value chain is an integral part of these two decisions regarding coordination and configuration. The Value Chain The typical MNE performs a variety of functions including: R&D, product design, process engineering, product fabrication, assembly, purchasing and materials management, marketing and distribution, advertising, and legal, accounting, and finance functions. All these value-adding activities, from initial product development through final sale and service, make up an industry's value chain. Each MNE must determine its own firm value chain, which could be as broad as the complete industry value chain, or, more likely, include only some of the activities in the industry chain. We can split the activities of the industry value chain into primary and support functions (Eden 1991c; Porter 1985; Shrank and Govindarajan 1993, ch. 4). Primary activities include resource extraction and processing, fabrication, assembly, sales, and distribution. Support activities include technology development, support services, and strategic management of the enterprise. Each firm must decide what parts of the industry value chain are performed inside the firm (i.e., are internalized) and which segments are purchased through various contractual arrangements with other firms. The activities in the value chain, and the particular types of intrafirm trade flows that accompany each activity, are described below. In terms of support activities the MNE can engage in: Strategic management: The long-run planning for the enterprise as a whole is the sine qua non of the parent firm. Foreign affiliates engage in planning and manage the day-to-day operations of their particular affiliate. Occasionally a large MNE will set up a regional headquarters to manage the operations of several affiliates in another region of the Triad (e.g., a U.S. multinational may designate its U.K. subsidiary as the regional headquarters overseeing its affiliates in Europe), but generally strategic management is the prerogative

142 Multinationals and Intrafirm Trade of the parent firm. The parent charges its affiliates management fees for these services. Support services: These activities include the finance and accounting functions, marketing and advertising, and occasionally central purchasing. Again support services are normally performed by the parent, but regional headquarters may be established to provide services to all the affiliates within a region. Alternatively, the MNE can contract out support services. These activities may be provided either to the group (group services) and/or to an individual MNE affiliate (specific services); normally service fees are charged. Technology development: Creation of intangibles may be done by a centralized R&D unit (usually part of the parent firm) or decentralized as part of oreign production plants (e.g., where MNEs practise simultaneous engineering). The R&D unit creates product and process technologies and licenses them to other divisions for royalty payments or licence fees. Occasionally the unit may license intangibles to outsiders, particularly if it is mature, off-theshelf technology. In terms of primary activities, the MNE can engage in: Resource processing: Upstream units are set up to extract and process natural resources (e.g., crude oil extraction, mining operations). Typically these affiliates extract raw materials that have internationally available prices that are classified by product characteristics (e.g., copper prices on the London Metals Exchange). Open market prices at the refining and smelting stages (fabrication) are less likely to be available. Manufacturing: Manufacturing affiliates are downstream from resource affiliates. Their activities consist of parts production, subassembly, and final assembly of goods. Intrafirm trade for these affiliates consists of trade in tangibles - that is, in intermediate parts and subassemblies while the good passes through the manufacturing stages - and then sale of finished goods to distribution affiliates. Distribution and sales: Affiliates at this stage are interposed between manufacturers and final consumers and are providers of services. These units perform transportation, importing, distributing, sales, and service functions. Box 3.2 shows the range of value-adding activities that can be performed within a typical MNE; each MNE determines its own value chain as some combination of primary and support activities.

The Multinational Enterprise as an Integrated Business

143

BOX 3.2 The Value Chain of a Typical Multinational

Head office (strategic management) Business services (accounting, finance, marketing) Technology development (product and process technology; basic and applied) Extraction of raw materials

Processing of raw materials

-resource processing

Fabrication of parts and subassembly

Final assembly

manufacturing-

Distribution, sales, and service distribution

The value chain of a typical multinational enterprise consists of three primary activities (resource processing, manufacturing, and distribution) and three support activities (strategic management, business services, and technology development). Vertically integrated multinationals have foreign affiliates located at different stages of the value chain in a number of countries. Horizontally integrated MNEs have more than one foreign affiliate located at the same stage of the value chain (for example, producing different product lines). Intrafirm trade occurs when goods, services and/or intangibles flow among the affiliates and the parent firm.

All of the above activities can take place in one plant or value-adding activities may be split between several plants. Vertical integration occurs when different stages of the value chain are performed by different plants with intrafirm transfers between them (e.g., one plant may extract bauxite, another process it into alumina, a third may be an aluminum smelter, and a fourth may make aluminum cans). Horizontal integration occurs when different plants produce the same or similar product lines with intrafirm trade occurring to fill excess demand or to supply niche markets (e.g., auto plants specializing in different product lines). Questions of vertical and horizontal integration lead us directly into a discussion of two major decisions about the value chain: how to coordinate the MNE's interfirm and intrafirm activities in the value chain, and how to configure these activities within the MNE. We discuss each in turn.

144

Multinationals and Intrafirm Trade

Coordinating the MNE's Value Chain Coordination refers to the organizational structure of the MNE - that is, how the enterprise organizes and coordinates its intrafirm (within the enterprise) and interfirm (outside the enterprise) relationships. The way a firm sets up its internal and external organizational structure depends on factors such as (1) the ownership and legal status of the firm; (2) its age and size; (3) the number and range of value-adding activities in which the firm is engaged; (4) the geographic spread of these activities; (5) the types of relationships the firm has with competitors, suppliers, and buyers; and (6) the firm's international product strategy (Dunning 1993, 211). Coordination of Intrafirm Activities There are several ways to organize the internal structure of an enterprise. For example, a small single-activity firm is likely to choose a 'U' or unitary structure, in which the head office makes all major decisions and closely coordinates the activities of the firm. A second common structure is the 'M' or multidivisional structure, in which the enterprise is separated into different divisions, organized usually by product (activity) or by function (area of decision making). Where the divisions are organized along product lines, each unit may have the same organizational structure. Within the unit, those in charge of individual functions report to the division product manager, with the product manager from each division then reporting to the group product manager or to the chief executive officer. If the divisions are organized along functional lines, each division is normally responsible for that function for the organization as a whole, with the board of directors of the enterprise playing the main planning and coordinating role. The more numerous and varied are the MNE's product lines (e.g., the auto industry), the more likely that the organizational structure will be based on product lines rather than on functions. The fewer the products and the more vertically integrated the MNE, the more likely is a functional organizational structure. When the enterprise first goes abroad, it may set up an international division designed to manage international sales and production activities (e.g., Xerox and Du Pont). As the MNE grows, this division may separate into several units organized along geographic lines, either national (e.g., Canada, Mexico, the United Kingdom) or regional (e.g., North America, Europe). Regional divisions are common among oil, tobacco, and food-processing MNEs. Some MNEs use structures that are organized by product line, with international and domestic operations grouped under their respective product lines (e.g., Rockwell International, Colgate, General Electric [Dunning 1993, 217]). These different struc-

The Multinational Enterprise as an Integrated Business

145

tures have different benefits and costs to the enterprise as a whole. Structures that are geographically based can pay more attention to local consumer demands and government regulations, allowing more autonomy to the local affiliate operations. Functionally organized MNEs are able to exercise greater command and control over their activities, facilitating worldwide specialization and greater productivity. Global firms, in order to obtain the advantage of specialization of activities while also attempting to be locally responsive, often adopt a matrix structure based on a combination of product, function, and geography. We will come back to this issue later in the chapter when we discuss the impact of the MNE's internal organizational structure on intrafirm trade patterns. Coordination of Interfirm Activities The MNE must also coordinate its interfirm activities, that is, its relationships with buyers, sellers, and competitors. Not all the activities in the value chain must be performed within the multinational enterprise (Eden 1991c; Shrank and Govindarajan 1993). Some will be conducted inside; some contracted out; some activities may be performed both by subsidiaries and also by arm's length subcontractors. Some affiliates may perform more than one function (e.g., final assembly and sales may be bundled together where production needs to be close to final market); others a limited contracting manufacturing role. The choice, for each value-adding activity, depends on the benefits and costs of integration. MNEs must trade off the costs of not being integrated against the governance costs of integration, as outlined above, in a particular activity or line of business. The MNE of the 1990s resembles an international network of firms, linked together more or less tightly through equity flows and organizational control exercised by the parent firm. The borders of the multinational are blurring as the number and variety of organizational forms increases. Figure 3.1 illustrates some of the alternative methods of organizing MNE activities in terms of control.10 Within the home country, the MNE is likely broken up into a headquarters unit, one or more R&D units, and various local resource, manufacturing, and distribution subsidiaries located around the country. For tax purposes, these units are generally treated as one entity and file one consolidated return. The MNE may then have a variety of contractual linkages with other firms, both at home and in foreign countries. Foreign affiliates may be organized either as a branch or as a subsidiary, the latter being either a wholly owned subsidiary (WOS) or a majority-owned foreign affiliate (MOFA). The legal definition of the multinational enterprise stops here - that is, the MNE is defined as the parent firm and its branches, and wholly and majority-owned affiliates, both domestic and foreign.

146 Multinationals and Intrafirm Trade FIGURE 3.1 Defining the Borders of a Multinational Enterprise

Outside this definition are alternative contractual forms that move across the spectrum from the hierarchy to the market. For example, the MNE may be engaged in an equity joint venture with a foreign firm, taking either a majority or a minority stake but still maintaining effective control. A joint venture is an independent legal entity, but in practice it may be closely tied to one or both of its parent firms and could be considered part of the hierarchy. Closer to the market mechanism are licensing and franchising. The MNE may give a licence or franchise to a foreign distributor or franchisee, or subcontract its production to an original equipment manufacturer (OEM). Tax authorities tend to treat such relationships as close to arm's length. Lastly and more recently, MNEs are engaging in strategicbusinessalliancesorpartnershipswhereby two firms cooperate, supply complementary assets, and develop products designed to further their joint competitive advantages. Strategic alliances may be pro- or anticompetitive, since they generally occur between large, oligopolistic firms in high-profit, high-risk activities such as aerospace, biotechnology, and informatics (Dunning 1993, ch. 9). Figure 3.1 also helps us to think about the types of flows that can occur within and outside the MNE network. There will be inflows of intermediate

The Multinational Enterprise as an Integrated Business

147

goods, capital equipment, primary factor services, and technology at each stage of production, with some purchases being made on external markets, others performed by close suppliers, and still others by the parent firm and/or its affiliates. There are flows of intermediate products downstream from one affiliate to another if the MNE is vertically integrated. If the plants are specialized by product line, there are exchanges of product lines between plants at same stage of the value chain (i.e., there is horizontally integrated trade). There are flows of finished products to importers/distributors/sales units, and flows of technology services provided by R&D units to affiliates. Lastly, there are flows of support and management services provided by parent and/or service units to affiliates. Transactions made on external markets normally take place at arm's length prices, but intrafirm trades take place at transfer prices set within the MNE. There will be many (possibly thousands) intrafirm sales of goods, services, and intangibles by supplier affiliates, with payments flowing in the reverse direction from purchasers, all transactions taking place within the multinational enterprise and at non-arm's length prices. In addition, for many intrafirm flows particularly in services and intangibles - there will be no recorded transactions, nor will payments be made. Thus the tax authorities are forced to create transactions and then to hunt for comparable, arm's length prices for these nonexistent transactions.11 Configuring the MNE's Value Chain Since multinationals engage in several activities, they must configure these activities, that is, determine the physical location of each plant within the MNE hierarchy. The configuration of the MNE is a map of the global activities of the enterprise, documenting where each plant is located and the functions it performs within the enterprise.12 We argue that multinationals invest abroad in order to engage in foreign production and intrafirm trade. Thus causation runs from the MNE's choice of the value-adding activity to its choice of plant location and thus to foreign direct investment in this plant, not vice versa. The location selected for an affiliate depends on the affiliate's role in the value chain of the firm. Plant function drives the FDI choice. We can classify foreign affiliates according to plant location strategy, that is, to the MNE parent's four basic motives for setting up a foreign plant: to find raw materials (resource seeking), to manufacture parts and assemblies at lowest cost (cost reduction), to access foreign markets (market access), and to provide support services to other parts of the MNE group (support services). Each of these motives affects the value-adding activities of the MNE: processing, man-

148 Multinationals and Intrafirm Trade

ufacturing, distribution/sales, and support, as outlined above in Box 3.2. The choice of affiliate location depends on which of the general motives for FDI is involved, the relative attractiveness of various host locations, and the availability/cost of alternative contractual arrangements. Locational or country-specific advantages (CSAs) are the key to determining which countries will become host countries for the MNE, depending on whether the basic nature of the investment is resource seeking, cost reduction, market access, or the provision of support services to the rest of the enterprise. Given these four general locational motivations for setting up a foreign plant, we argue that multinationals construct their overall production structure by choosing among a range of locational structures for their foreign affiliates. Each foreign subsidiary or branch plant normally fulfils one of these plant functions, grouped around the four general categories of resource extraction, cost reduction, service support, and market access. Box 3.3 illustrates this topology by placing each factory type in the relevant part of the value chain. The higher the vertical placement of the factory, the greater is the amount of technological innovation generally expected from the foreign plant. Each MNE, depending on the length of its value chain and the nature of the industry, can therefore be seen as a set of foreign affiliates, strategically located according to their underlying resource, cost, service, or market functions (see also Eden 1991c). We expand on each of these foreign plant roles below. The Resource-Seeking Foreign Affiliate We identify two possible types of resource-seeking plants. Extractors access natural resources essential to the production process, so the key factor driving location is the need to be close to the source of raw materials. Processors turn natural resources into fabricated materials. Where the weight-value ratio is high, economies of scale at the two stages are similar, and foreign tariffs on processed imports are not high, extracting and processing may occur in the same plant. The Cost-Reducing Foreign Affiliate In terms of cost-reducing plants, we identify two possibilities: offshores and source factories. Offshores use cheap local inputs, particularly labour, to assemble simple parts for the parent company. In the 1970s and 1980s many MNE investments in the Asian newly industrializing countries (NICs) and the Mexican maquiladoras were of this type. As wage rates have risen in the NICs, these offshore plants have started to move from country to country searching for even lower-wage sites. Source factories are a step up from offshore factories. They use low-cost

The Multinational Enterprise as an Integrated Business 149 BOX 3.3 Affiliate Roles within the MNE

Head office functions (strategic management) Support services

Server 12

Technology development

R&D Outpost 1 1 Lead factory 10

World product mandate 9

1 Extractor

Extraction

2 Processor

4 Source factory 3 Offshore

Processing

Subassembly

8 Min. replica 7 Focused factory 5 Local assembler Final assembly

6 Distributor

Distribution and sales

In the value chain, the higher the vertical placement of the affiliate, the more technologically sophisticated the value adding functions. Some affiliates can perform more than one function (e.g. R&D outpost and lead plant, or extractor and processor, but it is useful to conceptualize specific MNE affiliates in this fashion. We can think of each of the MNE's affiliates as performing one or more of the activities in the enterprise's overall value chain. These activities may include: strategic management of the affiliate technology development provision of support services extraction of raw materials processing of these resources manufacture and subassembly of parts and components final assembly distribution and sales

150 Multinationals and Intrafirm Trade

labour but also produce specific components for the MNE. Source factories are globally or regionally rationalized plants in a vertical integration sense since the plant produces one segment of the value chain which is then sold for further assembly and final sale elsewhere in the MNE network. Sometimes offshore and source activities are contracted out to unrelated firms such as OEMs (original equipment manufacturers), often in Asia. Both offshores and source factories are generally considered by tax authorities as contract manufacturers since unrelated firms are often willing to perform these manufacturing functions for an arm's length price. The Market Access Foreign Affiliate There are several possible types of market access plants depending on their product range and technological sophistication. At the final assembly stage but with a low level of technology are local assemblers, import competing factories that assemble subcomponents for domestic sale (e.g., bottling plants, drug packaging). Such assembly is often driven by government regulations requiring a local presence, but a local production base may also increase domestic sales. Where a local presence is not useful, this function (bottling, packaging) will generally be performed on the basis of lowest-cost location. At the sales and service stage, with a similar level of technology and functions as local servers, are the distributors, firms that buy local or imported goods from wholesalers and provide marketing, sales, service, and warehousing facilities. Usually the first step when a firm goes abroad is to set up a distributor plant to facilitate exports from the parent firm. The first Japanese transplant operations in the automotive industry were basically distributors. Higher up the technology ladder me focused factories, globally or regionally rationalized subsidiaries which produce one or two product lines in massproduction runs for final sale in both local and foreign markets, purchasing other product lines from sister affiliates. World product mandates (WPMs) are somewhat similar to focused factories. WPMs are plants with the full responsibility for all stages of the value chain for a single product line within the MNE. The WPM is different from the focused factory since the responsibility for technology development is shifted to the subsidiary under a WPM but not in a focused factory. WPMs are generated primarily by government policy (e.g., in Canada) since specialization advantages normally encourage worldwide sourcing rather than locating the entire value chain in one country. Miniature replicas, or copy-cat plants, are plants generally protected from foreign competition by high tariff barriers and nontariff barriers, that assemble and sell locally a full range of products similar to their parent's production line. They normally import parts from sister affiliates but do not engage in exports.

The Multinational Enterprise as an Integrated Business

151

These affiliates are likely to be high cost when domestic markets are small, with excess product variety and short production lines. Miniature replicas were a common plant structure in tariff-ridden economies such as Canada before the GATT Tokyo Round and the 1966 Auto Pact. Latin American countries such as Brazil and Mexico, with their historically high trade barriers, have attracted many foreign miniature replicas, which are now being rationalized as governments liberalize their economies. Lead factories are effectively equal partners with the parent firm and placed in strategic locations within the Triad (i.e., the United States, the European Community, and Japan). Lead plants are responsible for technology and product creation and distribution of lead products in each Triadic market. Lead factories are seen as true 'insiders' in each of these regions. The Support Services Foreign Affiliate In the support services category we identify two types of foreign affiliates: R&D outposts and servers. The MNE may wish to decentralize some of its technology development activities to an R&D outpost. These outposts can provide a window on competitors' technological activities in host countries, take advantage of differences in human capital endowments among countries (e.g., lower-cost engineers and computer scientists in India), facilitate foreign sales by adapting MNE products to local market conditions and tastes, and meet host country regulations in the clinical testing and health areas. Servers are affiliates that provide support services to parts or all of the MNE family, such as standalone financial, centralized purchasing, or accounting firms. These affiliates, organized by function, can also be set up to serve a regional network, for example, all the MNE's affiliates within the European Community. The larger the regional core network, the more likely that the MNE has server affiliates. Summary: The 1990s Multinational Enterprise In summary, each MNE has a value chain of primary and support activities. The firm's chain is part of the complete industry value chain of activities that spans initial conceptualization of a product through to the product's delivery and after-sales service. Each firm decides how extensive its own value chain is relative to the complete industry value chain. Its network of affiliates is organized around the four functions an affiliate can perform within the MNE: resource seeking, cost reduction, market access, and support services. Within each general function are several types of plants which vary in their degree of technical sophistication, range of product lines, and linkages with sister affiliates and the parent firm.

152 Multinationals and Intrafirm Trade The more heavily the MNE engages in integrated international production, the more complex and numerous are the intrafirm flows linking the various parts of the MNE family group (UNCTAD 1993). Historically, U.S. multinationals used to be organized like the spoke of a wheel, with the parent firm at the centre, its wholly owned foreign affiliates spread around the rim of the wheel, and intrafirm trade flows forming the wheel's spokes. In the new global firms (e.g., Ford, IBM, ABB) the multinational looks like a cobweb or network model with nodes at different places (the parent firm, regional headquarters, and lead firms), intrafirm trade flows going in all directions, and close linkages to other network partners such as suppliers and buyers (Rugman and D'Cruz 1994; Stopford and Strange 1993; UNCTAD 1993). What factors can affect this networked MNE, its linkages and the intrafirm trade patterns that support them? These questions form the basis of the next section. Multinationals and Intrafirm Trade We now turn to characterizing and analyzing the intrafirm trade patterns of multinationals. First, we argue that MNEs can be characterized as one of four types depending on where their main value-adding activities lie. These four types have implications for the patterns of intrafirm trade within the MNE. Second, we examine several factors that can affect the pattern of intrafirm trade, including internal factors (the age of the affiliate, the degree of vertical or horizontal integration of the MNE, and the MNE's choice between a local versus a global strategy) and external factors (the country-specific advantages of the host country, regional integration, technological change). Multinational Types and Their Intrafirm Trade Patterns We can use the value chain concept as a way to categorize types of firms according to whether their main activities lie in technology, resources, manufacturing, or services. In other words, what is the major value-adding activity of the enterprise: resource extraction and processing, manufacturing, technology development, or supplying business services? These four categories are illustrated in a general way in Box 3.4, and explored below.13 Technology MNEs: These are high-tech firms specializing in production of new technology in products and/or processes, either for sale to other firms or as developed in their own products for sale to customers. Intrafirm trade in intangibles is the main type of interaffiliate trade. Information technology

The Multinational Enterprise as an Integrated Business

BOX 3.4 Typical Types of Multinational Enterprises

Resource-based MNEs

Extract and process natural resources

Service MNEs

Distribution, importing, sales and services

Manufacturing MNEs

Parts, subassembly, and final assembly

Technology MNEs

Technology development

The distribution of value added and the types of intrafirm trade flows are likely to vary in different types of multinationals. The value added in resource-based MNEs is typically earned upstream in the extraction and processing stages. Value added in manufacturing MNEs is normally highest in the subassembly and final assembly stages. Service MNEs generally earn most of their value added in the downstream stages of distribution and sales, while technology MNEs normally earn their largest portion of value added in producing process and/or product technologies. Over time, however, the value added within manufacturing MNEs has changed from being primarily assembly related to sales and service. This is also true for resource MNEs, which now earn most of their income from marketing differentiated products to industrial consumers rather than from extraction and processing. In addition, all MNEs are seeing the knowledge intensity of their production processes increasing. Thus value added is shifting from upstream to downstream functions, and the value added by support functions is rising. As a result, the value of services and intangibles is rising relative to tangible goods.

153

154 Multinationals and Intrafirm Trade BOX 3.5 The Value Chain of a Typical Pharmaceutical MNE

Head office management Purchasing/accounting/finance Basic R&D Production of fine chemicals

Marketing

Product development

Clinical testing

Preparation and packaging

Distribution and sales

The typical pharmaceutical multinational enterprise is a knowledgeintensive, vertically integrated firm. There are four stages in the value chain: the R&D stage (which includes basic and applied research and clinical testing), the production of fine chemicals, preparation and packaging into dosage form, and marketing and sales. The research stage is usually carried out near the parent's headquarters where products are developed for sale on the world market. Clinical testing is usually done in the final sales market, unless the country accepts clinical results from elsewhere. The production of fine chemicals is characterized by substantial economies of scale such that one or two plants can satisfy the global demand for a particular active ingredient. It is the most capitalintensive stage, located usually in the home country and in one or two others, depending on costs and available materials. The preparation and packaging (blending the fine chemicals with other ingredients and production in dosage form) stage is footloose since it has few economies of scale. Preparation/packaging is usually done in the least-cost location or in the domestic market, depending on the content policy of the local government. Marketing and sales are domestic functions which are located near the final consumers. Thus, of the four stages, all but the last can be imported. In most host countries, at least the research and chemical stages are imported.

The Multinational Enterprise as an Integrated Business

155

industries such as computer software (Microsoft, WordPerfect Corporation) and Pharmaceuticals (Glaxco) are good examples. Pharmaceutical MNEs are primarily engaged in technology development in basic and applied research and clinical testing; almost all costs are incurred before drug production actually occurs. As an example of a technology-intensive MNE, Box 3.5 outlines the various stages in the pharmaceutical value chain and Figure 3.2 provides a different, more disaggregated illustration of the same MNE's activities that shows the linkages between the stages of production. Resource-Based MNEs: Multinationals in the mining and metals sectors (e.g., Brascan, Alcan Aluminium) and oil and gas sectors (Imperial Oil, Mobil, Ontario Hydro) are typical examples of resource-based MNEs. These are vertically integrated multinationals, with their major value-adding activities in extraction and processing of raw materials. They may be horizontally integrated at the upstream end (diversifying source of supplies) and at the sales end (by consumer market). Product prices for these goods are often available on global markets such as the London Metals Exchange. The industry is usually oligopolistic and dominated by a price leader. Intrafirm trade flows are therefore primarily in commodities and shipping. The value chain for a typical petroleum MNE is illustrated in Figure 3.3. Manufacturing MNEs: Manufacturing multinationals are engaged in fabrication of parts, subassemblies, and assemblies for final sale in industrial and/or consumer markets. MNEs in the motor vehicle (General Motors, Ford), computer and office machines (IBM), and aircraft (Boeing, General Dynamics) industries are examples. These MNEs are generally both horizontally and vertically integrated. Their main value added lies in the production of parts, components, subassemblies, and final assembly, but technology development and sales are also important value-adding activities. Typical intrafirm trade flows are intermediate, semi-finished manufactures, horizontal exchange of finished product lines, and royalties for intangibles. The value chain in a typical automotive assembly MNE, such as Ford or General Motors, is illustrated in Figure 3.4. Service MNEs: These firms are driven by their need to be close to their customers, whether they are the general public or other firms. Examples of consumer-oriented service firms are in the retail industry (Wal-Mart, Loblaws, Sears), hotels (Holiday Inn), telecommunications services (AT&T, BCE), finance (Royal Bank, American Express), insurance (Sun Life), and transportation services (Air Canada, American Airlines). Professional service MNEs can be found in the international accounting, legal, and consulting industries (McKinsey, Coopers & Lybrand, H&R Block). The major value added by service MNEs is downstream, in the development and provision of

FIGURE 3.2 The Value Chain in a Typical Pharmaceutical Multinational Enterprise

FIGURE 3.3 The Value Chain in a Typical Petroleum Multinational Enterprise

FIGURE 3.4 The Value Chain in a Typical Automotive Assembly Multinational Enterprise

FIGURE 3.5 The Value Chain in a Typical Airline Multinational Enterprise

160 Multinationals and Intrafirm Trade services. Intrafirm trade flows generally entail the exchange of services and intangibles such as goodwill. As an example of a service MNE, we illustrate the value chain in a typical airline MNE in Figure 3.5. Stylizing MNEs in this fashion will help us to identify the transfer pricing problem areas where disputes are likely to arise between the MNE and the tax authorities, whether in terms of the pricing of tangibles, intangibles, or services. In fact, of course, most MNEs embody all or most of these activities to a greater or lesser extent, and so transfer pricing disputes occur in all three categories.14 Factors Affecting the Pattern of Intrafirm Trade In addition to the general type of MNE, several other factors, both internal and external to the firm, can influence intrafirm trade patterns within the MNE. First, the age of the affiliate may affect its intrafirm imports and exports. Second, firm-level factors such as the degree of vertical and/or horizontal integration within the enterprise and the overall strategy of the MNE - whether it is integrative or nationally responsible or attempts to be both (sometimes called a mixed or multi-focal strategy) - are important determinants. At the national level, particular home and host country-specific advantages (CSAs) such as trade barriers, transport costs, corporate income taxes, and FDI regulations can encourage or discourage intrafirm trade flows. At the regional level, preferential trading arrangements and tax treaties can affect the nature and volume of intrafirm trade flows. Finally, technological change can also be a factor influencing the level and direction of intrafirm trade. We look first at the internal factors, and then at external influences. Age of the Affiliate and Intrafirm Trade The type and volume of intrafirm trade may partly depend on the age of the affiliate. New factory plants are usually extractors, offshores, or distributors depending on their strategic function within the MNE (resource, cost, or market based). Their intrafirm trade patterns are, however, very different. Extractors and offshores are upstream plants, feeding low-tech intermediate goods to downstream affiliates. Distributors are downstream plants, importing finished products for local sale. Over time, the strategic function of a plant may change as the plant grows and matures and has the capabilities to undertake new functions. If the subsidiary is allowed relatively autonomous development within the MNE, this growth in functions is more likely. Therefore as the foreign affiliates mature, extractors may take on processing functions, offshores become source factories, and distributors become focused factories. As a result, their

The Multinational Enterprise as an Integrated Business 161 technological sophistication rises and their intrafirm import and export patterns become more complex. Support service affiliates generally come later, as the MNE network spreads and additional units are needed to supply financial, advertising, and other business services on a regional basis. Vertical and Horizontal Intrafirm Trade Whenever a firm is vertically integrated, intrafirm trade is inevitable. Vertically integrated MNEs tend to locate plants based on economic motivations, with upstream plants being run as cost centres, directing their outputs to downstream plants for further processing. Transfer pricing decisions tend to be made by the financial unit at the MNE's headquarters and set so as to maximize MNE profits net of taxes and tariffs. Intrafirm trade in intermediate products may be in commodities, such as copper or wood pulp, which are standardized and where external markets exist. Prices in these external markets represent the opportunity costs to the firm of trading inside as compared with trading outside the firm. Such external prices tend to be visible and noncontroversial (e.g., on the London Metals Exchange or prices of various grades of crude oil or wood pulp).15 On the other hand, the upstream plant may be producing specialized intermediate components for which no external market exists. A common relation in this regard is for the parent firm to supply intangibles to the foreign affiliate in return for royalties, with the affiliate producing the product for resale back to the parent at cost plus a mark-up. Where there are no external suppliers the question is what the price would be were the MNE to contract production with an outside supplier. The problem is that if the intangible knowledge needed for production is difficult to patent, the MNE may be reluctant to move to outside contracting. When an MNE is horizontally integrated, affiliates are located at the same stage of the value chain and may or may not trade with one another. If plants are specialized in individual products or product lines, they may exchange these with sister affiliates either to fill gaps in product lines or to handle temporary periods of plant excess supply or demand. On the other hand, if the foreign plants were set up behind a tariff wall to satisfy a local market, the plants may be miniature replicas of their parents, with clearly segmented markets and little intrafirm trade between them. Similarly, at the resource-extraction stage, plants may be in direct competition with one another, with no intrafirm trade between them and all their output directed downstream. Organizational Structure and Intrafirm Trade Multinationals must choose an organizational structure for the enterprise that answers the question: How much autonomy should the affiliates have? There

162 Multinationals and Intrafirm Trade are at least two clear choices. The answer affects the nature of intrafirm trade within the affiliates of the MNE. In a globally integrated strategy the MNE is a tightly connected organizational network, with the head office closely monitoring and integrating affiliate activities so as to maximize global profits. Resource-seeking and cost-reducing strategies tend to be integrative strategies since affiliates are closely linked through vertical intrafirm trade and financial flows. In mature, global industries such as automobiles and consumer electronics, multinationals tend to use integrative strategies that are cost driven, with offshores and server factories to divide up the production process among their affiliates and subcontractors. In a locally responsive strategy the individual affiliates of the MNE have more local autonomy, often being run as profit centres. Market-driven strategies may be either integrative (e.g., if affiliates specialize in one product line and import other lines from affiliates) or nationally responsive (e.g., if affiliates are run as autonomous, miniature replicas of their parents). Where it is important for the firm to be seen as an insider, such as in government-controlled industries like telecommunications and aircraft, MNEs tend to adopt more nationally responsive strategies such as miniature replicas and focused factories. Affiliates may be miniature replicas or copy-cats of their parents, with little intrafirm trade flows between them. Focused factories are another relatively autonomous and nationally responsive plant structure. In a locally responsive strategy, both economics and politics affect plant output, sales, and trade decisions since the firm needs to be seen as an insider. Transfer pricing decisions may be left to the individual plant managers to negotiate among themselves. A new form of business strategy in the 1990s is the regional core network in which the MNE incorporates both integrative and locally responsive strategies at a regional level. These 'regionally integrated, independently sustainable networks of overseas investments [are] centered on a Triad member,' according to the UN Center for Transnational Corporations (UNCTC 1991,42). Each network tends to have a lead plant in a member of the Triad plus affiliates located in regional spoke countries (e.g., lead plants in the United States with cluster plants in Canada and Mexico). By setting up a core network the multinational ensures that it has access to, and that its affiliates become regional insiders in, each of the three Triadic regions. The UNCTC argues that regional core networks have both economic motives (e.g., specialization based on comparative advantage, economies of scale) and political motives (e.g., avoidance of tariff and nontariff barriers, qualifying as insiders for government subsidy purposes). Such a strategy is, within the region, both integrative because it takes advantage of regional specialization and economies of scale, and locally responsive because it is attentive to government policies and domestic interest groups within the region.

The Multinational Enterprise as an Integrated Business

163

Country-Specific Factors and Intrafirm Trade Economic, social, and political host country factors can each affect the volume and pattern of intrafirm trade by a particular affiliate. As the OLI paradigm shows, national factors such as transport costs, weight-to-value ratios, trade barriers, and wage rates will encourage or discourage the siting of MNE activities in particular locations. In addition, CSAs can affect the choice between worldwide versus domestic sourcing of intermediate products or between local processing or simple export of raw materials for processing elsewhere. The size of the local market can affect the choice of a simple distributor plant or a more complex focused factory, with consequent implications for the type of intrafirm imports and exports. Government regulations that require a local presence or purchases of local inputs can have an impact upon intrafirm trade patterns. Regional Integration and Intrafirm Trade One important CSA at the regional level which has influenced both plant location and intrafirm trade patterns is the recent proliferation of regional trading agreements such as the Canada-U.S. Free Trade Agreement (FTA), which became law on 1 January 1989, and the North American Free Trade Agreement (NAFTA), which took effect on 1 January 1994. Under a free trade agreement crossborder tariffs are eliminated, but the member countries are free to maintain their own tariffs against nonmember countries. A customs union carries this one step further by requiring that the members adopt a common external tariff; an example of a customs union is the European Union. The key to investment and production decisions in response to free trade agreements is the reduction in policy risk for firms within the region due to increased security of access to the markets of member countries (Eaton et al. 1994a). With policy risk reduced, the strategic responses of firms are partly determined by factors such as: (1) whether the firm is headquartered inside or outside the area; (2) whether the firm has significant investments inside the region; and (3) the industry(ies) in which the firm competes. Eden (1994b) identifies these three categories of firms as the veterans (well-established multinationals located inside a free trade area with significant investments in the partner countries prior to the agreement), the outsiders (foreign firms outside the area, which may have been exporting into the area or may have investments inside the area), and the domestics (local firms inside the area without significant investments in the other partner countries; they may or may not already be exporting to these countries). The responses of each should vary since the insiders should be better positioned to take advantage of the benefits from the reduction in crossborder trade barriers (Eden 1994b, Vernon 1994b, Dunning 1994). We explore this issue later in this chapter in the context of NAFTA.

164 Multinationals and Intrafirm Trade Technological Change and Intrafirm Trade16 The shift from mass to lean production that is ongoing throughout North America can and is having a major impact on intrafirm trade patterns, one of which is the impact on-configuration of R&D activities as noted above by Eaton et al. (1994a). U.S. multinationals built their success on mass-production techniques: the use of simple interchangeable parts, the division of labour and specialization of tasks, the assembly line, the stocking of inventory, and the substitution of capital for labour (Womack et al. 1990). The search for ever lower costs led to large plants built to achieve minimum efficient scale, the upstream ones close to the source of raw materials and the downstream ones close to the consumer. The middle stages of production, where footloose, were located where costs were the cheapest. Thus MNEs engaged in large volumes of intrafirm trade, primarily in intermediate products being shipped around the world for further processing in low-cost locations before their final sale, mostly in the developed market economies (Eden 199la). The shift to lean or just-in-time production is changing the configuration of MNE activities and thus the pattern of intrafirm trade. The key characteristics of lean production are outlined in Box 3.6 below. A lean production strategy uses skilled labour, flexible technology, and economies of scope to shift up the value chain into higher-value products. The emphasis is on achieving both low cost and higher quality in a product that closely meets consumer needs. Lean production therefore provides a way to escape the 'stuck-in-the-middle' strategy outlined by Porter (1995) since it involves both cost reduction and product differentiation. A lean production strategy means the firm must be located near the consumer, and this draws first- and second-tier suppliers in close to the downstream activities of the MNE. Thus some of the footloose stages of production, allocated under a mass-production strategy to low-labour-cost countries in East Asia and Latin America, return to the country of final sales. The MNE should rely less on offshores and source factories and more heavily on focused and lead factories and R&D outposts. Economies of scale at the plant level become less important so that plants of different sizes can be equally efficient; economies of scale at the firm level, however, can become more important due to the high overhead costs of technology upgrading. Quality control becomes an essential part of an informationmediated MNE (Kenney and Florida 1993; Shrank and Govindarajan 1993). Since the costs of robotics and information technology necessary to achieve market success at the regional or global level are high, firms may engage in selling off unrelated activities (i.e., become lean enterprises) and use strategic alliances and/or outsourcing to first-tier suppliers to increase their flexibility and ability to manage change.

The Multinational Enterprise as an Integrated Business

165

BOX 3.6 Key Characteristics of Lean Production

Characteristic

Explanation

Demand-driven production

The philosophy of production shifts from producing to stock to producing to order. Production is in smaller batches with greater variety.

Minimization of downtime

Quick changeovers and setups are essential. Production workers are trained to work on a variety of machines.

Pull-through work flow

Factory layouts are changed to encourage smooth flow-through of batch production.

Inventory reduction

Firms switch from 'just-in-case' storage of inventories to 'just-in-time' inventory control. This means that suppliers must locate close to their downstream buyers in order to ensure quick delivery of inputs.

Zero-defect components

Components are of perfect quality in order to maintain pull-through work flow.

Total quality control

Expenditures are made to control quality, including prevention costs (including quality circles), appraisal or monitoring costs, costs of internal failure (costs of fixing bad quality before it leaves the factory), and costs of external failure (warranty claims, customer ill will).

Knowledge-intensive production

Workers are multi-skilled and paid according to their skill level and output quality.

SOURCE: Based on UNCTC (1988)

Summary: Multinationals and Intrafirm Trade We have argued that MNEs can be characterized as one of four types depending on whether their main activities lie in technology, resources, manufacturing, or services. Each type tends to have a different pattern of intrafirm trade. In addition, several other factors can affect the pattern of intrafirm trade such as internal factors (the age of the affiliate, the degree of vertical or horizontal integration of the MNE, and the MNE's choice between a local versus a global strategy) and external factors (the country-specific advantages of the host coun-

166 Multinationals and Intrafirm Trade try, regional integration, and technological change). We now move to a case study that illustrates these propositions. Multinationals and the North American Free Trade Agreement In this section we apply our theory of the impacts of regional integration and technological change on MNE intrafirm trade patterns to a case study of the North American Free Trade Agreement. We look first at the theory of how the FTA and NAF TA should affect MNE investment, production, and trade patterns and then at some evidence provided by a Conference Board of Canada survey. Multinationals after NAFTA Under the NAFTA, over the next 15 years, all tariff and nontariff barriers among Canada, the United States, and Mexico are to be either eliminated or harmonized. Each country can maintain its own trade barriers vis-a-vis nonmember countries such as Japan and the United Kingdom, but intracontinental trade barriers should fall dramatically. Thus valuation of imported goods for customs duty purposes should cease to be a major factor influencing intrafirm trade; however, firms, in order to qualify for duty-free status must meet certain rule-of-origin tests. Where these tests are not met, tariffs continue to apply. In addition to eliminating tariffs and reducing most nontariff barriers over 15 years, the NAFTA contains several investment provisions. The key commitment in terms of investment is to national treatment (NAFTA partners must be treated at least as well as domestic investors) together with most-favourednation treatment (NAFTA investors must be treated at least as well as any foreign investor). Nondiscrimination applies to all North American investments and investors, including firms controlled by non-North Americans. There is a list of proscribed performance requirements, and most existing requirements must be phased out over ten years. Restrictions on capital movements, including all types of payments and profit remittances, are forbidden except for balance of payments reasons. Expropriation is outlawed, except for a public purpose and on a nondiscriminatory basis, and full and prompt payment of fair compensation is required. Investors can seek binding arbitration against a host government for violations of NAFTA obligations (Eaton et al. 1994b; Kudrle 1994; Gestrin and Rugman 1993, 1994; Graham and Wilkie 1994). Eden (1994b) asserts that U.S. MNEs are best placed to take advantage of the falling tariff and nontariff barriers that the FTA and NAFTA will bring because they are already located in all three countries. The configuration of U.S. subsidiaries in North America was historically based on the 'blocks' that governments

The Multinational Enterprise as an Integrated Business

167

had positioned on the North American 'chessboard.' With governments now removing these blocks, the underlying economic factors will have more impact on location decisions. Because MNEs are international oligopolists, concerned about their shares of global markets, they will change the configuration of their activities so as to increase their international competitiveness. Eden (1994b) argues that these veterans will locate, close, and/or expand their plants with the whole North American market in mind. This should lead to reduced numbers of product lines in various plants and increasing horizontal trade among plants. MNEs are also likely to segment their production process among plants so that more vertical intrafirm trade takes place. As a result, there should be more crossborder vertical and horizontal intrafirm trade flows taking place. Certain product lines, industry segments, and plant functions will shift among the three countries, and these will cause job losses and plant closures in certain locations. Which changes will occur depends on a complicated array of factors, some of which are exogenous to the firms involved (such as factor prices and transportation costs) and others which are firm specific (such as the nature of the products produced and the ingenuity and animal spirits of the individuals involved). Veteran multinationals that are resource seeking are likely to use the FTA and NAFTA to relocate extractor and processor affiliates where resources are relatively more abundant and capital and energy costs lower. Cost-reducing affiliates may be induced to relocate to North American countries with lower unit labour costs. Market-driven affiliates are likely to rationalize production in existing plants, creating more vertical integration through source factories and more horizontal integration through focused factories. MNEs that have not yet opened up branch plants in potential NAFTA members may do so. Therefore the responses of parents and affiliates will involve relocation and expansion, taking advantage of lower trade and investment barriers to develop a more integrative regional strategy. In the short run, existing plants are unlikely to be closed, but in the longer term economic efficiency will determine locational decisions within North America. The key here is rationalization of demand for the regional market as a whole, and of supply to capture specialization and economies of scale (Eden 1994b). Outsiders - non-North American MNEs that have already established transplant operations within North America - if not deterred by investment barriers, are likely to also expand and rationalize their investments to take advantage of the larger market size. If rules of origin are tightened in order to meet North American content, transplants may be forced to upgrade production and source more inputs locally. Thus parts plants may be induced to follow distributors and assembly plants. Outsiders that are currently exporting to North America may

168 Multinationals and Intrafirm Trade shift to foreign direct investment. They are likely to be drawn to the larger market (the U.S. market or the hub) unless cost differentials make location in the spokes (Canada and Mexico) more attractive and/or interregional barriers are completely eliminated. For domestics, firms without established links to other potential NAFTA members, a free trade area will be seen as both an opportunity (i.e., new markets, access to lower cost inputs) and a threat (i.e., more competition). Such firms, with encouragement, may start or increase their exports within North America and possibly open up distributors or offshore plants where market size or costs warrant. They will, however, have to face the difficult task of breaking into established distribution networks of domestics and MNEs in the North American markets. In terms of the location of research and development activities in MNE networks, Eaton et al. (1994b) find that there are strong forces favouring the centralization of R&D at the parent firm's headquarters; however, significant decentralization of R&D has occurred over the past ten years for some countries and industries as the knowledge base becomes more geographically dispersed. If the FTA and NAFTA encourage rationalization of firm activities and reduce the autonomy of foreign manufacturing subsidiaries, the authors conclude that the production of local R&D by subsidiaries in Canada and Mexico may also be reduced. Where inside North America firms are likely to locate as a result of the FTA and NAFTA, in addition to being affected by country-specific factors, is also partly dependent on agglomeration economies, or the advantages of firms clustering in one location. Eaton et al. (1994a) develop a theory of agglomeration which explains geographic concentration of business activity as the outcome of two opposing forces: economies of scale at the plant level (encouraging concentration) and transportation and communication costs (discouraging concentration). Restrictive trade policies can encourage agglomeration; the reverse, however, does not necessarily follow. The authors conclude that the probable impacts of the FTA and NAFTA on agglomeration are unclear since scale economies and asset specificity discourage quick dissolutions, and thus have opposite effects to falling trade barriers. In addition, new technologies of production have both reduced the importance of labour and transport and communications costs and increased the need for supplier firms to locate proximate to their downstream customers. As a result, technological change may lead to increased diversity in patterns of industrial location. The 1992 Conference Board Study: Multinationals after the FTA Some evidence on the responses of MNEs to internal and external factors

The Multinational Enterprise as an Integrated Business

169

TABLE 3.1 Plant Rationalization by Multinationals in Canada Have already rationalized

Have already rationalized

Anticipate rationalizing in future

Anticipate rationalizing in future

in the area of:

Parents

Subs

Parents

Subs

Marketing New opportunities Specialization - sophistication of product Scale of production Location Quality Upgrading inputs Codes and standards Alternative supplies and services Changes to new suppliers

43 34

30 35

34 31

34 30

29 27 23 21 18 11

27 44 20 24 18 16

29 39 34 14 13 19

24 34 16 18 10 19

7 7

21 19

9 7

19 17

Per cent of rationalized MNFs that haH ratinnaliypH

SOURCE: Based on data in Krajewski (1992, 1-9)

affecting intrafirm trade is presented in a 1992 Conference Board of Canada survey of multinationals in Canada. This study examined the changing nature of intrafirm trade flows and international production patterns in North America, after the FTA but before NAFTA (Krajewski 1992, 1-9). The Conference Board sent out a sample survey to 1,000 firms in Canada and supplemented this with interviews. The study looked at two groups: 250 Canadian parents with U.S. subsidiaries (PARENTS) and 750 Canadian subsidiaries with U.S. parents (SUBS); the per cent of respondents in each group is not clear in the report. The report defines intrafirm trade (IFT) as importing and exporting between affiliated companies in Canada and the United States. The results are summarized in Table 3.1. For manufacturing firms, the study shows that for PARENTS, exports by U.S. affiliates as a per cent of total reported company exports fell from 52 per cent in 1985 to 33 per cent in 1990, while imports by U.S. affiliates as a per cent of total reported company imports fell from 76 per cent in 1985 to 57 per cent in 1990. It is interesting that the data show a drop in intrafirm trade as a per cent of all Canada-U.S. trade; so firms are selling more to or buying more from outside suppliers. This is consistent with the move to cut business lines, do away with in-house parts suppliers, and downsize that is happening under just-in-time production. The report also says PARENTS exported more in intrafirm trade to

170 Multinationals and Intrafirm Trade

BOX 3.7

Krajewski (1992) Survey of Factors Motivating Canada-U.S. Intrafirm Trade Motivating factors

Explanation of motivating factor

Corporate strategy

The overall production strategy of the MNE, the desire to rationalize production among the affiliate family, and the need to adjust particular plant functions in the light of declining tariff barriers are important motivations for changes in intrafirm trade patterns.

Production strategy of the MNE

The overall production strategy of the MNE matters; i.e., the different locations of plants within North America allow flexibility to switch production between plants, providing a 'back-up plan'.

Strategic rationalization

Firms are concerned with eliminating duplication within the MNE, on a continental and worldwide basis. A foreign plant must justify its existence by producing products that are in demand and/or are not made elsewhere within the MNE; otherwise, the plant will be closed.

Plant organization along product lines

The respondents say that product mandates are replacing geographic mandates. Geographic mandates at the national level had been the traditional driver of plant function, e.g., miniature replica plants in Canadian manufacturing sector. In the 1970s manufacturing plants in Canada produced for the Canadian market only. With tariffs coming down, these plants are reorganizing on a continental basis. These focused factories now produce specific product lines for exchange regionally or globally within the MNE.

Cost advantages

The next highest-ranked factors are lower costs and economies of scale as a motivation for rationalizing production.

Internal supply lines

Subsidiaries in Canada tend to supply their U.S. parents with inputs or certain specialized products made only by the affiliate. One reason may be that an arm's length supplier might not be able to match quality and specifications, so the MNE would choose a Canadian supplier for reliability and integrate it into the corporate network. Also, just-in-time production might encourage trade between affiliated companies. Lastly, the internalization motivation could also explain the use of internal supply lines.

The Multinational Enterprise as an Integrated Business 171

BOX 3.7 (concluded) Motivating factors Government policies

Explanation of motivating factor Government policies are not ranked as significant motivating factors in the IFT decision. Some firms do not see government policy as affecting their trade; however, a majority of firms admit that the FTA was one of the most important developments for their North American activities. Thus, while firms may not perceive government policies in general as affecting their IFT decisions, specific policies that affect trade flows such as regional trading agreements are seen as important.

SOURCE: Based on information in Krajewski (1992)

their U.S. subsidiaries than they imported in terms of volume and share of total company sales. Since IFT exports were less concentrated than IFT imports, this must mean exports in total are much larger than imports. The Conference Board study does not provide similar data for SUBS but does say that SUBS have IFT export ratios similar to their IFT import ratios, which is different from PARENTS data. The Conference Board study then looked at the motivating factors behind Canada-U.S. intrafirm trade and concluded that several factors are important variables affecting IFT. The results of this survey are summarized in Box 3.7 above. The study also asked firms if they had undergone any significant change since 1989 (the date when the FTA came in); 75 per cent of the SUBS and 82 per cent of the PARENTS said yes. PARENTS had rationalized primarily on the demand side, through changes in marketing techniques (43 per cent), looking for new market opportunities in the U.S. market (34 per cent), and through changes in product specialization and/or sophistication (29 per cent). SUBS had rationalized primarily on the supply side through changes in scale of production (44 per cent) as well as through the demand side. In terms of anticipated rationalizations, PARENTS expect changes on both the demand and supply sides, with 39 per cent anticipating a change in scale of production and 34 per cent a change in plant location. This can be seen as PARENTS reacting to the FTA and to the anticipation of trilateral free trade with Mexico through shifting plant locations and plant sizes, their reactions being due both to the threat of new

172 Multinationals and Intrafirm Trade competitors as well as to the opening of new market opportunities.17 SUBS anticipate continued rationalization through economies of scale, marketing, and new opportunities. Thus the FTA can be seen as the primary driver, in addition to globalization of markets in general, that is causing multinationals in Canada to rationalize their production and sales. The study argues that the FTA has shaken up the traditional FDI balance between Canada and the United States. PARENTS now sense new opportunities and/or feel that they must compete globally in order to survive. The FTA has given them business opportunities through a more open door to the U.S. market (as NAFTA will do to the Mexican market). The report argues that Canadian firms need to be more aggressive, both to protect the Canadian market and to penetrate the U.S. market. The unspoken fear in the report is that rationalization may mean relocation of PARENTS to the United States in the long run. These firms already have plants in the United States (by definition of the sample); so the growing importance of the location variable (from 23 to 34 per cent of the PARENTS) should be a worry for Canadian policy-makers. Since 1989 SUBS have been continually reevaluated by their U.S. parents in terms of the need to keep manufacturing in Canada. The days of 'copy-cat' or miniature replica plants in Canada are now gone; a Canadian plant has to fit into the existing structure of the MNE by playing a role in its overall strategy. The question is whether FDI in Canada will be curtailed in the long run. There is evidence that product lines are being scaled back in favour of filling niche markets for the entire parent group. U.S. subsidiaries in Canada are now in a critical position; they have to carve out and justify their existence as an integral component of the MNE. The responsibility to maintain manufacturing operations lies within company itself; the SUB has to justify its existence and place within the MNE. Conclusions In this chapter we have outlined a theory of the multinational enterprise as an integrated business, first reviewing the OLI paradigm that explains why such global enterprises exist and why they are successful, and, second, examining how MNEs formulate and implement global strategies: In particular, we concentrated on two types of strategies: coordination and configuration. Using the concept of the value chain, we explained the factors that affect MNE coordination and configuration choices. We then argued that MNEs can be characterized as one of four types depending on whether their main activities lie in technology development, resources, manufacturing, or support services, and that the particular type of MNE affects its choice of intrafirm trade patterns. We exam-

The Multinational Enterprise as an Integrated Business

173

ined several internal and external factors that affect intrafirm trade, focusing in particular on regional integration. We concluded with a look at regional integration in North America and the Conference Board study of Canada-U.S. intrafirm trade. Our analysis suggests that intrafirm trade will be of increasing importance in the North American economy as MNEs restructure their upstream activities to take advantage of differences in country-specific endowments and restructure their downstream activities around a continental market. Technological change, in the form of new information technologies and just-in-time production, complicate decision making, raising the importance of location close to market and reducing the importance of labour and materials costs. As firms downsize, restructure, and reconfigure their activities, it should become increasingly difficult to regulate the complex intrafirm transactions that make up the industry and firm value chains. This is the microeconomic theory of the firm that lies behind the concept of the MNE as an integrated business. We have outlined the factors that encourage internalization, what the internal market of the MNE looks like, how it is reflected in trade patterns, what factors can affect the volume and nature of this trade, and how regional free trade schemes such as the FTA and NAFTA can influence the MNE's production, sales and intrafirm trade patterns. Finally, we reviewed a recent survey that looked at the factors affecting intrafirm trade patterns in North America. For statistical evidence on the size and direction of these in North America, we turn to Chapter 4.

4

Multinationals and Intrafirm Trade in North America

Introduction In this chapter we examine the roles played by multinationals, in particular by U.S. MNEs, in the North American economy. We first examine U.S. and Canadian trade and investment patterns, focusing on their linkages with each other, and with Mexico, Japan, and the European Community. Next we look at the top 20 multinationals in Canada and the United States. Our focus then shifts to estimates of the size and pattern of Canadian and U.S. intrafirm trade in goods and services, using estimates from the U.S. Department of Commerce and Statistics Canada. We find that about two-thirds of Canada's trade and investment flows are conducted with the United States. Intrafirm (i.e., non-arm's length) trade accounts for approximately half of total Canada-U.S. trade, both in goods and in services. In some categories (e.g., automotive tooling charges and royalty payments by Canadian subsidiaries to their U.S. parents) the share of intrafirm trade is almost 100 per cent. We explore the reasons for these differences in the light of the integrated nature of the multinational enterprise, as outlined in Chapter 3. North American Trade and Investment Patterns The Hub-and-Spoke Relationship Perhaps a useful starting point for our discussion of intrafirm trade in North America is to focus on the general nature of the relationship. We argue that silent integration, the multiplying of linkages and interdependencies among economies, has long been part of the economic history of North America (Eden andMolot 1992b).

Multinationals and Intrafirm Trade in North America

175

American multinationals have been the major investors in both Canada and Mexico since the late 1800s. Canada historically has had, and still has, the highest degree of foreign ownership and control of any industrialized country. Penetration by U.S. multinationals in Mexican resource sectors was extensive until the 1917 Mexican revolution led to the nationalization of the petroleum industry and the expulsion of most multinationals. After the interruption in foreign direct investment (FDI) caused by two world wars, integration between the United States, and its northern neighbour, Canada, and its southern neighbour, Mexico, proceeded rapidly as U.S. multinationals began serious penetration of the Canadian and Mexican economies in the 1950s. Initially, FDI was resource seeking or designed to supply the domestic market behind high tariff walls, but as tariff barriers came down through the various GATT rounds, U.S. multinationals began to rationalize production on a continental basis. While MNEs, through their intrafirm trade, technology, and investment flows, were creating intracontinental linkages, their governments alternatively tried to facilitate or to impede this silent integration. Facilitating integration have been policies such as the Canada-U.S. Auto Pact and the Mexican maquiladoras program. Restrictive investment regulations and nationalizations, on the other hand, have slowed the integration process but not reversed it. Since the mid-1980s the integration pace has again quickened, especially since the passage of the Canada-U.S. Free Trade Agreement (FTA), Mexico's joining the GATT, and the recent signing of the North American Free Trade Agreement (NAFTA).1 Silent integration has created a hub-and-spoke economic relationship among the three economies, one that is visible in trade and in investment patterns.2 Merchandise trade flows are clearly in the form of a hub-and-spoke pattern, with the United States as the hub or centre of the economy and Canada and Mexico as its northern and southern spokes, respectively. The United States is the largest partner for each country in terms of inward and outward foreign direct investment and in terms of exports and imports for both Canada and Mexico. Some Statistics on the Hub-and-Spoke Relationship North American gross domestic product (GDP) in 1991, was 29 per cent of world GDP, for a total of US$6,464 billion. The U.S. share of the North American gross domestic product was 88 per cent, with Canada generating 9 per cent, and Mexico 3 per cent. In 1991, total North American trade (exports plus imports), at US$1,260 billion, was 18 per cent of world trade; intraregional trade was 19 per cent of the North American total. North America's 1991 FDI stock (inward and outward) was 28 per cent of the world FDI stock. Intraregional FDI was 11 per cent of the North American FDI stock.3

176 Multinationals and Intrafirm Trade TABLE 4.1 The Hub-and-Spoke Relationship in North America, Early 1990s United States Total GDP, 1992/(US$ billions) Country share of North American GDP (%) Total population, 1992 (millions) Country share of North American population (%) Total merchadise trade (exports plus imports), 1991 (US$ billions) Country share of North American trade (%) Total FDI stock (inward plus outward), 1991 (US$ billions) Country share of North American FDI stock (%) Number of top 1,000 North American firms based in this country* Country share of top 1,000 firms (%) Number of parent MNEs based in this country** Number of foreign affiliates based in this country **

North America

Canada

Mexico

6038.0

537.0

329.0

6,904.0

87.5

7.8

4.8

100.0

255.4

27.4

85.0

367.8

69.4

7.5

23. 1

100.0

908.8

245.4

98.6

1,252.8

72.5

79.6

7.9

100.0

857.8

197.2

18.3

1,073.3

79.9

18.4

1.7

100.0

823

158

19

1,000

82.3

75. 8

1.9

100.0

2,972

1,396

N.A.

N.A.

15,341

6,328

8,420

N.A.

*List of the top 1,000 firms is from Knubley et al. (1994). **UNCTAD (1994, 4) data on MNEs are from different years: U.S. (1991), Canada (1992), Mexico (1993). SOURCE: Author's calculations based on data in Griffin and Pustay (1995, 44), Knubley et al. (1994, 150-540), and UNCTAD (1994, 4)

Table 4.1 provides some additional statistics on the hub-and-spoke relationship in the early 1990s, in terms of GDP, population, trade, FDI, and numbers of MNEs. In general, the statistics reflect the strength of the U.S. economy as the economic hub, and the undeveloped state of the Mexican economy relative to its northern neighbours (compare Mexico's population share to its economic shares of trade and FDI, for example). We examine the U.S. and Canadian data on trade and FDI patterns, and roles played by MNEs in both countries, in more detail below.

Multinationals and Intrafirm Trade in North America

177

Canada-U.S. Merchandise Trade Patterns Table 4.2 provides data on U.S. and Canadian merchandise trade (trade in goods only) with each other, and with Mexico, Japan, and the European Community (EC), over the 1980-92 period. Over the 1980-92 period, the general direction of Canadian exports shifted away from Europe and towards the United States. Canadian imports, however, shifted towards the EC and away from the United States until the start of the 1990s, when the situation began to reverse. In 1980, 63 per cent of Canadian exports went to the U.S. market, compared wit 12.7 per cent to the EC, and Canada bought 68.5 per cent of its imports from th United States, compared with 8.1 per cent from the EC. By 1992, 77.3 per ce of Canadian exports were sold to, and 65.2 per cent of its imports came from, the United States. Only 7.1 per cent of its exports, and 9.8 per cent of its imports, came from the European Community. The Canada-U.S. trade spoke, for Canada, has therefore strengthened over the past decade. This pattern has also been true for the United States, but not to as great an extent. In 1980, 17.7 per cent of U.S. exports went to Canada, and 22.9 per cen went to the EC. In terms of U.S. imports, 17 per cent came from Canada, and 14.5 per cent from the EC. By 1992, 20.4 per cent of U.S. exports and 19 per cent of U.S. imports were traded with Canada. Another 24.3 per cent of its exports, and 17.5 per cent of its imports, were traded with the EC. As barriers to trade within North America have fallen in the wake of the 1978 GATT Tokyo Round, the FTA and the recent NAFTA, one would expect intracontinental trade flows to increase, and the Canada-U.S. trade relationship to deepen. Thus the general trend towards deeper integration of the two economies, at least in the area of crossborder trade in goods, is clear. Let us now look at the pattern in terms of investment. Canada-U.S. Foreign Direct Investment Patterns Table 4.3 provides parallel information on U.S. and Canadian foreign direct investment (FDI) stocks, measured on a historical cost basis, for 1985-91. In 1991, 63.3 per cent of the inward FDI stock in Canada was controlled by U.S. firms; this amount represented 15.2 per cent of all U.S. direct investment abroad. At the same time, 58 per cent of Canadian direct investments abroad were held in the United States, representing 9 per cent of all inward FDI in the United States. Both inward and outward FDI in Canada have diversified away from the United States and into Europe over the past ten years. The same is true for the United States: the European shares of U.S. inward and outward investment have risen relative to the Canadian shares (Knubley et al. 1994; Niosi 1994). In 1991, 23 per cent of the FDI stock in Canada was held by investors from the European

178 Multinationals and Intrafirm Trade TABLE 4.2 Trends in the Direction of North American Merchandise Trade (nominal dollars) Part A: The direction of Canadian exports 1989 1980 1991 1985 Exports (Can$ millions) U.S. share (%) Mexican share (%) EC share (%)* Japanese share (%)

76,158 63.25 0.65 12.7

5.7

1 19,474 77.9

138,701 73.2

145,924 75.1

162,596 77.3

0.3 5.8 4.8

0.5 8.6 6.4

0.4 8.0 4.9

0.5 7.1 4.6

Part B : The origin of Canadian imports 1989 1980 1985 Imports (Can$ millions) U.S. share (%) Mexican share (%) EC share (%)* Japanese share (%)

1992

1991

1992

135,334 63.8

147,994 65.2

1.9 9.8 7.3

69,273 68.5

104,355 69.0

135,191 65.2

0.5 8.1 4.2

1.3 9.8 6.5

1.3

1.9

11.0

10.8

7.1

7.6

Part C: The direction of U.S. exports 1980 1985 1989 Exports (US$ millions) Canadian share (%) Mexican share (%) EC share (%)* Japanese share (%)

233,966 17.7

214,424 25.1

360,465 22.1

415,962 20.4

6.5

6.2

6.8

8.0

22.9

21.1 10.3

23.5 12.1

24.3 11.4

8.9

Part D: The origin of U.S. imports 1980 1985 Imports (US$ millions) Canadian share (%) Mexican share (%) EC share (%)* Japanese share (%)

1991

249,308 17.0

338,863 21.0

1989

1991

475,329 18.8

489,398 19.0

5.0

5.6

5.7

6.4

14.5 12.5

18.5 19.4

18.0 19.7

17.5 18.7

* Includes Spain and Portugal after 1986. SOURCES: (A) Statistics Canada (1989a, 1992b); (B) Statistics Canada (1989b, 1992c); (C) U.S. Department of Commerce, Sui-vey of Current Business, various years; (D) U.S. Department of Commerce, Statistical Abstract of the United States, various years

Multinationals and Intrafirm Trade in North America

179

TABLE 4.3 Trends in the Direction of North American Investment (nominal dollars) Part A: Canadian foreign direct investment abroad

Total stock (Can$ millions) U.S. share (%) Mexican share (%) EC share (%)* Japanese share (%)

1980

1985

1989

1991

26,967 69.5 0.6 15.8 0.4

54,123 68.5 0.4 12.6 0.4

80,779 62.3 0.3 18.8 0.5

94,435 57.9 0.2 21.2 1.8

Part B: Foreign direct investment in Canada

Total stock (Can$ millions) U.S. share (%) Mexican share (%) EC share (%)* Japanese share (%)

1980

1985

64,708 77.8 0.002 16.9 0.9

87,226 75.7 0.006 17.0 2.2

1989

1991

118,958 65.8 0.003 23.1 3.4

131,630 63.6 0.001 23.4 4.1

Part C: U.S. foreign direct investment abroad

Total stock (US$ millions) Canadian share (%) Mexican share (%) EC share (%)* Japanese share (%)

1980

1985

1989

1991

215,375 20.9 2.8 36.0 2.9

230,250 20.4 2.2 35.3 4.0

372,419 17.2 2.0 40.1 5.0

450,196 15.2 2.6 41.9 5.1

Part D: Foreign direct investment in the United States

Total stock (US$ millions) Canadian share (%) Mexican share (%) EC share (%)* Japanese share (%)

1980

1985

1989

1991

65,483 15.0 n.a. 57.8 6.4

184,615 9.3 0.3 58.0 10.5

368,924 8.2 0.1 57.6 18.2

414,358 9.0 0.2 54.0 22.4

*Includes Spain and Portugal after 1986, except Part (B), which does not include Portugal. SOURCES: (A) and (B) Statistics Canada (1992a, 67-72, 101-7); (C) OECD, International Direct Investment Statistics Yearbook, various editions; (D) U.S. Department of Commerce, Survey of Current Business, various issues

180 Multinationals and Intrafirm Trade Community, whereas Canadian firms held 21 per cent of their investments in the Community. Forty-two per cent of U.S. outward FDI was invested in the EC, whereas investors from the Community held 54 per cent of the U.S. inward FDI stock. As Table 4.3 shows, Canada-U.S. cross investments in dollar amounts have been rising over the 1985-91 period, but each country's percentage share in the other's FDI stock has continued to decline over the past ten years in spite of the 1978 GATT Tokyo Round, FTA, and NAFTA. These trends are also illustrated in Table 4.3. In terms of inward FDI into Canada, the U.S. share has fallen from 77.8 to 63.6 per cent over the 1980-91 period. Canada's share of inward FDI into the United States has similarly contracted from 15 to 9 per cent. In terms of outward FDI, only 15.2 per cent of the U.S. FDI stock in 1991 was invested in Canada, compared with 20.9 per cent in 1980. The trend for Canadian FDI in the United States is similar: from 69.5 per cent in 1980 to 57.9 per cent in 1991. There are several possible explanations for this downward trend in CanadaU.S. cross investments. 4 The most likely reasons are: The fall in tariffs over the 1980s, particularly since the FTA, together with unexploited economies of scale, have led to some rationalization of activities on a continental scale by existing U.S. and Canadian MNEs, and thus to plant closures and the replacement of FDI by exports from the remaining plants. The 1989-91 recession, together with intense competition from Asian imports and transplant production into the U.S. market, have forced U.S. MNEs to consolidate their domestic operations, close Canadian plants and serve the Canadian market from exports. Regional integration in the European Community has attracted both U.S. and Canadian FDI into Europe and out of their North American partners. FTA and NAFTA have attracted inward, export substituting investments by European and Asian firms, themselves recent entrants to the multinationals category, and this has diversified investment partners for Canada and the United States. The evidence we have presented on crossborder trade patterns is consistent with all of these hypotheses. If MNEs are closing some plants and substituting exports for local production, one would expect to see trade linkages increasing, which does appear to be happening. However, it is clearly too soon to tell, given the lags in data availability. The evidence from the 1992 Conference Board of Canada survey of MNEs in Canada, reviewed in Chapter 3, also provides some support for these hypotheses. The sectoral pattern of Canada-U.S. bilateral investment has also been

Multinationals and Intrafirm Trade in North America

181

changing. In 1985, 27 per cent of Canada's FDI stock in the United States was in the manufacturing sector; by 1990, this had risen to 34 per cent. U.S. FDI in Canada in manufacturing rose from 46 to 49 per cent of total FDI over the same period, while the U.S. share in petroleum sector fell from 16 to 22 per cent. Manufacturing is the largest single sector held by both countries; in both cases autos and auto parts is the largest industry in the sector.5 Looking at statistics on trade flows and foreign direct investment stocks gives us a macroeconomic snapshot of the hub-and-spoke relationship between Canada and the United States. In the next section, we move to the micro level and look at the individual firms that are engaged in these crossborder flows. The Largest Firms in North America This section helps answer the question 'who is doing what?' at the firm level in Canada and the United States. Tables 4.4 and 4.5 provide some information on the largest multinationals6 in the two countries. This data was gathered by Industry Canada as part of a survey on the top 1,000 multinationals in North America (Knubley et al. 1994).7 The top 20 multinationals in the United States in 1991, as ranked by sales, fall into seven sectors: resources, food, transportation, electrical/electronics, communications, retail trade, and finance. Total sales range from US$122 billion (General Motors) to US$25 billion (Amoco). Table 4.5 provides similar information on the top 20 multinationals in Canada. Note that several of these large firms are foreign controlled (e.g., all four firms in the transportation sector have U.S. parents). The firms can be grouped into five sectors: resources, food, transportation, communications, and banking and insurance. MNEs in Canada are more heavily represented in the financial sector than their U.S. counterparts. The largest MNE in Canada (BCE, i.e., Bell) has sales of US$17.2 billion, and the smallest (Manufacturers Life) has sales of US$5.8 billion. The Canadian MNEs are therefore much smaller than their U.S. counterparts. The top U.S. firms, on average, are twice the size of, but are less outward oriented8 than, the top Canadian firms; Mexican firms are smaller and less outward oriented, according to the Industry Canada study. This Canada-U.S. size difference is obvious from comparing the sales figures in the two tables. Knubley et al. (1994) then calculate national indexes of revealed comparative advantage,9 and investigate the productivity, growth, and R&D performance of these top firms. Lastly, data on the average size and outward orientation of the firms are provided. The authors find that Canadian MNEs have a revealed comparative advantage in resources and resource-intensive manufacturing and

182

Multinationals and Intrafirm Trade

TABLE 4.4 The Top 20 Multinationals in the United States, 1991 Company name

Industry

The resource sector multinationals Mining 102,847 Mining 56,042 Petroleum refining 38,695 Mining 37,271 Mining 36,461 Chemicals and allied products 29,362 Mining 25,325

Exxon Corporation Mobil Corporation Du Pont (E.I.) De Nemours and Company Texaco Inc. Chevron Corporation The Proctor and Gamble Company Amoco Corporation Philip Morris Companies, Inc. General Motors Corporation Ford Motor Company Chrysler Corporation The Boeing Company

Total sales (US$ millions)

Rank by assets

2 8

8 12

11 12 13

13 16 14

17 20

17 15

9

10

122,081

1

2

88,286

2

3

29,370 29,314

16 18

11 19

4 6

7 4

5

9

7 10 14

6 20 18

15

1

195

5

The food sector multinationals Tobacco 48,064

The transportation sector Motor vehicles and equipment Motor vehicles and equipment Motor vehicles and equipment Aircraft and parts

Rank by sales

multinationals

The electrical and electronics multinationals International Business Machines General Electric Company

Computer and office equipment Electrical products

64,792 59,379

The communications sector multinationals American Telephone and Telegraph

Communications

63,089

The retail sector multinationals Sears, Roebuck and Company Wal-Mart Stores, Inc. K Mart Corporation Citicorp American Express Company

Retail trade Retail trade Retail trade

57,242 43,886 34,580

The financial sector multinationals Depositary institution 31,839 Securities and brokers 25,763

SOURCE: Knubley et al. (1994, 165)

Multinationals and Intrafirm Trade in North America

183

TABLE 4.5 The Top 20 Multinationals in Canada, 1991 Company name

Industry

Sales (US$ millions)

Rank by sales

Rank by assets

Imperial Oil Ltd. Brascan Ltd. Alcan Aluminium Ltd. Noranda Inc. Ontario Hydro

The resource sector multinationals Mining 7,994 Mining 7,979 Primary metals 7,748 Lumber and wood 7,118 Utilities 6,179

11 12 13 16 19

11 16 12 10 6

George Weston Ltd. Loblaw Companies Ltd. Seagram Company Ltd.

The food sector multinationals Wholesale trade 9,316 Retail trade 7,381 Food and products 6,242

5 14 18

17 20 13

2

15

4 8

18 9

15

19

1

5

10

14

3

1

6

2

7

3

9

4

General Motors of Canada Ltd. Ford Motor Company of Canada Ltd. Canadian Pacific Ltd. Chrysler Canada Ltd.

BCE Inc. (Bell) Northern Telecom Ltd.

Royal Bank of Canada Canadian Imperial Bank of Commerce Bank of Montreal The Bank of Nova Scotia Sun Life Assurance Company of Canada Manufacturers Life Insurance Company

The transportation sector multinationals Motor vehicles and equipment 16,847 Motor vehicles and equipment 10,531 Transportation 8,711 Motor vehicles and equipment 7,157 The communications sector multinationals Communications 17,200 Communications equipment 8,182 The financial sector multinationals Depositary institution 12,414 Depositary institution 9,176 Depositary institution 8,861 Depositary institution 8,287 Insurance

6,856

17

7

Insurance

5,845

20

8

SOURCE: Knubley et al. (1994, 166)

184

Multinationals and Intrafirm Trade

financial services, U.S. MNEs in technology-intensive manufacturing and commercial services, and Mexican MNEs in resources and in the low-skill parts of resource-intensive and technology-intensive manufacturing. 10 This is reflected in tables 4.4 and 4.5 in terms of the larger concentration of Canadian multinationals in the resource and financial services areas. The authors conclude that increased economic integration within North America should cause the largest firms, and the three countries, to further increase their specialization along the lines of their revealed comparative advantage, generating pressures on Canadian firms to rationalize and restructure their operations, as we outlined in Chapter 3. How Large Is Intrafirm Trade in North America? Since the purpose of this book is to examine the regulation of transfer pricing by Canadian and U.S. tax authorities, the volume, industry distribution, and trends in intrafirm trade flows are important inputs to our study. If intrafirm trade is a small percentage of Canada-U.S. trade flows, then large deviations in transfer prices will have little impact on national treasuries or the bilateral balance of payments. However, where non-arm's length trade is large, small changes in pricing policies can have marked effects on tax revenues, industry returns, and the Canada-U.S. exchange rate. So the volume of intrafirm trade has important implications for the effects of transfer price manipulation - and transfer price regulation. As we have seen from the previous section, the Canadian and U.S. economies have been silently integrating through most of the twentieth century. With the ETA and NAFTA, we expect this trend to continue. Over two-thirds of Canada's exports and imports are tied to the United States; the equivalent figures for the United States are smaller (in the 20 per cent range) but rising. How large a percentage of this trade is conducted by multinationals and what per cent is between related affiliates? As we show in this section, the answer is: close to half of all Canada-U.S. trade in goods is intrafirm, and in some business services categories the figure is closer to 100 per cent. Our sources of statistics, however, are imperfect, the Canadian ones perhaps even poor. We have two basic sources at present. First, the U.S. government keeps statistics on activities of U.S. parents and their majority-owned foreign affiliates (MOFAs) and on U.S. affiliates with foreign parents - foreign affiliates (FAs), or foreign controlled corporations (FCCs) as they are called by the U.S. Department of Commerce (1993a,b). These data are compiled from the form 5471 and 5472 reporting requirements that the Internal Revenue Service imposes on all firms that engage in intrafirm trade.11

Multinationals and Intrafirm Trade in North America

185

Second, a few sources of information on intrafirm trade are available in Canada. The Canadian government, through Statistics Canada, is now releasing information on intrafirm trade in business services, based on the T106 reporting requirement form that all MNEs in Canada must similarly provide to Revenue Canada (Statistics Canada 1991). Two other recent studies use unpublished Statistics Canada data to develop estimates of Canadian intrafirm trade in goods (Covari and Wisner 1993, Mersereau 1992). We turn first to the U.S.-provided data, and then to the Canadian sources. Evidence from the United States MOFAs and Foreign Affiliates: How Many? In this section we examine U.S.-Canada trade and investment relationships as viewed through two sets of lenses: U.S. multinationals and their majorityowned foreign affiliates (MOFAs) in Canada and elsewhere, and foreign affiliates located in the United States with parents (defined as their 'ultimate beneficial owner') headquartered in Canada and elsewhere. The first set focuses on U.S.-owned subsidiaries located in Canada, the second on Canadian-owned subsidiaries located in the United States. Some general statistics on both sets are provided in Table 4.6. First, note that the last line in the table shows that the ratio of the share of U.S. FDI in Canada to Canadian FDI in the United States is 2.15; that is, Canada's share of the U.S. outward FDI stock in 1990 was 15.81 per cent, approximately twice the size of Canada's share of the U.S. inward FDI stock of 7.36 per cent. In 1990, U.S. multinationals had approximately 15,500 MOFAs around the world; about 1,800 or 12 per cent of these were in Canada. Since Canada represents about 16 per cent of the U.S. stock of outward FDI, this means that MOFAs in Canada on average either are somewhat larger and more capital intensive and/or had a higher proportion of equity to debt than U.S. MOFAs in other countries. U.S. FDI in Canada tends to be clustered in the upstream stages of capital-intensive natural resource and manufacturing sectors, so size and capital intensity explain this pattern. In 1990, there were about 10,000 affiliates of foreign MNEs (FCCs) operating in the United States; that is, about two-thirds as many as the number of U.S. MOFAs operating abroad. Canada's share of FCCs (at 11.5 per cent) was roughly equivalent to its share of MOFAs (11.7 per cent), but well above its share of U.S. inward FDI (7.4 per cent). Thus each Canadian affiliate on average either is smaller and less capital intensive and/or held a smaller proportion of equity to debt than the average foreign affiliate in the United States. To the extent that Canadian FDI in the United States tends to cluster in the downstream

186 Multinationals and Intrafirm Trade TABLE 4.6 U.S. Majority-Owned Foreign Affiliates and Foreign Affiliates in the United States, 1990 Host country in which MOFAs are located:

Number of MOFAs Country share of total MOFAs Country share of stock of direct investment abroad, historical cost basis, 1990

All Countries

Canada

15,532 100.00%

1,814 11.68%

113 0.73%

138 0.89%

6,831 43.98%

100.00%

15.81%

2.22%

4.95%

41.89%

Mexico

Japan

EC

Country of ultimate beneficial owner: All Countries

Number of foreign affiliates Country share of total foreign affiliates Number of companies consolidated as foreign affiliates Country share of total companies Average number of companies per affiliate Country share of direct investment stock in the U.S., historical cost basis, 1991 Ratio of share of MOFAs to share of foreign affiliates Ratio of share of U.S. direct investment abroad to inward FDI in the U.S.

Canada

Mexico

Japan

EC

10,142

1,162

166

2,142

3,747

100.00%

1 1 .46%

1.64%

21.12%

36.95%

31,388 100.00%

4,262 13.58%

256 0.82%

4,799 15.29%

14,012 44.64%

3.09

3.67

1.54

2.24

3.74

100.00%

7.36%

0.15%

21.26%

56.92%

1.00

1.02

0.44

0.04

1.19

1.00

2.15

14.86

0.23

0.74

SOURCE: Author's calculations based on U.S. Department of Commerce (1993a, Table A-l; 1993b, Table POSN-91; 1993c, Table 90-2; 1993d, Table POSN-90), as reported in the National Trade Data Bank - The Export Connection

Multinationals and Intrafirm Trade in North America

187

labour-intensive stages (marketing and distribution) and in the financial services industries, this is not surprising. Each Canadian affiliate also consists of a larger number of companies (3.67) than the average for all FCCs (3.09). MOFAs and Foreign Affiliates: Intrafirm Sales MOFA Sales. Table 4.7 looks at sales of goods and services by destination and by nature of the trade (with related parties, with unrelated parties), focusing on U.S. MOFAs located in Canada, Mexico, the EC, Japan, and All Countries.12 Although MOFAs in Canada represent only 11.7 per cent of all MOFAs, their sales are 14.9 per cent of all MOFA sales worldwide, and their share of U.S. sales is a huge 33.4 per cent of all MOFA-related U.S. sales. Thus these affiliates are important components of U.S. MNE networks. MOFAs located in Canada are primarily designed to sell within the Canadian market; local sales at US$129.7 billion are 73 per cent of all sales made by these affiliates. Given the fact that MOFAs located in Canada sell 23 per cent of their total output to the United States, whereas the world average is only 10.4 per cent, we can see how significant U.S. sales are for MOFAs in Canada. The intrafirm nature of these sales is clear from the table. Worldwide, 21 per cent of total sales made by MOFAs in Canada are to affiliated persons (i.e., intrafirm trade); this is below the worldwide average of 24 per cent. Four of every five dollars in sales (81.3 per cent) to the United States goes to U.S. parents. Within Canada, these firms sell through arm's length channels; only 1.69 per cent of local sales were interaffiliate. If we compare the intensity of intrafirm trade of U.S. MOFAs in Canada relative to the average for MOFAs worldwide, we see that the ratio for Canada to All Countries is well below one in terms of intrafirm sales that are worldwide (0.87), local (0.35), third country (0.31), and non-U.S. (i.e., third plus local, 0.14). In general, therefore, MOFAs in Canada are less intensively engaged in intrafirm trade than are other U.S. MOFAs around the world (see also Encarnation 1994). Foreign Affiliate Sales. In terms of Canadian affiliates in the United States, Table 4.8 provides data on the intrafirm nature of sales of services.13 These firms represent 11.5 per cent of all FCCs in the United States, and conduct a similar share of all U.S. sales made by FCCs. Foreign affiliates, in general, are heavily engaged in the sale of goods, rather than services; Canada much less than other FCCs (73.8 per cent of all sales are goods). Almost all the services sold by Canadian FCCs are sold within the United States; of sales outside the United States, about 61 per cent are intrafirm. This is considerably higher than the average for all FCCs in the United States (44.1 per

188

Multinationals and Intrafirm Trade

TABLE 4.7 Sales of Goods and Services by Majority-Owned Affiliates of U.S. Companies, 1990 (in US$ millions) Host country in which MOFAs are located:

Sales to all destinations, total Sales to all destinations, to affiliated persons Sales to all destinations, to unaffiliated persons Sales to parents and other affiliates as % of all sales Host country relative to All Countries Local sales, total Local sales, to other foreign affiliates Local sales, to unaffiliated foreigners Sales to other affiliates as % total local sales Host country relative to All Countries Sales to the United States, total Sales to the United States, to U.S. parents Sales to the United States, to unaffiliated U.S. persons Sales to U.S. parents as % of total sales to U.S. Host country relative to All Countries Sales to the other countries, total Sales to the other countries, to other foreign affiliates Sales to the other countries, to unaffiliated foreigners Sales to other affiliates in other countries as % of total sales to other countries Host country relative to All Countries

All Countries

Canada

1,191,832

177,200

19,330

62,117

615,192

286,829

36,907

6,483

9,507

154,066

905,003

140,294

12,847

52,610

461,126

24.07% 1.00

20.83% 0.87

33.54% 1.39

15.30% 0.64

25.04% 1.04

795,244

129,740

13,461

55,048

412,295

37,857

2,188

864

3,265

20,327

757,387

127,552

12,596

51,783

391,968

4.76% 1.00

1.69% 0.35

6.42% 1.35

5.93% 1.25

4.93% 1.04

123,801

41,404

5,066

3,280

22,129

98,574

33,673

4,985

3,171

17,045

25,227

7,732

81

109

5,084

79.62% 1.00

81.33% 1.02

98.40% 1.24

96.68% 1.21

77.03% 0.97

272,787

6,056

803

3,789

180,768

150,397

1,046

633

3,071

116,694

122,389

5,010

170

718

64,074

55.13% 1.00

17.27% 0.31

78.83% 1.43

81.05% 1.47

64.55% 1.17

Mexico

Japan

EC

Multinationals and Intrafirm Trade in North America

189

TABLE 4.7 (concluded) Host country in which MOFAs are located:

Total sales to foreign countries (i.e., local sales plus other sales), total Total sales to foreign countries, to other foreign affiliates Total sales to foreign countries, to unaffiliated foreigners Sales to other affiliates in foreign countries as % of total sales to foreign countries Host country relative to All Countries

All Countries

Canada

Mexico

1,068,030

135,796

14,264

58,837

593,063

188,254

3,234

1,498

6,335

137,021

879,776

132,562

12,766

52,501

456,042

17.63% 1.00

2.38% 0.14

10.50% 0.60

10.77% 0.61

23.10% 1.31

Japan

EC

SOURCE: Author's calculations based on U.S. Department of Commerce (1993c, Tables £0-20 and 90-40), as reported in the National Trade Data Bank - The Export Connection

cent). This suggests that the affiliates of Canadian MNEs are located in the United States in order to service the U.S. market. MOFAs and Foreign Affiliates: Merchandise Trade Tables 4.9, 4.10, and 4.11 provide data on U.S. merchandise trade conducted by multinationals. Table 4.9 focuses on merchandise trade conducted by U.S. MOFAs around the world, while Table 4.10 examines trade conducted by FCCs in the United States. Table 4.11 puts these data together to provide a picture of the share of U.S. merchandise trade that is intrafirm. MOFA Trade. According to Table 4.9, in 1990, MOFA exports exceeded MOFA imports by about US$13 billion. Over three-quarters of merchandise trade in both directions occurs between MOFAs and their U.S. parents. In the case of MOFAs in Japan, this rises to almost 100 per cent. Although only 11.7 per cent of all MOFAs are in Canada, 36.3 per cent of all U.S. merchandise exports to MOFAs worldwide are shipped to MOFAs in Canada (US$36.9 billion), and 45.2 per cent of U.S. imports from MOFAs worldwide come from Canadian affiliates (US$40 billion), for a net U.S. trade deficit with Canada of approximately US$3.2 billion, according to Table 4.8.

190 Multinationals and Intrafirm Trade TABLE 4.8 Sales of Goods and Services by Foreign Affiliates in the United States, 1990 (in US$ millions) Country of ultimate beneficial owner: All Countries

Canada

Mexico

1,168,490 971,169

127,097 93,839

Goods as % of total sales Home country relative to All Countries

83.11% 1.00

Sales of services, total Sales of services to U.S. persons U.S. sales as % of total sales of services Home country relative to All Countries

Japan

EC

2,851 2,704

313,138 285,412

494,936 424,802

73.83% 0.89

94.84% 1.14

91.15% 1.10

85.83% 1.03

164,479 155,221

27,590 26,588

117 112

17,449 15,684

61,933 59,266

94.37% 1.00

96.37% 1.02

95.73% 1.01

89.88% 0.95

95.69% 1.01

9,259 3,766 318 5,175

1,002 520 95 387

4 2 0 3

1,765 n.a. 69 n.a.

2,667 1,432 111 1,124

44.11% 1.00

61.38% 1.39

50.00% 1.13

n.a. n.a.

57.86% 1.31

Investment income

32,842

5,668

30

10,276

8,201

Investment income as % of services sales Home country relative to All Countries

19.97% 1.00

20.54% 1.03

25.64% 1.28

58.89% 2.95

13.24% 0.66

Total sales Sales of goods

Sales of services to foreign persons, total - to foreign parent group - to foreign affiliates - to other foreign persons Intrafirm (parent + affiliate) sales as % of total services sales to foreigners Home country relative to All Countries

SOURCE: Author's calculations based on U.S. Department of Commerce (1993a, Table E-13), as reported in the National Trade Data Bank - The Export Connection

Thus, a very high per cent of all U.S. MOFA-related trade takes place with MOFAs in Canada, reflecting the silent integration that has taken place between the two economies since the 1950s. Trade with the U.S. parent generates 83.0 per cent of total MOFA-related U.S. trade with Canada, in terms of both imports and exports; these percentages

Multinationals and Intrafirm Trade in North America

191

TABLE 4.9 U.S. Merchandise Trade with U.S. Majority-Owned Foreign Affiliates, 1990 Host country in which MOFAs are located: All Countries

Canada

Mexico

Japan

EC

U.S. export!3 to MOFAs (US$ mil lions) Exports shipped to affiliates, total - by U.S. parents - by unaffiliated U.S. persons

101,661 89,649 12,012

36,857 30,599 6,259

7,428 7,062 365

7,361 7,098 263

29,145 26,598 2,547

Exports by U.S. Parents as % of total Host country relative to All Countries

88.18% 1.00

83.02% 0.94

95.07% 1.08

96.43% 1.09

91.26% 1.03

U.S. imports from MOFAs (US$ millions) Imports shipped by affiliates, total - to U.S. parents - to U.S. unaffiliated persons Imports by U.S. parents as % of total Host country relative to All Countries

88,607 75,364 13,243

40,017 33,210 6,807

7,239 7,164 74

1,859 1,799 60

13,442 11,156 2,286

85.05% 1.00

82.99% 0.98

98.96% 1.16

96.77% 1.14

82.99% 0.98

SOURCE: Author's calculations based on U.S. Department of Commerce (1993c, Tables 90-20 and 90-59), as reported in the National Trade Data Bank - The Export Connection

are slightly below the worldwide MOFA intrafirm average (88.2 per cent for exports, 85.1 per cent imports). Thus MOFAs in Canada, on average, tend to engage in slightly less intrafirm trade than MOFAs in general (see also Encarnation 1994). These statistics are well below the intrafirm trade percentages recorded by U.S. MOFAs in Mexico and Japan (see also Encarnation 1994). Foreign Affiliate Trade. In terms of U.S. merchandise trade through U.S.based affiliates of foreign multinationals, Table 4.10 shows that FCCs run a substantial deficit in merchandise trade; with imports of US$180.7 billion, almost double the level of exports at US$91.1 billion. The import figure is more than double the imports recorded by U.S. MOFAs. Almost half of FA imports come from Japanese-controlled firms in the United States.14 Canadian affiliates export less to their parents as a per cent of all U.S. exports than do affiliates from other countries (18.4 per cent compared with an average of 41.5 per cent); the parent firms also export less to their affiliates (64.2 per

192

Multinationals and Intrafirm Trade

TABLE 4.10 U.S. Merchandise Trade of U.S Affiliates of Foreign Companies, 1990 Country of ultimate beneficial owner: All Countries

Canada

Mexico

Japan

EC

U.S. exports by foreign affiliates (US$ millions) Exports shipped by affiliates, total Exports shipped by affiliates to foreign parent group Exports shipped by affiliates to other foreigners, total - to other foreign affiliates - to non-affiliated foreigners

91,137

6,207

157

39,155

31,169

37,795

1,139

64

22,480

8,698

53,343 7,097 46,246

5,068 1,925 3,143

93 0 93

16,674 1,283 15,392

22,471 2,874 19,597

Exports to foreign parents as % of total

41.47%

18.35%

40.76%

57.41%

27.91%

1.00

0.44

0.98

1.38

0.67

49.26%

49.36%

40.76%

60.69%

37.13%

1.00

1.00

0.83

1.23

0.75

Foreign country relative to All Countries Exports to foreign parents and other affiliates as % of total Foreign country relative to All Countries

U.S. imports by foreign affiliates (US$ millions) Imports shipped to affiliates, total Imports shipped to affiliates by foreign parent group Imports shipped to affiliates by foreigners, total - by other foreign affiliates - by non-affiliated foreigners Imports from foreign parents as % of total Foreign country relative to All Countries Imports from foreign parents and other affiliates as % of total Foreign country relative to All Countries

180,674

10,541

811

87,712

49,443

136,672

6,763

588

73,180

31,613

44,002 4,384 39,618

3,778 712 3,066

223 0 223

14,532 728 13,803

17,830 2,148 15,682

75.65%

64.16%

72.50%

83.43%

63.94%

1.00

0.85

0.96

1.10

0.85

78.07%

70.91%

72.50%

84.26%

68.28%

1.00

0.91

0.93

1.08

0.87

SOURCE: Author's calculations based on U.S. Department of Commerce (1993a, Table G-2), as reported in the National Trade Data Bank - The Export Connection

Multinationals and Intrafirm Trade in North America

193

cent, compared with 76 per cent). When trade with other affiliates is added to trade with the Canadian parent, the Canadian figures (49.4 per cent for exports and 70.9 per cent for imports) are much closer to levels for FCCs from all countries. This suggests that U.S. affiliates of Canadian MNEs tend to export primarily to other affiliates of the MNE, rather than to their Canadian parent. This is a very different export pattern from other FCCs in the United States, which are much more heavily involved in exporting to their foreign parents than to other affiliates in the MNE group. U.S.-Canada Intrafirm Trade. One important question that we can partially answer using tables 4.9 and 4.10 is the question: How large is intrafirm trade as a per cent of total U.S. merchandise trade? We make a first attempt to measure the size of intrafirm trade in Table 4.11. By reorganizing the types of trade into two categories by affiliation (arm's length or non-arm's length) and comparing these data with total U.S. merchandise trade, we can provide a lower estimate on the size of intrafirm trade as a per cent of total trade. There are some problems with this type of analysis.15 First, countries are listed according to the Ultimate Beneficial Owner of the firm. Thus, a two-tiered affiliate (e.g., a Japanese MNE owns a Canadian affiliate, which owns a foreign subsidiary in the United States) would show up in the Bureau of Economic Analysis's FCC data set as a subsidiary of the top-tiered parent (in this case, as a Japanese-owned FCC). Second, the MOFA and FCC data are organized by industry, not by product, code. Thus intrafirm trade is organized in these data sets by the industry of the parent firm, and not by the types of products trade. As a result, of these differences, our estimates are likely to be biased and should therefore be treated cautiously. The top half of Table 4.11 refers to U.S. exports, the bottom half to U.S. imports. The table shows three important percentages: • the percentage of MOFA plus FCC trade (i.e., trade by domestic and foreignowned multinationals in the United States) that is intrafirm; • MOFA and FCC trade as a percentage of U.S. total trade; and • intrafirm MOFA plus FCC trade as a percentage of U.S. total trade (this last statistic is the product of the first two). The figures are surprisingly high. Seventy per cent of MNE merchandise exports from the United States to all countries are intrafirm, as are 80 per cent of MNE merchandise imports. For U.S. trade with Canada conducted by MNEs, 78 per cent of U.S. exports and 80 per cent of U.S. imports are intrafirm transactions. Therefore, multinationals are primarily moving goods within the MNE network.

194 Multinationals and Intrafirm Trade TABLE 4.11 U.S. Intrafirm Merchandise Trade, 1990 (in US$ millions) All Countries

Canada

Mexico

Japan

EC

U.S. intrafirm merchandise exports Exports shipped to MOFAs, total 101,661 36,857 7,428 Exports shipped by foreign affiliates, total 91,137 6,207 157

7,361

29,145

39,155

31,169

Total MOFA and foreign affiliate exports

192,798

43,064

7,585

46,516

60,314

Non-arm's length U.S. exports shipped abroad by: - U.S. parents to MOFAs abroad - foreign affiliates to foreign parents - foreign affiliates to other foreign affiliates

89,649 37,795

30,599 1,139

7,062 64

7,098 22,480

26,598 8,698

7,097

1,925

1,283

2,874

Total non-arm's length U.S. exports

134,541

33,663

7,126

30,861

38,170

Intrafirm exports as % MOFA and FCC exports This country relative to All Countries

69.78% 1.00

78.17% 1.12

93.95% 1.35

66.34% 0.95

63.29% 0.91

Total U.S. exports, customs value

374,537

78,218

27,468

46,138

93,087

MOFA and FCC exports as % total U.S. exports 51.48%

55.06%

27.61% 100.82%

64.79%

Intrafirm exports as % of total U.S. exports 35.92% This country relative to All Countries 1.00

43.04% 1,20

25.94% 0.72

66.89% 1.86

41.00% 1.14

U.S. intrafirm merchandise imports Imports shipped by MOFAs, total 88,607 40,017 7,239 Imports shipped to foreign affiliates, 180,674 10,541 811 total

1,859

13,442

87,712

49,443

Total MOFA and foreign affiliate imports

269,281

Non-arm's length U.S. goods, imported by: - foreign-based MOFAs to U.S. parents 75,364 - foreign affiliates from foreign parents 136,672 - foreign affiliates from other foreign affiliates 4,384 Total non-arm's length U.S. imports

216,420

50,558

8,050

89,571

62,885

33,210

7,164

1,799

1,156

6,763

588

73,180

31,613

712

0

728

2,148

40,685

7,752

75,707

34,917

Multinationals and Intrafirm Trade in North America

195

TABLE 4.11 (concluded) All Countries

Intrafirm imports as % MOFA and FCC imports This country relative to All Countries Total U.S. imports, customs value FOB MOFA and FCC imports as % of total U.S. imports Intrafirm imports as % of total U.S. imports FOB This country relative to All Countries

Canada

Mexico

Japan

EC

80.37% 1.00

80.47% 1.00

96.30% 1.20

84.52% 1.05

55.53% 0.69

491,332

91,198

29,506

89,612

87,540

54.81%

55.44%

27.28%

99.95%

71.84%

44.05% 1.00

44.61% 1.01

26.27% 0.60

84.48% 1.92

39.89% 0.91

SOURCE: Author's calculations based on U.S. Department of Commerce (1993a, Table G-2; 1993c, Table 90-59; 1993e), as reported in the National Trade Data Bank - The Export Connection

The second statistic shows the percentage of U.S. trade that is conducted by multinationals, either intrafirm or to unrelated firms. Worldwide, over half of U.S. exports and imports go through multinationals; the figures are slightly higher for trade with Canada.16 Multiplying these two figures together, Table 4.11 estimates that, in 1990, intrafirm trade represented 35.9 per cent of total U.S. merchandise exports and 44.1 per cent of U.S. imports with all countries. In terms of the composition of intrafirm trade, the percentage distribution for exports is as follows: U.S. parent exports to MOFAs (66.6 per cent), FCCs to their parents (28.1 per cent), and FCCs to other affiliates (5.3 per cent); and for imports: U.S. parent imports from their MOFAs (34.8 per cent), FCC imports from their foreign parents (63.2 per cent), and FCC imports from other affiliates (2 per cent). In terms of U.S. intrafirm trade worldwide, it therefore appears that trade flows are dominated by affiliate imports from their parents, whether these are U.S. parents exporting to MOFAs or FCCs importing from foreign parents. For U.S. merchandise trade with Canada, 43.0 per cent of exports and 44.6 per cent of imports were at non-arm's length in 1990. This means that 43 per cent of Canadian imports from the United States (since U.S. exports are Canadian imports) and 45 per cent of Canadian exports to the United States (since U.S. imports are Canadian exports) were traded within MNEs. The only country with a significantly higher percentage of intrafirm trade is Japan, with

196 Multinationals and Intrafirm Trade intrafirm trade representing 67 per cent of U.S. exports to, and 84 per cent of U.S. imports from, Japan. In terms of the composition of U.S.-Canada intrafirm trade, the distribution for exports is: U.S. parent-MOFA (90.0 per cent), FCC-foreign parent (3.4 per cent), and FCC-FCC (5.7 per cent); and for imports: U.S. parent-MOFA (81.6 per cent), FCC-foreign parent (16.6 per cent), and FCC-FCC (1.8 per cent). Therefore U.S. parent-MOFA trade dominates both U.S. intrafirm exports and imports with Canada by a wide margin. We hypothesize that this is because Canadian MNEs engage less heavily in intrafirm trade than do their U.S. counterparts. (Some additional evidence on this is presented in tables 4.12-4.14 below.) While total FCC-parent imports represent 63 per cent of U.S. intrafirm imports, FCC imports from their Canadian parents are only 17 per cent of all U.S. intrafirm imports from Canada. In summary, multinationals account for well over half of Canada-U.S. total merchandise trade; about 80 percent of the trade conducted by multinationals is intrafirm. Therefore slightly less than half of Canada-U.S. trade is intrafirm. As multinationals extend their presence into the other NAFTA countries, we expect these percentages to increase. We now turn to the Canadian evidence. Evidence from Canada Canadian sources for statistics on intrafirm trade are scanty, to say the least. In the 1970s, the federal government used to publish a voluntary survey of foreignowned subsidiaries in Canada, which were referred to as the FOSC surveys.11 Eight surveys were issued, covering the years 1966 through 1981. Their purpose was to provide information about, and to monitor the performance of, nonfinancial MOFAs in Canada with assets in excess of $5 million. In 1981, 274 respondents voluntarily completed the questionnaire on behalf of approximately 1,000 firms. The survey covered revenues, expenditures, international transactions, investments, and liabilities, broken down into several industry groups. Unfortunately the survey was discontinued. The last one, published in 1983, covered the 1980-81 period, which is too out of date for our purposes. In terms of more recent sources, we were able to find three: (1) a study by Mersereau (1992) of Canadian merchandise imports for 1978-86, (2) a study of merchandise imports by multinationals in the manufacturing sector in Canada in 1988 (Covari and Wisner 1993), and (3) a recent publication by Statistics Canada of business services trade in 1989 (Statistics Canada 1991). Canada: Intrafirm Merchandise Imports The Mersereau (1992) study builds a longitudinal panel of Canadian importers

Multinationals and Intrafirm Trade in North America

197

TABLE 4.12 Canadian Imports by Industry and Country of Control, 1986 Total foreign

U.S.-controlled Per cent of imports by value

Per cent of firms All industries All manufacturing Automotive manufacturing Other manufacturing Wholesale trade All other industries

69.3 80.8 98.0 61.8 57.1 38.1

10.9 14.4 25.9 14.0 10.0

6.1

Per cent of imports by value

Per cent of firms

53.2 71.1 94.5 45.4 19.2 32.5

7.4 10.5 21.3 10.1

5.6 4.3

SOURCE: Calculated from Mersereau (1992, 22) TABLE 4.13 Interaffiliate Imports by Industry and Country of Control - Canada, 1986

All industries All manufacturing Automotive manufacturing Wholesale trade All other industries

Canadiancontrolled firms

Foreigncontrolled firms

10.3 11.8

68.1 65.8 73.0 77.0 67.3

6.0 10.7

5.9

SOURCE: Mersereau (1992, 18)

for the 1978-86 period. The author estimates that his database covers 98 per cent of all Canadian merchandise imports (1992, 34). We focus only on the data for 1986, and in particular on the data for intrafirm trade. The heavy dominance of Canadian imports by foreign-controlled firms is clearly evident in Table 4.12. Although foreign-controlled firms make up only 10.9 per cent of all importing firms in Canada, they represent 69.3 per cent of imports by value. U.S. firms dominate this trade: at 7.4 per cent of all importers, they bring in 53.2 per cent of the imports. Thus Canadian-controlled firms, the bulk of the importers (89 per cent of the firms), do only 30.7 per cent of the importing by value. In terms of manufacturing, and in particular automotive manufacturing, the numbers are even higher. As Table 4.12 shows, 98 per cent of manufactured auto imports are brought in by foreign-controlled firms. Not only do foreign-controlled firms dominate Canadian merchandise imports, as Table 4.13 shows they are much more likely to engage in intrafirm

198 Multinationals and Intrafirm Trade trade. Non-arm's length imports represent only 10.3 per cent of imports by Canadian-controlled firms, but 68.1 per cent for foreign-controlled firms. The numbers for all manufacturing are roughly similar, while for wholesale trade, the intrafirm share for foreign firms rises to 77 per cent.18 In Table 4.11 we estimated that 43 per cent of Canadian imports from the United States were intrafirm in 1990. By combining the information from tables 4.12 and 4.13 we can calculate that 50.4 per cent of Canadian merchandise imports from all countries, not just the United States, were intrafirm imports in 1986; Mersereau (1992, 19) notes that this is a total of Can$47.4 billion out of Can$92.4 billion.19 A more recent study conducted for Investment Canada by Covari and Wisner (1993) focuses specifically on multinational enterprises in the Canadian manufacturing sector in 1988.20 For all manufacturing, Covari and Wisner find that imports represent 171.1 per cent of shipments made by Canadian-controlled firms, but only 34.3 per cent of shipments by foreign-controlled firms. The share of intrafirm imports in total imports also varies significantly with Canadian MNEs bringing in only 14.5 per cent of their imports through affiliate trade, whereas foreign-controlled firms bring in 63.3 per cent of their imports via their foreign parents or other affiliates. This reinforces the hub-and-spoke nature of the Canada-U.S. economic relationship, as we outlined above. See Table 4.14 below. As Table 4.14 shows, the numbers vary significantly between industries, with intrafirm trade shares tending to be highest for foreign firms in miscellaneous manufacturing and machinery, and for Canadian firms in machinery, primary metals, and fabricated metals. In all cases (except wood at 10 per cent for both), the Canadian intrafirm trade share is well below its foreign counterpart in the same industry. In summary, the evidence on merchandise trade, while scanty, suggests that (1) most imports are brought in by foreign-controlled firms, (2) Canadiancontrolled firms engage in very little intrafirm trade relative to their foreign counterparts, and (3) as a result, most Canadian intrafirm imports are by foreign MNEs. The next section shows that these facts are also true for Canadian trade in business services. Canada: Intrafirm Trade in Business Services Starting in 1989, Revenue Canada has required all firms in Canada to report their interaffiliate international transactions on form T106 (see Chapter 10 for more details). The T106 form is very detailed in terms of the breakdown of business services information it requires (e.g., imports and exports of items such as royalties, advertising expenses, and management fees), but little detail

Multinationals and Intrafirm Trade in North America

199

TABLE 4.14 Intrafirm Imports by Country of Control, Canada, 1988 Canadian-controlled MNEs

Manufacturing industry Food and beverages Rubber and plastics Textiles Wood Furniture and fixtures Paper and allied products Printing and publishing Primary metals Fabricated metal Machinery Transportation equipment Motor vehicles Other transport Electrical products Nonmetallic mineral products Petroleum and coal Chemicals Miscellaneous manufacturing Total manufacturing

Imports as per cent of total shipments

Intrafirm imports as per cent of total imports

Foreign-controlled MNEs Intrafirm Imports as imports as per cent of per cent of total shipments total imports

23.02 98.31 318.5 19.81 n.a. 7.51 244.25 394.47 71.5 427.58 335.04 365.85 196.11 111.71

7.7 2.29 10.62 9.5 11.47 8.3 2.37 23.27 20.77 51.06 9.68 8.1 13.47 19.5

17.68 64.39 93.25 27.11 n.a. 9.22 11.97 33.83 73.29 147.65 94.81 221.58 48.7 69.84

37.26 69.01 70.38 10.17 75.84 35.89 45.43 59.6 41.54 80.49 64.93 65.55 59.62 46.6

301.49 93.63 65.1

3.25 0.06 15.11

17.85 106.74 34.96

60.71 15.72 58.8

119.38 171.07

6.74 14.51

82.94 34.31

88.45 63.29

SOURCE: Covari and Wisner (1993, 59)

is required in regard to intrafirm trade in goods. The data on trade in business services are now available in an annual publication by Statistics Canada, Canada's International Transactions in Services, Catalogue 67-203.21 All figures reported below are in Canadian dollars. Business Services Trade by Region. Table 4.15 shows Canada's largest international transactions in business services with the United States for 1991.22 Canadian receipts from exports of business services to the United States totalled $5.25 billion, while imports of business services totalled $8.88 billion, for a net deficit of $3.6 billion. The largest deficit contributors in the Canada-U.S. business services category are royalties, patents, and trademarks ($1.4 billion);

TABLE 4.15 Canada-U.S. Trade in Business Services, 1991 Canadian receipts for service exports

Business services category (only major categories included) Research and development Royalties, patents, and trademarks Total intangibles (R&D, royalties, patents, & TMs) Consulting and other professional Transportation related Management and administrative Insurance and brokers Other financial Tooling and other auto Total trade with U.S. in all business services

Total receipts from U.S. (Can$ mill.)

U.S. share of total Canadian receipts (%)

Canadian payments for service imports Intrafirm share of receipts from U.S. (%)

Total payments to U.S. (Can$ mill.)

U.S. share of total Canadian payments (%)

Intrafirm share of payments to U.S. (%)

874

87.1

96.2

658

86.0

95.7

83

46.1

68.7

1,484

82.2

88.0

957

80.9

92.0

2,142

83.3

90.4

264 336

32.1 34.0

46.2 0.3

336 306

53.3 32.7

56.5 2.9

449 869 238 546

70.3 67.8 48.8 87.1

95.8 22.8 0.4 100.0

1,557 1,149 432 483

89.1 60.8 43.6 100

98.1 23.4 2.8 99.6

5,245

59.4

48.7

8,877

68.5

61.9

SOURCE: Author's calculations based on Statistics Canada (1993, tables 2 and 8)

Multinationals and Intrafirm Trade in North America

201

management fees ($1.1 billion); and insurance ($0.3 billion). Canada had small surpluses in the R&D, transport-related services and automotive retooling categories.23 Almost 60 per cent of Canadian receipts, and 69 per cent of Canadian payments, for business services, came from trade with the United States. In terms of Canadian receipts, the U.S. share is highest in tooling (87%) and R&D (87%) and lowest in consulting (32%) and transportation services (34%). In terms of Canadian payments, the U.S. share is highest in tooling (100%) management fees (89%) and R%D (86%), and lowest in transportation services (33%). Thus the U.S. share tends to be highest in the R&D, management fees, and automotive tooling categories. In terms of royalties, however, Canada earns only 46 per cent of receipts in the form of royalties, patents, and trademarks from trade with the United States, whereas 82 per cent of its payments in this category go to the United States. The Canadian figures also show us what per cent of business services trade with the United States was conducted within the MNE family as intrafirm trade, as compared with arm's length trade. For all business services trade with the United States, 49 per cent of Canadian receipts and 62 per cent of Canadian payments in 1991 were intrafirm trade flows.24 The shares vary widely by category, ranging from almost zero to 100 per cent. However, the numbers have a pattern. Trade in automotive tooling (both exports and imports of business services) is effectively all intrafirm. Similarly, management fees and R&D payments are almost 100 per cent conducted with affiliated firms. On the other hand, transport-related and other financial services tend to be almost 100 per cent arm's length crossborder transactions. If we consider R&D and royalties, patents, and trademarks as returns for intangibles, and the other business service categories as services, Table 4.15 shows us that the United States represented 81 per cent of all Canadian receipts, and 82 per cent of all Canadian payments, for intangibles in 1991. Over 90 per cent of U.S. trade with Canada in intangibles is intrafirm transactions. Since Canada-U.S. trade in intangibles is almost wholly intrafirm and payments far outweigh receipts, the picture that emerges is a large net outflow to U.S. multinationals as payments for technology transfers. Business Services Trade by Country of Control. Table 4.15 focuses on the direction of trade, but tells us nothing about who controls the corporations that determine Canada's business services transactions. Table 4.16 provides some information by looking at U.S.-controlled firms in Canada and their receipts and payments; Table 4.17 does the same for Canadian-controlled firms. Here the question is country of control or which country owns the firm in Canada that is doing the trading rather than region of the world with which Canada is trading.

202

Multinationals and Intrafirm Trade

Thus the information in these two tables is the share of total Canadian trade with the world that is conducted through firms of different nationalities. As Table 4.16 shows, U.S.-controlled firms in Canada received 23 per cent of all Canadian business services receipts, but made 45 per cent of all payments, in 1991. U.S.-controlled firms consistently run a deficit, with payments exceeding receipts, in all categories except transport and automotive tooling. The deficit is particularly large in royalties, patents, and trademarks ($1.3 billion) and management and administrative fees ($1.3 billion), which is not surprising given that such payments normally flow from subsidiaries to their parent firms and thus would be expected to flow outside of Canada. The percentage of Canadian services trade conducted inside U.S.-controlled corporations is very high: 71 per cent of their receipts and 81 per cent of their payments were non-arm's length transactions.25 The percentages are close to 100 per cent in almost all categories in Table 4.16, with the notable exception of transport (2%) and insurance related services (27%). In terms of intangibles, U.S.-controlled firms are responsible for 28 per cent of all receipts and 76 per cent of payments, creating a large deficit of $1.5 billion on Canadian business services trade. Over 90 per cent of the trade in intangibles made by U.S.-controlled firms, in both directions, is conducted inside the MNE. Clearly, U.S. subsidiaries are remitting payments for intangibles to their U.S. parents. Table 4.17 paints a different picture. It focuses on Canadian-controlled firms and their contribution to Canada's business services trade. About two-thirds of all Canadian receipts are received by Canadian firms, whereas they are responsible for only 43 per cent of payments. Thus, Canadian-controlled firms tend to run a surplus, with receipts exceeding payments. This is true for all categories except royalties, insurance, and other financial services. The percentage of such trade that is in-house also tends to be much lower than for U.S.-controlled corporations. Only 25 per cent of Canadian firms' receipts, and 17 per cent of their payments, were conducted within the MNE; the remainder were arm's length transactions.26 Thus the question of transfer price manipulation of business receipts and payments, through under- or overinvoicing or timing, is primarily an issue - for both Revenue Canada and the Internal Revenue Service - of U.S.-controlled, rather than Canadian-controlled, MNEs. If we separate trade in intangibles from other business services trade, Canadian firms were responsible for 56 per cent of all receipts, but only 11 per cent of all payments. The share that was intrafirm is high for receipts (89%), but much lower for payments (61%). This is not surprising, since receipts for intangibles normally flow from subsidiaries to parent firms (i.e., inward to the Canadiancontrolled firms).

TABLE 4.16 Canada's Trade in Business Services, Conducted by U.S.-Controlled Firms, 1991 Canadian receipts from business service exports

Canadian payments for business service imports

U.S.-controlled Intrafirm share U.S.-controlled Intrafirm share Business services Receipts to firms' share of of receipts by Payments by firms' share of of payments by category (only major U.S.-controlled total Canadian U.S.-controlled U.S.-controlled total Canadian U.S.-controlled categories included) firms$ mill.)receipts (%)firms(%)firms($ mill.)payments (%) firms (%) Research and development Royalties, patents, and trademarks Total intangibles (R&D, royalties, patents & TMs) Consulting and other professional Transportation related Management and administrative Insurance and brokers Other financial Tooling and other auto Total trade in all business services

330

32.9

98.5

505

66.0

98.6

51

28.3

62.7

1,369

75.8

92.8

381

32.4

93.2

1,874

72.9

93.7

106 47

12.9

87.7

2.1

189 36

30.0

4.8

78.8 13.9

115 348 1 568

18.0 27.1

1,392

822 76 483

79.7 43.5

90.6

100.0 27.0 100.0 99.1

100.0

99.0 32.1 100.0 99.6

1,992

22.6

71.4

5,843

45.1

80.6

0.2

SOURCE: Author's calculations based on Statistics Canada (1993, table 9)

3.8

7.7

TABLE 4.17 Canada's Trade in Business Services, Conducted by Canadian-Controlled Firms, 1991

Business services category (only major Categories included) Research and development Royalties, patents, and trademarks Total intangibles (R&D, royalties, patents, & TMs) Consulting and other professional Transportation related Management and administrative Insurance and brokers Other financial Tooling and other auto Total trade in all business services

Canadian receipts from business service exports

Canadian payments for business service imports

CanadianReceipts to controlled Canadianfirms' share of controlled firms total Canadian

Payments by Canadiancontrolled firms

($mill.)

Intrafirm share of receipts by Canadiancontrolled

($mill.)

CanadianIntrafirm share controlled of payments by firm's share of Canadiantotal Canadian controlled firms (%) payments (%)

receipts (%)

firms (%)

562

55.9

94.0

170

22.2

85.3

105

58.3

62.9

110

6.1

22.7

667

56.3

89. 1

280

70.9

60.7

606 940

73.6 95.2

13.5

317 897

50.3 95.7

12.9

0.1

424 768 487 0

66.4 59.9 99.8

95.0 27.1

11.1 43.6 92.3

79.4 29.1

0

n.a.

194 824 914 0

0

n.a.

5,936

67.2

24.6

5,619

43.4

17.4

1.4

SOURCE: Author's calculations based on Statistics Canada (1993, table 9)

0

3.1

Multinationals and Intrafirm Trade in North America 205 Summary: Canada's Trade in Business Services These numbers on Canadian business services trade reinforce our theory about the hub-and-spoke nature of North American trade and investment. Canada runs a deficit on its business services trade with the United States. The deficit is particularly large in the intangibles categories (R&D, royalties, patents, and trademarks) since the majority of MNEs in the region are headquartered in the United States and have subsidiaries in Canada. These U.S-controlled subsidiaries are heavily engaged in intrafirm business service transactions with their MNE parents and affiliates. Canadian-controlled firms, on the other hand, are less heavily engaged in intrafirm trade, and tend to generate trade surpluses in business services. How Big? A Summary In this chapter we have drawn together estimates of trade in goods and services by multinationals in North America, both trade with their affiliates and with unrelated parties, in order to provide some estimate of the size of intrafirm trade flows. Looking first at U.S. merchandise trade made by U.S. MNEs with majorityowned foreign affiliates, we found that U.S. exports to MOFAs worldwide totalled US$101.7 billion, while imports were US$88.6 billion in 1990 (see Table 4.9). U.S. exports to MOFAs in Canada were 36 per cent of the total (of which 83 per cent were by U.S. parents); U.S. imports from MOFAs in Canada were 45 per cent of the total (of which 83 per cent were by U.S. parents). Thus Canada's share of U.S. MOFA trade far exceeds its share of U.S. MOFAs (of 11.7 percent). Looking at U.S. merchandise trade in terms of U.S. affiliates with foreign parents, we then found that total U.S. exports shipped by foreign affiliates were US$91.1 billion compared with U.S. imports of US$180.7 billion (see Table 4.10). Of this total, U.S. affiliates with Canadian parents accounted for 6.8 per cent of total exports (49 per cent of which was intrafirm) and 5.8 per cent of total U.S. imports, made by foreign affiliates (71 per cent of which was intrafirm), even though they represent a larger 11 per cent of all FCCs in the United States. In terms of Canadian trade with the United States in business services, we found that Canada ran a deficit of Can$ 5.2 billion in 1991 (see Table 4.15). About 50 per cent of Canadian receipts, and 62 per cent of Canadian payments, were intrafirm transactions with the United States. U.S.-controlled corporations received Can$ 2.0 billion for exports, of which 71 per cent was intrafirm, and paid Can$ 5.8 billion for imports, of which 81

206 Multinationals and Intrafirm Trade per cent was intrafirm (Table 4.16). Canadian-controlled firms received Can$5.9 billion from exports, and paid Can$ 5.6 billion for imports, of business services in 1991 (Table 4.17). Twenty-five per cent of the receipts and 17 per cent of payments by Canadian-controlled corporations were between affiliated firms. Thus U.S.-controlled corporations in Canada conduct a higher percentage of trade in business services as interaffiliate trade than do Canadian controlled corporations. We can put these statistics on goods and services trade together, but must be cautious since they are drawn from different sources and are not for the same year. In 1990, 43 per cent of Canadian merchandise imports from the United States and 45 per cent of Canadian goods exports to the United States were traded among related firms. In 1991, 49 per cent of all Canadian receipts for business services exports to the United States were intrafirm, as were 62 per cent of all payments for business services imports from the United States. Thus intrafirm trade represents about 45 per cent of Canada-U.S. merchandise trade, 50 per cent of Canadian business service exports to the United States, and 60 per cent of Canadian business services imports from the United States. In sum, about half of all trade in goods, services, and intangibles that flows between Canada and the United States is non-arm's length trade. Clearly, multinationals are important to Canada-U.S. trade patterns. And, in particular, U.S. multinationals are critically important to Canada-U.S. intrafirm trade. As Table 4.9 shows, MOFAs in Canada are heavily engaged in intrafirm merchandise trade: about four-fifths of U.S. exports to MOFAs in Canada, and four-fifths of U.S. imports from MOFAs in Canada are U.S. parent-MOFA trade flows. Table 4.16 shows that intrafirm trade in business services is at a similar level: 70 per cent of Canadian business service exports to the United States made by U.S.-controlled firms are intrafirm, as are 81 per cent of their imports from the United States. Thus Canada-U.S. intrafirm trade is primarily a story of U.S. multinationals and their foreign affiliates in Canada. This has obvious ramifications for tax authorities on both sides of the border, to which we return in later chapters. Conclusions This chapter had two purposes. First, we showed the economic linkages between Canada and the United States, and the importance of multinationals in cementing and deepening those linkages. We found that Canada's economic relationship with the United States is its premier relationship, and is so by a large margin. Canada, on the other hand, is less important to the United States. This fact of life is captured by our hub-and-spoke metaphor: the United States is

Multinationals and Intrafirm Trade in North America 207 the hub around which and through which Canada (and Mexico) engage in most international economic transactions. It is also clear from the statistics that U.S. multinationals dominate the trade and investment components of the hub-andspoke relationship. Second, we explored the intrafirm nature of this trade. We gathered together the currently available statistics on intrafirm trade generated by multinationals in Canada and the United States. It is clear from the statistics that U.S., not Canadian or third-country, multinationals dominate trade and investment patterns in both countries, and that U.S.-controlled intrafirm trade makes up a huge percentage of trade in both goods and business services (and particularly in trade in intangibles). Given that most trade is conducted at non-arm's length, this means that the prices for these flows are transfer prices, set within the multinational enterprise for reasons that may have to do with internal motivations of managers and/or external reasons such as reducing the total tax and tariff burden on the enterprise. Changes in the transfer pricing policies of all MNEs, but in particular U.S. multinationals, in response to changes in either internal or external conditions, therefore, can have an enormous impact on the CanadaU.S. balance of payments, the Canada-U.S. exchange rate, gross domestic products in both countries, and the tax revenues received by the Internal Revenue Service and Revenue Canada. It is therefore critical to have an understanding of how these large corporate giants are likely to respond to tax and tariff barriers. How much tax do these firms pay? What evidence is there that they have used transfer price manipulation to reduce their tax bills? We turn to these questions in chapters 5, 6, and 7. Chapter 5 develops a theoretical model to explain how profit-maximizing MNEs should set prices for intrafirm trade in goods, services, and intangibles. Chapter 6 looks at taxing multinationals in theory; it extends the models in Chapter 5 to encompass the impacts of taxes and trade barriers on these pricing choices. Chapter 7 reviews the empirical work that has been done on transfer price manipulation, investigates the recent claim that foreign affiliates are underpaying their U.S. corporate income taxes, and provides new evidence on taxes and transfer pricing for multinationals in North America. We therefore turn to Part III of Taxing Multinationals, Transfer Pricing and Taxation.'

This page intentionally left blank

P A R T III: T R A N S F E R P R I C I N G A N D T A X A T I O N

This page intentionally left blank

5 The Simple Analytics of Transfer Pricing

Introduction In order to explain the theory behind transfer pricing regulation and to use the theory to critique existing legislation, it is important to understand how economists believe MNEs behave, and in what ways their behaviour differs from that of purely domestic firms. There is a well-developed economics literature on transfer pricing which explains why multinationals choose certain transfer prices, how these prices are affected by government regulatory barriers such as tariffs and income taxes, and the impacts of these regulations on MNE output, sales, and intrafirm trade levels. The microeconomic literature on transfer pricing is reasonably small, technical, and primarily concerned with the efficiency of transfer pricing in response to exogenous constraints such as taxes and tariffs. See, for example, Batra and Hadar (1979), Besanko and Sibley (1991), Bond (1980), Chalos and Haka (1990), Copithorne (1971, 1976), Das (1983), Diewert (1985), Eden (1976, 1978, 1983, 1985), Ghosh and Grain (1993), Gould (1964), Grace and Berg (1990), Halperin and Srinidhi (1987, 1991), Harris et al. (1993), Hines (1990), Hines and Rice (1994), Hirshleifer (1956, 1957), Horst (1971, 1973, 1977), Itagaki (1979, 1981, 1982, 1985), Kant (1988a,b, 1990), Katrak (1977, 1980, 1981), Prusa (1990), Samuelson (1982, 1985, 1986), and Stewart (1986). Several readings are drawn together and/or reviewed in Dunning (1993, 51224), Plasschaert (1993), and Rugman and Eden (1985). Less technical pieces with microeconomic foundations are Berry et al. (1992), Chandler and Plotkin (1993), Higinbotham et al. (1987), and Witte and Chipty (1990). In most of this literature, the MNE is modelled as an integrated business, as we have argued in Chapter 3. The multinational operates and controls divisions or firms with production plants in several countries and sells in several markets.

212 Transfer Pricing and Taxation The firms are under common control, have common goals, and can draw from a common pool of resources. The goal of the enterprise is to maximize profits net of taxes and other regulatory barriers for the enterprise as a whole, not for each affiliate. Achieving this goal has implications for the MNE's allocation and pricing decisions. Integration is therefore key to understanding intrafirm trade and transfer pricing. In this book we want to develop a simple and general theory of transfer pricing behaviour by multinational enterprises that builds upon the microeconomic literature on transfer pricing. Such a model should take into account that: (1) MNEs engage in intrafirm trade simultaneously in tangibles, intangibles, and services; (2) intrafirm trade flows go in both a vertical direction (i.e., to firms either upstream or downstream from this firm) and a horizontal direction (i.e., trade in similar or identical products between firms at the same stage of production); and (3) MNEs are exposed to a wide variety of government interventions such as taxes, tariffs, foreign exchange restrictions, and so on. How can we model the integrated business, its intrafirm trade flows, and transfer pricing decisions when the enterprise is subject to regulatory barriers? Answering the question 'how to model the integrated business?' is the purpose of this chapter and the next chapter. In this chapter, we develop a simple microeconomic model of transfer pricing by a two-division multinational enterprise, assuming there are no trade or tax barriers. We examine how the MNE, as an integrated business, chooses its output, sales, and trade levels, and sets its transfer pricing policies. While MNEs do trade raw materials, intermediate components, and finished goods among units of the integrated business, two very large sources of intrafirm trade in the 1990s are in intangibles and business support services. We examine all three categories - goods, services, and intangibles - and divide our analysis into the two cases of horizontal and vertical integration. In each case, we compare the MNE's choice of an efficient transfer price with the transfer pricing rules laid down by the OECD in its various reports, including the new draft guidelines being issued over the 1994-6 period. We also examine three problems associated with the OECD pricing methods for tangibles: the continuum price problem, location savings, and affiliate losses. In the next chapter, we turn our attention to taxation and MNEs, examining the theory of how the firm's transfer pricing decisions can change in response to various government policies such as tariffs and corporate income taxes. Because they are less closely tied to our main subject, we omit the following topics from our survey in this chapter and the next: behavioural aspects of transfer pricing, incentive-compatible transfer pricing regulatory schemes, foreign exchange risk, demand or supply uncertainty and transfer pricing, revenue-maximizing taxes

The Simple Analytics of Transfer Pricing

213

and tariffs, bilateral bargaining between divisionalized firms over transfer prices, endogenous transfer pricing, transfer pricing where there are three or more divisions of the MNE involved, and impacts on world welfare levels.1 The materials in these two chapters consist primarily of economic models of transfer pricing, together with a discussion of the parts of the OECD transfer pricing guidelines that pair with the various cases for tangibles, intangibles, and intragroup services. In each case, the analysis is presented in three ways (mathematically, graphically, and in text) to make it easier for the reader to follow the analysis.2 A glossary of economic terms can be found at the end of this book for the benefit of readers who may be unfamiliar with microeconomic terminology. Transfer Pricing in a Horizontally Integrated Multinational Horizontal integration occurs when two affiliates of a multinational enterprise produce the same product or product line at the same stage in the value chain. Horizontal integration generally occurs when the MNE sets up foreign plants in order to earn rents on the MNE's intangible assets (e.g., patents on technology) through selling in foreign markets.3 For example, let us take the case of Xerox, a U.S. multinational specializing in photocopying machines, which has set up subsidiaries around the world to produce and sell photocopy machines that use the parent's technology. Initially, these affiliates were miniature replicas of their parent firm, making similar product lines and concentrating all sales within their local market with little exchange of product lines between national markets. More recently, in Xerox as in other MNEs, regional integration and lean production techniques are changing these horizontally integrated affiliates. The MNE is becoming a complex, integrated network with substantial intrafirm horizontal trade moving within its affiliates (UNCTAD 1993, ch. V). Horizontal intrafirm trade will usually take place once the foreign affiliates are established inside each foreign market, for two reasons. First, even though two or more affiliates may be producing the same product, the cost and revenue structures of the affiliates will not be the same. This provides the MNE with arbitraging possibilities that tend to generate intrafirm trade. For example, it may be cheaper to assemble a large number of copying machines in Xerox's Spanish affiliate and supply some fraction of these to Xerox's French affiliate rather than to have all the French requirements met through local production. Or, affiliates of Xerox may agree to supply one another when shortages develop. Second, the affiliates are likely to specialize by product line (i.e., in different sizes and qualities of copying machines) and then trade these lines, giving each affiliate a broader array of products for sale than it produces on its

214

Transfer Pricing and Taxation

own. As a result, the products going between Xerox's affiliates are not identical; the trade is at the same level of the value chain, and therefore horizontal, but it is not in identical goods. In the simple models below we assume the MNE has one affiliate which produces the same product as its parent, so we are focusing on the first reason for intrafirm trade. The Multiplant Multimarket Monopoly Microeconomic theorists, starting with a seminal article on transfer pricing by Thomas Horst (1971), have generally modelled the horizontally integrated MNE as a multiplant, multimarket monopoly. This means that the MNE produces the same product (usually modelled as a physical good like a car or refrigerator) in more than one plant and sells that product in more than one market. The MNE is assumed to have monopoly power in both markets (i.e., the firm faces downward-sloping demand curves) and the ability to price discriminate between the markets. Transport costs are generally ignored. Assume initially that we have a multinational enterprise consisting of two divisions or firms: firm 1 (the parent) and firm 2 (its wholly owned subsidiary). The profit function for firm i (i = 1.2) is:

where TC^ represents the total profit of firm i, Y; represents the volume of local sales by firm i, Pt is the consumer price of Ys, Rj(Y;) is total local sales revenue of firm i (where P4 times Y, equals Rj), Q; is the output of plant i and C^Qj) is the total cost of producing Qj in plant i. The economic or pure profit of a firm - what we are calling 71 - is an economic concept; it is defined as the excess of actual profit over normal profit. Actual profit is the difference between all revenues and all expenditures in one period. Normal profit is an opportunity cost concept; it represents the minimum amount necessary to compensate the entrepreneur for his or her time and risk and to keep him or her from exiting the industry. Thus pure or abnormal profit represents additional returns or rents left over after all factors of production, including the owner of the business, have been compensated for their opportunity costs. In a perfectly competitive economy, pure profits would be competed away by free entry. However, where firms have market power in either product or factor markets, pure profits can persist over time. Since we assume our multinational enterprise has market power in its product markets (but not in its primary factor markets for labour and capital), the enterprise should earn rents in excess of

The Simple Analytics of Transfer Pricing

215

opportunity costs, and in the absence of entry these rents will persist over time. (This, of course, is one of the long-standing complaints about MNEs in host countries: that they earn excess profits and repatriate them to the home country.) Equation (1) says that the pure profit of division i equals its sales revenue minus its total costs of production. Implicit behind this equation is that each firm is restricted to selling only what it produces; it can neither engage in intrafirm trade with its sister affiliate, nor buy nor sell on an external market. Therefore Qj = Yj. Assume initially that each firm maximizes its own profit function. Rewriting equation (1), adding in the Lagrangian constraint that all output must be sold, we have each division's objective function (the function to be maximized):

Differentiating (2) with respect to Qj and Y4 and setting the results equal to zero, we have:4

Putting these two conditions together and rearranging:

That is, each firm should maximize profits by setting its marginal revenue from local sales equal to its marginal production costs. This is the standard condition for a profit-maximizing firm. Note that there is no reason for A,, to equal X2, however, so that marginal revenues are not equated between the markets, nor are marginal costs equated across the plants. The firms behave solely as if they were independent, nontrading units. The contribution of economists like Horst (1971) was to show that it is almost always more profitable for the MNE to maximize its joint profits over its divisions, and that such joint maximization will normally lead to intrafirm trade flows. The MNE should be seen as a multiplant enterprise (since it has several production sites), a multimarket enterprise (since it sells in several markets that are usually differentiated from one another), and an enterprise with some market or monopoly power (since it sells a somewhat unique product and therefore faces downward-sloping demand curves in its markets). The MNE is a multiplant enterprise because it has more than one production

216

Transfer Pricing and Taxation

location. The multiplant monopoly shows that it is more profitable for one firm which owns two plants to allocate output between the two plants such that the marginal cost of production (including transport costs to market) is the same in the two plants. Total profit can be increased by shifting output from the highcost plant to the lower-cost plant. As output falls in the high-cost plant, marginal cost (MC) falls; as output rises in the low-cost plant, MC rises, until at the equilibrium the marginal costs are equalized.5 Thus, a multiplant monopoly should allocate output between two plants such that MCj equals MC2 for any two plants 1 and 2 in which the MNE makes the same product. The MNE can also be considered as a multimarket monopoly because it sells or distributes its output in at least two countries where the MNE has market power. It will always be more profitable for the enterprise to allocate sales to the higher-revenue market because the return on sales is greater. As sales are shifted out of the low-revenue market, marginal revenue rises; as sales are shifted into the high-revenue market, MR falls. At the equilibrium the two marginal revenues are equalized. Thus, a multimarket monopoly should allocate its sales so as to equate MRj to MR2 for any two markets 1 and 2 in which the MNE sells the same product and where, for various reasons, the MNE can price discriminate between them.6 In general, a multiplant multimarket monopoly will engage in intrafirm trade, with one division being the selling or export division and the other the buying or importing division. In both this chapter and the next one, for simplicity, we assume that firm 1 is the exporter and firm 2 the importer. Let X be the volume of intrafirm trade and p be the transfer price (the internal price set by the MNE) per unit of X. As a multiplant multimarket monopoly, the multinational enterprise's joint profit function looks like:

where the first square bracket represents the total proms or the parent firm and the second square bracket is the total profits of its foreign subsidiary firm 2. Since firm 1 is the exporter, X must equal Qj — Yj (since exports must equal production minus local sales) and also equal Y2 - Q2 (since imports equal local sales minus local production).7 With these constraints taken into account, the objective function to be maximized is:

The Simple Analytics of Transfer Pricing

217

Notice that the pX terms will simply cancel out in (7); thus neither the transfer price nor the volume of trade appear explicitly in the MNE's total profit function. We differentiate the MNE's total profit function with respect to Y4, Qi5 and X to determine the optimal amounts of output, sales, and intrafirm trade. This process leads to the following three profit-maximizing first-order conditions. Differentiating (7) with respect to Qj and Y,:

Differentiating (7) with respect to Q2 and Y2:

Differentiating (7) with respect to X:

Putting these three first-order conditions together we have:

Equation (11) can be read as four separate statements: (1) the MNE, as a multimarket monopoly, should equate marginal revenues from local sales in the two markets; (2) the MNE, as a multiplant monopoly, should equate marginal costs of production in the two plants; (3) the MNE, as a profit-maximizing enterprise, should equate marginal revenue with marginal cost in each firm; and (4) the MNE, for efficient resource allocation within the enterprise, should set each of these terms equal to the shadow price X where X = A,j = -X2. That is, the transfer price should equal the marginal cost of the exporting firm (Horst 1971, 1973). As equation (11) shows, the first-order conditions for a global profit maximum will normally imply intrafirm transfers between the two divisions of the MNE (that is, X is positive). Production will be shifted towards the low-cost plant and sales shifted towards the high-revenue market, and there is no reason for these two locations to be the same. The exporting division will be the relatively low-cost, high-revenue firm in the no-trade (autarky) situation, which we have assumed (for simplicity) to be firm 1. This is illustrated in Figure 5.1. Figure 5.1 shows the marginal cost and revenue curves for the parent firm in the left-hand graph and those for the foreign affiliate in the right-hand graph. The demand curve Di and the market price P> corresponding to each marginal revenue curve are also shown. If the two firms were unrelated and did not

218

Transfer Pricing and Taxation

FIGURE 5.1 Transfer Pricing in a Horizontally Integrated Multinational Enterprise

engage in trade, as equation (5) shows, each firm would set MRj = MC; (at point a as shown in the figure), sell all of its own production (so Y; = Q;), and set its market price Pt corresponding to its volume of sales (point b). Now assume the two firms are part of the same MNE, the goal of the integrated business is to maximize its global profits, and the firms are allowed to engage in intrafirm trade. The middle graph in Figure 5.1 shows the internal market of the MNE, the market where the product is traded between the two divisions. The selling or exporting division (firm 1) has a marginal cost of exports curve MCX which slopes upward, reflecting the additional costs of providing X to firm 2.8 The buying or importing division (firm 2) has a marginal revenue from imports curve MRX which slopes down, reflecting the diminshed revenues from an additional unit of imports.9 If the two divisions trade they will set MRX = MCX (point v), trading volume X at transfer price p. Tracking back horizontally into the left-hand and right-hand graphs,10 we can determine the new volumes of production (as determined by point d) and domestic sales (point c) and the new market prices (point e) for each division.

The Simple Analytics of Transfer Pricing 219 We can easily prove that intrafirm trade is more profitable than no trade. Net profits for any firm can be proxied by the area under the marginal revenue curve and over the marginal cost curve (in the absence of fixed costs).11 Thus, in the case of two separate, nontrading firms, the total profits are represented by the area Oaf in the left-hand graph plus the area gaf in the right-hand graph. Once the firms are allowed to trade, total profit for division 1 is area Oaf (its original profit) plus area acd (the profits from intrafirm trade) in the left-hand graph, while the profits for firm 2 are the area gaf (its original profit) plus the area acd (its profits from intrafirm trade) in the right-hand graph. The net gain is represented by the sum of the two triangles acd, which in total just equals the area under the MRX curve and over the MCX curve in the middle graph (triangle kvh). This triangle represents the net profit gain to the MNE as a whole from intrafirm trade.12 Triangle pvh is the gain to the importer and triangle kvp the gain to the exporter, where the transfer price p divides the gain between the two parties.13 This simple model of the multinational enterprise therefore predicts that horizontally integrated businesses will decouple the location of production from the location of sales and create intrafirm trade. The sine qua non of a horizontally integrated business is intrafirm trade and transfer pricing. How is the transfer price determined in this model? Setting the Transfer Price: No External Market According to several studies of transfer pricing practices of multinationals (which we reviewed briefly in Chapter 1) the typical MNE values its intrafirm trade flows within the enterprise either by having the prices set by headquarters or through bargaining among divisions. The prices are normally set for internal reasons only, and the MNE's planning department is generally not involved. Pricing tends to be on a cost plus basis or a price adjusted between the divisions. The more centralized the MNE, the more likely the transfer price is to be set by headquarters. If an outside market price exists, some reference is often made to that price; however, not all MNEs allow their divisions to buy or sell on the outside market or use these external prices even if the firms do trade outside. As we saw in the simple model of a horizontally integrated MNE, in the absence of any taxes or tariffs or other constraints, the transfer price simply vanished from the objective function. What division 2 pays for X exactly equals what division 1 receives for X, so that when the two profit functions are amalgamated these terms cancel out. How then should the transfer price be set? To answer this question we have to distinguish between three different prices: the

220

Transfer Pricing and Taxation

BOX 5.1 Transfer Price Definitions

Variable

Term

Definition

X

Efficient transfer price

The opportunity cost of the transaction; what it would earn in its next best alternative use; the shadow price that ensures all of the MNE's output is sold.

w

Money, or accounting, transfer price

If the MNE can keep different sets of books, this is the transfer price used to allocate profits between divisions for external purposes; it has no direct effect on MNE resource-allocation decisions.

P

Profit-maximizing The transfer price that maximizes the transfer price MNE's global profit net of taxes and tariffs.

F

External price

An exact comparable uncontrolled price on the external market against which the MNE's transfer price can be compared; the external price may or may not exist.

W

Regulated transfer price

The transfer price imposed by regulatory authorities on the multinational; it may be a floor or a ceiling price. If an exact comparable exists, this is the comparable uncontrolled price (CUP), and the MNE will select this price, according to the Hirshleifer rule.

efficient transfer price, the money or accounting transfer price, and the profitmaximizing transfer price. These concepts are shown in Box 5.1 and discussed below. The Efficient Transfer Price The efficient transfer price A, is based on the concept of opportunity cost, that is, what the next best alternative would be to each division of the MNE. To quote Erwin Diewert, a Canadian economist, in perhaps the best theory paper written on transfer pricing:

The Simple Analytics of Transfer Pricing

221

The efficient transfer price ... arose when solving the firm's global profit maximization problem when there were no tax distortions. This price is the value to society of the intermediate good that is being traded between the various divisions of the firm ... The efficient transfer price is the optimal Lagrange multiplier for the intermediate goods constraint in the undistorted profit maximization problem. It may be used to evaluate divisional performance if there are no tax distortions. (Diewert 1985,76) Each firm should look at the alternative opportunities available to itself; for example, the importer firm could make the product itself or contract the process out to external firm, and the exporter could sell the product itself or contract the sales to an outside distributor.14 All possible options, both inside and outside the MNE, are to be considered here. The next-best alternative will differ depending on whether an external market with exact or inexact comparables exists.15 Where no outside market exists, the MNE should still use the opportunity cost of the exporter as the transfer price for efficiency reasons. When the product is produced and sold by both divisions, the efficient transfer price for the exporter firm should be its marginal revenue since the opportunity cost of exporting it, for the producer, is to sell the product itself.16 Similarly, from the importing firm's point of view, the efficient transfer price should be its marginal cost since the opportunity cost of importing the product for the importer is the cost of making the product itself.17 In the model we have outlined above there is no external market for the product sold internally within the MNE. As a result, the efficient transfer price is A-, the shadow price on the Lagrangian constraints, as outlined in the first-order profit condition (equation 11). As we saw in Figure 5.1, the efficient transfer price is also determined by the intersection of the MRX and MCX curves. If p were set above A, (see the middle graph in Figure 5.1), there would be an excess supply of X; similarly, a price below X would imply an excess demand for X. For the internal market to be cleared, the efficient transfer price must equal the shadow price X.18 The Money or Accounting Transfer Price The money, or accounting, transfer price (let us call it w) is the official price used by the MNE to allocate profits between the divisions of the enterprise. Usually the reason for setting this price is to avoid taxes, tariffs, or foreignexchange controls. The transfer price w will be set either high or low depending on which saves most on these constraints.19 In the model we have outlined above, the money transfer price has no fixed value. There are no costs to the MNE of decoupling the transfer price for resource allocation purposes (X) from the official transfer price used to deter-

222

Transfer Pricing and Taxation

mine the level of accounting profits in each division (w). Thus the MNE is free to keep two sets of books: one for internal resource allocation, and a second for determining the allocation of profits between the two divisions. Since there are no trade barriers to minimize, w can be set at any level. The Profit-Maximizing Transfer Price The third concept is that of the profit-maximizing transfer price (let us call it p*). This is the transfer price which maximizes the MNE's global profits, net of taxes and tariffs, over all its divisions. Where there are no tax or tariff distortions, the profit-maximizing transfer price can be set at any level. We can see this in equation (7) where the variable pX, intrafirm sales, simply falls out of the equation when the profit functions of the two firms are added together. This can be seen also by partially differentiating equation (7) with respect to p. Since 3ft/3 p = 0, the assumption is that p can be set at any level and n not be affected.20 Thus p* is indeterminate in this model (Copithorne 1971, Diewert 1985). Setting the Transfer Price: the Hirshleifer Rule These results change if there is an external market price for an exactly comparable product. If such an external price (let us call it pe) does exist, that price should be taken into account in determining the efficient transfer price. This is the Hirshleifer rule (Hirshleifer 1956, 1957). Hirshleifer proved that, if an exact comparable uncontrolled price (CUP) exists on the open market, MNE should use it as long as there are no interdependencies between the divisions. That is, the two divisions should be allowed to buy and/or sell on the external market at the going market price. In such a case, if the two divisions bargain between themselves over the price, the profit-maximizing transfer price for the MNE as a whole will be the external price. Even though one of the divisions will suffer a fall in profits, the gain to the other(s) more than offsets the loss. We first provide a simple proof of the Hirshleifer rule for a horizontally integrated multinational. Suppose there exists an external market where the final good can be bought or sold at the external price pe and the MNE allows both affiliates to either buy or sell at pe. Suppose that pe is higher than the initial transfer price p = A,, so that division 1, which initially was exporting the intrafirm traded good, will now want to export to the external market. Letting X be the initial volume of intrafirm trade and S be the volume of outside sales (which could be either positive or negative), we have:

The Simple Analytics of Transfer Pricing

223

where the first square bracketed term is the profit of the exporter, firm 1, and the second square bracketed term is the profit of the importer, firm 2. Firm 1 sells Y! at price Pl in its local market, transfers X in intrafirm trade to its affiliate at internal transfer price p, and sells volume S on the open market at external price pe, where Qj = Yj + S + X. Firm 2, the importer, buys X from its parent at transfer price p, produces Q2 itself and sells Y2 = Q2 + X at price P2 in its local market. Simplifying the MNE's profit function, and noting that all output must be sold either inside or outside the MNE, we have:

Note that all terms involving pX cancel out and are gone from the MNE's objective function, equation (13). Thus the money transfer price w and the profit maximizing transfer price p* are indeterminate as we saw earlier. We differentiate equation (13) with respect to Qi? Y i? X, and S to find the first-order profit maximizing conditions. Differentiating (13) with respect to Q, and Y, and S, we have:

Differentiating (13) with respect to Q2 and Y2, we have:

Differentiating (13) with respect to X we have:

Putting these three conditions together gives us:

Therefore, when an external market exists, the MNE will maximize profits by setting the efficient transfer price, A, = A,j = -A^, equal to the external price pe. This will ensure that all output is sold so that the Lagrangian constraint is satisfied. Figure 5.2 illustrates the Hirshleifer rule where the external price is both lower and higher than the initial internal transfer price. In the left-hand graph in Figure 5.2 we assume that the external price pe is below the initial transfer price p = X. Initially, the MNE is at point a, setting

224 Transfer Pricing and Taxation FIGURE 5.2 The Hirshleifer Rule: Set the Transfer Price Equal to the Comparable Uncontrolled External Price

MR2 = MQ = p. However, the MNE is better off if it allows the importing division to purchase the good in amount S on the open market, setting p = pe so that its total sales rise to Y2, even though intrafirm trade falls to Q t . In the right-hand graph, the internal transfer price in the absence of an external market (represented by transfer price p at point a) lies above the external price. However, the MNE's total profits are higher if the exporting division sells less to its affiliate (Y2) but produces more in total (Q^, exporting the difference (S) to the external market. In both cases, the total profits to the MNE rise by the triangular area abc. It is easy to prove that, where the external price is below the initial price, the profit gain to the importing division exceeds the loss to the exporting division, with a net gain to the MNE of area abc in Figure 5.2. (Similarly, it is easy to show that where the external price is higher, the profit gain to firm 1 exceeds the loss to firm 2 by area abc.) In the first case, in the initial equilibrium, firm 1's profits are the triangular area Opa, while firm 2 receives the area pad. Open-

The Simple Analytics of Transfer Pricing

225

ing up to the external market means the effective transfer price becomes the external price pe. The exporter's profit falls to area Opec, while the importer's profit rises to area pebd. Effectively, two things have happened. Some of the exporter's profit, represented by the area pacpe, is transferred to the importing affiliate, and, secondly, there is a net gain to the MNE of area abc. This latter area represents the gain to the MNE as a whole from participating in the external market. Regardless of whether the external price is higher or lower than the in-house transfer price, the MNE should use the external price. The parent should allow the exporting division to sell on the external market if the external price is higher (the alternative to selling the product inside the MNE is to sell it outside) or the importing division to buy on the external market if the external price is lower (the opportunity cost of buying inside is buying outside). The efficient transfer price, in the presence of an exact comparable outside price, is therefore the outside price. Where there are costs of using the external market, or where small differences exist between the internal and external products, the external price should be adjusted for these differences. The OECD Rules for Product Comparables: The CUP Method We have made the argument above that the multinational enterprise, of its own volition, will use the external arm's length price as the efficient internal price if external markets exist. The Hirshleifer rule implies that if interdependencies among units either do not exist or are small, and if an exact comparable product is traded on the external market, the MNE will choose the arm's length transfer price. Therefore there are two caveats to the Hirshleifer rule: interdependencies and comparability. We address each in turn. First, in terms of interdependencies, Hirshleifer (1957) argued that demand or supply interdependencies can easily arise in practice. On the demand side, affiliated divisions of the MNE may have related demands if the additional sales of one division (e.g., Chevrolets) negatively or positively affect the sales of another division (e.g., Pontiacs). There can also be supply interdependencies in the form of large economies of scale at the upstream stage of production or fixed costs which are common to all divisions of the MNE (e.g., overhead costs). In such cases, the external price is not a true reflection of the overall costs or revenues to the multinational since the affiliated firms, if left to bargain alone, do not take into account the effects of their actions on other parts of the enterprise. Such interdependencies are most likely to arise in the case of support services (common supply) and group intangibles (joint demands). We therefore leave the discussion of the impacts of interdependencies to later in this chapter. Suffice it to say that simple marginal pricing rules (e.g., set the transfer price

226

Transfer Pricing and Taxation

equal to the external price) may provide inefficient solutions because the divisions ignore their interdependency. Where interdependencies exist, for efficient resource allocation, the MNE must look at the overall impact of the affiliate on MNE profits; that is, a comparison of total costs and total revenues is required (Hirshleifer 1956, 1957). Second, in terms of comparability, if exact comparable prices exist in the external market, these are the prices that the Hirshleifer rule predicts would be used by the multinational enterprise. The MNE would pick these prices for efficiency reasons; that is, it is more profitable for the enterprise as a whole if the profit-maximizing transfer price reflects true opportunity costs in terms of available opportunities to the buying and selling divisions. If the exact comparable is available, it is always more profitable for the MNE as a whole to allow the divisions to access the product at the going market price even if one of the divisions suffers a fall in profits as a result. For traded, undifferentiated commodities such as metals (steel, copper, bananas), there may be an exact comparable price on the open market. Where products are differentiated, such as for consumer goods (cars, computers), there is likely to be an inexact comparable price in the external market, even if no exact comparable exists, so that the Hirshleifer rule has practical relevance for the MNE's transfer pricing policy. And, as we saw in Chapter 1, about 40 per cent of MNEs do set their transfer prices on some variant of the market price, and another 15 per cent allow their divisions to bargain over the transfer price. Therefore, where an external price for an exact comparable exists, the Hirshleifer rule says that it is more profitable for the MNE as a whole to allow its affiliates to either buy or sell on the external market at the going external price. This is the same rule for the firm as the fundamental principle underlying the OECD 's tax rules on transfer pricing; that is, transfer prices should approximate the arm's length price which two unrelated parties would have chosen if the transaction had taken place in the external market. The specific OECD definition of the arm's length standard for transfers of tangibles is: prices paid for goods transferred between associated enterprises should be, for tax purposes, those which would have been paid between unrelated parties for the same or similar goods under the same or similar circumstances. (OECD 1979, 28)

To satisfy the arm's length standard, the OECD recommends transactional methods be used, with the highest priority being given to the comparable uncontrolled price (CUP) method. The new proposed OECD guidelines define the CUP method as follows:

The Simple Analytics of Transfer Pricing 227 The CUP method compares the price for property or services transferred in a controlled transaction to the price charged for property or services transferred in a comparable uncontrolled transaction in comparable circumstances. (OECD 1994b, par. 92, 173).

In order for this comparison to be useful, the 'economically relevant characteristics of the situations being compared must be sufficiently comparable' (OECD 1994b, par. 32, 163). In the case of the CUP method, the OECD says that the comparability criterion for related and unrelated party transactions is satisfied in one of two cases: (1) there are no differences between the transactions or between the parties that could materially affect the price in the open market,21 or (2) there are differences, but they are quantifiable, so their effect can be eliminated (OECD 1994b, par. 92, 173). Tax authorities should look at the price of goods sold on markets that are 'economically comparable'; that is, the goods are sold at the same point in the value chain and are physically identical: similar volume, similar time period, similar terms of sale and payment, and so on (OECD 1979, 35-7; 1994b, par. 36, 164). The CUP method is therefore a good proxy for the transfer price that a profitmaximizing multinational would pick if there are no interdependencies in demand or supply among the affiliates, and in the absence of external motivations such as taxes and tariffs. In these circumstances, CUP and the efficient transfer price are one and the same. Summary: The Horizontally Integrated MNE This, then, is a simple microeconomic model of how a horizontally integrated multinational chooses its profit-maximizing output, sales, and trade levels, and sets its efficient transfer price for the traded good. Where an exact comparable uncontrolled price exists in the external market, and there are no interdependencies within the MNE family, the Hirshleifer rule implies that the MNE will select that price as the efficient transfer price. In the absence of an exact CUP, the MNE will set its transfer price at the shadow price that clears the internal market. The money transfer price and the profit-maximizing transfer price are indeterminate in this model. We turn now to analysing transfer pricing in the second major type of MNE: the vertically integrated producer. Transfer Pricing in a Vertically Integrated Multinational Although most microeconomic models of intrafirm trade and transfer pricing in the transfer price literature assume the MNE is horizontally integrated and trades in an identical product, the most common case in practice is trade among

228

Transfer Pricing and Taxation

vertically integrated firms.22 By far the largest volume of intrafirm trade in tangible goods consists of parts, components, and subassemblies being shipped from upstream affiliates to downstream affiliates in manufacturing industries such as automobiles and consumer electronics, and in natural-resource industries like petroleum and mining. In such cases foreign direct investment and intrafirm trade are complements: MNEs set up foreign plants in order to segment the production process, generating as a result higher volumes of intrafirm trade in intermediate products. The reasons for siting production processes in different countries have to do with the need to use resources that vary geographically in their availability, quality, and cost. As we saw in Chapter 3, resource MNEs, which go abroad to extract and process raw materials, are attracted to locations with abundant, cheap materials; low costs for energy, capital, and transportation; and few governmental barriers to trade. Manufacturing MNEs tend to segment the production process, subcontracting the footloose stages of production such as parts and subassemblies in developing countries, and the stages where market presence is important close to final sales. If service MNEs can segment the value chain they may also shift routine functions to developing countries. High-tech MNEs, however, are less likely to move production offshore since they need to be close to their major markets, particularly for new products and processes where the need to tailor products and consumer tastes is high. The Vertically Integrated Multinational Assume that we have a vertically integrated multinational consisting of two plants in different countries, one producing a raw material (firm 1 producing a mineral like bauxite, for example) and selling this raw material to a firm located in a different country. This firm (firm 2) processes the raw material and sells the final product (e.g., aluminum) to consumers in its country or elsewhere. How should the volume of intrafirm trade and the transfer price be determined? Assume that one unit of the raw material Q[ is used to make one unit of the finished good Q2 so that Qj = Q2-23 The MNE's problem is how to maximize overall profits given that there are two costs involved in production, Cj(Qj), the cost of the raw material, and C2(Q2), the cost of processing the raw material into a finished good. Assume a transfer price of p is initially charged per unit of the raw material. Then firm 1 makes profits of:

where Y, = Q, = X by definition since all primary output is sold to the downstream MNE affiliate. Firm 2 makes profits of:

The Simple Analytics of Transfer Pricing

229

Our production constraints are that (1) all output of the intermediate good must be sold internally - that is, Q} = Yj = X = Q2 - and (2) all output of the final good must be sold - that is, Y2 = Q2. The MNE maximizes the overall profit function:

Rearranging, and adding in the Lagrangian constraints, we have:

Note first that all pX terms have vanished from the objective function. Thus the money and profit-maximizing transfer prices are again indeterminate in this model. Now differentiating (21) with respect to Y2, Qj, and Q2, and rearranging, profits are maximized where:

or, rewriting (23) in terms of the net marginal revenue of the importing firm, we have

Equation (22) determines the efficient transfer price X for the intermediate good as the marginal cost of the exporting division MCf, this ensures that the Lagrangian constraint that all output of Ql is sold to firm 2 is satisfied. Equation (23) says the MNE maximizes global profits when it equates the marginal revenue from final sales (MR2) to the summed marginal costs of producing these sales (MC, + MC2), and that both of these conditions should equal the shadow transfer price MCj, the MC, curve shifts even further downwards as the tax deduction for intrafirm imports increases. We assume the tax authorities in country 2 are aware of this and set a floor on the acceptable transfer price of W. The new curve is therefore MCj - t2W so that the new equilibrium is at point s with a higher output level Q,. The new net-of-tax profit of the MNE is the area hzs, which is larger than the old aftertax profit of gbz by the area hgbs. The gain to the MNE from transfer price

Taxing Multinationals in Theory

291

manipulation can be calculated as follows. First, overinvoicing saves on tax costs equal to the distance sf times the volume of intrafirm trade; that is, the tax saving is the parallelogram hgfs. Second, the expansion in trade causes a misallocation in resources for the MNE that lowers profit by the triangle bfs. The net impact of these two factors is a rise in after-tax profits equal to the area hgbs. It is therefore profitable for the MNE to overinvoice tax-deductible items like intrafirm imports, and, where possible, to underinvoice taxable items such as revenues from intrafirm transactions.14 The Residence Principle If the home country follows the residence principle, firm 1 is taxed on its worldwide income, that is, on total MNE profits wherever earned. If country 1 taxes on an accrual basis (i.e., tax deferral is not permissible) then all profits are taxed at the same rate t2 and the MNE cannot avoid the tax through transfer price manipulation. This is clear from looking at a revised version of equation (10), where t2 now applies to all MNE profits:

From the rearrangement of equation (19) it is clear that the first-order condition for a profit maximum is the same as in the no-tax situation, MR, = MC, = MR2 = MC2, and that the profit-maximizing and money transfer prices have no role to play in this model. That is, the MNE cannot avoid the tax in the same way in which a monopoly firm cannot avoid a pure profit tax. The multinational simply absorbs the cost of the tax. On the other hand, if the taxing authority allows deferral of the tax for profits retained in the host country, the situation changes. In this case the MNE has an incentive to retain profits abroad. Assume the MNE retains the fraction 1 - (3 of subsidiary profits in the host country and remits (3 to the parent firm. Rewriting (19) to allow for deferral we have:

Rearranging and adding in the Lagrangian constraints we have the MNE's objective function:

292 Transfer Pricing and Taxation If p = 1 (full repatriation), then equation (21) collapses to equation (19); that is, the home country taxes on the residence principle and all profits wherever earned are taxed as earned. If p = 0 (no repatriation), equation (21) collapses to equation (11); that is, the home country taxes on the source principle. The first-order conditions for a profit maximum in the case when 0 < P < 1 are the following:

Rearranging (23) we have:

That is, the after-tax net marginal revenue to the importing firm should be equated to the after-tax marginal cost of the exporting firm. There are two effects, the ad valorem tax effects on the marginal revenue and cost curves, and the specific tax effect of the transfer price. Again, the net incentive is to raise the transfer price as long as P < 1 so that some profits are retained in the host country, where they face a lower effective tax rate.15 We can see this by partially differentiating equation (21) with respect to p and using the envelope theorem:

Thus the MNE should overinvoice transfers into high-tax locations in order to shift profits to lower-taxed affiliates.16 Both Countries Tax Pure Profits The Source Principle Now assume that the host country also taxes the MNE's pure profits. If the home country continues to follow the source principle, both countries tax only profits earned within their borders. In this case the MNE's objective function becomes:

where Yj = Q, = Q2. Rearranging and adding in the Lagrangian constraints, we have the MNE's objective function:

Taxing Multinationals in Theory

293

The first-order conditions for a profit maximum are:

Again, the MNE compares the after-tax returns to the importer, firm 2, to the after-tax costs of the exporter, firm 1. The profit taxes have two effects, ad valorem and specific. The profit-maximizing transfer price is determined by:

That is, if the home tax rate (t2) is higher than the host rate (t,), MNE profits are increased if the transfer price is set high; if t2 is less than t, the transfer price should be set low. In either case, the profit-maximizing price p* is set so that profits are shifted to the lower-taxed affiliate. The Residence Principle In our last case we assume both countries tax pure profits on the residence principle, so that t2 applies to MNE profits as a whole while t, applies only to host country profits. We assume as before that the MNE remits the fraction (3 of its subsidiary profits to the parent firm. This allows us to model three cases at the same time since setting (3 to different levels is equivalent to different tax regimes in the home country. We identify these cases as: (1) home country taxation on the residence principle with no tax deferral so (3 = 1, (2) home country taxation on a source basis so p = 0, and (3) home country taxation on a residence basis with deferral so 0 < (3 < 0. We assume that the home country gives a foreign tax credit up to the level of the home tax rate for any taxes the MNE pays to the host country. We explain the calculation as follows. n\ is the subsidiary's total before-tax profit on which taxes of tlnl are paid to the host country, leaving (1 - t^TCj in aftertax profits. Of this amount, (3(1 — t])7t] is remitted to the home country while (1 - |3)(1 - t,)7Cj represents the subsidiary's retained earnings. When the repatriated earnings (the dividends) are remitted to the parent, firm 1 brings them into its income and grosses them up by the amount of foreign taxes paid so that they are now on a pre-tax basis, as follows: [3(1 - t,)7t,/(l - t,) = (371,. The govern-

294 Transfer Pricing and Taxation ment then calculates the home country tax due on the dividends as t2$n\ anc* gives a credit for the foreign taxes paid, so that the net effect of the residence principle with deferral and a foreign tax credit is: t2$Ki ~ *i P71! - (*2 ~ t i) P71! where, if tj > t2, no additional tax is due in the home country. In effect, the MNE pays tj in tax on all of its foreign subsidiary's profits and an additional tax equal to the difference between the home and host tax rates on any profit remittances to the parent firm. The effective tax rate on subsidiary profits is therefore tj + (3(t2 -1,), which has a floor of t t and a ceiling of t2. The MNE pays the floor if P is zero (no remittances) or if t2 < tj (the home tax rate is lower than the host rate); the MNE pays the ceiling if p = 1 (no deferral) and t2 > tj (the home rate is higher than the host rate). Note that either condition can hold to satisfy the floor but both conditions must hold to satisfy the ceiling. We assume in what follows that the home tax rate is higher than the host rate since if the host rate is higher, no additional taxes are due when the subsidiary remits its profits. Putting this altogether, the profit function for the MNE is:

Rearranging and adding in the Lagrangian constraints, we have the MNE's objective function:

If P = 1 (full repatriation), then equation (32) collapses to equation (19); that is, the home country taxes on the residence principle and all profits wherever earned are taxed as earned. If p = 0 (no repatriation), equation (32) collapses to equation (27), in which both countries tax on the source principle. The firstorder conditions for a profit maximum when 0 < P < 1 are the following:

That is, the after-tax net marginal revenue to the importing firm should be equated to the after-tax marginal cost of the exporting firm. Partially differentiating equation (32) with respect to p and using the envelope theorem:

Taxing Multinationals in Theory

295

Therefore the MNE has an incentive to overinvoice its exports to the home country, shifting profits to the subsidiary, if the effective tax rate on remitted dividends is lower in the host country. The profit-maximizing transfer price in this case should be set at its upper bound, that is, p* = max W. Where the tax rates are the same or where the home country rate is higher and all profits are taxed as earned, the profit-maximizing transfer price is indeterminate. And where the host tax rate is higher, p* should be set at the lower bound determined by the tax authorities, that is, p* = min W. These regulatory bounds will be based on the tax auditor's estimate of the arm's length price, based on the application of the various transfer pricing methods we described in chapters 1 and 5 (e.g., CUP, cost plus, resale price). Note also that, given the choice, the profit-maximizing deferral rate is zero, that is, no remittances should be made whenever the home rate is higher than the host rate. We can see this by differentiating (32) with respect to (3 as follows:

The expression in square brackets in (36) is the subsidiary's total pre-tax profit. As long as the home rate is higher and profits are positive, the MNE will reduce its dividend remittances. Some Related Issues In this section, we deal with four issues that extend the models we have used so far: (1) whether the MNE uses one or two sets of books, (2) minority shareholders, (3) exchange rate changes, and (4) the combination of taxes and tariffs. The first issue answers the general question of whether the MNE takes external constraints such as taxes and tariffs into account in setting its profit-maximizing transfer price. The last three issues deal with the impacts of common constraints on the MNE's transfer pricing policies. One or Two Sets of Books? The issue of whether MNEs use one or more sets of books has been frequently discussed in the transfer pricing literature (Rugman and Eden 1985, Eccles

296 Transfer Pricing and Taxation 1985, Cummins et al. 1995, McGuinness 1985). Two sets of books' can have different meanings to different readers. The first meaning is the practical reality in accounting for MNE profits (Cummins et al. 1995). In many countries, such as the United States, MNEs are required to keep two sets of accounting books. U.S. companies are required to publish financial statements that conform to generally accepted accounting principles (GAAP) as set by the Financial Accounting Standards Board (FASB). The published set of books is designed to provide financial information on the firm to potential and actual investors and creditors. At the same time, the IRS has quite different reporting requirements for tax purposes; for example, depreciation rates are different. In other countries, e.g., Germany, for example, one set of books is required for financial and tax purposes. In fact, the MNE may keep several sets of books: one for tax purposes, another for customs valuations, a third for public view by minority shareholders and union groups, and a final one for private allocation decisions. Several sets of books are necessary where different government regulatory authorities require different types of accounting. The second meaning is quite different; that is, should the MNE keep one set of books for resource allocation decisions, and a second set for tax purposes? The common perception is that MNEs keep several sets of books, and that real decisions and financial decisions are made separately (and by different departments). The evidence, however, is mixed. Surveys of firm pricing policies tend to show MNEs as setting transfer prices without the advice of the MNE's tax department; many studies also document that most MNEs do not have an explicit transfer pricing policy (e.g., Wilson 1993; U.S. Treasury 1988). In addition, keeping different sets of books has real administrative costs, can have create mixed signals for managers, and can have negative impacts on managerial performance where bonus packages are tied to subsidiary returns (Eccles 1985; McGuinness 1985). Firms that do keep two sets of books and engage in creative financial manoeuvres, in practice, are unlikely to face severe tax penalties in most countries.17 This encourages transfer price manipulation, particularly of fungible activities such as management fees. Auditing of tax books also has a time dimension as well, since customs valuations are often made years before tax officials may begin a corporate income tax (CIT) investigation, although this is changing as tax and customs authorities, both in Canada and the United States, are now more frequently exchanging transfer pricing information. Firms, especially smaller MNEs, see the probability of a tax audit as small; the audits take place one to three years after the event; and the probability of the dispute's going to court and being settled in favour of the tax authority is very small.18

Taxing Multinationals in Theory

297

Hence the incentive to use different sets of books to avoid taxes - that is, to keep a money transfer price policy different from the firm's profit-maximizing transfer price policies - appears clear. Note, however, that the models we have used above - and most of the economics literature in this area - implicitly assume that the multinational does not keep two sets of books; that is, it uses the same transfer price for internal decisions as it uses for tax purposes. Unless the MNE does keep different books for the government authorities and for private allocation decisions, nominal tax rates can and do influence real output, sales, and trade decisions for the integrated firm. Why do the microeconomic models of the MNE assume one set of books when MNEs appear in practice to keep several for different purposes? We argue that it is more profitable for the MNE to choose one transfer price, a price that takes into account both tariff and tax rates and the impact on resource allocation.19 If under- or overinvoicing pays in terms of tax relief, the MNE can increase this relief by expanding the volume of intrafirm trade; however, the MNE needs to balance off at the margin the savings from reducing tax penalties against the costs of such misallocated resources. Keeping two or three sets of books also incurs administrative time and effort and increases the risk of state monitoring of MNE transactions. Let us illustrate why one set of books is more profitable for the MNE than two. Assume that the MNE can keep two sets of books: one in which the transfer price is set at w (the money or accounting transfer price) for tax/tariff purposes and a second where the transfer price is set at A, = MC, (the shadow transfer price) for resource allocation purposes. We use the simplest tax model the home country taxes on a source basis, equations (10)-(14) - as our example. In this case the first-order condition (14), substituting w for p, is:

Taxes can be reduced and after-tax profits of the MNE increased through two methods. First, holding the volume of intrafirm trade constant, net profits are higher if the primary firm overinvoices its exports to the downstream parent. We saw this earlier: the MNE benefits if it overinvoices, shifting profits to the lower-taxed subsidiary. The gain to the MNE is t2(w - MC,) per unit of intrafirm trade, where w - MC, represents the amount of overinvoicing relative to the shadow transfer price X = MCj in the no-tax situation. In Figure 6.2, this is equivalent to the tax savings represented by area hgbm. Second, given that overinvoicing is possible, holding the amount of overinvoicing (w - MC,) per unit constant, taxes can be further reduced by expanding the volume of intrafirm trade. Since each unit of intrafirm trade saves the

298 Transfer Pricing and Taxation MNE t2(w - MCj) in taxes, the larger is X the greater total tax savings of t2(w - MCj)X. In Figure 6.2 this is shown as the move from Q0 to Qj; the additional gain to the MNE is the area mbs. So, raising the volume of intrafirm trade, with its subsequent implications for output and sales in each affiliate, is more profitable than leaving output at its pre-tax levels. Thus, using the same transfer price for resource allocation purposes as for tax/tariff purposes is more profitable than using separate transfer prices. One set of books is preferable! The money transfer price should be the profit-maximizing transfer price, which in this case is the highest possible transfer price allowed by the tax authority in country 2, which is W. Note that this does not mean there are no costs to the MNE from using only one set of books. Choosing a transfer price different from the efficient transfer price p = MCj does mean that output and sales are misallocated within the multinational enterprise. Manipulation of the first-order condition clearly shows this:

The left-hand side of (38) shows the gain to the MNE from overinvoicing, assuming the MNE only uses one set of books; the right-hand side shows the after-tax cost to the MNE from misallocating its resources. If w > MC t , the firm will choose to set MCl > MR2 - MC2; that is, the MNE will expand output, and thus the volume of intrafirm exports, beyond their initial no-tax levels. The misallocation of resources in Figure 6.2 is shown by the fact that output is expanded beyond the point where NMR2 = MC, (point a). Past that point, in pre-tax terms, marginal cost of the exporter exceeds the net marginal revenue of the importer.20 In post-tax terms, expanding output adds the area mbfs in Figure 6.2 in tax savings, but costs the area bfs in misallocated resources. In net-of-tax terms, the MNE is clearly better off by triangle mbs. Thus it pays to keep one set of books. In theory then, if not in practice, multinationals maximize their global profits net of taxes and tariffs. In doing so, they find it more profitable to choose transfer prices that are either higher or lower than those the no-tax situation would suggest. Over- or underinvoicing, per unit of intrafirm trade, creates tax savings and higher after-tax profits for the MNE. These tax savings are magnified when the firm increases the volume of intrafirm trade. The MNE will therefore increase its trade volume in both under- and overinvoicing cases.21 Thus the money transfer price w and the profit-maximizing price p* are the same: the MNE keeps one set of books.

Taxing Multinationals in Theory

299

Minority Shareholders: A Tax on the Multinational? Does it make a difference to the MNE if the foreign subsidiary has minority shareholders? The answer is: it depends. Where a host government imposes minority shareholder requirements on the MNE (as many developing countries have done and continue to do as part of their FDI regulations), the firm no longer has complete control over its activities and its affiliate profits must be shared with unrelated foreign parties. In effect, such regulations act as a tax on host country profits earned by the MNE. One would expect the outcome therefore to be similar to that of a pure profits tax levied by the host country on the affiliate's income. On the other hand, this is only true if minority shareholders are imposed on the MNE. If the enterprise freely enters into a strategic alliance or a joint venture, then presumably the MNE perceives the benefits (access to the partner's capital, knowledge of the local market, and so on) as outweighing the costs (sharing the profits with the foreign partner). In such cases, we would not expect the minority shareholder to be seen as a tax.22 In this section, we assume the host government requires the MNE to have minority shareholders, and investigate the reactions of the MNE in terms of transfer pricing. Assume the foreign affiliate in country 2 is partly owned by host country nationals, and the proportion of foreign minority ownership is k, where 0 < k < 1. Assume the foreign shareholders must be paid the fraction k of the profits declared by the subsidiary in the host country.23 Then there is a clear incentive to use transfer pricing manipulation to shift profits out of the host country. Assume we have a horizontally integrated multinational in which firm 1, the MNE parent, is exporting a finished good to firm 2, the joint venture, as we modelled in Figure 5.1 and in Figure 6.1. The global profit function for the MNE, taking minority ownership into account, is:

which can be rearranged as the objective function:

from which the profit-maximizing condition is:

300 Transfer Pricing and Taxation The presence of minority shareholders therefore has two effects on the MNE. First, it acts like an ad valorem tax on profits earned in the host country, since the MNE only receives 1 - k of the profits. This ad valorem effect tends to reduce the volume of intrafirm trade. On the other hand, there is also a specific tax effect; the kp term means that if the MNE charges a high transfer price for parent exports to its affiliates, these costs are deductible in country 2 and thus reduce declared profits in the host country. The profit-maximizing transfer price is positive since:

So there is an incentive to overinvoice subsidiary imports and underinvoice its exports in order to shift profits from the subsidiary to the parent firm. We can see this more clearly by examining the equation for net profit on the marginal unit of intrafirm trade:

which is positive (negative) if the MNE over- (under-) invoices. If the MNE is free to set its transfer price it will therefore shift profits out of the minority affiliate where possible. Transfer Pricing and Exchange Rate Changes Assume first that we have a vertically integrated MNE in which the foreign subsidiary, firm 1, exports a raw material for processing and final sale by its parent firm, firm 2. Let all transactions take place in the home country currency and the exchange rate e be used to convert the foreign currency. The exchange rate e is the price of the home currency divided by the price of the foreign currency, so the subsidiary's profit TI, (measured in units of the foreign currency) multiplied by e gives us eftj (measured in units of the home currency). When e rises, the home currency depreciates, which means one unit of the home currency buys less of the foreign currency, and the host currency appreciates. When e falls, the reverse occurs: the home currency appreciates and the host currency depreciates. As before, but now with the exchange rate added, the MNE maximizes the overall profit function:

Taxing Multinationals in Theory

301

Rearranging, and adding in the Lagrangian constraints, we have:

It is clear from equation (45) that the exchange rate e has no immediate link with the transfer price since the p terms have disappeared. The first-order conditions are basically unchanged since the MNE still sets NMR2 = eMC[; however, they are now adjusted for the exchange rate e. What the exchange rate does affect is the cost of production in the foreign country. Taking the partial differential of (45) with respect to e and using the envelope theorem, we have:

Therefore increases in e (which mean the host currency appreciates relative to the home currency) raise the cost of production in the foreign subsidiary and have a negative impact on MNE profits. Intuitively, if production becomes more expensive abroad because the foreign currency has appreciated relative to the home currency, the MNE will react by curtailing production abroad and importing less. However, the exchange rate shock cannot be avoided through transfer pricing since the money and profit-maximizing transfer prices are not determinate in this model. To pin down the transfer price, we need to introduce an additional constraint such as a tariff. Let the home country levy a tariff on imports from firm 1.24 The MNE's objective function is:

In this case, the effect of the exchange rate change is:25

which again is negative if there is a tariff. To the extent that the MNE underinvoices to avoid the tariff (which it would do if possible), the exchange rate shock is also moderated. This is clear from deriving the profit-maximizing transfer price:

302 Transfer Pricing and Taxation

The transfer price p* should be set as low as possible to avoid the tariff, as before. The higher the exchange rate, the more tariff revenue is saved from underinvoicing. We can show this by looking at the marginal profit per unit of intrafirm trade, which is:

Thus the lower is p the greater the marginal profit, and, vice versa, the higher is e the lower the marginal profit per unit of intrafirm trade.26 Transfer Pricing, Taxes, and

Tariffs

The last case we want to look at before turning to a full-blown model of transfer pricing and the corporate income tax is a model that includes both a tariff and a tax on pure profits. Let us take our second model in this chapter (the ad valorem tax) and add in taxes levied by both countries on a source basis. The MNE's objective function becomes:

which, adding in the Lagrangian constraint, reduces to:

Leaving the first-order conditions for the reader as an exercise, we turn directly to the profit-maximizing transfer price. Partially differentiating (52) with respect to p and using the envelope theorem, we have:

which, after some rearranging, says that the MNE should set its transfer price high (low) if the tax differential (t2 - tj) exceeds (is less than) the effective tariff rate adjusted for the tax saving i(l - t2). Thus the MNE trades off the tax differential against the tax-adjusted tariff cost in determining whether the profit-

Taxing Multinationals in Theory

303

maximizing transfer price should be set high or low. Alternatively, the MNE's pricing policy should depend on whether the tax gap ratio (t2 - t,)/(l - t2) is greater or less than the tariff T. If the tax differential exceeds (is less than) the tariff, the MNE should over- (under-) invoice the transfer price. This is the standard theoretical result in the tax transfer pricing literature (Eden 1976, 1985; Horst 1971). Transfer Pricing and the Corporate Income Tax We turn now to a more realistic model of MNE transfer pricing. In the section below the corporate income tax (CIT) is modelled as falling both on pure profits and on the return to equity capital. The MNE is faced with an ad valorem tariff, two corporate income taxes, and various withholding taxes on remittances to the parent. This model more closely approximates the problems facing multinationals as they are taxed in the real world. Setting Up the Model Assume we have a horizontally integrated multinational enterprise, consisting of a parent firm, firm 1, and a subsidiary, firm 2. The subsidiary produces a finished good and also imports the same good from its parent, depending on relative costs and the transfer price charged on these intrafirm imports, for sale in its domestic market. The subsidiary pays head office fees to cover various services provided by the parent, and also makes dividend payments to the parent. Each firm produces output, Qi? for sale locally as Yj, or for export X, where i = 1,2. Parent sales are therefore Yj = Qj — X with revenues R^Y,), while affiliate sales are Y2 = Q2 + X with revenues R2(Y2). Country 2 levies a tariff at rate i on imports. Since intrafirm imports are priced at transfer price p for a total trade value of pX, customs revenues are xpX. We assume the exchange rate between the two countries is e and that all variables are measured in the home currency. Each firm's net profit function is based on its taxable income, defined as its economic profit minus tax-deductible expenses. The initial tax payable is the corporate income tax (CIT) rate times taxable income, from which tax credits are subtracted to determine the actual tax bill. Subtracting the actual tax bill from the economic profit determines the net profit of the firm, TI*J. We look first at the foreign subsidiary's, and then at its parent's, after-tax profit function. The net profit function for the foreign subsidiary is:

304

Transfer Pricing and Taxation

where Ck2 is the net-of-CIT cost of capital. We ignore labour costs for simplicity. There are three variables that complicate this model and that are not present in our earlier tax models: capital costs, head office fees, and dividends. We outline the impact of each variable below. First, we assume that each firm owns P^K} in physical capital in the form of machinery and equipment. The capital stock depreciates at a uniform rate d, and the MNE must invest to replace and expand Pj^Kj.27 The annual economic cost of capital to the firm is its opportunity cost, that is, its real return plus the depreciation rate. The real return is assumed given (the firm is a price taker in the capital market), and equals r, the going world rate of return. Arbitrage between countries ensures that the rate of return is equal across markets so that r = TJ = r2.28 The cost of capital is therefore (r + d^^Kj to each firm in the absence of taxation. This is the pre-tax cost of capital. The MNE will purchase units of capital as long as the economic benefits of an additional unit exceed the costs of an additional unit.29 The economic benefit to the firm can be measured as the extra output an additional unit of capital can produce (i.e., the marginal product of capital) times the additional revenue earned when that output is sold (i.e., the marginal revenue); this is called the marginal revenue product of capital (MRP^) or the firm's demand for capital. The economic cost of capital is its opportunity cost; this is the price of capital to the firm. In the absence of taxes, the MNE would therefore invest in capital up to the point where MRPj^ equals (r + d)?^ that is, where the demand and supply curves of capital intersect. The corporate income tax complicates the firm's estimate of capital costs because interest costs on debt capital are tax deductible, whereas the costs of equity capital are not.30 The tax deductibility of interest costs reduces the true opportunity cost of capital to the firm by the CIT rate times the leverage ratio, Lj (the ratio of long-term liabilities to long-term liabilities plus equity). The after-tax cost of capital to each firm is therefore C^ = (r + d - tjLji^Pj^Kj.31 Second, in terms of head office fees, we assume that the subsidiary is charged for H as its share of head office services (these may be specific services or some share of group services - see Chapter 5), and that H is deductible against the host country's CIT. The subsidiary remits H to the parent firm after paying a withholding tax at rate wh to the host government, state 2. Third, we treat dividends as a residual payment out of after-tax subsidiary profits; the subsidiary remits its dividends, D, to the parent, net of a withholding tax wd paid to state 2. The after-tax profit function of the parent firm is also calculated in a manner similar to equation (54), as the CIT times taxable income, minus tax credits. Taxable income equals domestic economic profits plus remittances from the subsidiary (after grossing up the dividends by the host CIT) minus other tax-

Taxing Multinationals in Theory

305

deductible expenses. Both the CIT and withholding tax paid by the subsidiar) are creditable up to the level of the home country CIT rate. The profit functior of the parent is therefore:

where C kl is the net-of-tax cost of capital to the parent. The variable / must be either zero (if the subsidiary has a surplus of foreign tax credits) or positive (a deficit of credits). Now we put these two profit functions together to get the MNE's overall profits. We assume the goal of the MNE is to maximize global net profits, TI* = (71*2 + rc*])' subject to the constraints that EQ; = EYj (everything is sold) and r2 = TJ = r (the net return to capital is everywhere equal, so capital markets are perfect), where n* is:

Given this objective function, what should the multinational do? The Various Transfer Pricing Choices In this particular model the MNE has several possible transfer pricing alternatives, broadly defined, one for each type of intrafirm flows. First, the standard case is the transfer price on the tangible, the final good imported by the foreign affiliate from the parent firm. As we have seen above, both the transfer price and the volume of trade can be manipulated to shift profits out of the high-tax location. The tax saving, however, if it involves overinvoicing, must be traded off against the extra tariffs paid to the customs authorities. Second, the parent firm provides support services to the subsidiary, for which the affiliate pays head office fees. Third, the subsidiary pays dividends to the parent firm. A fourth possibility, which we have not modelled, is that the parent firm lends financial capital to the subsidiary, so there are intracorporate interest charges to be considered. As a result, the MNE has several degrees of freedom in terms of choosing which flows it wants to manipulate, in what direction, and by what amount. The goal of the tax and customs authorities, therefore, is to prevent these various types of transfer price manipulations. Let us now turn to the MNE's choice of transfer pricing policies.

306

Transfer Pricing and Taxation

The real decision variables for the MNE are K2 and Kj (the capital stocks and therefore output levels of each firm) and X (the volume of intrafirm trade and therefore, indirectly, sales and output levels of each firm). The financial decision variables for the MNE are H (the price of head office services), D (the amount of repatriated dividends), and p (the transfer price on the tangibles transaction). We turn first to the real decision variables. Differentiating with respect to K2, Kj, and X, we have the following conditions for maximizing MNE overall profits net of taxes and tariffs. The first two conditions determine the optimal amounts of plant investment and output levels in the two plants:

This says that the marginal revenue product of capital divided by its price should equal CGi, the tax-adjusted or gross cost of capital per dollar of capital expenditures. Note that although capital arbitrage ensures that the net return, r, is equalized between the firms, the gross costs of capital are unlikely to be equalized, since CIT rates, tax deductions, and credits are likely to differ between countries, and depreciation and leverage ratios to differ between firms. The third first-order condition determines the optimal volume of intrafirm trade (and hence sales levels):

which says that the MNE should balance the subsidiary's net marginal revenue from imports against the parent's net marginal cost of exports. The marginal revenue from imports equals affiliate marginal revenue from domestic sales, eMR2, net of importing costs, (1 + i)p, in after-tax terms. The net marginal cost of exports equals the parent's forgone marginal revenue on local sales, MR1? minus its marginal earnings from exports, p, after tax. The profit-maximizing financial decisions concerning D and H (what Brean [1985] calls the fiscal transfer prices) and profit-maximizing transfer price p* are found by partially differentiating the profit function (56) with respect to p, D, and H, and using the envelope theorem. First, the optimal level of D depends on whether a surplus or deficit of tax credits applies to dividends. In the deficit case, since I, > t2 and / > 0, the optimal amount of remitted dividends to the parent firm is determined by:

Taxing Multinationals in Theory

307

In the surplus case where / = 0, the optimal amount of dividend repatriation is:

Thus the important tax variables influencing dividend repatriation are the statutory CIT rate and the dividend withholding tax rate. In both deficit and surplus of credits cases, the MNE maximizes profits by setting dividends at their lowest possible level. The optimal amount of head office charges, H, is:

which implies that normally, where t2 exceeds (is less than) t,, head office charges should be raised (lowered) since they are tax deductible in the host country and taxable at home. The withholding tax usually has no effect on the optimal amount of H because wh is so low relative to t t that the tax is normally all deductible when head office fees are remitted. The net cost of H to the subsidiary therefore is e(l - t2 + wh)H and to the parent is e(-l + t t - wh)H, for a total net cost to the MNE of e(tj - t2)H. The profit-maximizing transfer price p* is determined by:

The square-bracketed term may be either positive or negative depending on the tax and tariff costs; for example, a higher host country CIT rate than in the home country tends to encourage overinvoicing, whereas host country tariffs encourage underinvoicing. If the host CIT exceeds (is less than) the combined home CIT rate plus the tax-adjusted host tariff, the MNE should over- (under-) invoice its exports to the foreign affiliate. The general conclusions from this more complicated model of the impacts of corporate income taxes, withholding taxes, and tariffs on MNE behaviour are that (1) marginal rates of tax tend to affect the real decision variables such as investment and output; (2) nominal or statutory tax and tariff rates affect financial variables such as the choice of transfer price, repatriation of dividends, and the remittance of head office fees; (3) the MNE chooses its transfer pricing policy for intrafirm trade in goods by comparing the tax differential with the taxadjusted tariff rate; and (4) the MNE should attempt to minimize financial payments, and maximize tax-deductible expenses, in high-tax locations through financial transfer pricing, if the goal of the enterprise is to maximize global

308 Transfer Pricing and Taxation after-tax profits. We come back to these conclusions in the next chapter when we explore the impact of tax differentials and tariffs on Canada-U.S. intrafirm trade. Tax Penalties for Transfer Price Manipulation In this chapter we have developed several cases in which the MNE has an incentive to manipulate transfer prices. Transfer price manipulation (TPM), as we have used the term so far in this chapter, means to set a price different from the shadow price of intrafirm trade in order to reduce tax and/or tariff payments to the government. TPM occurs because the multinational increases its global net-of-tax-and-tariff profits by over- or underinvoicing its transfers and expanding the volume of intrafirm trade. Governments historically have been concerned about transfer prices for just this reason, that is, the ability of MNEs to alter their prices so as to reduce their overall tax and/or tariff costs. So far we have explored one way in which governments can affect the incentives for TPM: the regulated price W. By setting a floor or ceiling on the MNE's transfer price, governments constrain the tax and/or tariff savings that are available to the enterprise from over- or underinvoicing. Whether or not the constraint is successful depends on (1) the government's ability to predict which way the MNE will want to move the transfer price (higher or lower), (2) whether W is set correctly (i.e., in the right direction), and (3) whether or not the constraint binds (i.e., the MNE would have preferred to set p outside the regulated price).32 In this section we look at another way to reduce TPM. The United States has recently introduced penalties for misstatement of transfer prices; that is, if the transfer price lies too far outside the regulated price a penalty is levied on the differential. This is a different view of TPM than the one we have been using. In our economic models in this chapter, TPM occurs when the profit-maximizing transfer price p* is different from the shadow transfer price X. The federal government, however, defines TPM as a transfer price declared by the MNE for tax purposes or when paying customs duties (this will be either the profit-maximizing price p* or the money transfer price w, depending on whether the MNE keeps one or two sets of books) that is different from the price determined by the regulator (i.e., that is different from the regulated transfer price W). So the regulator looks at the gap between p* and W (assuming one set of books), not between p* and A,. Let us call the regulator's view of TPM regulatory transfer price manipulation (RTPM) and the economist's view of TPM economic transfer price manipulation (ETPM). Clearly, RTPM and ETPM will only be the same if W = A, and

Taxing Multinationals in Theory

309

if the MNE keeps one set of books. Since the regulatory concept is the arm's length standard, the key question is: Will the MNE, in the absence of taxes and tariffs, select the arm's length price? We know that the answer to this question depends on the answers to three subquestions, that is, whether: (1) an exact comparable exists in the open market, (2) the related parties are free to buy or sell on the external market, and (3) there are no interdependencies in demand or supply within the MNE. If the answer to all three questions is yes, then the Hirshleifer rule says that the MNE will choose the external price. In this case the arm's length price is the efficient shadow price and W = A. When the MNE reacts to a tax or tariff by moving the profit-maximizing price away from A,, the amount of ETPM is exactly the same, and in the same direction, as the amount of RTPM. However, in practice, we would not expect this to occur. Exact comparables seldom occur in real life; even in the case of undifferentiated tangibles such as crude oil and wood pulp, MNEs and tax authorities are forced into making adjustments to an inexact comparable to estimate a transfer price. In the case of differentiated goods, group services, and intangibles, the chance of finding a good comparable is even less likely. Therefore the regulator's price may be far away from the free trade shadow price A, and the transfer price manipulation the tax authority is trying to measure will be quite different from that an economist would be estimating. We first quickly summarize the U.S. rules in section 6662 (see Chapter 9 for more details). A substantial valuation misstatement (SVM) occurs if the transfer price p is 50 per cent or less, or 200 per cent or more, than the regulated transfer price (W); a tax penalty of 20 per cent of the amount of estimated underpayment of tax is due in this case. A gross valuation misstatement (GVM) occurs if p is 25 per cent or less, or 400 per cent or more, than W; if this occurs, a 40 per cent penalty is levied on the amount of underpaid taxes. We model this as follows. Let a be the penalty rate levied by country 2 on the underpayment of taxes. Under section 6662, a equals 20 per cent for an SVM and 40 per cent for a GVM. We ignore these complications and simply assume a is the same percentage regardless of the size of the misstatement or its direction. Let the percentage gap between p and W that triggers the penalty be $. We assume (j) is some fixed percentage and does not vary with the size of the misstatement or its direction (e.g., the penalty applies if the transfer pricing gap is plus or minus 50 per cent of the regulated price). The total penalty amount (call it ^) is a times the additional tax owed by the MNE, which is a times t2 multiplied by the absolute value of the change in the transfer price l(p - W)l times the volume of intrafirm trade X. Thus the penalty is:

310 Transfer Pricing and Taxation

which is positive if the absolute value of the transfer pricing gap p - W equals or exceeds the percentage § of the regulated transfer price. In order to model the impacts of penalty regulation on the MNE's choice of transfer pricing policy, it is critical to know whether or not ETPM and RTPM are the same measures. We make the following assumptions to simplify the analysis. Since there is no outside market, the internal shadow price X is the profit-maximizing transfer price for the MNE in the absence of taxes or tariffs. The regulatory authorities accept this transfer price as the arm's length price, so A, = W. Therefore ETPM and RTPM are one and the same in this model. Now assume we have a horizontally integrated MNE in which both governments levy taxes on their own firm's pure profits on a source basis; assume firm 1 is the exporter. Note that this model is similar to that in equations (51-3), but without the tariff. The MNE's objective function with the penalty is:

where a is positive only if l(p — W)l > t t , or underinvoice if t2 < tj. With the penalty, however, the incentive to manipulate the transfer price is reduced if the manipulation moves p sufficiently (higher or lower than W so as to trigger the penalty. For example, suppose t2 is 40 per cent, tj is 34 per cent, and a is 20 per cent. Since t2 exceeds tj, the MNE would like to overinvoice p, in the sense of charging a transfer price higher than the free trade marginal cost of the exporting firm (the shadow price X). However, if doing this triggers the penalty, the MNE must pay an additional eight per cent tax (20 per cent times 40 per

Taxing Multinationals in Theory

313

cent). Thus the gain from transfer price manipulation is reduced by the amount of the additional tax. We show this in Figure 6.3 by shifting the after-tax MRX curve downward to reflect the smaller (or possibly zero) net gain to the MNE from transfer price manipulation. Thus the penalty can be an effective way to reduce incentives to over- and underinvoicing. We turn now to another proposal that would reduce the multinational's incentive to manipulate transfer prices: unitary taxation. Transfer Pricing and Unitary Taxation Unitary taxation is taxation of the worldwide income of a unitary business. A unitary business consists of all the related affiliates of an enterprise that do business within the taxing jurisdiction. For example, if the jurisdiction is California, and one affiliate of IBM is located in California, all the related affiliates of IBM could be considered as a unitary business and the worldwide income of IBM taxed by the state of California. Normally, unitary taxation is based on a formula apportionment or worldwide combined reporting method whereby California IBM's share of certain factors (e.g., employment, sales, capital stock - the socalled 'apportionment factors') as a percentage of the worldwide IBM amount of these factors, however weighted, multiplied by the total worldwide income of IBM, is used to compute the tax to be paid by IBM to the state of California. Separate accounting, on the other hand, defines the borders of a firm (a permanent establishment) according to national boundaries, the so-called 'water's edge.' Domestic affiliates are consolidated with the parent for tax purposes (as are foreign branches), but foreign subsidiaries and other affiliates of the MNE are treated as separate firms. Transfer price rules are used to ensure that such transactions approximate arm's length prices. In this section we investigate the implications of unitary taxation for transfer pricing. We concentrate on the case in which one jurisdiction adopts a formula apportionment approach, and the other jurisdiction does not, and examine the resulting distortions and opportunities for transfer price manipulation.33 Setting Up the Model Assume the MNE is a horizontally integrated multinational consisting of two firms that share joint overhead costs, where firm 2 exports an intrafirm traded good to firm 1. The pre-tax global profit function of the MNE is:

314

Transfer Pricing and Taxation

where 71 is pre-tax global MNE profit, R^Y;) is total revenue of firm i from sales Yj, pX is the value of intrafirm trade, WjLj is the wage bill in firm i, and F is overhead costs (i = 1, 2). We assume, for simplicity, that all costs are labour costs; therefore, total cost Q equals WjLj for each firm i. Assume country 2 follows the water's-edge principle and taxes only profits arising in its jurisdiction, whereas country 1 applies formula apportionment to the worldwide income of its residents. Assume the ratio used to determine the share of MNE profits taxable in country 1 is firm 1's share of worldwide labour costs of the MNE.34 The MNE's objective is to maximize its after-tax global profit function:

where n* is post-tax MNE global profit and a} is the tax-deductible share of overhead costs allocated to jurisdiction i where (Xj + cc2 = 1. Firm 1's taxable income is calculated as worldwide MNE income, net of expenses, multiplied by the labour factor ratio P,35 the weighting factor used to determine firm 1's taxes in country 1, where:

Thus the first line in (74) represents pre-tax global profits of the MNE (re), the second line the taxes paid by firm 1, and the third line the taxes paid by firm 2. The Various Transfer Pricing Choices What should the MNE do to maximize its after-tax global profits in these circumstances? It is sufficient to look at L1? L2, and X in order to determine optimal output, sales, and trade volumes. Looking first at the national factor markets, where the two firms hire units of labour, we differentiate (74) with respect to L i? recalling that Y, = Q, + X and Y2 = Q2 ~ X:

Each firm should hire units of labour up to the point where the marginal reve-

Taxing Multinationals in Theory

315

nue product of labour36 (MRPLi) equals the wage rate Wi? both measured on an after-tax basis. Putting the two equations together, we have:

Equations (76, 77) say that the MNE as a whole should allocate labour between the two firms such that the after-tax marginal revenue product, net of the wage rate, is equalized between the two firms. Note that the tax rate for firm 1 is fit], whereas the tax rate on firm 2's profits is t2 + (3tj. The reason for this is straightforward. All profits wherever earned are taxed at (3tj in country 1; in addition, profits earned by firm 2 are taxed in country 2 at rate t2. Unless country 2 credits country 1's tax, or vice versa, this double taxation of MNE profits on intrafirm trade persists as long as p > 0. Therefore the effective tax rate is higher on firm 2's profits, by the amount t2, as long as country 1 practises unitary taxation. Since (3 is a fraction, it can lie either above or below tj, depending on firm 1 's share of total MNE labour income. Under separate accounting, the MNE would have paid taxes of tjTij to country 1; under unitary taxation the MNE pays t,P(Jij + 7C2). Thus, when country 1 moves from separate accounting to unitary taxation, the tax rate on firm 1's profits falls (as long as p < 1), but country 1 now taxes firm 2's profits at the rate tjp\ which is positive as long as p > 0. Whether the effective tax rate on firm 1's profits (i.e., the total tax divided by firm 1's profits) rises or falls depends on whether (3, the apportionment factor, is larger or smaller than ^/(KJ + 7i2).37 This situation has two effects. First, as long as the MNE has the choice between locating in country 1 or country 2, the firm will not choose to locate in the unitary tax jurisdiction because doing so in effect opens its other MNE affiliates up to taxation by that jurisdiction, since the 'water's edge' is ignored. Unless either taxing authority is willing to provide a full foreign tax credit for the double taxation of firm 2's profits, the MNE's overall tax costs rise. Second, if the MNE is already located in country 1, and therefore is paying unitary taxes, the enterprise will shift activities out of the other jurisdictions to avoid the additional taxation by country 1 unless country 2 credits the unitary taxes or country 1 credits the taxes already paid in country 2 on 7i2.38 Turning to the effects on intrafirm trade, we differentiate (74) with respect to X, the volume of intrafirm trade:

316 Transfer Pricing and Taxation FIGURE 6.4 Transfer Pricing under Mixed Systems: Unitary Taxation and Separate Accounting

which says the MNE should balance firm 1's after-tax net marginal revenue from imports, (1 - ptj)MR x , against firm 2's after-tax net marginal cost of exports (1 - Ptj -t2)MCx-t2p), or:

If MCX = A, (the transfer price is set equal to the shadow price on intrafirm trade), equation (80) collapses to MRX = MCX and the volume of intrafirm trade does not change from its pre-tax level. We illustrate the impact of setting the transfer price in Figure 6.4, which is based on the middle graph in Figure 5.1. In the pre-tax situation, the MNE chooses the volume of intrafirm trade where MRX = MCX = p at point a in Figure 6.4 (also labelled as point v in Figure 5.1. When unitary taxes are levied on firm 1, and separate accounting on firm 2, the MNE reacts by setting the after-tax marginal cost of exports to the after-tax marginal revenue from imports, as in equation (70). The MRX curve rotates downward to (1 - (3t!)MRx, and the MCX curve rotates downward to (1 - fkj - t2)MCx; these are the ad valorem tax effects. This intersection is labelled point b. The marginal cost curve then shifts upward by t2p; this is the specific tax effect.

Taxing Multinationals in Theory

317

Assume initially that p is set equal to MCX = X so the MNE absorbs the tax. The new equilibrium is at point c with the same output as before, X0. The after-tax global MNE profits are the triangle dee.39 These are the first-order conditions for an after-tax profit maximum; however, the MNE can affect its overall tax payments in three ways: changes to (1) (3, the weighting factor in the unitary tax formula, (2) oc2, the share of overhead costs allocated to the country using the separate accounting approach, and (3) p, the transfer price. First, if the MNE can affect p, the weighting factor should be set as low as possible, as can be seen from the equation below, where we differentiate (74) with respect to (3 and use the envelope theorem:40

Therefore, if the MNE can reduce the factor ratio used to determine its effective tax rate, overall MNE profits are higher.41 If the MNE can affect the allocation of the fixed costs F between the two countries, the MNE should set oc2 as high as possible:

When one government gives a tax deduction for an affiliate's share of overhead expenses and the other government does not, it makes sense to maximize the affiliate share in the country with the deduction. Lastly, the MNE should set its transfer price p as low as possible:

Manipulation of the transfer price p no longer affects the taxes paid in country 1 (since (3tj applies to pre-tax profits for the MNE as a whole), but still affects the taxes paid in country 2. Since firm 2 is the exporter, any income it makes from intrafirm trade is taxable. Therefore the MNE should minimize the transfer price to reduce its overall tax bill. This is clear from Figure 6.4, where the lower is p the lower the effective tax on the MNE and the greater the after-tax profits. Underinvoicing is shown as the new price p* < X, causing a downward shift in the after-tax marginal cost of exports curve; this causes an expansion of X to X*, and an increase in after-tax profits represented by the shaded area ecfg. Thus, where one country taxes on the basis of formula apportionment, while the other country or countries follow traditional separate accounting methods, there are still ways in which transfer price manipulation can be used to reduce

318

Transfer Pricing and Taxation

MNE tax payments. Would unitary taxation work if all countries used this approach? Our immediate reaction is to say 'yes.' We show below, however, that there is still at least one 'slip between the cup and the lip' that allows MNEs to avoid paying taxes under a global unitary tax sys*tem. Global Unitary Taxation Assume both governments follow a unitary tax approach and that they define and measure the MNE's global pre-tax income identically and accurately as n, and each government taxes a share P; of the worldwide income of the MNE where the country allocation factors sum to unity so that P t + P2 = I.42 Assume initially that the tax rates t, and t2 differ. The MNE's after-tax global profit function is:

Since {32 = 1 - P 1? we can rewrite (84) as follows:

The first-order conditions are straightforward (and are left to the reader). With all profits wherever earned taxed at the same rate, the MNE simply absorbs the tax and does not change its output, sales, or trade volumes. As a result, there are no resource allocation effects, or deadweight losses, imposed on the MNE; the tax is completely neutral. This is one of the key benefits argued by the proponents of unitary taxation (see Chapter 12). The MNE, however, still has some ability to manipulate its tax payments. Differentiating (85) with respect to Pj, p, and a;, we have:

which is positive (negative) if t2 exceeds (is less than) tj, thus, the MNE should raise (lower) P, whenever country 1's tax rate is less (greater) than country 2's rate. Therefore differences in tax rates can still be exploited by manipulating the factor allocation ratio at the national level, reducing it in high-tax countries and raising it in low-tax countries. The allocation of fixed costs between the two countries however makes no difference to the total tax paid, as long as the costs are deductible in both countries; see equation (87) below:

Taxing Multinationals in Theory

319

Lastly, the transfer price p also no longer affects total tax payments:

The MNE therefore has its degrees of freedom significantly reduced, but not eliminated. In practice, where formula apportionment is applied, generally a three-factor formula is used. That is, the multi-factor ratio for country i (Fj) is an arithmetic average of all the factors from 1 through n, defined by:

where Fni is factor n in country i and w ni is the weight (0 < wni < 1) attached to Fni. Once F; is calculated, the MNE's pre-tax profit in jurisdiction i is estimated as Fj times total MNE profit, or Oj = Fj n. The taxes paid in jurisdiction i are then determined as the actual tax rate multiplied by the estimated MNE profit. As before, FIj, the estimated profit, may be greater or less than n-r the profit actually declared by firm i. The Fj formula makes it clear that MNEs can manipulate their taxes in several ways, such as: (1) misrepresenting the size of variables that carry a high weight in the formula, (2) physically moving high-weight activities out of hightax jurisdictions, and (3) lobbying governments to reduce the weights and/or change the factors in the formula. The only way to eliminate transfer price manipulation is for all governments to use exactly the same formula, applied to all global income sources, at the same tax rate - an unlikely occurrence at best. Conclusions This concludes Chapter 6 on the theory of taxing multinationals. The chapter developed a general microeconomic theory of transfer pricing behaviour by multinational enterprises in response to taxes and trade barriers. We showed that a profit-maximizing MNE will attempt to arbitrage the imperfections in product and factor markets induced by government regulations, such as tariffs, profit and corporate income taxes, and minority shareholder requirements. The models explained the MNE's choice of transfer pricing policy, both for intrafirm trade in tangibles and for financial manoeuvres such as dividend repatriation and head office fees. We now turn to the empirical work that has been done on taxing multinationals, focusing in particular on the tax treatment of MNEs in North America. We will see that there is some evidence supporting the theoretical predictions of the above model, but that the data are, for the most part, inconclusive.

7

Taxing Multinationals in Practice

Introduction How much taxes do multinationals pay? How often do they engage in transfer price manipulations designed to reduce their tax bills? How much tax revenues are lost due to transfer price manipulation? In this chapter, we address these issues in four parts. First we review the relevant literature on transfer price manipulation (TPM) in response to tax differentials. Second, we examine the recent (and ongoing) U.S. debate over the alleged tax abuse by foreign multinationals in the United States. Third, we provide additional light on the U.S. tax debate, using our own estimates of U.S. and foreign taxes paid by multinationals. Fourth, we analyse the incentives to manipulate transfer prices due to income tax differentials within North America, including Mexico in our analysis. Transfer Price Manipulation: A Literature Review The belief that multinationals manipulate transfer prices so as to minimize tax and/or tariff payments is widespread. And, indeed, there is evidence that this has occurred, although the evidence, as Dunning (1993, 516) notes, is 'largely fragmentary, circumstantial, and highly industry and/or country specific'. Only a few studies review the literature in this area. For example, UNCTAD (1978, ch. IV) surveys several early industry and country studies. Dunning (1993, 512-25) provides a more recent review. However, most empirical work on transfer pricing has been written in the form of journal articles or doctoral dissertations.1 The empirical work that has been done in this area can be grouped by industry, country, and type of intrafirm transfer (good, service, intangible). In this section we provide an overview of the empirical work, organized into

Taxing Multinationals in Practice

321

four groups: country studies, cross-country studies, industry studies, and studies of fiscal transfer pricing. This breakdown is for convenience; in fact, many of the studies cross more than one category. Country Studies In terms of particular countries, by far the most work has been done on the United States since it has the best available data on intrafirm trade. A list of the studies using U.S. data include Altshuler and Newlon (1991), Benvignati (1985), Bernard and Weiner (1990), Eden (1988b, 1991b), Grubert et al. (1991), Grubert and Mutti (1991), Harris et al. (1993), Hines and Hubbard (1990), Hines and Rice (1994), Horst (1977), Jenkins and Wright (1975), Kopits (1976a), Mathewson and Quirin (1979), Mutti (1981), Oneal and Strength (1992), Pak and Zdanowicz (1994), and Slemrod (1990). The U.S. studies can be further subdivided into those that focus on transfer pricing by foreign affiliates (FAs or FCCs) within the United States and those that focus on pricing behaviour by U.S. majority-owned foreign affiliates (MOFAs) abroad. We first look at studies of transfer price manipulation in developing countries, in particular at Colombia, Brazil, and the group of southeast Asian countries (ASEAN). In terms of the U.S. cases, we restrict our analysis to Pak and Zdanowicz (1994), but come back to other U.S. studies of TPM later in this review. Transfer Price Manipulation in Developing Countries Most of the non-U.S. research on transfer price manipulation has been done on developing countries. Hines and Rice (1994) have looked at tax havens, Ellis (1981) at Central America, ESCAP (1984) at Thailand, Lecraw (1985) at ASEAN, Natke (1985) at Brazil, Vaitsos (1974) at Colombia, and Lall (1973) and UNCTAD (1978) at developing countries in general. In this section, we briefly review the TPM studies of Colombia, Brazil, and ASEAN. Transfer Price Manipulation in Colombia. The Colombia study is among the oldest empirical work on transfer price manipulation (Vaitsos 1974, Lall 1973). In 1967, Colombia imposed foreign-exchange controls. Because the government was worried that MNEs would overinvoice imports into Colombia as a way of evading the exchange controls, the customs authorities collected data on prices in the pharmaceutical, rubber, chemical, and electrical industries over the 1967-70 period (reported in Lall 1973). To calculate the amount of over- or underinvoicing, the government used prices for comparable commodities paid by locally owned firms, by other Latin American countries, and on world markets, taking an average of the available price quotations and adding in transport

322 Transfer Pricing and Taxation costs and a 20 per cent margin for error. Compared with the average benchmark price, transfer prices for certain imported Pharmaceuticals were overinvoiced by 87-155 per cent, rubber products by 44 per cent, chemicals by 25 per cent, and electrical components by 54 per cent. The conclusion was that foreign MNEs were using transfer price manipulations to evade Colombia's foreignexchange controls, and the government responded by taking the firms to court and levying fines on them. Transfer Price Manipulation in Brazil Paul Natke's 1985 study looks at import pricing of 127 products by 141 foreign and domestic manufacturing firms in Brazil in 1979 (Natke 1985). During this period, Brazil had extensive regulations on MNEs, including price controls, profit repatriation restrictions, tariffs, credit controls, and high taxes on profits. Natke's study assumes that prices paid by domestic firms reflect world, arm's length prices, and then uses these prices as a benchmark for comparison with import prices paid by foreign firms. Using a paired means test, he finds that foreign MNEs paid more for imports than did domestic firms, with the overinvoicing ranging from 21 to 39 per cent; foreign prices were also more variable than domestic prices. The results support the hypothesis of transfer price manipulation to avoid government regulations. Natke qualifies this conclusion by noting that these price differences could have been due to factors other than transfer price manipulation; for example, the MNEs' imported goods could have been of higher quality than the domestic imports. However, government regulators must 'make their decisions based on the prices they see' (Natke 1985, 221). Transfer Price Manipulation in ASEAN. The 1985 study by Donald Lecraw is based on interviews of managers of 153 foreign subsidiaries in ASEAN in 1978 about their pricing policies. The firms were in six light manufacturing industries, with 111 parents headquartered in four locations: the United States, Japan, Europe, and developing countries. Intrafirm trade was used extensively by the U.S. affiliates (68 per cent of subsidiary exports and 53 per cent of subsidiary imports) and the Japanese affiliates (79 per cent of exports; 84 per cent of imports) trading within the MNE family (Lecraw 1985, 226). Only for LDC affiliates were the percentages below 50 per cent (23 per cent of exports; 37 per cent of imports). The majority of subsidiaries reported that they used nonmarket prices for intrafirm transactions (both imports and exports), whereas external transactions were conducted at market prices or on a cost plus basis. This varied by nationality. U.S. affiliates used market-based prices for 68 per cent of intrafirm transactions, whereas Japanese affiliates were the lowest at 25 per cent. The average

Taxing Multinationals in Practice

323

for all subsidiaries was 42 per cent of transactions using market-based prices (Lecraw 1985,231). The subsidiaries apparently also had more control over setting export prices than import prices, and for pricing external transactions compared with intrafirm trade. The parent firms set prices for intrafirm exports for 55 per cent of respondents (45 out of 82 firms), and for external exports for 33 per cent of the cases (31 out of 94 firms). In the remaining cases the subsidiaries set the prices. The numbers for imports were quite different: the parent firms set prices for intrafirm imports in 79 per cent of the cases (112 of 142 firms), but prices for arm's length imports in 47 per cent of the cases (71 out of 152 firms). To explain these findings, Lecraw used discriminant analysis to explain the choice of market versus nonmarket prices. He concluded tariffs, relative tax rates, price and foreign exchange controls, and country risk were significant variables explaining pricing behaviour. The presence of a local joint venture partner tended to reduce the use of nonmarket-based prices. Lecraw concluded: The results of the analysis in this paper support the position often taken by host governments in LDCs and industrialized countries alike that MNEs engage in a widespread and systematic use of transfer prices that differ from market prices to increase their global profits, reduce risk, move funds across national boundaries and allocate them between subsidiaries. (Lecraw 1985, 238)

In sum, these three studies suggest that MNEs with foreign affiliates in host countries can and do engage in transfer price manipulation to avoid foreignexchange controls, tariffs, and corporate income taxes. Pak and Zdanowicz (1994): Abnormal U.S. Trade Prices Pak and Zdanowicz (1994) begin by stating that, in their view, 'transfer pricing is the most significant international compliance issue currently facing the U.S. government, foreign governments, and business firms engaged in international trade' (50). They note, however, the available methods for determining evidence of TPM are poor. At present, tax auditors can use two techniques: random audits of corporate income tax forms and character-based audits whereby firms with certain characteristics (e.g., firms in the pharmaceutical industry with affiliates in tax havens) are audited. Pak and Zdanowicz suggest a third technique: transactions-based audits, that is, an examination of actual international trade transactions using the U.S. Merchandise Trade Base produced by the Department of Commerce. The U.S. Merchandise Trade Base, available on 24 CD-ROM disks, contains information on over 18 million import transactions and 13 million export trans-

324

Transfer Pricing and Taxation

actions for years since 1989; it is updated monthly. The database contains 10-12 gigabytes2 of information on price and quantity data on imports and exports at the ten-digit harmonized commodity code (HCC) level,3 insurance and freight costs, means of transport (air or sea), exporting country, and port of entry. Pak and Zdanowicz use this database to estimate under-reported taxable income from exports and imports. Since underinvoicing of export revenues and overinvoicing of import costs reduce the tax base, by estimating the amount of under- and overinvoicing, and multiplying by the CIT rate, the authors can estimate the total tax loss. They find that the U.S. government lost total tax revenues of US$28.6 billion in 1992 and US$ 33.1 billion in 1993 due to unreported taxable income. Their assessment of unreported taxable income is calculated as follows. For each ten-digit HCC classification, they calculate the average U.S.-Country price, separately for exports and for imports. They then look for outliers, prices that are more than 50 per cent above the average price (for imports) or 50 per cent below the average price (for exports). Selecting only these outlier or abnormally priced cases, they sum the dollar value of these transactions and multiply by the CIT rate, ending with an estimate in the $30 billion range.4 Unfortunately, Pak and Zdanowicz do not have data on which transactions are intrafirm and which are between unrelated parties. As a result, they cannot identify what proportion of the estimated tax loss was due to transfer price manipulation per se versus other motivations. Nor do the authors provide evidence on the other side, that is, the dollar amount of overpayment of taxes based on the sum of the values of abnormally priced transactions at the other end of the distribution (i.e., those transactions in which the export price is 50 per cent above the average and the import price is 50 per cent below the average). Thus it is difficult to tell whether or not their analysis is useful in terms of the transfer price manipulation debate. In order to answer this question properly, the authors would have needed accurate information on the percentage of intrafirm trade by country and by commodity. The authors also note that the data can be used to provide CUPs, that is, inexact comparable, for specific ten-digit products. We explain how this can be done as follows.5 First, one chooses a product with a particular HCC code. This, in effect, standardizes for certain product characteristics since products with the same ten-digit code have certain identical characteristics. By also controlling for exogenous factors that would affect the product price (e.g., transport and insurance costs, country of export, date of export), it is possible to determine an average estimated price for all transactions that have the same values of the exogenous variables. That is, one can predict the product price using an equation such as the following:

Taxing Multinationals in Practice

325

PRICE = CONSTANT + IB^ PRODUCT CHARACTERISTICS + B2 SHIPPING + B3 INSURANCE + B4 EXPORTING COUNTRY + ... + Bn OTHER (1) where the E{ (i = 1, ..., n) terms are the coefficients on the exogenous variables that affect product price and B^ (j = 1, ..., m) are the various characteristics specified in the HCC code. We can visualize the procedure one could use to calculate a CUP, using equation (1) as follows. First, multiple regression analysis is used to estimate the B; coefficients in equation (1). Second, simple average values of the exogenous variables are calculated. Then the estimated values of the B; coefficients are multiplied by the average values for the exogenous variables, and these totals are summed. This gives an estimated average price that should apply to all transactions that 'fit' these characteristics (e.g., product characteristics as determined by the HCC code, date, insurance and shipping costs, country of export, and so on). This price is the 'benchmark' price. The benchmark price is illustrated in Figure 7.1. This figure shows a simple OLS regression of one exogenous variable X (for example, the sum of freight and insurance costs) against the endogenous variable, product price P, for a particular HCC code. The regression equation is of the form P = A + BX, where A is the constant term and B is the estimated coefficient. The asterisk points on the graph are various values for pairs of (X, P) combinations. The straight line drawn through the (X, P) pairs is the fitted regression equation that minimizes the sum of the squared differences between the points. Assume that the result is P = 10 + ViX. Then for each particular value of X, it is possible to calculate the benchmark price using this formula; for example, for X equals 30, the benchmark price is $25. This benchmark price can then be treated as an inexact comparable in terms of estimating a transfer price. That is, given particular values for the exogenous variables (the X variables in equation [1]), one can calculate the benchmark price corresponding to these variables for a particular HCC code. Thus the method can produce statistical evidence that could be used to support the CUP method, either by MNEs or by tax authorities, for a particular product. The 1994 U.S. final section 482 regulations allow the use of statistical comparisons if the data have been adjusted to ensure that the transactions are comparables (see Chapter 8), so this method could be legitimately used for corporate income tax purposes.6 In sum, one purpose to which this database could be put could be to provide evidence on a benchmark price, that is, on the price of an inexact comparable. This method can also be extended to identify abnormally priced import and

326 Transfer Pricing and Taxation FIGURE 7.1 Benchmark Prices and Outliers in a Regression Analysis

export transactions for particular products; that is, to look for outliers, product prices that significantly deviate from the average price that would have been predicted based on this transaction's specific characteristics. To do this, one would compare the benchmark price for a particular ten-digit product with the actual price for the transaction. By comparing the predicted average price with the actual price, one can isolate residual differences in prices, that is, differences that are not due to the characteristics of the product specified in the HCC code or to the specified exogenous variables. Outliers, as defined by this method, could be transactions with prices that are more than 50 per cent above or below the benchmark for a product with these characteristics, although other percentages (e.g., 25, 75, or 90) could be used. In terms of Figure 7.1, one is looking for P values that lie vertically above or below the regression line by some selected deviation from the line, for different values of X. Parallel lines above and below the regression line in the figure are used to show the 'range' or band within which the prices are considered to be 'normal'; prices outside these upper and lower bounds are outliers. The outliers

Taxing Multinationals in Practice

327

are circled in the figure; further investigation can be done to determine whether or not these are truly abnormal prices. Summing the value of all such transactions for a particular HCC code gives the value of merchandise trade that is abnormal. By summing the values of all such transactions for all ten-digit code numbers, it is possible to get an estimate of the total size of the outliers on both the merchandise export and import sides of the U.S. balance of payments. These statistics can then be used to calculate the total amounts of over- and underreporting (and thus of net over- or under-reporting) of income from merchandise trade transactions for tax and customs duty purposes. To be useful for estimates of transfer price manipulation, however, it is necessary to have evidence on which transactions are between related parties and which are not, and then to use this evidence to test for evidence of statistically significant differences between the related and unrelated party transactions.7 In the absence of such information, it is not possible to argue that these estimates are TPM estimates. Cross-Country Studies A few studies compare the effects of corporate income tax and tariff rates on MNE incentives to manipulate intrafirm commodity prices and/or alter trade and investment decisions.8 These studies are the empirical counterpart to the tariff-tax theoretical models we developed in the previous chapter. Some of the studies include Eden (1988b, 1991b), Feldstein, et al. (1995), Grubert and Mutti (1991), Horst (1977), Mathewson and Quirin (1979), and Wilson (1993). In this section we review Mathewson and Quirin (1979). Mathewson and Quirin (1979): Canada-U.S. Trade In the late 1970s, Frank Mathewson and G. David Quirin, two economics professors at the University of Toronto, wrote a monograph on tax transfer pricing for the (now-extinct) Ontario Economic Council. The book, Mathewson and Quirin (1979), reviews previous work on transfer pricing, discusses legal issues, outlines the Canada-U.S. tax and tariff situation, and also discusses the implications of a value-added tax for transfer pricing. Mathewson and Quirin develop a theoretical model of MNE transfer pricing behaviour in response to tax and tariff rates, and then use a pseudo-empirical technique to apply the model to MNEs in Canada and the United States. The model assumes a vertically integrated, two-division MNE operating under constant returns to scale in production with perfect competition in the final market. Using numbers based on average Canada-U.S. data to proxy for the representative MNE's labour and capital shares, sales revenues, and output volumes, the

328

Transfer Pricing and Taxation

FIGURE 7.2 The Impact of a U.S. Tariff on Canada-U.S. Location Decisions

authors simulate the impacts government regulation would have on the enterprise's location and transfer pricing decisions. The authors assume nine stages in the value chain, sequenced so that average cost rises monotonically by stage. The goal of the MNE, given that the final product price is fixed, is to minimize total overall cost of production. The relevant question therefore is which stages should be located in which country? The authors solve the model for the free trade case and then study the effects of Canadian and U.S. taxes, tariffs, and transfer pricing regulations on the allocation of production between the two countries, using numerical percentages as proxies for the Canadian and U.S. rates. We illustrate their model in Figure 7.2. In Figure 7.2, stages of production are arranged along the X axis with downstream stages being further along the axis. Unit production costs are on the vertical axis. Both average cost curves, COSTCAN and COSTUS, slope upwards, reflecting rising per-unit costs at downstream stages. Assume that unit costs of production are lower in Canada than in the United States at the upstream stages

Taxing Multinationals in Practice

329

of production, but that this differential is eliminated and, in fact, switches at the downstream stages. The figure therefore shows that a cost-minimizing firm would allocate the first five stages to Canadian plants and the downstream four stages to U.S. plants in the absence of taxes or tariffs, as determined by the intersection between the two unit cost curves (point a). Now assume the U.S. government levies a tariff on imports from Canada. This tariff raises the effective cost of production in Canada, shifting up the COSTCAN curve to COSTCAN plus the U.S. tariff.9 The new intersection is at point b, implying that it is now more expensive to produce in Canada. The firm therefore shifts stages 3,4, and 5 from the Canadian to the U.S. plant. To the extent that a multinational is allowed to manipulate transfer prices (i.e., if the tariff is ad valorem and customs regulations are not strict), the MNE can avoid the tariff by underinvoicing its Canadian exports to the U.S. downstream plants. This is equivalent to shifting the COSTCAN + U.S. TARIFF curve downward by the amount of underinvoicing per unit of exports. If tariff rates and the volume of trade are substantial, so that the impact on plant location is non-negligible, the MNE is more likely to keep stages of production in Canada than are unrelated parties which must pay the U.S. tariff. If there were corporate income taxes to consider as well as tariffs, the results are more complicated since we would be back to the situation in Chapter 6 with two taxes and a tariff. In this case, the firm must compare the tax differential, (ty,, - tc)/(l - t^), to the tariff rate, I, to determine which plant location at each stage is the least costly. In addition, the MNE must also consider whether or not transfer price manipulation is worthwhile. If (t^ - tc)/(l - t^) > I, the MNE should overinvoice its transfers; if the tax differential is less than the tariff, overinvoicing is more profitable. We conclude that there can be differences in location decisions between the related and unrelated firm cases, depending on (1) the actual levels and types of government regulations, (2) the ability of the MNE to manipulate transfer prices, and (3) the volume of crossborder transactions. In Mathewson and Quirin's empirical section, the authors conclude that the distributions of cost-minimizing locations that would be chosen by independent firms (i.e., if the affiliates were unrelated) and by multinationals, in the presence of taxes and tariffs, are approximately the same. That is, the same stages of production are allocated to each country regardless of ownership structure of the firms. The reason is that U.S. and Canadian corporate income taxes and tariffs in the two countries tend to have offsetting impacts on MNE location decisions; that is, they have a 'self-policing' impact. In terms of Figure 7.2, when both CITs and tariffs are considered, the net impact on the cost curves is offsetting.

330

Transfer Pricing and Taxation

From this the authors conclude that if either taxes or tariffs are absent, transfer price manipulation may increase and there might be differences in the related and unrelated firms' choices. Mathewson and Quirin argue that the incentive to manipulate transfer prices is particularly true for services and intangibles flows where tariffs do not apply. Tariffs are deductible costs for the importer, but not creditable against the home country CIT. Withholding taxes, which do apply to business services flows, are the fiscal equivalent of tariffs, except that such taxes are generally creditable against home country taxes. Thus services and intangibles flows may be more prone to transfer price manipulation (what the authors and Donald Brean [1985], among others, call fiscal transfer pricing) than flows of tangibles. Goods that flow duty free, such as raw materials, are also likely to be overinvoiced into, and underinvoiced out of, high-tax jurisdictions. The authors provide estimates, in an appendix, of the volume of intrafirm trade and the likely importance (high, low) of transfer price manipulation by industry. They conclude that the most likely candidates, in terms of tangibles, for transfer price abuse in Canada are forest products, crude petroleum, bauxite, and auto parts moving under the Canada-U.S. Auto Pact because tariffs are low or nonexistent, good comparables do not exist, and MNEs in these industries operate in markets that are imperfectly competitive. The book concludes that 'multinationals have a relatively restricted scope for transfer price manipulation ... [because] tax systems and tariff systems provide incentives for transfer-price manipulation which tend to be mutually offsetting' (Mathewson and Quirin 1979, 87). They caution that if Canada and the United States should move to a free trade area, the scope for transfer price manipulation will rise as crossborder tariffs disappear and crossborder trade volumes increase. Free trade, according to Mathewson and Quirin, should therefore be accompanied by a move to harmonization of Canada-U.S. tax levels.10 Industry Studies There have been several industry studies of transfer pricing. They tend to focus on transfer prices for goods, comparing intrafirm prices with those available on the external market. The best known of these studies are in industries such as bananas (Ellis 1981), copper (Lamaswala 1981), petroleum (Bernard and Weiner 1990, 1992; Bertrand 1981; Eden 1990; Jenkins and Wright 1975; Rugman 1985; Rugman and Mcllveen 1985), pharmaceuticals (Eden 1989; ESCAP 1984; UNCTAD 1978, ch. IV; Vaitsos 1974), as well as for all manufacturing (Benvignati 1985; Eden 1988b; Lall 1973; Mathewson and Quirin 1979; Natke 1985). We review the U.S. and Canadian petroleum studies below.

Taxing Multinationals in Practice

331

Transfer Pricing in the U.S. Petroleum Industry11 Jean-Thomas Bernard, a Canadian economist, and Robert Weiner, a U.S. economist, use OLS regressions to test for evidence of over-underinvoicing of crude petroleum imports into the United States during the period 1973-84 (Bernard and Weiner 1990). The data are from a petroleum industry monitoring program run by the U.S. Energy Information Administration, and include information on dates of loading and importation, exporting country, port of landing, FOB and landed prices, sulphur and gravity, credit terms, volume, and transaction type. The key variable in the analysis is 'transaction type,' which is broken down into four categories: interaffiliate imports (I), unrelated imports from host governments (H), third parties (T), and unidentified firms (U). The value of imports is the price P multiplied by the volume Q of imports for each transaction type. Transfer prices apply to the first category, interaffiliate imports (I); the other three types can be considered arm's length transactions. Therefore it is possible to test for evidence of transfer price manipulation by testing for differences between the I transactions and the others.12 The authors first use OLS regressions to isolate cases in which the gap between P(I) and P(T) is statistically significant. Their regressions are of the form:13 P(I) - P(T) = CONSTANT + B, SULPHUR + B2 GRAVITY + B3 CREDIT + B4 VOLUME + B5 EXPORTER + B6 DATE + B7 PORT + B8 TRANSACTION TYPE + B9 SHIPPING + B10 (TAXUS - TAXEXPORTER) (2) where the Bj terms are the coefficients on the exogenous variables. The results of the initial regression runs look quite good; the R2 is high and many of the variables are significant. The authors then test four hypotheses about the P(I) - P(T) price gap: • Hypothesis 1: Manipulation of interest rate charges (the CREDIT variable) can be used to reduce tax payments since interest costs are a tax-deductible expense for the U.S. importer. Therefore credit terms should be overinvoiced on the Q(I) transactions related to transactions with unrelated parties. • Hypothesis 2: The scope for transfer price manipulation fell over the period due to the rise of the oil spot market and the increased U.S. government monitoring of crude oil import prices. Therefore the price gap P(I) - P(T) should decline over time. • Hypothesis 3: Manipulation of transportation costs can be an alternative route to transfer pricing as a way of reducing taxation since such costs are tax

332

Transfer Pricing and Taxation

deductible for the U.S. refiner and represent income for the shipping affiliate. Therefore shipping costs should be overinvoiced on the Q(I) transactions relative to transactions with unrelated parties. Hypothesis 4: Differences in tax rates between the exporting country and the United States lead to substantive over-underinvoicing. Therefore the price gap should be correlated with differences in tax rates between the exporting country and the United States. We illustrate Bernard and Weiner's hypotheses about the opportunities for transfer price manipulation in Figure 7.3 below.14 Assume a downstream oil refinery buys crude oil from three sources: a foreign affiliate (I), a third-party firm (T), and a host government firm (H). (We ignore the 'Unidentified' party transactions.) Crude oil is shipped to the refinery through intermediary shipping firms. In the case of the related-party crude oil producer, the shipper is another affiliate of the MNE, whereas for the other two routes the parties are unrelated. Therefore, in Figure 7.3, the left-hand components of the value chain (the I transactions at the crude oil, shipping, and refining stages of production) are intrafirm transactions, whereas the T and H chains are at arm's length. There are at least three opportunities for transfer price manipulation (TPM) in the lefthand chain: (1) the price of crude oil, (2) the interest charge for credit terms, and (3) the shipping costs. Evidence of TPM in each case can be found by comparing the price of the I transaction with its comparables: the T and H transactions. In fact, Bernard and Weiner look for evidence of TPM by comparing only the I and T transactions. Now let us examine Bernard and Weiner's empirical results. They test hypothesis (1) first. Interest rate charges are proxied by number of credit days. Using multiple regression techniques on equation (2), the authors find that the CREDIT term for 30 days and under has a negative sign, while the CREDIT term for 60 days and up has a positive sign. This indicates that the longer the number of credit days, the higher the transactions price. However, the t-statistics are generally poor, implying that one cannot have much confidence in the results of testing this hypothesis. Bernard and Weiner then test the second hypothesis. They use the Bonferroni t test to see whether the differences between the coefficients on the T (thirdparty) and I (intrafirm) variables, by country, are significantly different from zero. In many cases there are significant differences; for example, looking only at the major crude oil suppliers, there is evidence of repeated overinvoicing through Nigeria and Indonesia and underinvoicing through Saudi Arabia. Most of the cases of apparent transfer price manipulation, however, involve small amounts of imports. In general, there appears to have been more manipulation

Taxing Multinationals in Practice FIGURE 7.3 Transfer Price Manipulation in the Oil Industry

P(I) Q(I) + C(I) + S(I) = Cost of intrafirm imports P(T) Q(T) + C(T) + S(T) = Cost of third-party imports P(H) Q(H) + C(H) + S(H) = Cost of import from governments

333

334

Transfer Pricing and Taxation

of transfer prices across the board in the 1976-81 period than in 1973-5 or 1982-4. Thus TPM apparently fell over the period, providing some support to the second hypothesis. The authors then multiply the estimated price gaps by the volume of imports from each country for each year. The sum of these numbers is compared with the total value of imports for an estimate of the amount of over-underinvoicing. The result is positive and small; there was net underinvoicing into the United States, equal to less than 2 per cent of the value of affiliate imports. (This is dominated by the large underinvoicing in 1979 by Abu Dhabi and in 1981 by Saudi Arabia [US$527 million].) This same analysis is then repeated for transport costs to test the third hypothesis. More significant differences in transport costs occur than in product prices for the large suppliers. In general, the reverse occurs: Nigeria, Libya, and Algeria undercharge and Saudi Arabia overcharges; Indonesia is mixed. The total value of net transfers is an overcharge of less than 1 per cent of interaffiliate imports (dominated by a huge overcharge by Saudi Arabia, also in 1981, of US$354 million). The final part of the paper tests the fourth hypothesis by regressing the transactions price differential against differences in tax rates on oil production income. The tax rates are proxied by income taxes paid as a fraction of taxable income. These 'average effective tax rates' are generally higher than the U.S. rate of 30 per cent so that the foreign affiliate generally has an excess of foreign tax credits which is not creditable against U.S. taxes. The regression results are very poor. The authors repeat the same technique for transport cost differentials (i.e., correlating shipping differentials with tax rate differentials) and the results improve slightly. The overall conclusion of the paper is that small amounts of transfer price and transport cost manipulations by foreign affiliates of U.S. petroleum companies appear to have occurred over the 1973-84 period. This TPM was, at best, weakly related to corporate tax rate differentials. The authors hypothesize that so little manipulation apparently occurred either because the IRS successfully enforced U.S. regulations, or because the oil MNEs engaged in tax avoidance through other channels, or because data and statistical problems confounded the results. Transfer Pricing in the Canadian Petroleum Industry15 The studies of transfer pricing in the Canadian petroleum industry date back to 1981, when R.J. Bertrand, in charge of the Combines Investigation Act, published a seven-volume report on the state of competition in the industry (Bertrand 1981). He concluded that the big oil multinationals incurred excess costs

Taxing Multinationals in Practice

335

to Canada of $12.1 billion from 1958 to 1973, which included overcharging Canadian consumers $3.2 billion, measured in 1980 dollars, for crude oil imports purchased by foreign-owned oil subsidiaries in Canada from their offshore affiliates. Bertrand calculated the petroleum transfer pricing overcharge as follows. First, he had data on the actual landed GIF prices for crude oil to Eastern Canada. From these prices he subtracted his estimate of freight and insurance costs to determine implicit FOB prices for crude oil. The gap between his estimated freight rate costs and the data provided by the petroleum MNEs provided his first source of transfer price manipulation: overinvoicing of freight charges.16 He then compared his estimated FOB prices with those available on the open market for third-party sales. The gap between his FOB prices and his estimates of arm's length prices provided a second source of TPM, based on overinvoicing of crude oil per se.17 In Figure 7.4, we illustrate Bertrand's transfer pricing calculations for the case of Arabian light crude in 1965, as representative of his calculations, which were done for many companies, many different types of petroleum imports, and several different locations.18 As the figure shows, the landed CIF price for Arabian crude in Eastern Canada in 1965 was $2.26 per barrel. The Average Freight Rate Assessments (AFRA) shipping conference rate used by the oil MNEs was 77.6 cents per barrel;19 Bertrand estimated that the true shipping costs were 59.9 cents, for an overinvoicing estimate of 17.7 cents per barrel in shipping charges. The second estimate of TPM comes from a comparison of the estimated FOB price for crude ($2.26 - 0.599 = $1.661) with the price of crude on the open market. Bertrand estimated that the arm's length price was $1.33 per barrel, creating a second TPM of 33.1 cents per barrel. The total amount of overinvoicing was therefore 50.8 cents per barrel, which is 22.5 per cent of the landed CIF price. Bertrand redid these calculations for other types of crude oil, multiplied his estimates of TPM per barrel by the number of imported barrels, and reached the conclusion that Canadian consumers had paid $3.2 billion too much for their oil imports. His estimates did not go unchallenged, however. Rugman (1985) and Rugman and Mcllveen (1985) were harshly critical of the report on several grounds, most notably the calculation of the freight rate charge (e.g., $0.599 in Figure 7.4) and of the arm's length price for crude oil (e.g., $1.33 in the figure). Rugman argued that Bertrand had underestimated the freight rate charge by using the lowest quoted shipping price for selected charters. He further argued that Bertrand had selected an arm's length price from a small and thin spot market in which prices were lower than in third-party sales. When Rugman recalculated

336 Transfer Pricing and Taxation FIGURE 7.4 Bertrand's Estimate of Overinvoicing Crude Oil Imports

SOURCE: Based on 1965 data for prices of Arabian light crude oil in Rugman (1985)

Bertrand's estimates, correcting for these and other deficiencies, he found no evidence of transfer price manipulation, concluding that the Bertrand Report makes selective use of the data available, misinterprets data and makes incorrect implications without any basis in (efficiency based) economics. Unfortunately, the unscientific nature of this work is representative of the Bertrand Report... Bertrand's analysis is ... meaningless and without any basis in fact on either theoretical or empirical grounds. (Rugman 1985, 191)

In 1986, a second study by the Restrictive Trade Practices Commission again

Taxing Multinationals in Practice

337

examined competition in the Canadian oil industry and concluded that there were excess costs, but that they were lower than Bertrand had estimated (RTPC 1986). Rugman's critical assessment of Bertrand's report also led two other economists, Jean-Thomas Bernard and Robert Weiner, to reopen the issue some years later, providing 'new evidence on Bertrand versus Rugman' (Bernard and Weiner 1992). The authors had access to an individual cargo-based data set of crude oil prices for Canadian imports between 1974 and 1984, collected under the Canadian Oil Import Compensation Program.20 Their data set provided detailed information on crude oil shipments, including oil gravity, sulphur content, number of credit days, shipment volume, monthly pricing data, and nature of seller (host government, third-party sales, affiliate sales). This data set allowed the authors to control for quality and transaction-specific data that could affect the transfer price. The authors used multiple regression analysis to calculate the difference between estimated third-party (arm's length) and affiliate (non-arm's length) FOB prices, controlling for differences in oil and transactions characteristics, and for differences in freight rate charges, much as they had done in their earlier study of TPM in the U.S. oil industry. Thus, their main equation looks like equation (2) above, but applied to Canadian rather than U.S. data. The conclusion they reached was more than a little surprising: When other differences are properly controlled for, we find that affiliates charged transfer prices that were on average lower than third-party prices, exactly the opposite of the claims by Bertrand. This means that Canada has in general been a beneficiary of transfer pricing practiced by multinational oil companies through the higher taxes paid to the Canadian government. Furthermore, the phenomenon appears to grow in importance through time. (Bertrand and Weiner 1992, 34-5)

Not only was Bertrand wrong, according to Bernard and Weiner, he had it backwards ! The two Bernard and Weiner studies are among the most technically sophisticated of the transfer pricing studies ever completed. Their results raise doubts (in this author's mind at least) about the robustness of the estimates of transfer pricing manipulation in earlier studies. Where the data are available (we note parenthetically that the paucity of good data in this area, particularly non-U.S. data, is a real and persistent problem), the lessons from the Bernard-Weiner studies are that earlier transfer pricing analyses need to be subjected to this same careful scrutiny, and that without such scrutiny earlier results should be treated with caution. There is room for optimism, however. By combining the U.S. Merchandise

338

Transfer Pricing and Taxation

Trade Data Base used by Pak and Zdanowicz (1994) with the methods used in Bernard and Weiner (1990, 1992), it should be possible to look at the question of TPM more closely, at least for U.S. merchandise trade flows. We therefore anticipate that economists will be able to provide much better evidence (for or against) transfer price manipulation in the foreseeable future. Fiscal Transfer Pricing Studies Most of the industry studies listed above focus specifically on transfer pricing of traded commodities such as bananas and copper. However, many other studies have focused on the relationship between taxation and financial transfers such as dividend repatriation, royalty payments, and management fees. Donald Brean (1985) has called these fiscal transfer pricing studies. They question the extent to which taxes affect financial flows. The studies most closely related to transfer pricing include Altshuler and Newlon (1991), Alworth (1988), Brean (1985), Grubert and Mutti (1991), Harris et al. (1993), Hines and Hubbard (1990), Hines and Rice (1994), Horst (1977), Kopits (1976a,b), Mutti (1981), and Slemrod (1990). See also the chapters in Feldstein, et al. (1995). We look at Hines and Rice (1994) and Grubert and Mutti (1991) below.21 Hines and Rice (1994): Tax Havens Hines and Rice (1994) use data from the 1982 Benchmark Survey of U.S. FDI to examine the distribution of assets, equity, and employment by U.S. MNEs among three locations: tax havens, other developing countries, and industrialized countries outside the United States. Their research asks the question: Do MNEs shift movable, taxable assets to low-tax locations? Tables 7.1 and 7.2 show some of their findings. Table 7.1 shows the number of U.S. parent firms in different industries, together with data on their affiliates worldwide. In all industries, in 1982, there were 2,245 parent MNEs with a total of 18,339 affiliates, for an average of 8.17 affiliates per MNE. Of these affiliates, just under 16 per cent were located in tax haven countries such as the Bahamas, Bermuda, and Switzerland. That percentage rises to 38.5 per cent for MNEs in the banking industry and to an astonishing 87 per cent in the shipping industry. The use of flags of convenience by the shipping industry is of course well known, and the correlation between countries with flags of convenience and tax havens is high; this explains the high incidence of tax haven affiliates in the shipping industry. Some evidence on the percentage of MNE total assets, equity, and employment located in tax havens in 1982 is provided in Table 7.2. In this table, industries are divided into three categories: banking, nonbank financial (adding these two together we have financial), and nonfinancial. In terms of the percentage of

Taxing Multinationals in Practice

339

TABLE 7.1 U.S. Multinationals and Their Tax Haven Affiliates, 1982

Industry of parent firm Banking Shipping Petroleum Nonfinancial services (excluding hotels) Manufacturing Wholesale trade Nonbank financial ALL INDUSTRIES

Number of U.S. parent firms

133 12 143

Number of affiliates worldwide

1,061

69 2,475

160

722

1,215

11,240

908

168 234

1,152 18,339

2,245

Average number of affiliates per

MNE

Percentage of affiliates in tax havens

7.98 5.75 17.31

38.5 87.0 18.2

4.51 9.25 5.40 4.92 8.17

13.4 12.3 19.4 16.8 15.8

SOURCE: Based on data in Mines and Rice (1994) TABLE 7.2 Percentage Distribution of U.S. FDI by Region, 1982 Tax haven countries

Other industrial countries

Other developing countries

Banking Nonbank financial Nonfinancial

37.5 46.0 9.7

Percentage distribution of assets 50.1 49.3 69.1

12.5 0.5 21.2

Nonbank financial Nonfinancial

56.1 9.1

Percentage distribution of equity 39.9 68.5

4.0 22.4

Banking Nonbank financial Nonfinancial

13.1 37.5 4.2

Percentage distribution of employment 48.4 38.5 50.1 12.5 67.6 28.2

SOURCE: Calculated from data in Mines and Rice (1994)

total assets that are allocated to tax havens, the highest is nonbank financial (46 per cent), followed by banking (37.5 per cent), and trailed by nonfinancial industries (9.7 per cent). Regulatory restrictions on U.S. banks are probably responsible for the lower share of total assets in tax havens, relative to nonbank financial FDI. Nonbank financial MNEs also make significantly greater use of tax havens as a location for employment (37.5 per cent) compared with banks

340

Transfer Pricing and Taxation

(13.1 per cent) and nonfinancial FDI (4.2 per cent). Comparing nonbank financial and nonfinancial FDI in terms of equity, we find that tax havens are the location for 56.1 per cent of nonbank financial equity but only 9.1 per cent of nonfinancial equity.22 This suggests that U.S. MNEs in the financial sector shift a larger share of their financial assets offshore to tax havens than do nonfinancial firms. Some additional evidence in this regard is provided by our last study. Grubert and Mutti (1991): Taxes and Investment One of the most widely cited of the fiscal transfer pricing studies is Grubert and Mutti (1991), which uses 1982 data on manufacturing affiliates of U.S. MNEs in 33 countries to examine the relationships between taxes, tariffs, and transfer pricing. Because this study examines the impact of taxation on profits, capital investment, and trade flows, using multiple regression techniques, it is broader than the industry or country studies we have covered above and more appropriately belongs with the cross-country studies. Their work has three parts. First, Grubert and Mutti test whether MNEs actually do shift taxable income to low-tax countries. Their general equation is of the form:

where tj is either the statutory or effective tax rate on corporate income; 7U*iA'i is either after-tax affiliate book income (profit) as a percentage of sales net of any purchases from the parent firm, or book income as a percentage of equity as the dependent variable; C is a constant term; %A GDP is the percentage change in gross domestic product; and i is the host country where the affiliate is located. An example of their results (Grubert and Mutti 1991, 287), using profit over net sales and statutory tax rates, is:

Equation (4) says that 60 per cent of the variance in after-tax profit rates can be explained by a constant term, the statutory tax rate and the rate of change of GDP. The tax rate is negative and statistically significant, implying that higher tax rates correlate with lower after-tax profit rates.23 For example, (6) implies that an affiliate in a host country with a tax rate of 40 per cent and a 10 per cent growth rate in GDP will, on average, report a net profit ratio of 4.26 per cent, whereas an affiliate in a host country with half the tax rate will report a profit

Taxing Multinationals in practice 341 return of 11.66 per cent, almost three times as high even though the tax rate is only one-half that of the first country.24 Mutti and Grubert next investigate the impact of taxes on investment. We argued in Chapter 6 that MNEs would allocate capital internationally such that risk-adjusted marginal after-tax returns were the same across all alternatives. This implies that effective tax rates should be negatively related to new investment decisions. Grubert and Mutti proxy this in an equation of the following form:

where all variables are reported in natural logs (so that coefficients represent percentage changes), NPE is the affiliate stock of net plant and equipment, GDP/POP is per capita gross domestic product, i is the weighted average tariff rate on manufactures, te is the average effective tax rate on equity, D is a dummy variable for distance, and P is a dummy variable for investment policy. An example of their results, for MOFAs only, is (Grubert and Mutti 1991, 289):

All the independent variables, except D, are significant in equation (6). Increases in GDP and in GDP per capita are positively related to new investment, as expected. The higher the tariff, the larger the predicted amount of net investment (this is the tariff-jumping motivation for manufacturing investment). Higher tax rates, however, cause a slower growth in new investment; in fact, Grubert and Mutti estimate that, on average, a cut in the effective tax rate from 20 to 10 per cent causes, ceteris paribus, a rise in net investment of 65 per cent (1991, 290). Lastly, the authors examine the impact of taxes and tariffs on U.S. exports to, and imports from, MOFAs, but these results are generally not significant and are not reported here.25 Weighing the Evidence John Dunning's conclusion, after reviewing many of these studies, is that 'while

342

Transfer Pricing and Taxation

each has made some contribution to our understanding of the phenomenon, it is still very difficult to form a view about the extent of TPM [transfer price manipulation], how far it differs from arm's length prices or its welfare implications' (Dunning 1993, 519). This is also our judgment based on the reviews above. It is clear that MNEs have a cost-reducing motivation for using nonmarket-based prices, since over- or underinvoicing may reduce the total tax and regulatory burden on the MNE group. However, there are costs involved in TPM, including bookkeeping charges, possible negative impacts on managerial performance, and the costs of resource misallocation if the MNE keeps one set of books. Also, it is clear that, at least over the past ten years, the incentives for manipulating transfer prices to avoid taxes and/or tariffs have been decreasing. First, general government policies have been and are being liberalized (e.g., through harmonization of tax rates, reduction in tariff rates, the formation of free trade areas, reduction in cross-border restrictions of all kinds), and this reduces the incentives for transfer price manipulation. Second, FDI policies are also being liberalized (e.g., the removal of investment barriers such as capital repatriation constraints, key sector legislation, and performance requirements). Third, tighter regulation by tax and tariff authorities (e.g., increased informational reporting, more strict regulations on transfer pricing, new penalties for over-/ underinvoicing) has raised the costs of such manipulation. On the other hand, Jeffrey Owens, head of fiscal affairs at the OECD, recently listed three reasons why MNEs still had incentives to shift profits, even though over 90 per cent of current transfer pricing disputes involve two developed, high-tax countries: First, there may be reasons other than the effective tax rate for shifting profits, for example, to move cash to other countries without paying withholding taxes or to improve financial reports to shareholders in an unconsolidated reporting system. Second, whilst tax rates in the OECD area have been reduced, there has been little tendency to convergence so that profit-shifting can still produce benefits from substantial rate differentials. Also, there are quite a few tax havens around, both within and outside of the OECD area. Thus, both the income allocation and tax avoidance aspects of transfer pricing are important. (Owens 1994, 877-8)

At the same time, the ability of MNEs to successfully manipulate transfer prices has probably risen. The volume of intrafirm trade has risen markedly, and the ways in which MNEs can engage in real and financial transfers (e.g., transfers over telephone lines and via satellite) have proliferated, making it more difficult for national authorities to regulate transfer pricing in all its many forms.

Taxing Multinationals in Practice

343

The likely expansion of inward FDI in the transition economies of Eastern Europe and Russia, and into Asian markets such as China and Vietnam, all of which have significant nontariff barriers and other government-induced market imperfections, imply that the financial returns to the MNE from engaging in transfer price manipulation in intrafirm trade with certain countries can be large. In addition, sectors in these countries that have been traditionally closed to FDI, such as telecommunications and banking where intangible assets and business services are a large part of the value-adding activities, may offer opportunities for transfer pricing that are less available in commodity-type industries. As a result, the tension between the desire to attract inward FDI and the perception that foreign MNEs may engage in tax abuse is still a real tension for national governments. Nowhere is this tension more evident than in the United States, where the U.S. tax authorities are convinced that substantial transfer price manipulation is taking place, particularly by foreign MNEs in the United States, in order to avoid paying U.S. taxes. The IRS versus Foreign Multinationals: Tax Grab or Tax Abuse? How much taxes do North American multinationals pay, and are they using transfer price manipulation to underpay their taxes? This has been a contentious topic for quite some time, particularly in the United States. In this section we investigate the U.S. evidence on this issue. Until the early 1980s the issue of foreign multinationals and their payment or nonpayment of U.S. income taxes was of interest only to the Internal Revenue Service and a few academics, tax lawyers, and accountants. The reasons are straightforward. Before 1980 there was little inbound FDI in the United States (Krugman and Graham 1992). The IRS had traditionally been concerned with U.S.-controlled MNEs and their use of tax havens for tax avoidance purposes. The government's goals were to ensure that more taxes were paid in the United States, profits were not diverted to tax havens, and that U.S. MOFAs were not unfairly taxed abroad. Most U.S. tax court cases dealt with U.S. MNEs and their pricing of outbound transfers (e.g., technology) to foreign subsidiaries. With a small presence inside the United States, foreign affiliates were generally ignored. In the 1980s, however, foreign MNEs moved onshore in large numbers and their presence became more visible. Intrafirm exports from foreign parents to their new affiliates rose, substituting for the previous arm's length sales to U.S. distributors and manufacturers. Overinvoicing of inbound intrafirm transfers became a possible way to shift profits out of the United States and into lower-

344

Transfer Pricing and Taxation

taxed locations. Foreign affiliates, setting up new operations in a foreign (to them) location, incurred substantial start-up costs, and often showed little or no profits.26 The rapid rise in the share of the U.S. market held by Japanese products (both imports and products of transplants) raised another issue: Were the firms underinvoicing final sale prices to U.S. consumers to increase their market share, with the losses being financed by the 'deep pockets' of their Japanese parents? Were the losses just due to the costs of new start-ups? Japanese MNEs argued that their FCCs were running short-run losses in order to penetrate the U.S. market (which was consistent with the longer-run focus in Japan on market share rather than the U.S. corporate focus on stock market performance); as a result, the Japanese MNEs argued that no transfer pricing abuse was taking place. In addition, the firms argued that Japanese trading corporations, soga shoshas, typically run on razor-thin margins. To the extent that the foreign-controlled corporations (FCCs) were units of trading corporations (e.g., those in the wholesale and retail trade industries), one would expect small profit margins on sales. The 1990 Pickle Hearings Evidence of these thin margins was provided in July 1990 by Patrick Heck, assistant counsel to the Subcommittee on Oversight of the House Committee on Ways and Means in testimony before the subcommittee (a.k.a. the Pickle hearings, after the senator who chaired the meetings). Heck was reporting on the results of a nine-month investigation by the subcommittee into the level of taxes paid by foreign-owned companies in the United States. The investigators conducted interviews and examined ten years of confidential company documents, including tax returns, for 36 foreign-owned distributors: 18 electronic distributors and 18 auto/motorcycle distributors. Twenty-five of the firms had Pacific Rim parents and 11 had European parents. Heck testified that more than half of the 36 companies paid little or no federal income tax over the ten-year period. The 18 electronics distributors reported US$116 billion in gross receipts but only paid 0.5 per cent, or US$654 million, in U.S. taxes. Only nine of the firms reported positive taxable income over the period. All but two of the 18 auto and motorcycle firms paid some tax, but loss carry-overs meant that 12 actually paid tax of less than 1 per cent of their gross receipts. The summary data provided by the Heck (1990) study are reported in Table 7.3. The study concluded that inflated transfer prices and the performance by the FCCs of functions not properly compensated for by the foreign parent were the main causes for the low U.S. profits and tax payments. In addition, tax pay-

Taxing Multinationals in Practice

345

TABLE 7.3 Heck (1990) Study of Transfer Pricing by 36 Foreign-Controlled Corporations

Number of returns reviewed Gross receipts Cost of goods sold Gross profits Total income Total deductions Taxable income Total U.S. tax paid Proposed IRS adjustments Tax as a per cent of gross receipts Tax as a per cent of gross profits Tax as a per cent of taxable income

Electronics equipment distributors

Automotive and motorcycle distributors

106

116

(dollar figures in US$ billions) 116.4 212.8 199.8 90.5 28.5 25.9 32.0 27.5 26.1 22.6

1.6 0.6 0.2 0.52% 2.32% 37.5%

9.2 4.4 2.5 2.07% 15.44% 47.8%

SOURCE: Heck (1990, tables 1 and 2)

ments often were negated by net operating loss carry-backs from later years. The main transfer pricing issue, according to Heck, involved inbound transfers in which the distributor paid too high a price for goods purchased from the overseas parent; for example, rebates by foreign-owned manufacturers were often not reflected in the transfer price paid by the FCCs. These affiliates also used the allocation of advertising and warranty costs to reduce their taxes, shifting income out of the United States through excessive freight, insurance, interest, and other fees and charges. Not only were the firms avoiding the CIT, in addition, transfer prices were manipulated to avoid paying U.S. customs duties. Duty avoidance schemes included the overstatement of freight and insurance costs on imported products; this had the additional benefit of reducing tax payments since freight charges were tax deductible. In addition, Heck reported that the IRS's international enforcement program suffered from a lack of coordination, difficulty retaining personnel, and shortage of resources. The IRS was 'outgunned' by the foreign MNEs (Heck 1990, 3). Transfer pricing cases were difficult to examine because the issues were factual and complex, and some firms refused to provide the IRS with information. Difficulties in obtaining information caused delays of up to ten years for an auto audit, and four to six years for a cycle or electronics audit. The study noted that

346 Transfer Pricing and Taxation FCCs, by decentralizing their U.S. operations into separate businesses, were able to move out of the IRS large-case-file program and therefore avoid detailed IRS scrutiny. Multiple transactions could also be used to set up elaborate paper trails. The 1992 IRS and GAO Reports Since the purpose of the Heck study was fact finding, the report did not end with recommendations. However, the U.S. Congress responded by strengthening the international enforcement program of the IRS (see also Chapter 9). Reporting and record-keeping requirements were expanded, penalties associated with section 482 valuations were expanded, and the statute of limitations on tax assessments was lengthened (U.S. Committee on Ways and Means 1993, 238). The Secretary of the Treasury was directed to further study and make recommendations on section 482. In April 1992, the Subcommittee on Oversight held a second hearing to receive the Treasury's report. The IRS testified before the committee that, in 1989, 28 per cent of FCCs reported U.S. taxable income compared with 41 per cent of the U.S. firms. Committee members charged that underpayment of U.S. taxes was costing more than US$30 billion annually, and that such tax evasion was undermining U.S. competitiveness by robbing the government of resources needed to rebuild the country's infrastructure and critical technologies. The committee also updated its 1990 report and found that of the 36 FCCs, 40 per cent of the electronics firms and 28 per cent of the auto/motorcycle firms paid no U.S. tax in 1989. IRS commissioner Shirley Peterson, in testimony before the committee, admitted that there was a compliance problem. While there were insufficient data to reliably estimate the size of the 'tax gap' resulting from overall noncompliance by FCCs, she concluded that the gap due to noncompliance with section 482 was about $3 billion per annum; however even this number was uncertain. The estimate of the tax gap was made as follows. The IRS calculations were based on a Grubert et al. (1991) study which concluded that as much as 50 per cent of the differential in rates of return between foreign and domestically controlled corporations was due to transfer pricing noncompliance. In 1989, according to Statistics of Income (SOI) data, the rate of return on assets of FCCs was 0.6 per cent, compared with a rate of 1.8 per cent for U.S. corporations (USCCs). The differential in pre-tax rates of return is therefore 1.2 percentage points. Assuming that half of the differential was due to transfer price manipulation (as argued in Grubert et al. 1991), the tax gap is the additional taxes that would be paid if the FCCs had reported an additional

Taxing Multinationals in Practice

347

TABLE 7.4 GAO Estimates of Profit as a Per Cent of Total Sales for USCCs and FCCs, 1982-1989

1982

1983

1984

1985

1986

1987

1988

1989

USCCs FCCs

1.6 0.4

2.3 0.5

2.7 1.0

2.6 0.6

-0.3

3.1 0.8

3.7 1.4

3.1 0.9

FCC relative to USCC

0.25

0.22

0.37

0.23

-0.11

0.26

0.38

0.29

2.7

USCCs are U.S.-controlled corporations; FCCs are foreign-controlled corporations. SOURCE: Calculated from data in Cushman (1992, 9). Data are from the U.S. General Accounting Office.

50 per cent of the rate-of-return gap (i.e., one-half of the 1.2 percentage points, or 0.6 points) on their assets, bringing their pre-tax returns to 1.2 per cent. The total assets of FCCs in 1989 were 1.429 trillion U.S. dollars. Raising their pretax rate of return from 0.6 to 1.2 per cent of $1.429 trillion in assets, at an average tax rate of 34 per cent, would have increased the FCCs' taxable income by US$8.57 billion, generating an additional US$2.92 billion in tax revenues for the U.S. government (i.e., 0.6% x US$1.429 trillion x 34%). There were several problems with the IRS estimate, not the least of which was the aggregate level at which it was calculated, ignoring differences in industry and asset mix, size of firm, riskiness of investments, and so on, which can influence pre-tax rates of return on assets. In addition, the 50 per cent estimate of Grubert et al. (1991) was a ballpark figure, which may or may not been right. A more conservative estimate of 25 per cent of the tax gap being due to transfer price manipulation would have cut the estimated additional tax revenue to US$1.46 billion. Compared with total U.S. receipts in 1989 from the corporate income tax of US$103.29 billion, however, either estimate is an insignificant amount (Committee on Ways and Means 1993, 86). In addition, U.S. MNEs were probably also using transfer price manipulation to understate their U.S. taxable income. Since section 482 issues can and do arise on both FCC and USCC transactions, and the Heck study ignores this by concentrating only on tax abuses of foreign MNEs, its estimates are one-sided. In June 1992, the General Accounting Office (GAO) released a study comparing FCC and USCC profits as a percentage of total sales revenues. See Table 7.4 above. The GAO study confirmed that foreign-controlled corporations were paying taxes, as a per cent of sales, well below those paid by domestically controlled firms, for the 1983-9 period. As the table shows, the ratio varies from 0.22

348 Transfer Pricing and Taxation to 0.38, excluding 1986 when the FCCs in aggregate recorded losses. The GAO also reported that the IRS internal appeals process, over the fiscal period 1987-9, had received less than 26.5 per cent of the US$757 million it sought in tax adjustments. Also in 1992, increased U.S. concern with the U.S.-Japan trade deficit drew attention to the high percentage of that trade conducted through Japanese MNEs and the impact that ownership might have on trade volumes and prices (see, for example Encarnation 1992, 1994). Dennis Encarnation, in Rivals beyond Trade, argued that most Japanese subsidiaries in the United States were wholly owned, whereas most U.S. subsidiaries in Japan were joint-venture arrangements. His data (the U.S. Department of Commerce data we used in Chapter 4 and later in this chapter, which are derived from IRS information reporting requirements) showed that wholly owned affiliates tended to engage more heavily in trade in general, and in intrafirm trade with their parents in particular, than did partly owned affiliates. He concluded that differences in ownership were partly responsible for the U.S.-Japan trade deficit. Encarnation's work has increased the interest of U.S. policy-makers in intrafirm trade, and the perception that it needs closer scrutiny. Clinton's Campaign Promise Testimony before the Pickle hearings in 1990 and 1992 caused a stir on Capitol Hill and was widely quoted. It was also used as part of Governor Bill Clinton's election campaign for president of the United States. He promised to raise US$45 billion from foreign corporations, over four years, from tighter enforcement of existing transfer price regulations. Clinton pledged that he would start 'cracking down on foreign companies that prosper here and manipulate tax laws to their advantage' (quoted in Wartzman 1992, A16). Where the figure of US$45 billion came from has never been clear. One possibility is as follows. Clinton aides noted that in 1988, FCCs made US$825.6 billion in receipts but paid only US$5.8 billion in taxes. Combining this data with information from the GAO study above, showing that profit as a percentage of total sales for FCCs in 1988 was 1.4 per cent compared with 3.7 per cent for USCCs, we can estimate the additional tax FCCs would have paid in 1988 had they declared pre-tax profits at the same rate as the USCCs. On receipts of US$825.6 billion, at a profit rate of 3.7 per cent, taxable profits for the FCCs would have been US$30.55 billion. At a tax rate of 34 per cent, that would represent total tax payments of US$10.39 billion. Over four years and if we allow for inflation, this number does approximate US$45 billion. However, it ignores the taxes already paid by the FCCs; that is, using this method double counts the

Taxing Multinationals in Practice

349

TABLE 7.5 Organization for International Investment's Estimates of FCC Tax Payments, 1983-1988

1983 1984 1985 1986 1987 1988

Net FCC income (USS billions)

FCC taxes (US$ billions)

1.8 4.5 3.0 -1.5 5.6 11.2

3.4 4.5 3.6 5.6 4.6 5.8

Gross FCC income (USS billions)

5.2 9.0 6.6 4.1 10.2 17.0

Taxes as a per cent of gross income income 65.38 50.00 54.55 136.59 45.10 34.12

SOURCE: Calculated from data in Wartzman (1992, A16)

taxes already paid by the FCCs in 1988. Recalculating, the net additional tax revenue is only (0.037 - 0.014) times US$825.6 billion times 34 per cent, or US$6.46 billion. Scaling this over four years gives us an estimate of US$30 billion, which is the 1992 estimate of the Pickle Subcommittee - an estimate more than double the US$3 billion per annum proposed by IRS commissioner Peterson. An Alternative View: Porter (1993) The view that FCCs have substantially underpaid U.S. taxes has not gone unchallenged. Catherine Porter, tax counsel for the Organization for International Investment (a U.S. lobby group for foreign MNEs), argued that the study was not statistically valid since it looked only at 36 unnamed companies, whereas there are 46,000 FCCs in the United States (Porter 1993). She suggested that high start-up costs, currency fluctuations, debt-equity ratios, the desire to increase market share, newness of the investment, managerial skill and experience could all have important influences on taxable income. She noted that a 1992 KRMG Peat Marwick survey of FCC tax payments, conducted for the Organization for International Investment, showed that over the 1983-8 period tax payments for FCCs had risen from US$3.4 to 5.8 billion. The average effective tax rate, measured as tax payments over pre-tax or gross income, varied from 34 to 65 per cent, for the years where the FCCs recorded gross profits (see also Wartzman 1992). See Table 7.5 above. In addition, Porter argued that a 1993 General Accounting Office study, using 1989 tax data, concluded that the percentages of very large FCCs and USCCs (assets of more than US$250 million) that did not pay any U.S. tax were similar

350

Transfer Pricing and Taxation

TABLE 7.6 IRS Examinations of Foreign-Controlled Corporations, 1992 Tax Year Country of ultimate beneficial owner: Canada Number of FCC cases Number of FCC tax returns Total income adjustment (US$ millions) Average income adjustment per tax return (US$ thousands) Estimated additional taxes (34 per cent of total income adjustment) Additional taxes unrelated to income adjustment (US$ millions) Penalties unrelated to income adjustment (US$ millions) Total taxes and penalties (US$ millions) Average taxes and penalties per FCC tax return (US$ thousands) U.S. merchandise imports to FCCs, 1990 (US$ millions) Taxes and penalties as a per cent of total imports

Mexico

Japan

United Kingdom

69

46

175

121

109

82

314

229

133.77

9.67

507.76

462.97

1,227.20

117.90

1,617.10

2,021.70

45.48

3.29

172.64

157.41

2.51

0.67

45.80

75.05

0.90 48.89

0.19 4.14

0.24 218.68

0.85 233.31

448.50

50.50

696.40

1,018.80

10.54

0.81

87.71

13.23

4.64

5.10

2.49

17.64

SOURCE: Polinsky (1993, 1419), Ernst & Young (1993, 74-5)

(30 versus 33 per cent, respectively) as were the percentages for smaller (assets of less than US$100 million) FCCs and USCCs (88 versus 93 per cent, respectively).27 Finally, commenting that FDI in the United States had slumped to its lowest level since the mid-1970s, Porter concluded that the government could not afford to discourage inward FDI through overzealous taxation. The 1993 Ernst & Young Study Ernst & Young did a study of FCC and USCC tax payments, using examination data obtained from the IRS under the U.S. Freedom of Information Act (Ernst & Young 1993; Polinsky 1993). The Ernst & Young study looks at the taxes assessed on FCCs by the Internal Revenue Service in fiscal year 1992. Data on FCCs with parents in Canada, Mexico, Japan, and the United Kingdom are provided in Table 7.6 above. The study shows that tax understatements were found on 55 per cent of the 3,357 FCC tax returns examined by the IRS. The ten top trading partners of the

Taxing Multinationals in Practice

351

TABLE 7.7 Average Tax Rates for U.S. Manufacturing, 1989-1990 Average tax rate on FCCs (per cent)

Average tax rate on USCCs (per cent)

Differential in tax rates (per cent)

Taxes/receipts, 1989 Taxes/receipts, 1990 Change, 1989 to 1990

0.84 0.97 0.13

1.25 1.18 -0.07

0.41 0.21 -0.20

Taxes/assets, 1989 Taxes/asets, 1990 Change, 1989 to 1990

0.64 0.77 0.13

1.16 1.08 -0.08

0.52 0.31 -0.21

SOURCE: Amerkhail (1993, 49)

United States, which include the four countries in the table, had total understatements of US$1.36 billion. Assuming a 34 per cent tax rate, the additional taxes amount to US$460 million. Together with US$150 million in other taxes and penalties, the maximum total revenues the IRS could raise if it were successful in these adjustments is US$610 million (Polinsky 1993, 1410). The amount that could be raised from Canada, Mexico, Japan, and the U.K. totals most of this amount (US$505 million, about 80 per cent). Estimated tax understatements by Japanese FCCs, while large (US$219 million, or an average US$696,400 per return), are below those of U.K. FCCs. We conclude that the Ernst & Young study indicates that the total amount that could be raised by additional enforcement of Treasury regulations is probably small. The Amerkhail (1993) Study A recent salvo in this 'tax war' has come from Valerie Amerkhail, a member of the Coopers and Lybrand tax policy economics group. She argues that appropriate comparisons cannot be made between FCCs and USCCs without correcting for obvious differences such as size of firm and industry mix (Amerkhail 1993). She re-estimates the tax gap correcting for these two sources of bias. As evidence of the first bias, Amerkhail provides data on the ratio of taxes to sales receipts and of taxes to assets for FCCs and USCCs. (see Table 7.7). While the USCC ratio is higher in both cases, the tax gap is wider for the ratio of taxes to assets; both ratios also fell between 1989 and 1990. The explanation for this result is the higher proportion of FCCs in wholesale and retail trade, a less capital-intensive activity than manufacturing, where a larger proportion of USCCs tend to cluster. Wholesalers typically have higher

352 Transfer Pricing and Taxation sales-to-assets ratios than manufacturers. For example, in 1989, FCCs averaged US$ 1.48 of assets per dollar of sales, while the corresponding ratio for USCCs was 1.84 to 1. As a result, USCCs tend to be more capital intensive than foreign affiliates.28 This difference affects their taxes-to-assets ratios (making the ratios higher for manufacturing and thus for USCCs than for FCCs in general) and taxes-to-sales ratios (making them higher for wholesale trade and thus for FCCs relative to USCCs in general). Thus the tax gap should be wider when taxes-toassets is used as a base comparison than if taxes-to-sales is the criterion. This prediction is borne out by the evidence. Table 7.8 summarizes the results of correcting for industry bias in 1990. The estimated total tax gap for all FCCs, using the general USCC rate on sales receipts of 1 per cent, is US$3.4 billion. This additional tax represents a 45.9 per cent increase over the actual taxes paid by FCCs in 1990. Amerkhail then individually estimates a tax gap for manufacturing; trade; fire, insurance, and real estate (FIRE); and other industries. The estimated gap using the tax-tosales ratio, as a percentage of actual FCC taxes paid, varies from a low of 7 per cent (FIRE) to 104 per cent (other industries). Industry mix is taken into account in the 'sum of four classes' row of the table, which shows the industryadjusted total for all corporations. When the mix is taken into account in the total calculation, the additional tax drops from US$3.4 to US$2.3 billion. The second set of calculations in Table 7.8 uses taxes-to-assets as the benchmark. In this case, the total additional tax is US$1.8 billion when industry structure is ignored (a 24 per cent increase over actual payments). When mix is taken into account the tax gap is US$4.3 billion, an increase of 57 per cent over actual payments. This is because the tax gap, holding industry constant, is much larger; the lowest gap is 20 per cent (FIRE) rising to 205 per cent (other industries). A second source of bias is firm size. This reflects the fact that a relatively small number of large corporations, U.S. and foreign, own most of the assets, receive most of the sales income, and pay most of the taxes. In the top bracket (firms with assets over US$250 million), the 1993 GAO study had estimated that 70 per cent of FCCs paid U.S. taxes compared with 67 per cent of USCCs. Since the percentages do not differ by much, if total taxes paid do differ it must be because the large FCCs are paying on average less taxes than the large USCCs. Amerkhail provides data for 1989, showing that these large FCCs paid US$4.193 billion in taxes; if they had paid taxes at the same rate on sales receipts as was paid by large USCCs (i.e., 0.5 per cent), their total taxes would have been US$5.929 billion, or 41.4 per cent higher. Correcting for each quintile, Amerkhail estimates an overall tax gap of US$2.06 billion, if the ratio of taxes to sales is the benchmark, or US$4.09 billion if taxes-to-assets is the benchmark, in 1989.

Taxing Multinationals in Practice

353

TABLE 7.8 The Underpayment of U.S. Taxes by Foreign Corporations, 1990 The tax gap measured in terms of business receipts, 1990

Business type Manufacturing Trade Finance, insurance, and real estate Other industries Sum of four classes All corporations

FCC tax if paid at USCC rates (US$ mill.)

Change in tax due on FCCs (US$ mill.)

Change in tax as per cent of FCC tax paid

USCC rate on receipts (per cent)

FCC tax paid (US$ mill.)

1.2 0.5

4,340 1,416

5,245 2,232

905 816

20.85 57.63

1.1 1.3 1.0 1.0

1,154 528 7,438 7,438

1,233 1,077 9,787 10,852

79 549 2,348 3,414

6.85 103.98 31.57 45.90

The tax gap measured in terms of business assets, 1990

Business type Manufacturing Trade Finance, insurance, and real estate Other industries Sum of four classes All firms

FCC tax if paid at USCC rates (US$ mill.)

Change in tax due on FCCs (US$ mill.)

Change in tax as per cent of FCC tax paid

USCC rate on receipts (per cent)

FCC tax paid (US$ mill.)

1.1 1.1

4,340 1,416

6,096 2,601

1,756 1,185

40.46 83.69

0.2 1.1 0.6 0.6

1,154 528 7,438 7,438

1,387 1,609 11,693 9,224

233 1,081 4,255 1,785

20.19 204.73 57.21 24.00

SOURCE: Amerkhail (1993, 48)

Amerkhail's work proves that estimates of the underpayment of taxes by foreign-controlled corporations, calculated by comparing them with payments by domestic firms, are problematic at best. Ignoring simple factors that could influence payments, such as industry mix, size of firm, and using sales receipts or assets as the benchmark, can yield wildly different results. Amerkhail's estimates for 1989 range from a low of US$1.9 billion (ignoring industry mix, using taxes-to-assets, correcting for size of firm) to a high of US$5.3 billion (ignoring industry mix, using taxes-to-receipts as a base, correcting for size of firm). The highest estimate is more than double the lowest estimate; the average

354 Transfer Pricing and Taxation across all the estimates for 1989, however, is US$3.4 billion,29 which is close to the US$3 billion estimate of IRS commissioner Peterson. The GAO (1995) International Taxation Study In April 1995, the General Accounting Office issued an updated report on transfer pricing and tax compliance by U.S. and foreign-controlled corporations (U.S. General Accounting Office 1995). The GAO report states that 73 per cent of all FCCs and 62 per cent of all USCCs paid no CIT in 1991. This finding was comparable with data for the past five years, - that is, most corporations paid no tax, and a larger per cent of foreign firms paid no tax than did domestic firms. The firms that did pay tax, in general, were large corporations; for example, the 27 per cent of FCCs that paid U.S. income tax had 63 per cent of the FCC assets and 69 per cent of the receipts. Similarly, the 38 per cent of large U.S. corporations that paid income tax had 80 per cent of the assets and 81 per cent of the gross receipts. The report notes that in 1993 the IRS proposed total section 482 income adjustments of $2.2 billion for 369 corporations. The proposed adjustments were $0.9 billion for 247 U.S. corporations and $1.3 billion for 122 foreign-controlled corporations. Most of these adjustments applied to a few MNEs: 18 firms accounted for $0.7 of the $0.9 billion proposed adjustments for USCCs and 33 firms for $1.1 of the $1.3 billion for FCCs. The most common reasons cited by the IRS for the proposed adjustments were the following: interest (14 per cent), royalties (12 per cent), pricing of tangible goods (11 per cent), general allocation of income and deductions (11 per cent), and service charges and fees (7 per cent). The GAO cautions that proposed adjustments do not, however, necessarily result in tax payments since the taxpayer can go to appeals or to tax court. As of June 1994, the IRS had 114 open cases (i.e., cases beyond examination but not settled because they are either in appeals or litigation) of large taxpayers, with total proposed adjustments under section 482 of $14.4 billion. Of the 82 transfer pricing cases that were settled in fiscal year 1993, only 27 per cent ($249 million) of the proposed $926 million in income adjustments were sustained by the IRS. The GAO notes that the sustentation rate (i.e., the percentage of the proposed adjustment actually sustained by the IRS when the case is settled) has varied widely from year to year, and tended to be higher for USCCs than for FCCs.30 The reasons why the IRS settled for a reduced amount are interesting: in 63 per cent of the cases closed in fiscal year 1993 the rationale given by the IRS was hazards of litigation; that is, the Service was concerned about whether a tax court would interpret the facts or evidence in the same manner as the IRS had done. In terms of the transfer pricing methods used to determine arm's length

Taxing Multinationals in Practice

355

prices, the GAO notes that the major obstacle in enforcing section 482 was the difficulty IRS examiners had in finding comparable transactions. Over the 1990-2 period, the three basic methods (CUP, resale price, cost plus) were used in only about half of the completed cases; similarly, they were used only 38 per cent of the time in the Advance Pricing Agreements issued over this period.31 The GAO report concludes that recent IRS experience with transfer pricing has been mixed. Based on the evidence of large amounts of proposed income adjustments, the low sustentation rate, and the small percentage of cases using transactional transfer pricing methods, one could conclude that tax compliance with section 482 was quite poor. On the other hand, the IRS now has new rules (the final 482 regulations) and procedures (advanced pricing agreements, contemporaneous documentation, inaccuracy penalties, arbitration) that should improve compliance. While the success of these new rules and procedures is not yet clear, the GAO report implies that the Service is making progress in combatting tax avoidance by related parties. Summary The United States has historically been a home country. As such, its focus has been to ensure fair and equitable treatment of its foreign affiliates by host countries, together with protection of the U.S. tax base from erosion by tax havens. With the inflow of FDI in the 1980s, U.S. concern has shifted towards ensuring that FCCs in the United States pay their 'fair' share of taxes. The U.S. concern with these underpayments is overstated. As the various estimates show, each new estimate has been smaller than the last in terms of the probable amount of actual income that could be raised through tighter enforcement of the U.S. tax laws. However, the jury is still out on this matter since estimates of tax losses from transfer pricing manipulation by foreign-controlled corporations in the United States vary from zero to US$ 11 billion a year. The U.S. Congress, on the other hand, is not waiting for new estimates, but is proceeding with its attempts to raise tax revenues from multinationals. One example is the Foreign Tax Compliance Act, introduced in the House in July 1994. The sponsors argued that 73 per cent of FCCs did not pay U.S. taxes in 1991, with a loss to the U.S. Treasury of $10 billion a year. In addition, 'runaway plants' that relocate overseas to take advantage of lower costs and taxes cost the Treasury another US$1.6 billion over five years (Kirchheimer 1994, 419). The bill would have repealed tax deferral for U.S. multinationals that produced abroad for sale in the U.S. market (in order to discourage runaway plants) and taxed the 'imported property income' of U.S. shareholders of a foreigncontrolled corporation on an accrual basis, with a new separate foreign tax

356 Transfer Pricing and Taxation credit limitation on imported property income. In addition, the bill recommended using a simple formula approach in cases where the arm's length transactions rules do not work. While the bill did not become law, it does suggest that the tax payments of multinationals in the United States, both domestic and foreign, are under increasing scrutiny. This concludes our historical review of the recent U.S. controversy over tax payments made by foreign multinationals in the United States. We now turn to two topics which are directly related to the controversy: the tax incentives to manipulate transfer prices within North America, and evidence on taxes paid by multinationals in the United States. In the first section, we outline the differences in CIT and withholding tax rates and show how these are likely to affect MNE incentives to engage in TPM. We also examine the impact of the North American Free Trade Agreement (NAFTA) on these incentives to shift MNE income. In the second section, we use two sets of U.S. Department of Commerce data, the first on U.S. multinationals and their foreign-owned subsidiaries (MOFAs) and the second on foreign-controlled corporations (FCCs) in the United States, to see whether there is any evidence of income shifting due to these tax differentials that either supports or refutes the general view that foreign MNEs are underpaying taxes in the United States. Incentives to Manipulate Transfer Prices in North America In this section, we explore the incentives MNEs in North America have to manipulate transfer prices based on differences in corporate income tax and withholding tax rates. The Statutory Tax Rates We can see the tax incentives for transfer price manipulation through a closer look at the CIT and withholding tax rates in tables 2.4 and 2.5. As the tables show, the combined federal/state CIT in the United States is 46 per cent; the dividend withholding tax raises this to 51 per cent. The combined Canadian rate depends on whether the income is from manufacturing or nonmanufacturing activities. For manufacturing, the CIT rate is 38 per cent; with the withholding tax this increases to 44 per cent. For nonmanufacturing activities, the CIT rate is 44 per cent, rising to 50 per cent with the withholding tax. The tax rate in Mexico is 34 per cent, plus a nondeductible, 10 per cent profit-sharing arrangement, and a 2 per cent business assets tax (that functions only as a minimum tax); there is no withholding tax on dividends. The equivalent Mexican tax rate is therefore 44 per cent.32

Taxing Multinationals in Practice

357

When taxes exclusive of withholding taxes are compared, the rates are: Canada (36 per cent manufacturing, 44 per cent nonmanufacturing), Mexico (44 per cent), and the United States (46 per cent). The U.S. rate therefore exceeds the Mexican and Canadian rates. The tax rates inclusive of the withholding tax are: Canada (42 per cent manufacturing, 50 per cent nonmanufacturing), Mexico (44 per cent), and the United States (51 per cent). Again, the U.S. rate exceeds that of the spokes. Because the analysis is complicated, we first outline the incentives for U.S. and Canadian firms, and then expand the analysis to look at U.S., Canadian, and Mexican multinationals. Canada-U.S. Tax Incentives for Transfer Price Manipulation The corporate income tax differentials between Canada and the United States can be decomposed into four possible cases, depending on whether the parent firm is a Canadian or U.S. multinational. We look first at the 1995 case, and then at the impact of the 1995 Canada-U.S. tax protocol on these differentials. In each case, we assume that tax rates in New York State and Ontario are used to calculate the subfederal rate.33 1995 Canada-U.S. Tax Differentials The incentives to manipulate Canada-U.S. intrafirm trade flows, as of 1995, are presented in Table 7.9. In case #1, a U.S. multinational (USCO) with a Canadian nonmanufacturing subsidiary (CANCO) pays a U.S. CIT rate (federal plus state) of 46 per cent on profits declared in the United States, whereas the Canadian CIT rate (federal plus provincial) plus a withholding tax is 50 per cent on profits declared in Canada and remitted to the United States. When the Canadian subsidiary remits the income, the parent grosses up the income by the Canadian taxes and receives a tax credit up to the level of the U.S. tax rate. Since the Canadian tax rate is higher than the U.S. rate, no further taxes are due in the United States. In this case, since the tax rate on CANCO is higher than the tax rate on USCO, transfer price manipulation can be used to reduce the MNE's overall tax bill. The MNE should shift taxable profits to the low-tax location (the United States) and tax deductions to the high-tax location (Canada). However, if the subsidiary's profits are retained in Canada and not remitted to the United States, only the Canadian CIT applies, since no U.S. taxes are levied on deferred foreign-source income. In this case, CANCO's tax rate is only 44 per cent, which is well below the 50 per cent rate on dividends. Thus the MNE has an incentive to not remit dividends to the parent in order to avoid the additional Canadian withholding tax. Where dividends are not remitted, the

358 Transfer Pricing and Taxation TABLE 7.9 The Canada-U.S. Tax Incentive for Transfer Price Manipulation, 1995 (tax rates in percentage terms) Taxes on MNE income

U.S.

Canada

Federal and state CIT, combined rate

General Manufacturing

46 46

44 36

Federal and state CIT + 10% dividend withholding tax

General Manufacturing

51 51

50 42

The tax incentive to manipulate transfer prices U.S. tax rate on the parent's profits U.S. parent with a Canadian wholly owned subsidiary

Canadian parent with a U.S. wholly owned subsidiary

Retained earnings

Dividends

Case#l: General subsidiary

46

44(C)

50(C)

Case # 2: Manufacturing subsidiary

46

36(C)

42(C) 46(US)

Retained earnings

Dividends

Canadian tax rate on the parent's profits

Case #3: General parent

46(US)

51(US)

44

Case #4: Manufacturing parent

46(US)

51 (US)

36

Conclusion: Except for dividends in cases #1 and #2, the combined statutory tax rate in 1995 was lower in Canada than in the United States. Therefore the MNE's overall tax bill could be reduced by shifting profits from the United States to Canada. There are several ways this could be done: USCO could underinvoice transfers to CANCO, CANCO could overinvoice transfers to USCO, and the MNE could shift tax-deductible expenses to USCO. NOTES: (1) The lowest tax rate in each case is in bold print. (2) The table ignores tariffs and any other tax breaks (e.g., R&D credits) that might affect this decision. SOURCE: Calculations based on the Canadian and U.S. tax rates in Table 2.4

Taxing Multinationals in Practice

359

effective tax rate is lower in Canada than in the United States and the MNE can further reduce its tax bill by shifting taxable income from USCO to CANCO. In case #2, the U.S. parent has a Canadian subsidiary in manufacturing which qualifies for the low Canadian CIT rate on manufacturing profits. If the subsidiary retains its income in Canada, the effective tax rate on CANCO is only 36 per cent, compared with the U.S. rate on USCO of 46 per cent. Transfer price manipulation in this case can be used to shift taxable income to CANCO and reduce the MNE's overall tax bill. On the other hand, if CANCO remits its profits to its parent firm, because the effective Canadian tax rate is 42 per cent, below the U.S. rate of 46 per cent, an additional 4 per cent tax is due in the United States, raising the effective tax rate on CANCO's dividends to 46 per cent. In this case, there is no tax incentive to manipulate transfer prices since the tax rates on CANCO and USCO are the same. The situation changes in cases 3 and 4 because the home country is now Canada. In case #3, if a U.S. subsidiary of a Canadian parent remits profits to its parent, the U.S. CIT plus withholding tax both apply. Therefore the effective U.S. tax rate on USCO is 51 per cent on dividends. However, assuming the income of the subsidiary is treated as active business income, no tax is due in Canada when the dividends are repatriated because Canada exempts foreign source-income from taxation if it is active business income. Since the U.S. rate of 51 per cent is higher than the parent's rate of 44 per cent, the exemption is of no benefit to the foreign affiliate. Therefore, if USCO remits profits to its parent, the effective tax rate is higher on USCO than on CANCO, and the MNE can reduce its tax bill by shifting taxable income to Canada. If profits are retained in the United States, the subsidiary pays only the U.S. CIT of 46 per cent; however, the tax differential between the two countries, although reduced, still favours Canada. Given the choice between repatriating profits and reinvesting the earnings, USCO should choose reinvestment since the tax rate is lower (46 versus 51 per cent). Case #4 is similar to case #3, except that the Canadian tax rates are even lower because of the reduced Canadian CIT on manufacturing activities. In summary, there are three general patterns in this table: 1. a U.S. parent with a Canadian nonmanufacturing subsidiary remitting profits to its parent, where the tax rate on CANCO is higher than on USCO; 2. a U.S. parent with a Canadian manufacturing subsidiary remitting profits to its parent, where the two tax rates are the same; and 3. all other cases, where the tax rate on USCO is higher than on CANCO. In most situations, if a subsidiary does not remit dividends to its parent, the

360 Transfer Pricing and Taxation effective tax rate, for both Canadian and U.S. MNEs, is lower in Canada than in the United States. The tax incentive is therefore to shift taxable income to Canada and tax-deductible expenses to the United States. If taxes were the only external factor affecting transfer prices and the dividend-retained earnings choice, we would expect to see these patterns in the transfer pricing of tangibles, services, and intangibles between related firms across the Canada-U.S. border. In practice, given tariffs, rules of origin, tax credits for R&D expenses, location incentives, and government services provided to business, our simple calculation must be taken with several caveats. The Impact of the 1995 Canada-U.S. Tax Protocol These tables show the tax situation as of 1995. We note, however, that the new Canada-U.S. protocol which took effect on 1 January 1996, reduces the withholding tax on dividends from 10 to 5 per cent (see Table 2.6). Redoing the calculations for remitted profit, we find that the effective CIT plus withholding tax on USCO's dividends falls from 51 to 49 per cent.34 The effective rate on dividends of a Canadian nonmanufacturing subsidiary falls from 50 to 47 per cent, and on manufacturing profits from 42 to 39 per cent. As a result, the 'tax penalty' for remitting dividends versus retaining income drops in certain cases. For example, in the United States, retained earnings of a U.S. subsidiary with a Canadian parent, post-protocol, are taxed at 46 per cent and dividends at 49 per cent. For nonmanufacturing subsidiaries in Canada the gap is 44 per cent on retained earnings versus 47 per cent on dividends. In these two cases, we expect more dividends to be remitted to parent firms since the withholding tax penalty is lower. For manufacturing subsidiaries the comparison is 36 per cent versus 39 per cent; however, unless the U.S. parent has extra foreign tax credits, additional tax is due in the United States. Therefore, with the U.S. rate at 46 per cent, manufacturing affiliates in Canada continue to have a deficit of foreign tax credits. As a result, there is no additional incentive to remit profits in this case. In addition, the incentive for transfer price manipulation also falls because in most cases the effective tax differential declines or is eliminated for dividends. The tax situation vis-a-vis retained earnings, per se, remains unchanged. Lowering the withholding tax, therefore, where it reduces the Canada-U.S. tax differential, reduces the incentive to shift profits to Canada through transfer price manipulation. North American Incentives for Transfer Price Manipulation Table 7.10 shows our estimates of the tax incentives to manipulate transfer

Taxing Multinationals in Practice

361

TABLE 7.10 The North American Tax Incentives for Transfer Price Manipulation, 1995 (tax rates in percentage terms) Canada

Combined federal/state corporate income tax rate CIT plus withholding tax

U.S.

Mfg.

Non-Mfg.

Mexico

46

36

44

34 *44

Canada 51 Mexico 49

42

50

44

50

46

44

44

44

44

44

44

50

44

44

44

Effective tax rates on profits earned by: U.S. parent with remitted profits from Canadian and Mexican affiliates 46 46 U.S. parent with no profits remitted from Canadian and Mexican affiliates 36 46 Canadian parent with remitted profits 36 from U.S. and Mexican affiliates 51 Canadian parent with no remitted profits from U.S. and Mexican affiliates 36 46 Mexican parent with remitted profits 42 from Canadian and U.S. affiliates 49 Mexican parent with no remitted profits 36 from Canadian and U.S. affiliates 46

* Includes a 10 per cent profit-sharing arrangement tax. NOTE: The lowest tax rate in each category is in bold. SOURCE: Calculations based on Canadian, U.S., and Mexican tax rates in Table 2.5

prices and intrafirm trade flows within North America in 1995. There are now six possible cases to consider. The first comparison is between a parent firm in one country and its affiliates in the other two countries. This gives us three cases. In each of these cases, the effective tax rate if profits are remitted to the parent (i.e., the tax on dividends) must be distinguished from the nonremittance case (i.e., the tax on retained earnings). This gives us six cases. For a U.S. multinational with foreign affiliates in Canada and Mexico that repatriate profits to their parent, the appropriate comparisons are: (1) the U.S. tax rate on the U.S. parent firm of 46 per cent; (2) the Canadian tax rates inclusive of withholding taxes of 42 per cent on manufacturing (which becomes 46 per cent if the MNE has a deficit of tax credits since an additional 4 per cent is due upon repatriation), and 50 per cent on nonmanufacturing (these fall to 39 and 47 per cent under the 1995 Canada-U.S. tax protocol); and (3) the

362 Transfer Pricing and Taxation Mexican tax rate of 44 per cent (46 per cent with a deficit of tax credits). Three of the four cases thus have the same effective statutory tax rate. If the foreign affiliates do not repatriate profits to their U.S. parent, then the host country rates are the CIT rates exclusive of withholding taxes: Canada (36 per cent manufacturing, 44 per cent nonmanufacturing) and Mexico (44 per cent), which are lower than the U.S. tax rate. In Table 7.10, the lowest tax rate in each case is in bold numbers. Note first that the majority of these cases occur in the Canadian manufacturing column. In general, the Mexican and Canadian rates are lower than those due in the United States, and a U.S. MNE has an incentive to: (1) not repatriate profits earned in these affiliates to the parent, (2) shift income from the parent to the affiliates, and (3) shift deductible expenses from the affiliates to the U.S. parent. The only case in which this is not true is in nonmanufacturing activities of Canadian subsidiaries where profits are repatriated to the U.S. parent. Here the subsidiary has a surplus of tax credits since the Canadian rate (50 per cent) exceeds the U.S. rate (46 per cent); the incentives are therefore reversed: income should be shifted to the U.S. parent and deductions to the Canadian affiliate. A Canadian multinational with U.S. and Mexican affiliates is in a different situation. The parent's rate on nonmanufacturing activities is 44 per cent; on manufacturing it drops to 36 per cent. The U.S. affiliate that repatriates profits to its Canadian parent pays a 51 per cent U.S. tax rate (which falls to 49 per cent under the 1995 tax protocol); the Mexican affiliate pays 44 per cent in taxes. If the U.S. subsidiary does not remit profits, the rate is 46 per cent.35 Thus the Canadian parent may have the lowest tax rate; this is clearly true for manufacturing MNEs, but also true vis-a-vis the United States. As a result, there is an incentive to shift income to Canada and expenses to the United States. U.S. subsidiaries also have an incentive to not remit profits to their Canadian parents in order to avoid the U.S. withholding tax. A Mexican parent firm generally finds the tax rate on its U.S. subsidiary to be higher than the Mexican rate on the parent. Thus there is an incentive to shift deductible costs into, and income out of, the U.S. affiliate, and to avoid remitting profits to the Mexican parent. The tax rate on the Canadian subsidiary, on the other hand, is either lower or equal to the Mexican rate, with the exception of remittances from a nonmanufacturing affiliate where the Canadian rate is higher (50 per cent versus 44 per cent). Where the rate is lower, the incentives are reversed. If Canadian and Mexican corporate income tax rates generally continue to remain lower than U.S. rates, as they are at present, the incentive will continue to shift profits into the spokes and out of the United States. Mexican and Canadian (manufacturing only) affiliates with U.S. parents will continue

Taxing Multinationals in Practice

363

to have small deficits of foreign tax credits, bringing their effective tax rate upon repatriation up to the U.S. level. Nonmanufacturing affiliates in Canada, on the other hand, will continue to be taxed higher than elsewhere in North America. Affiliate tax rates are lower if profits are not remitted, so that avoiding remittance of dividends reduces overall MNE taxes. The reduction of the Canadian withholding tax to 5 per cent under the 1995 Canada-U.S. tax protocol increases this deficit of U.S. foreign tax credits for manufacturing subsidiaries, providing more reason to not remit profits. Thus eliminating the Canadian withholding tax will not encourage more dividend repatriation to U.S. parents. The effective tax rate determining the North American investment decisions of U.S. multinationals, assuming profits are not repatriated, should therefore continue to be the domestic tax rate. However, eliminating the Canadian withholding tax on dividends would mean a loss in tax revenues from those MNEs that did remit profits to their foreign parents. Thus repatriation decisions are not affected, but domestic tax revenue falls, assuming CIT rates remain at present levels. On the other hand, a reduction in the U.S. withholding tax rate (as in the 1995 Canada-U.S. protocol) does make a difference to the repatriation decisions for Canadian and Mexican multinationals. Since Canada exempts active business income earned in tax treaty countries from tax, the effective rate is the host country tax rate. Lowering the U.S. withholding tax rate provides a real income gain to Canadian MNEs. Similarly, since the Mexican tax rate is below the U.S. one, remitting dividends to a Mexican parent does not incur additional Mexican tax. Thus the incentive for both spokes is to press for lower dividend withholding tax rates on a bilateral basis with the United States. A lower U.S. withholding tax rate could therefore encourage more outflows to 'spoke' parents along with lower tax revenues. Thus dividend withholding taxes, used in the past by Canada and Mexico as a 'backstop' to the corporate income tax, could cease to be an effective revenue protector. The reduction in withholding taxes increases the competitiveness of firms within the 'spoke' regions (since inward FDI increases), but at the cost of lower tax revenues.36 One potential advantage, however, of the complete or almost complete elimination of withholding taxes should be less reliance on tax haven countries as a means to move income within North America.37 The incentive to not remit profits to U.S. parents remains, however, as long as the U.S. CIT rate is higher than Canadian and Mexican rates. As we know from Chapter 6, differences in corporate income taxes are only one factor affecting the MNE's incentive to manipulate transfer prices. Tariffs are also an incentive. We therefore need to bring tariffs, and the North American Free Trade Agreement (NAFTA), into our discussion.

364

Transfer Pricing and Taxation

Transfer Pricing Problems after NAFTA Mathewson and Quirin (1979, 93) caution that when countries move to a free trade area, the scope for transfer price manipulation generally rises as crossborder tariffs disappear and trade volumes increase. Their study finds that taxes and tariffs tend to have offsetting impacts on MNE location decisions so that the combination of CITs and tariffs is 'self-policing'; they reduce the incentive for transfer price manipulation. From this the authors conclude that if tariffs are absent (such as within a free trade area), transfer pricing manipulations may increase. Free trade should therefore be accompanied by a move to harmonization of national tax levels. In fact, the reduction in withholding taxes does help harmonize tax rates among the NAFTA member countries. Under the NAFTA, over the next 15 years, all tariff and nontariff barriers among Canada, the United States, and Mexico are to be either eliminated or harmonized. Each country can maintain its own trade barriers vis-a-vis nonmember countries such as Japan and the United Kingdom, but intracontinental trade barriers should fall dramatically. Thus valuation of imported goods for customs duty purposes should cease to be a major activity of the customs authorities in the three countries. However, these government agencies will not be without employment since firms, in order to qualify for duty-free status, must meet certain rule-of-origin tests. Does this mean the scope for transfer pricing after NAFTA is reduced? We investigate this issue in two parts by looking, first, at rules of origin and, second, at tax differentials and tariffs after NAFTA. Transfer Pricing and Rules of Origin: The Honda Civic Dispute. A recent example of a dispute over rules of origin in a free trade agreement is the wellknown case of Honda of America, in which about 100,000 engines made in Honda's plant in Anna, Ohio, were shipped yearly to its Canadian assembly plant in Alliston, Ontario. The Canadian assembler then shipped approximately 90,000 finished Honda Civics back to the United States for final sale. In 1991, after the Canada-U.S. Free Trade Agreement (FTA) was in place, a dispute arose between the U.S. government and Honda when U.S. customs authorities argued that the Civic did not meet the FTA rule-of-origin test (50 per cent of the value had to originate in North America), and levied customs duties on the imported cars. The U.S. customs assessed Honda with the 2.5 per cent U.S. tariff on finished vehicle imports, approximately US$200 per car for total duties of Can$22 million (US$17 million). 38 The determination of North American content depended on the value assigned to: (1) parts and components imported into Canada and the United States from outside North America, (2) parts and components bought from

Taxing Multinationals in Practice

365

Honda's affiliates within North America, (3) services provided by a member of the Honda family to the U.S. and/or Canadian affiliates, (4) intangibles such as technology transfers, and (5) intracorporate financial flows such as loans and dividend payments. All of these prices, to the extent that the products were traded among affiliates of Honda, were transfer prices. Manipulation of these prices could affect Honda's ability to satisfy the FTA rules of origin; for example, underinvoicing parts coming from outside North America and overinvoicing locally made parts would increase the North American content. The choice of transfer prices not only affects regional content but also the gross profit declared in Canada and the United States. For example, charging a higher transfer price for an automobile engine shipped from Anna, Ohio, to Alliston, Ontario, shifts profits out of Honda's Ontario plant and into its Ohio plant with subsequent effects on the division of taxes paid by Honda to the Canadian and U.S. governments. The Canadian government supported Honda's side in the dispute, and the two governments argued over what factors should or should not be included in calculation of value added (for example, whether pollution control equipment, company uniforms, and interest costs on buildings should be included).39 The issue was settled in the 1994 North American Free Trade Agreement (NAFTA) by more clearly defining what did and did not count in the definition of North American content, so that Honda did not have to pay the assessed duties. Less attention at the time, however, was paid to the question of whether Honda had set the 'appropriate' transfer prices on the intrafirm intermediate product flows that were used to determined regional content. We suspect that, as NAFTA phases in, the valuation of crossborder transactions will become a more important issue for customs authorities. They will have to focus their attention on transfer pricing manipulation as a tariff-avoidance mechanism, not just for inbound transfers of tangibles but for the broader determination of whether North American content rules are met in terms of the valuation of tangibles, intangibles, and services. Tax Differentials in a Free Trade Area: After NAFTA.Now let us return to the issue of tax differentials in a regional trade area. We jump forward to 2003 when NAFTA is supposed to be fully phased in, all tariffs eliminated, and most nontariff barriers either reduced or harmonized. There should be no or very low withholding taxes levied on dividends, royalties, or other intercorporate, crossborder financial flows. What are the implications for transfer pricing manipulation and for the tax revenues of the three countries? We outline one possible scenario as follows. The key factors affecting the answer to this question are: (1) the volume of crossborder transfers, (2) the level

366

Transfer Pricing and Taxation

of statutory and effective marginal tax rates in each country, and (3) the rigour with which national revenue authorities enforce transfer pricing regulations.40 First, we hypothesize that the hub-and-spoke nature of North American trade and investment patterns, documented in Chapter 4, should strengthen over time. Thus the role of U.S. multinationals in the Canadian and Mexican economies should increase. This means that both 'spoke' countries should experience high and increasing amounts of inbound transfers of intermediate parts and capital equipment, together with increasing outbound transfers of royalty payments, head office charges, dividends, and other service fees. We predict this will occur even in the absence of the tax treaties with their lower withholding tax rates, but the reduction of rates should encourage these crossborder flows. As the volume and complexity of crossborder activities increases, the ability of national taxing authorities to effectively measure and allocate, at the national level, the income and expenses associated with integrated continental production should significantly weaken. For example, if a U.S. multinational operates a fully integrated production network within North America, the search for comparable third-party transactions (that is, a comparable uncontrolled price, or CUP) for the variety of intermediate flows within the MNE becomes enormously difficult. In addition, deciding how to allocate the joint costs of decentralized R&D or service centres becomes a larger problem as these MNE activities increase. The risk of an arbitrary assignment of MNE costs and revenues by a national tax authority, with consequent double taxation, rises as integrated production networks expand (UNCTAD 1993, ch. X; Vernon 1994b). And, while the network of tax treaties helps to harmonize tax rates and bases, and to reduce withholding taxes on crossborder financial flows, different transfer pricing methodologies can be still used to shift profits between tax jurisdictions. For example, while both Canada and the United States follow the arm's length standard, the Canadian and U.S. interpretations of this standard differ. The U.S. regulations now allow for the comparable profits method and periodic adjustments, both of which are frowned on by Revenue Canada. The tendency for the relative importance of corporate tax revenues to fall in Canada and Mexico may be exacerbated by the Internal Revenue Service's attempts to use transfer pricing regulations and the penalty regulations as a way to raise revenues and reduce the U.S. budget deficit. 'Spoke' multinationals and U.S. subsidiaries in both spokes will be careful to choose and document transfer pricing methodologies that avoid the onerous penalty regulations. This may also result in a larger share of the tax revenues, from both U.S. and 'spoke' multinationals, shifting to the U.S. Treasury. As tariffs are eliminated among Canada, the United States, and Mexico, the responsibilities of customs authorities are reduced in one direction (e.g., concern

Taxing Multinationals in Practice

367

about underinvoicing of imported goods), but their responsibilities are increased in other directions (e.g., enforcing rules of origin that ensure goods sold within the free trade area meet North American content requirements). The latter requires a focus on both inbound and outbound transfers and on services and intangibles as well as goods inputs. Also, under NAFTA, customs authorities must provide advance pricing rulings, similar to those offered for tax purposes under the Advance Pricing Agreement (APA) process. Thus the responsibilities of customs authorities begin to look more similar to those of tax officials. It is therefore quite possible that section 1059A, which requires U.S. tax authorities to treat the customs valuation as a base price, may 'grow teeth' in the future, even if it is apparently toothless now. This could create additional tax problems. In the scenario we have painted above, tax tensions between the United States and its NAFTA tax partners may well increase, particularly if the U.S. Congress continues to threaten foreign multinationals with higher taxes. If the 'spoke' countries' share of total tax revenues from multinationals in North America should decline over time, while the 'hub' country's share rises, the resentment in Canada and Mexico against U.S. investors, long dormant in Canada and more recently in Mexico, may again become an issue. Conclusions: Transfer Pricing Incentives under NAFTA With a free trade area, tariffs and nontariff barriers are either eliminated or harmonized. This means their impact on MNE transfer pricing decisions should be reduced. Other factors, such as income tax differentials, can therefore be expected to play a larger role. While bilateral tax treaties are working to reduce and harmonize withholding taxes inside North America, effective corporate income tax rates do differ between the countries and across types of profits (e.g., dividends versus retained earnings). Thus taxes may play a larger role in MNE transfer pricing decisions within NAFTA than they have in the past. In the last chapter of this book, we return to this issue with suggestions for lessening the incentives to manipulate transfer prices and intrafirm trade flows within North America. New Evidence on Taxing Multinationals in North America In the section below, we examine U.S. Department of Commerce data for 1990 to see if we can shed additional light on income shifting by multinationals. We look first at foreign income taxes paid by U.S. MOFAs in other countries and then at U.S. income taxes paid by foreign affiliates in the United States. We turn first to the MOFA data.

368 Transfer Pricing and Taxation Foreign Income Taxes Paid by U.S. MOFAs The 1990 summary income statement compiled for all U.S. MOFAs is shown in Table 7.11. Gross income is almost entirely generated through sales, with minor sources of income from equity investments, capital gains, and other income. Subtracting cost of goods sold (COGS) and general expenses from gross income equals gross profit. As the table shows, COGS varies between 85 and 93 per cent of sales income. Subtracting other costs and expenses from gross profit gives us operating (or pre-tax) profit. This is the base on which foreign income taxes are calculated. Income taxes are deducted from operating profit, yielding net (or after-tax) income. We provide three measures of profitability in this table, all shown in percentage form. The ratio of operating profit to gross income measures pre-tax profit relative to total income. This varies from a low of 4 per cent for MOFAs in Canada to a high of 11 per cent for MOFAs in Mexico, with a worldwide average of 8 per cent. The Berry ratio (named after Charles Berry, an economist at Princeton and well-known transfer pricing expert who first developed the formula in the Du Pont case) provides another measure of profitability: gross profit relative to operating expenses. The Berry ratio reflects the same ranking as profit-toincome, with Canada being the lowest and Mexico the highest. Our last measure of MOFA profitability is the rate of return on capital employed, measured as operating (i.e., pre-tax) profits divided by operating (i.e., fixed, current) assets. The rate of return on capital employed is 15 per cent in Canada and 39 per cent in Mexico, compared with a worldwide average of 20 per cent. We also provide three measures of effective tax rates. The first measure is foreign income taxes paid divided by operating profit; this ranges from 24 per cent (Japan) to 52 per cent (EC), with the average for all MOFAs 30 per cent. This ratio should most closely approximate statutory tax rates. The next two measures are the same ones used in calculations of the tax gap between domestic corporations and foreign-controlled corporations in the United States, that is, taxes as a per cent of sales receipts and taxes as a per cent of total assets. The tax-to-sales ratio varies from a low of 1.5 per cent in Canada to 4.2 per cent in Mexico, with an average ratio of 2.6 per cent. The tax-to-assets ratio is roughly similar. This is not surprising since total MOFA assets are roughly equal to total MOFA sales, by country. The low taxes-to-sales and taxes-to-assets ratios for MOFAs in Canada probably reflect the lower profitability of these subsidiaries since taxes as a percentage of operating (pre-tax) profits are roughly similar or higher than in other countries. The relatively low profit levels may have several possible explanations. They may be a temporary phenomenon, due to the more pronounced

Taxing Multinationals in Practice

369

TABLE 7.11 Income Statement for Majority-Owned Foreign Affiliates of U.S. Parent Companies, 1990 Host country in which MOFAs are located: All Countries

Canada

Mexico

Japan

European Community

1,233,496 1,191,832 19,787 484 21,393

180,637 177,200 983 255 2,198

19,717 19,330 95 (95) 387

63,055 62,117 72 110 756

636,683 615,192 10,915 69 10,507

1,067,608

164,359

16,496

55,397

556,126

Gross profit (gross income cost of goods sold) Other costs & expenses

165,888 62,325

16,278 8,334

3,221 989

7,658 3,197

80,557 32,314

Operating profit (gross profit - other expenses) Foreign income taxes

103,563 30,658

7,944 2,658

2,232 807

4,461 2,330

48,243 11,564

72,906

5,285

1,425

2,131

36,679

528,668 1,263,457

54,201 182,063

5,736 13,993

36,289 61,696

300,799 659,920

8.4%

4.4%

11.3%

7.1%

7.6%

266.2%

195.3%

325.7%

239.5%

249.3%

19.6%

14.7%

38.9%

12.3%

16.0%

29.6%

33.5%

36.2%

52.2%

24.0%

2.6%

1.5%

4.2%

3.8%

1.9%

2.4%

1.5%

5.8%

3.8%

1.8%

Income Statement (US$ millions) Gross income Sales income Income from equity investments Capital gains or losses Other income Cost of goods sold & general expenses

Net income (operating profit - taxes) Operating assets (fixed, current assets) Total assets Ratio of operating profit to gross income Berry ratio (gross profit/operating expenses) Rate of return on capital employed (operating profit/operating assets) Average effective foreign tax ratios Foreign income taxes/operating profit Foreign income taxes/sales receipts Foreign income taxes/total assets

SOURCE: Author's calculations based on U.S. Department of Commerce (1993c, Tables 90-22 and 90-30), as reported in the National Trade Data Bank - the Export Connection

370

Transfer Pricing and Taxation

recession in Canada in 1990. The U.S. subsidiaries could be engaging in transfer price manipulation in order to reduce their Canadian tax payments (although this is unlikely, at least for manufacturing affiliates, given the lower tax rates in Canada; see Eden [1988b, 1991b] and tables 7.9 and 7.10). Alternatively, MOFAs in Canada may be higher-cost affiliates, due to factors such as higher taxes and wages or an overvalued Canadian dollar. If this is not a temporary phenomenon, it should be cause for concern as it will affect Canada's ability to retain its share of U.S. outward foreign direct investment in the future. U.S. Income Taxes Paid by Foreign Affiliates in the United States A similar income statement for U.S. affiliates with foreign parents (FAs or FCCs) is shown in Table 7.12. As we saw with U.S. MOFAs, sales income accounts for almost all income for FCCs in the United States. COGS relative to sales income is very high, ranging from 97 to 102 per cent; that is, in the case of Mexican-owned affiliates, ordinary costs exceed sales revenues! Clearly, at least in 1990, foreign affiliates in the United States are operating on razor-thin margins. In fact, once other costs are taken into account, as the table shows, Mexican and Japanese affiliates make pre-tax losses. All foreign affiliates in total earn only $5.367 billion in pre-tax profits on $1.19 trillion in gross income. These results are consistent with the newspaper reports that foreign MNEs in the United States earn little profits because they have high operating costs relative to sales revenues. The low profits of FCCs in the United States can have many explanations, including: (1) overinvoicing of inbound intrafirm transfers to avoid U.S. tax, (2) underpricing products to U.S. consumers in order to gain market share, (3) poor sales due to the 1990-2 U.S. recession, (4) start-up costs associated with new FCCs, and (5) the fall in the U.S. dollar vis-a-vis the Japanese yen and German mark. In terms of profitability measures, operating profit as a percentage of gross income is very low, with the highest being 2 per cent (Canada), excluding countries where overall losses were earned. The Berry ratios range from 88 to 151 per cent, again excluding Mexican FCCs (which incurred losses). The return on capital employed is also either low or negative, with an all-country average of only 2.3 per cent. Clearly, foreign controlled affiliates in the United States in 1990 are not doing well. Note that in two of the three cases, Canadian-owned affiliates record higher profitability measures than do firms from other countries. With the exception of the Mexican affiliates, the other FCCs, however, do pay U.S. income taxes and at high rates relative to their operating profits. As a percentage of operating profits, ignoring Japan and Mexico where the FCCs

TABLE 7.12 Income Statement for U.S. Affiliates with Foreign Parent Companies, 1990 Country of ultimate beneficial owner: All Countries

Canada

Mexico

Japan

European Community

15,318 1,145,232

131,019 127,097 1,974 160 1,787 123,023

2,864 2,851 1 (16) 28 2,895

317,392 313,138 831 89 3,334 312,271

503,732 494,936 1,898 63 6,835 479,895

Gross profit (gross income - cost of goods sold) Other costs & expenses

44,770 39,403

7,996 5,288

(31) 48

5,121 5,860

23,837 17,890

Operating profit (gross profit ± other expenses) U.S. income taxes

5,367 10,258

2,708 1,058

(79) (1)

(739) 1,452

5,947 6,265

Net income (operating profit - taxes)

(4,891)

1,651

(77)

(2,191)

(318)

238,859 1,529,778

33,639 227,509

123 3,045

67,079 369,669

68,172 583,184

0.45% 113.62%

2.07% 151.21%

-2.76% -64.58%

-0.23% 87.39%

1.18% 133.24%

2.25%

8.05%

-64.23%

-1.10%

8.72%

191.13% 0.88% 0.67%

39.07% 0.83% 0.47%

1.27% -0.04% -0.03%

-196.48% 0.46% 0.39%

105.35% 1.27% 1.07%

Income statement (US$ millions) Gross income Sales income Income from equity investments Capital gains or losses Other income Cost of goods sold & general expenses

Operating assets (fixed, current assets) Total assets Ratio of operating profit to gross income Berry ratio (gross profit/operating expenses) Rate of return on capital employed (operating profit/operating assets) Average effective U.S. tax rate measures U.S. income taxes/operating profit U.S. income taxes/sales receipts U.S. income taxes/total assets

1,190,002 1,168,490 5,873 320

SOURCE: Author's calculations based on U.S. Department of Commerce (1993a, Tables B-3, B-6, and E-2), as reported in the National Trade Data Bank - The Export Connection

372 Transfer Pricing and Taxation made overall losses, the effective tax rate ranges from a low of 39 per cent (Canada) to a high of 191 per cent (all countries). Japanese FCCs pay large taxes even though they show losses overall. This suggests that many of the costs incurred are not deductible or creditable against the U.S. corporate income tax. These results clearly do not square with the view that foreign MNEs in the United States pay little or no U.S. tax as a percentage of gross profits. The last two rows in this table provide data on taxes-to-sales and taxes-toassets ratios. In terms of sales, the tax ratios range from 0.5 to 1.3 per cent, excluding Mexican FCCs. In terms of assets, the tax ratios are lower than those for sales because assets-to-sales ratios are greater than unity for FCCs. Note that both ratios are much lower for Japanese affiliates than for all foreign affiliates. This reflects the losses of Japanese FCCs, due primarily to the high cost of goods sold relative to sales revenue (COGS is 99.7 per cent of sales). U.K. affiliates pay much higher taxes as a percentage of assets and sales, relative to all FCCs, because of their greater profitability. Comparing the Income Taxes Paid by MOFAs and FCCs How do the income statements, especially the profitability and tax measures, of U.S. affiliates abroad compare with the measures for foreign firms in the United States? We can provide some light on this by pulling together the information from tables 7.11 and 7.12 in Table 7.13. The contrast in profitability is stark. FCCs in the United States in 1990 are much less profitable than MOFAs are abroad, regardless of which measure is used. The gap is most noticeable for all countries in terms of profits as a percentage of gross income (8.4 versus 0.5 per cent) and rate of return on capital employed (19.6 versus 2.2 per cent). Despite these lower returns, taxes as a percentage of operating profits are actually higher on FCCs than on MOFAs, except in the Mexican case. This could reflect differences in statutory tax rates, profitability, tax credits and deductions, and/or transfer price manipulation. In practice, probably all are involved. Looking at the taxes-to-sales and taxes-toassets ratios, we see that the FCC ratios are below the corresponding MOFA ones, suggesting that the reason is lower profitability. Amerkhail (1993, 48) notes that the tax rate on sales for USCCs in 1990 is 1.0 per cent for all corporations; the taxes-to-assets ratio is 0.6 per cent. Thus taxes-to-sales are 0.9 per cent (FCCs), 1.0 per cent (USCCs), and 2.6 per cent (MOFAs), and taxes-to-assets are 0.6 per cent (USCCs), 0.7 per cent (FCCs), and 2.4 per cent (MOFAs). This suggests that a simple calculation comparing taxes paid with either sales receipts or total assets may not be a meaningful comparison, at least in the short run. Over the long run, however, such differ-

Taxing Multinationals in Practice

373

TABLE 7.13 Comparing Income Statements for MOFAs and Foreign Affiliates, by Country, 1990 (in percentage terms) All Countries

Canada

Mexico

MOFA FCC

8.4 0.5

4.4 2.1

11.3 -2.8

7.1 -0.2

7.6 1.2

Berry ratio

MOFA FCC

266.2 113.6

195.3 151.2

325.7 -64.6

239.5 87.4

249.3 133.2

Rate of return on capital employed

MOFA FCC

19.6 2.2

14.7 8.1

38.9 -64.2

12.3 -1.1

16.0 8.7

Taxes/operating profit

MOFA FCC

29.6 191.1

33.5 39.1

36.2 1.3

52.2 -196.5

24.0 105.3

Taxes/sales receipts

MOFA FCC

2.6 0.9

1.5 0.8

4.2 -0.0

3.8 0.5

1.9 13

Taxes/total assets

MOFA FCC

2.4 0.7

1.5 0.5

5.8 -0.0

3.8 0.4

1.8 1.1

MOFAs FCCs

100.0 100.0

8.7 10.3

2.6 -0.0

7.6 14.2

37.7 61.1

MOFAs FCCs

100.0 100.0

14.9 10.9

1.6 0.2

5.2 26.8

51.6 42.4

MOFAs FCCs

100.0 100.0

14.4 14.9

1.1 0.2

4.9 24.2

52.2 38.1

Japan

European Community

Profitability measures Operating profit/gross income

Taxation ratios

Country share in total Country share of taxes Country share of sales receipts Country share of assets

SOURCE: Based on data in tables 7.11 and 7.12

ences should lead to investment shifting from low-profit, high-tax locations to high-profit, low-tax locations. As far as country comparisons are concerned, MOFAs in Canada are much more profitable (for all three profit measures), and have higher taxes-to-sales and taxes-to-asset ratios, than Canadian FCCs in the United States. This suggests that the firms might be shifting profits to Canada through TPM. Another perspective paints a different picture, however. MOFAs in Canada pay 8.7 per cent of total MOFA taxes, receive 14.9 per cent of all MOFA sales receipts, and own 14.4 per cent of all MOFA assets. Therefore, relative to their share of sales and assets, MOFAs in Canada pay relatively less tax. This may be evidence of

374

Transfer Pricing and Taxation

transfer price manipulation by MOFAs to avoid Canadian taxes. Canadian FCCs, on the other hand, pay 10.3 per cent of all FCC taxes, receive 10.9 per cent of all sales, and own 14.9 per cent of all FCC assets. This suggests that Canadian FCCs may be more capital intensive than other foreign affiliates, although they do pay taxes in rough proportion to their sales revenues. Thus it is difficult to make a judgment as to whether there is any evidence of TPM, much less the direction (that is, are profits shifted north or south?) of TPM, based on tax comparisons between U.S. affiliates with Canadian parents and Canadian affiliates with U.S. parents. In the Japanese case, MOFAs in Japan are much more profitable and pay more taxes than Japanese FCCs in the United States. In comparisons with all affiliates, MOFAs in Japan pay 7.6 per cent of all MOFA taxes, receive 5.2 per cent of sales revenues, and own 4.9 per cent of assets; thus tax payments are slightly more than one would expect on the basis of assets and sales. Japanese FCCs, on the other hand, pay 14.2 per cent of taxes but receive 26.8 per cent of all FCC sales receipts and own 24.2 per cent of all FCC assets. This suggests either that short-run factors such as exchange rate or business cycle effects may be responsible, or possibly underpayment of U.S. taxes through transfer price manipulation (as is possible for MOFAs in Canada) may be occurring. In the EC case, MOFAs are also paying relatively less tax (37.7 per cent) in comparison with their shares of sales (51.6 per cent) and assets (52.2 per cent), vis-a-vis all MOFAs. EC affiliates, on the other hand, are paying substantially more in tax (61.1 per cent) than their sales (42.2 per cent) and asset (38.1 per cent) shares would suggest. It is possible that both the U.S. and EC multinationals may be shifting taxable income out of Europe and into the United States in response to U.S.-EC tax differentials. However, even if EC MNEs are overpaying U.S. taxes (the data are inconclusive), this does not necessarily imply they are underpaying EC taxes; income shifting may be occurring at the expense of a third country. The same is true for MOFAs in the Community. Without more information, it is difficult to conclude that tax abuse is or is not occurring. Taxing Foreign Affiliates in the United States: Industry Comparisons We can explore the 'foreign MNEs as tax abusers' hypothesis a bit further by looking at a breakdown of foreign affiliate tax payments and average effective tax rates on pre-tax profits, sales, and assets, by industry. Unfortunately, the U.S. Department of Commerce data do not provide as detailed a country breakdown as we have used in earlier tables, so that individual data for Mexico and the European Community are not available. We were able, however, to gather data for all FCCs and for Canadian, U.K., and Japanese affiliates in the United States.

Taxing Multinationals in Practice

375

For all foreign affiliates, positive pre-tax profits in the petroleum and manufacturing sectors (US$4.3 billion and US$5.6 billion, respectively), when paired with losses in most other sectors (especially real estate, services, and finance), result in much smaller profits overall (US$5.4 billion).41 This pattern is repeated for Canadian and U.K. affiliates. Japanese affiliates, on the other hand, show relatively large losses in the manufacturing sector (US$1.0 billion), with losses in most other industries except most notably in wholesale and retail trade and finance; the net impact is an overall loss of US$0.7 billion. The picture painted here shows large differences across industries in terms of pre-tax profits, with Japanese affiliates performing worse than the average for all FCCs. That Japanese affiliates in the manufacturing sector perform so poorly is surprising, given their high productivity (from lean production techniques, as we discussed in Chapter 3) and large share of the U.S. market in industries such as motor vehicles, steel, and electronics. Table 7.14 shows the total taxes paid by FCCs in the United States, by industry, together with three tax ratios: taxes-to-profits, taxes-to-sales, and taxes-toassets. The average corporate income tax rates, as a percentage of pre-tax profits, vary considerably across industries for MNEs from the same home country. For example, in most cases where the FCC shows a loss for 1990, the firm continues to pay U.S. income taxes.42 The tax ratios for the 'all industries' row are identical to those in Table 7.13. Looking specifically at the manufacturing sector, we see that Canadian affiliates pay the lowest taxes-to-profits rate (24 per cent), whereas the average for all FCCs is 94 per cent; Japanese affiliates pay 36 per cent of their losses in U.S. tax. The only industry making a negative tax payment is services, where Japanese firms receive net tax rebates of US$41 million on losses of US$695 million. Looking at the other two tax ratios, in the manufacturing sector, for all FCCs the ratio for sales is 1.4 per cent and for assets, 1.3 per cent. Canadian FCCs are higher (1.7 per cent, 1.4 per cent) as are U.K. FCCs (2.6 per cent, 2.2 per cent), but Japanese FCCs are significantly lower (0.6 per cent, 0.6 per cent). These figures compare with an average of 1.1 per cent for USCCs in 1990 (Amerkhail 1993, 48). Thus manufacturing FCCs as a group pay more than USCCs do, in terms of taxes as a percentage of assets and/or sales; Japanese FCCs are significant outliers in this regard. Comparing tax ratios for Japanese FCCs to those for all FCCs, we find that Japanese FCCs pay taxes that are, in general, higher than the average, relative to their sales and assets, in the following industries: food and kindred products, motor vehicle and equipment manufacturing, wholesale trade in motor vehicles, retail trade, finance, and insurance. Japanese affiliates pay relatively low taxes, compared with all FCCs, in the following industries: petroleum, all manufactur-

376 Transfer Pricing and Taxation TABLE 7.14 Taxation of Foreign Affiliates in the United States, by Industry, 1990 Country of ultimate beneficial owner All Countries

Petroleum Total manufacturing Food and kindred Chemicals & allied Primary & fab. metals Machinery Elect, equip. Other mfg. Motor vehicles Wholesale trade Motor vehicles Retail trade Finance, ex. banking Insurance Real estate Services Other industries All industries

Canada

Taxes (US$ millions)

Taxes as % of profits

Taxes as % of sales

Taxes as % of assets

Taxes Taxes (US$ as % of millions) profits

1,599 5,278

37.0 93.6 67.9 34.8 45.9 -17.2 -10.5 -531.1 -21.3 783.7 203.2 -39.3 -5.2 27.8 -8.9 -13.4 -51.4 191.1

1.4

1.6

1.4 0.6 2.3 1.0 0.6 0.4 1.4 0.7 0.4 0.4 0.4 0.3 1.0 1.0 0.7 1.1 0.9

1.3 0.6 1.9 1.0 0.7 0.5 1.3 1.0 0.8 0.8 0.6 0.0 0.3 0.2 0.3 0.8 0.7

130 1,000 (10) n.a.

266 2,558

503 502 196 1,450

110 1,348

386 274 73 652 178 248 607 10,258

77 n.a. n.a.

114 1 25 2 n.a.

10 n.a.

61 18 22 1,058

27.3 24.1 -2.1 n.a. 50.0 n.a. n.a. n.a. -2.8 357.1 -33.3 n.a. -12.3 n.a. -8.1 -11.0 -7.0 39.1

Taxes as % of sales 2.5

Taxes as % of assets 2.1

1.7

1.4

-0.2 n.a.

n.a. n.a.

1.0

1.6

n.a. n.a. n.a.

n.a. n.a.

0.1 0.3 0.6

0.7 0.1 0.5

n.a.

n.a. n.a.

0.4

0.0

n.a.

n.a.

0.9 0.8 0.2 0.8

0.2 0.6 0.1 0.5

SOURCE: Author's calculations based on U.S. Department of Commerce (Tables B-5, E-3, E-10, and E-ll), as reported in the National Trade Data Bank - The Export Connection

ing, metals, electronics and electrical equipment manufacturing, and other industries. It is perhaps not surprising that the industries where IRS scrutiny of Japanese FCCs has been the highest (wholesale and retail trade, autos, electronics) are the industries where tax ratios are relatively high. Thus, while the Japanese tax ratios overall are lower than for all FCCs, the ratios can be either higher or lower when comparisons are made on an industry-specific basis. This suggests that more detailed investigation is required before one could conclude that underpayment of U.S. taxes by Japanese MNE affiliates is, in fact, occurring. We conclude from these figures that a simple calculation of the tax gap between domestic and foreign-controlled affiliates in the United States is problematic, and at best likely to be misleading. FCCs in some cases pay higher effective tax rates than do USCCs; in other cases, the rates are lower. The tax ratios vary by country, industry, and time period, and are sensitive to profitabil-

Taxing Multinationals in Practice

377

TABLE 7.14 (concluded) Country of ultimate beneficial owner United Kingdom

Japan

Petroleum (Total manufacturing Food and kindred Chemicals & allied Primary & fab. metals Machinery Elect, equip. Other mfg. Motor vehicles Wholesale trade Motor vehicles Retail trade Finance, ex. banking Insurance Real estate Services Other industries All industries

Taxes (US$ millions)

Taxes as % of profits

Taxes as % of sales

Taxes as % of assets

Taxes Taxes (US$ as % of millions) profits

3 371 20 42 34 119 9 156 90 744 335 41 229 18 51

-6.5 -36.2 -69.0 -66.7 -485.7 -43.8 -4.3 -23.8 -80.4 98.0 93.6 89.1 68.8 40.9 -20.2

0.1

0.2

0.6 0.8 0.8 0.3 0.7 0.2 0.7 0.8 0.3 0.5 2.2 1.9 3.3 1.2

0.6 0.8 0.5 0.2 0.7 0.2 0.7 1.5 0.9 1.1 2.1 0.2 1.3 0.1

n.a. 2,113

5.9 -0.6 0.7 0.5

-0.2

(41)

36 1,452

37.1 -196.5

0.6 0.4

185 851 114 94 30 869 1 205 (5) 58 n.a.

127 12 120 362 3,796

n.a. 52.0 83.3 38.3 34.1 -38.5 -30.0 56.9 -7.7 -79.2 35.7 -18.4 n.a. 57.0 -4.0 -25.9 27.3 61.2

Taxes as % of sales

Taxes as % of assets

n.a.

n.a.

2.6 1.2 3.6 2.1 0.9 0.6 3.5 0.2 0.5

2.2 1.0 2.9 1.9 0.8 0.6 2.7 0.3 1.6

-0.5

-0.5

0.7

0.8 n.a 0.4 0.1 0.6 1.5 1.4

n.a.

1.0 0.8 1.1 3.2 2.0

ity measures. It is difficult, without further investigation, to conclude from the figures whether or not transfer price manipulation has occurred or is occurring. However, the low profits and tax payments, relative to assets and sales, across most industries, for Japanese FCCs are clearly worth more scrutiny. Summary: Taxation of Multinationals in North America In this section, we have provided new U.S. estimates of taxes paid, relative to income, sales, and assets, by multinationals in North America. In general, U.S. MOFAs are more profitable as a group than foreign affiliates in the United States, and pay relatively more taxes. However, when MOFA comparisons are made across host countries, their share of total MOFA taxes is generally lower in Canada and the European Community, relative to their shares of MOFA sales and assets. When FCC comparisons are made by home country, Japanese FCCs in particular pay low taxes relative to their shares of FCC assets and sales; EC FCCs pay significantly more.

378

Transfer Pricing and Taxation

Our estimates show that basing the definition of 'over' or 'under' payment of taxes on country-based comparisons can lead to strange results. In terms of possible transfer price manipulation, there are three noticeable patterns, relative to sales and assets: (1) underpayment of foreign income taxes by MOFAs in Canada and in the European Community, (2) overpayment of U.S. income taxes by EC affiliates, and (3) underpayment of U.S. taxes by Japanese affiliates. These patterns may be due to transfer price manipulation in response to income tax differentials or to factors totally unrelated to transfer pricing and taxation. Using these statistics to make policy can be particularly dangerous. For example, if the IRS believes that Japanese FCCs should be taxed more heavily because they are underpaying taxes in the United States, then the same argument can be used to justify higher taxation of U.S. MOFAs in other countries. For example, if the IRS were to decide on the basis of low tax ratios that Japanese FCCs should be made to pay more in tax, the rationale must be just as strong (and just as bad) for the European Community or Canada to argue that MOFAs are undertaxed and should be penalized, or that EC affiliates are overtaxed in the United States. What the statistics also suggest is that simple comparisons, focusing on taxes-to-sales or taxes-to-assets ratios for one group of MNEs - for example, FCCs in the United States - are likely to give a misleading picture of tax compliance or abuse. Such comparisons ignore other groups, are conducted at too high a level to be statistically useful, and can lead to policy error. Clearly, country- and industry-level data can only serve as general pointers to a possible problem; they should not be used to 'correct' for that 'problem' lest the correction be more distortionary than the problem it was designed to correct. At best, these statistics can be used as a 'smell test,' to flag outliers, and to point out where further investigation may be useful, but to take these numbers at their face value is likely to result in egregious errors. Conclusions This chapter has reviewed the empirical literature on transfer price manipulation, investigated the recent U.S. policy debate over tax abuse by foreign affiliates in the United States, calculated the income tax incentives to manipulate transfer prices within North America, and provided new evidence on the profitability and tax ratios for multinationals in North America. We have ranged over a large number of countries, industries, and topics in this chapter. Our literature review focused solely on evidence of transfer price manipulation and income shifting by MNEs. Such evidence comes from many sources, but primarily from host country studies of estimates of underinvoicing

Taxing Multinationals in Practice

379

outbound transfers or overinvoicing inbound transfers. The most detailed and sophisticated of these studies - the U.S. and Canadian oil industry studies failed to find any strong evidence of transfer price manipulation. Even the evidence on underpayment of U.S. taxes by foreign MNEs in the United States is mixed. Our own estimates suggest that any numbers must be carefully examined, counterfactuals investigated, and, in addition, taken with a grain of salt. This concludes Part II of Taxing Multinationals. We turn now to Part III, The Rules of the Game in North America,' where in four chapters we analyse the U.S. and Canadian tax rules as they apply to transfer pricing. Chapters 8 and 9 deal with U.S. regulations and procedures, respectively, while Chapter 10 covers the Canadian tax rules and procedures. Chapter 11 provides an extensive analysis of one tax court case on transfer pricing.

This page intentionally left blank

PART IV: THE RULES OF THE GAME IN NORTH A M E R I C A

This page intentionally left blank

8

The U.S. Tax Transfer Pricing Regulations Part I: Rules

Introduction U.S. law with respect to the pricing of intercompany transactions comes from four sources: 1. Legislation: The key piece of legislation is Internal Revenue Code (IRC) section 482 dealing with the allocation of income and deductions under the corporate income tax, but there are other IRC sections dealing with issues such as intrafirm transfers to Puerto Rico, valuing royalty payments, reporting requirements and penalties. 2. Regulations: The key ones are the U.S. Treasury 1968 regulations interpreting section 482 and the 1994 regulations (which went through three versions: 1992, 1993, and 1994), but other Treasury regulations dealing with intrafirm aspects of the Revenue Code are relevant, for example, those dealing with the advance pricing agreement process, and various Treasury publications (e.g., the 1988 White Paper on transfer pricing of intangibles). 3. Court rulings: The United States has had more Tax Court cases dealing with transfer pricing issues than any other country in the world. Normally these arise in disputes over income adjustments between a U.S. taxpayer and the U.S. Internal Revenue Service (IRS). Court decisions have been an important influence on the development of new regulations. 4. Bilateral tax treaties: The U.S. tax treaty network is extensive and follows the general pattern laid down under the OECD Model Tax Treaty. Each tax treaty includes articles dealing with transfer pricing issues, particular with dispute settlement under the Mutual Agreement Procedure. Our purpose in this chapter and the next is to provide a thorough historical

384 The Rules of the Game in North America review of U.S. treatment of transfer pricing under the corporate income tax from the first version of the legislation in 1917 up to the present. A chronological summary is provided in Appendix 8.1. We focus on three topics. This chapter deals with the general U.S. approach (underlying principles and norms, purpose, and scope) to tax transfer pricing, and the specific legislation and regulations that set out the rules for valuing intrafirm transfers. Chapter 9 focuses on the U.S. administrative procedures dealing with transfer pricing, particularly the dispute-settlement procedures. The U.S. Approach to Tax Transfer Pricing The United States has developed complicated rules for the pricing of intrafirm transactions in tangibles, intangibles (including technology), and services. We look first at the general regulations supporting IRC section 482 and at the methods for pricing tangibles developed in the late 1960s. The major thrust of the legislative changes in the 1980s was not on the valuation of goods and services but on the pricing of intangibles; we deal with intangibles next. We end with a discussion of the new section 482 regulations in their three versions: Mark I (1992, preliminary), Mark II (1993, temporary), and Mark III (1994, final). The centrepiece of the U.S. regulatory structure with respect to transfer pricing is Internal Revenue Code (IRC) section 482, 'Allocation of Income and Deductions among Tax Payers,' which states that: In the case of two or more organizations, trades, or businesses (whether or not incorporated, whether or not organized in the United States, and whether or not affiliated) owned or controlled directly or indirectly by the same interests, the Secretary may distribute, apportion, or allocate gross income, deductions, credits, or allowances between or among such organizations, trades, or businesses, if he determines that such distribution, apportionment, or allocation is necessary in order to prevent evasion of taxes or clearly to reflect the income of any of such organizations, trades or businesses. (U.S. Federal Tax Guide-2, Tax Tactics - Codes and Regulations, section 482, 5221J1

Section 482 has been accompanied by U.S. Treasury regulations which interpret it as requiring, and provide detailed instructions to ensure, that all transactions between related parties take place 'at arm's length,' that is, at the prices which uncontrolled taxpayers would charge in identical or similar circumstances. This is the arm's length standard which underpins the U.S. approach to taxing intrafirm transactions. In terms of scope, the courts have interpreted section 482 quite broadly. It covers almost any enterprise with independent tax significance, the definition

The U.S. Tax Transfer Pricing Regulations: Rules 385 of control is very general, and the transaction does not have to be tax motivated or lack business purpose. The commissioner has broad discretion and his or her allocation is presumed correct and cannot be overturned by the courts unless it is shown that he or she acted unreasonably, in an arbitrary manner or without justification. Thus the burden of proof rests with the taxpayer (Cole 1987, 12). The principles underlying section 482 are the same as those underlying the tax transfer pricing regime: neutrality and taxpayer equity. A neutral tax system, as we saw in Chapter 2, does not distort the taxpayer's choices, for example, between being an integrated business or not, or between wholly or partly owning or controlling one's affiliates. Section 482 places a controlled taxpayer on an even par with an uncontrolled taxpayer, making the corporate tax system in terms of valuation of transactions neutral with respect to organizational form. Taxpayer equity is also an underlying principle. Since section 482 applies both to intrafirm trade within the United States and in international transactions, resident taxpayers in principle receive the same treatment on both their domestic and international transactions. IRC section 482 has two purposes: to prevent tax evasion, and to ensure that the true taxable income of firms in a related group is reported for tax purposes. The IRS commissioner can only reallocate MNE income for these two reasons. The prevention of tax evasion was also identified in Chapter 2 as one of the purposes underlying the tax transfer pricing regime. The second purpose of the TIP regime was the prevention of double taxation of income. However, there is nothing in the wording of section 482 that implies a commitment to preventing double taxation of income. While only the commissioner can invoke section 482, the regulations that accompany section 482 do allow for appropriate offsetting income adjustments to other parties involved in the allocation in order to prevent double taxation.2 In fact, the second stated purpose of section 482 is to 'clearly reflect the income' of the taxpayers, which implies quite a different focus: the need to protect the U.S. tax base. The importance of this goal will be apparent as we proceed through our historical review of the U.S. rules for valuing intrafirm transactions. Section 482: Allocation of Income and Deductions3 In the United States, as in most countries, the actual legislation on transfer pricing is typically four or five lines; but the regulations consist of pages and pages of details. In Internal Revenue Code (IRC) section 482, the government expresses its commitment to the arm's length standard; in the Treasury regulations, the government spells out what the norm means in practice, where 'practice' involves all the thousands of ways in which companies engage in intrafirm

386 The Rules of the Game in North America transactions. We look first at the history of the 482 legislation, and then turn to the regulations. Section 482: From 1917 to 1968 Section 482 has been part of U.S. tax history since the United States adopted a corporate income tax in 1917. The 1917 tax code, in one sentence related to transfer pricing, authorized the commissioner to allocate income and deductions among related corporations. The 1928 Revenue Act added two rationales for reallocation of income: to prevent tax avoidance and to determine the true taxable income of the parties. In 1935, the U.S. Treasury published a regulation defining the standard the commissioner was to use in his or her allocation of income as the arm's length standard: that of an uncontrolled taxpayer dealing at arm's length with another uncontrolled taxpayer. Section 482 remained essentially unchanged from 1935 until, in 1986, the U.S. Congress added a sentence on the pricing of intangibles. Through most of the early part of the twentieth century, the U.S. government adopted a liberal attitude towards related party sales. International transactions were a small percentage of intercompany transactions because most firms did .not have overseas affiliates. Section 482 applied primarily to intrafirm sales within the United States where the net corporate income tax loss at the federal level from over- or underinvoicing transactions would be minimal. The IRS commissioner would simply reallocate income or deductions between the related entities, using correlative adjustments to leave the total amount of MNE income unchanged; that is, if firm A overinvoiced sales to firm B, the commissioner would reduce A's income and increase B's. Since there were so few offshore transactions, there was no strict prohibition against MNE transfers of tangibles or intangibles offshore. Tax assessments and appeals, including international cases, were handled within each district IRS office. There was little attempt to develop an overall departmental approach to transfer pricing. There were few court cases and the courts, when dealing with transfer pricing cases, used different standards (U.S. Department of the Treasury 1988, 6-8). After the end of the Second World War, U.S. multinationals rapidly expanded their businesses abroad, using their comparative strengths in research and development, brand names and advertising, and their large size to explore new markets and set up branch plants around the world. Most foreign direct investment was one of two kinds: market seeking, in which subsidiaries in industries such as autos and consumer durables produced the same products as their parents using U.S. technology for sale in foreign markets, or resource seeking, in

The U.S. Tax Transfer Pricing Regulations: Rules

387

which foreign subsidiaries used U.S. technology to extract raw materials (e.g., crude oil, wood pulp, minerals) and then shipped them to their parents for further processing and final sale. There were several types of international transactions through which aggregate taxes could be reduced. U.S. multinationals could underinvoice exports to, and overinvoice imports from, affiliates in tax haven countries, reducing U.S. income and raising foreign affiliate income. Costs of research and development could be declared in the United States (reducing U.S. taxes) while profits from these intangibles were kept abroad through underinvoicing royalty payments to U.S. parent firms. Other financial manoeuvres such as manipulating interest charges, leading and lagging on payments, and so on exacerbated the problem. The huge growth in outward FDI meant that by 1960 transfer pricing had become an important tax issue. Using transfer prices to shift income among related parties in different U.S. states was one thing; shifting income between the United States and a foreign country was clearly another. In 1961, IRS Commissioner Caplin testified before the House Ways and Means Committee that section 482 was not protecting the U.S. government's taxing jurisdiction. The Ways and Means Committee in 1962 proposed adding a subsection to 482 that would have used a formula to apportion taxable income between the parties based on their relative economic activities. The Senate Finance Committee disagreed, arguing that section 482 already contained broad authority to allocate income and deductions, and authorized the U.S. Treasury to develop regulations that would provide guidelines and formulas with respect to international income.4 In 1968, after several years of work, the Treasury issued the final version of its regulations on intercompany transactions providing detailed guidelines for the arm's length standard. The 1968 U.S. Treasury Regulations The 1968 Treasury regulations accompanying section 482 define the arm's length standard and show how firms are to apply the standard to five different types of transactions: loans or advances, business services, rental of tangible property, use or transfer of intangibles, and sales of tangibles. For the last category, four methods for calculating an arm's length price are defined: the comparable uncontrolled price (CUP) method, the resale price (RP) method, the cost plus (C+) method, and other methods. Let us look at these regulations in more detail. The first part of the regulations, 1.482-1, focuses on the arm's length standard in general. Treasury Regulation 1.482-1 (a) specifies which businesses are affected by section 482.5 The threshold test is that the parties must be owned or controlled by the same interests. Control is defined very broadly as 'direct or

388

The Rules of the Game in North America

indirect, whether or not legally enforceable, and however exercisable or exercised ... [T]he reality of the control... is decisive, not its form or the mode of its exercise' (1.482-1 [a] [3]). If income or deductions have been artificially shifted, section 482(1) presumes that control exists.6 The regulations then define the benchmark standard. Treasury Regulation 1.482-l(b)(l) says: The standard to be applied in every case is that of an uncontrolled taxpayer dealing at arm's length with another taxpayer.' Arm's length dealings are to be examined on a transactional basis. Section (c) elaborates: (c) Application. Transactions between one controlled taxpayer and another will be subjected to special scrutiny to ascertain whether the common control is being used to reduce, avoid, or escape taxes. In determining the true taxable income of a controlled taxpayer, the district director is not restricted to the case of a fraudulent, colourable, or sham transactions, or to the case of a device designed to reduce or avoid tax by shifting or distorting income, deductions, credits, or allowances. The authority to determine true taxable income extends to any case in which either by inadvertence or design the taxable income, in whole or in part, of a controlled taxpayer, is other than it would have been had the taxpayer in the conduct of his affairs been an uncontrolled taxpayer dealing at arm's length with the other uncontrolled taxpayer. (Treasury Regs. 1.482-l[c])7 Regulation 1.482-l(d)(3) defines the arm's length standard as: 'the amount which ... would have been charged in independent transactions with unrelated parties under the same or similar circumstances considering all the relevant facts and without regard to the rules found in Method 1.482-2.' Thus 482 is designed to put controlled taxpayers on a parity with uncontrolled taxpayers, and 'true taxable income' is defined in terms of the 'taxable income that would have resulted to a controlled taxpayer had it in the conduct of its affairs dealt with the other member or members of the group at arm's length' (U.S. IRS 1988, 517). The taxpayer is therefore expected to deal with related members the same as with unrelated parties. Section 1.482-2 then provides the detailed methods and guidelines for determining arm's length prices for five types of transactions. Regulation 1.482-2(a) deals with loans or advances by one member of a controlled group to another. If loans or advances are part of the regular business of the creditor, an arm's length interest rate is to be charged; otherwise a rate based on the facts and circumstances or a 'safe haven rate' is acceptable. Section 1.482-2(b), on the performance of services, states that, if the provision of services is part of the regular business8 of the member, costs plus a reasonable profit are allowed, whereas in other cases an arm's length charge covers only the costs, both direct and indirect, incurred by the member providing the

The U.S. Tax Transfer Pricing Regulations: Rules 389 service. Business services are often transferred by MNE parents in connection with the use or transfer of intangibles to foreign affiliates. Where services are rendered in connection with transfers of intangibles (i.e., are subsidiary and ancillary to the intangibles transfer), the value of the services should be built into the royalty charged for the intangibles transfer (see section 2[d] below) and not charged separately (Fuller 1989, 18-26).9 Section 1.482-2(c) deals with the use of tangible property and argues for an arm's length rental charge, generally equal to allowable depreciation plus 3 per cent per year on the original basis of the property plus direct and indirect expenses. Therefore in all three sections - a, b, c - if the activity is an integral part of the business a full arm's length charge including a profit component must be charged, whereas in other cases only costs are to be covered. Section 1.482-2(d), on the use or transfer of intangible property such as patents, copyrights, and trademarks, also requires the arm's length standard.10 However, since comparables are difficult to find, the section recommends that the commissioner make an allocation, taking into account twelve factors: (1) prevailing rates of return in the industry for similar property, (2) competing offers and bids, (3) terms of the transfer, (4) uniqueness of the property and the length of time it is expected to remain unique, (5) degree and duration of protection of property rights in the relevant countries, (6) value of services rendered by the transferor to the transferee, (7) prospective profits to be realized, or costs to be saved, through use or transfer of the property, (8) capital investment and start-up costs of the transferee, (9) availability of substitutes, (10) arm's length rates and prices paid by unrelated parties, (11) costs incurred by the transferor in developing the property, and (12) other relevant facts and circumstances (Hellawell and Pugh 1987, 186; King 1994b, 17-18). As an alternative to intracorporate licensing of technology transfers, the 1968 regulations allow cost-sharing arrangements in which R&D expenses incurred in developing the intangible are shared by the parties and the participants are deemed to have joint ownership of the intangibles so developed. If the related parties are engaged in a bona fide cost-sharing arrangement in which costs and risks for research and development of intangibles are shared on an arm's length basis, then the 1968 regulations do not require any further charge for intangible property.11 Section 1.482-2(e), perhaps the best known of the subsections of 482, deals with sales of tangible property between related taxpayers. Regulation 1.4822(e)(l)(i) notes that '[s]ince unrelated parties normally sell products at a profit, an arm's length price normally involves a profit to the seller.' Section 1.4822(e)(l)(ii) specifies the hierarchy of methods that must be used to determine the arm's length charge:

390 The Rules of the Game in North America (a) Comparable uncontrolled price (CUP) method: A comparison with sales to unrelated entities is deemed the most appropriate methodology. The major problem is how can a taxpayer validly prove a comparable uncontrolled price? Factors in doing so would include the terms of sale, the value of intangible property tied thereto, the timing of the sale, the conditions in the marketplace, and the seller's share of the market, quality differences, transportation costs, and the like. (b) Resale price (RP) method: Arm's-length price is computed by looking to a sale by a related buyer to its customer, reduced by an appropriate mark-up for the intermediary. (c) Cost plus (C+) method: The seller's price is computed by multiplying the cost of production by an appropriate gross profit percentage, with adjustments.

The three methods are to be applied in their stated order. Only if there are no comparable uncontrolled transactions can methods two or three be used, according to 1.482-2(e)(l)(ii). If none of the three methods 'can reasonably be applied under the facts and circumstances,' Regulation 1.482-2(e)(l)(iii) allows any other appropriate other or fourth method, such as pricing component parts, a profit split, computing the rate of return on investment, or comparison with industry-wide gross profit margins. The history of the various methods for pricing intrafirm transactions, as developed by the U.S. Treasury, is presented in Box 8.1. The left-hand column summarizes the 1968 regulations and serves as a useful benchmark of comparison with the changes that follow. The box looks at three areas: the definition of the arm's length standard, the methods for pricing of tangibles, and the methods for pricing intangibles. Applying the Arm's Length Standard to Tangibles The arm's length standard, as defined in the 1968 regulations, proposes a hierarchy of methods for valuing intercompany transactions: comparable uncontrolled price, resale price, cost plus, and fourth methods, with CUP dominating all other methods. In the section below, we look briefly at how these methods have been used by the Service. We then briefly describe the valuation methods used by customs officials. In 1986, Congress added the customs valuation as a fifth pricing method for intrafirm transactions in tangibles. The 1968 Treasury Regulations in Practice Neither the Service nor the U.S. taxpayer has found the pricing methods for tangibles simple to interpret in practice.12 In fact, the most popular fourth method at least in the U.S. tax courts - has been the profit split, in which total profits of

BOX 8.1 The Evolution of U.S. Methods for Pricing Intrafirm Transfers 482 Treasury regulations (1968)

Treasury White Paper (1988)

482 proposed regulations (1992)

482 temporary regulations (1993)

482 final regulations (1994)

The arm's length standard The standard is that of an uncontrolled taxpayer dealing at arm's length with another uncontrolled taxpayer. There is one arm's length price.

Arrangements made between unrelated parties if they could choose to have the costs of (i.e., use the same technology as) the related parties.

The standard is met if unrelated parties agree to the same contractual terms under the same economic conditions and circumstances as the controlled parties, given the knowledge and experience of the controlled parties. There is one arm's length price.

The result of a controlled transaction must be consistent with the results that would have been realized had uncontrolled taxpayers engaged in a comparable transaction under comparable circumstances. Transactions need not be identical, but must be sufficiently similar to provide a reasonable, reliable benchmark to determine if the controlled transaction was an arm's length result. There can be a range of arm's length prices.

Controlled taxpayers are expected to realize from their controlled transactions the results that would have been realized if uncontrolled taxpayers had engaged the same transaction under the same circumstancessss. An uncontrolled transaction need not be identical to the controlled one, but must be sufficiently similar that it provides a reliable measure of an arm's length result. A range of arm's length results is possible.

The arm's length standard applied to transfers of tangibles Methods ranked in priority: CUP, RP, C+, other methods.

Methods ranked in priority: CUP, RP, C+, other methods.

Methods ranked in priority: CUP first; RP and C+ have equal priority, followed by other methods.

Best method rule applies. All methods have equal priority except other methods. CUP expected to have first priority where comparables exist.

Best method rule applies. No priority given to methods but: (1) CUP takes priority where good comparables exist, (2) CPM and the profit split method are expected to be methods of last resort.

BOX 8.1 (continued) 482 Treasury regulations (1968)

Treasury White Paper (1988)

482 proposed regulations (1992)

482 temporary regulations (1993)

482 final regulations (1994)

CUP

CUP

CUP

CUP

CUP

RP

RP

RP

RP

RP

c+

C+

C+

C+

C+

Fourth methods

Fourth methods

CPM using a comparable profit interval (CPI). CPM must be used to verify non-CUP methods. Can replace non-CUP methods.

CPM, as check or replacement for non-CUP methods.

CPM. Expected to be method of last resort. Used where outside data on comparable transactions are unreliable.

Other methods

Other methods; includes profit split method (PS), which requires tax return disclosure, election, and contemporaneous documentation.

Profit split (PS). Expected to be method of last resort. Used where outside data on comparable transactions are unreliable. Unspecified methods.

The arm's length standard applied to transfers of intangibles Related parties should earn the same returns that unrelated parties would earn under comparable circumstances.

Commensurate with income (CWI) standard, with periodic adjustments

Commensurate with income (CWI) standard, with periodic adjustments. Methods are priorized: MTM, CATM, CPM. All methods require CPI test.

Commensurate with income (CWI) standard, with periodic adjustments. Equal priority for CUP and CPM. Other methods require tax return disclosure.

Commensurate with income (CWI) standard, with periodic adjustments required in most circumstances. Best method rule with equal priority for all methods.

BOX 8.1 (concluded) 482 Treasury regulations (1968)

Treasury White Paper (1988)

482 proposed regulations (1992)

482 temporary regulations (1993)

482 final regulations (1994)

CUP

Pricing method #1: exact comparables

Matching transactions method (MTM)

Comparable uncontrolled transaction method (CUT)

Comparable uncontrolled transaction method (CUT)

Pricing method #2: inexact comparables

Comparable adjustable transactions method (CATM)

Income method #1: basic arm's length return method (BALRM)

Comparable profits method (CPM), based on comparable profit interval (CPI).

Comparable profits method (CPM)

Comparable profits method (CPM)

Income Method #2: BALRM with profit split (BALRM-PS)

Total profit split or residual profit split, only for nonroutine intangibles

Profit split method (PS) (proposed only)

Profit split method (PS)

Fourth methods

Other methods

Other methods

Unspecified methods

A bona fide costsharing arrangement must be based on the cost/benefit principle (direct and indirect costs proportionate to the benefits expected).

A qualified cost-sharing arrangement means each participant must share in the costs proportionate to the expected benefit from exploitation of the intangible (based on 1988 White Paper).

Same as the 1992 proposed regulations for a qualified cost-sharing arrangement.

Final regulations, issued in late 1995, define a qualified cost-sharing arrangement as an agreement where the parties share the costs of development intangibles in proportion to their shares of reasonably anticipated benefits that arise from individually exploiting the assigned interests in the intangibles developed under the arrangement.

If there is no CUP, an analysis of 12 factors applies. The factors are not priorized or weighted.

A bona fide costsharing arrangement must reflect the good faith efforts of the participants to share respective costs and risks of development on an arm's length basis.

SOURCE: Based on U.S. Department of the Treasury (1988), U.S. Internal Revenue Service (1992, 1993b, 1994b), Carlson et al. (1994).

394

The Rules of the Game in North America

TABLE 8.1 Surveys of IRS Transfer Pricing Methods, 1972-1987 Percentage of cases where various pricing methods were used Survey report 1972 Conference Board report on corporate experience with section 482 1973 Treasury Report on the IRS's international cases involving section 482 1980 corporate survey by Jane Burns on how IRS applies Section 482 1981 General Accounting Office report on the IRS and protecting U.S. tax interests 1984 IRS survey of IRS international examiners on frequency of use of various transfer pricing methods 1987 IRS survey of IRS international examiners on frequency of use of various methods - all transactions - tangible property Average, all 482 surveys

CUP method

Resale price

Cost plus

Fourth methods

28

13

23

36

20

11

27

40

24

14

30

32

15

14

26

47

41

7

7

45

32 31

8 18

24 37

36 14

27

11

23

39

SOURCE: U.S. Department of the Treasury (1988, 22)

the related parties are divided among the individual members according to some formula (Hellawell and Pugh 1987, 178). Table 8.1 provides some evidence on the relative use of the four transfer pricing methods by the Service in its audits, and summarizes the results of different surveys of IRS examiners completed between 1973 and the present. For all types of transfers, the ranking of usage rates was fourth methods (on average, 39 per cent of the time), CUP (27 per cent), cost plus (23 per cent), and the resale price method (11 per cent).13 Clearly, CUP is not the most common method for evaluating intrafirm transactions (although note the aberration in the 1984 survey). The CUP method has clear appeal because, on paper, it looks simple to implement. The method uses marketplace results as its benchmark: all one has to find is a price two unrelated parties reached in arm's length negotiations for similar goods sold in similar circumstances and use that price instead of the transfer price in the related party transaction. Thus comparables are the key to using the CUP method.

The U.S. Tax Transfer Pricing Regulations: Rules 395 Comparability, however, is also CUP's key weakness since what are similar transactions in the eyes of the IRS, may not be in the eyes of the taxpayer, the courts, or expert witnesses, and vice versa. How comparable are the comparables? If there are some differences, are they minor, and can the market prices be adjusted for these complications? Standards of comparability depend on the facts and circumstances of the case, and these can, and will often be, interpreted differently by different parties.14 Thus disagreements are bound to arise, even from as apparently simple a method as the CUP. As we show below, because of these problems, the IRS has allocated a significant amount of the new 482 regulations to defining standards of comparability for the various transfer pricing methods. Whereas CUP looks to firms with comparable arm's length transactions, the RP and C+ methods look to firms performing comparable functions. The resale price method is normally appropriate if there are no comparable uncontrolled transactions, the buyer acts as a distributor of the seller's product, and the buyer only adds a small amount to the value of the product (e.g., packaging, labelling). Under the RP method, the buyer's resale price is reduced by an appropriate mark-up percentage to determine the arm's length price. Comparables are established by reference to resales by other firms in the same or a similar market. The criteria for comparability include type of property, functions performed and intangibles used by the re-seller, and the geographic market of the re-seller. The cost plus method is normally appropriate if there are no comparable uncontrolled transactions (implying that CUP cannot be used), and one firm sells an intermediate product to a downstream firm. The downstream firm engages in substantial manufacturing, processing, or assembly prior to final sale, so that RP cannot be used either. In the C+ method, the cost of production is determined and a margin representing gross profit percentage is added to determine the fair market value. Comparability is established by reference to similar products being produced under similar circumstances, where similarity depends on type of property, functions performed by the seller, intangibles involved, and geographic market of the seller. Section 1059A: Customs Valuation as a Fifth Method15 In the 1986 Tax Reform Act, Congress introduced a fifth method for valuing tangibles. Section 1059A was designed to link the transfer price for customs purposes with the transfer price for income tax purposes. Historically, customs authorities and tax authorities in the United States worked independently to value intrafirm transactions and did not use the same pricing techniques. They were in different departments, but the reasons went deeper than geographic location and function.

396 The Rules of the Game in North America Tariff and tax officials tend to have competing objectives, since a high transfer price on imports raises per-unit and probably total customs revenues (and offers protection to domestic import competing firms) but lowers taxable income and tax payments, ceteris paribus. Thus tax officials worry about overinvoicing of inbound transfers while customs authorities worry about underinvoicing. In addition, customs valuations occur much sooner than tax evaluations, since tax audits often take place several years after the goods have crossed borders but customs valuations must be completed within three to six months. In the 1970s, U.S. customs used to apply nine separate methods to value imports, including several designed to raise tariff revenues and offer protection to local firms. Since the Tokyo Customs Valuation Code was introduced into U.S. customs law in 1980 as U.S. Customs Code section 1401, the nine methods have been reduced to four, in descending order of priority: (1) transactions value, (2) deductive value (similar to the resale price method), (3) computed value (similar to the cost plus method), and (4) derived value (an 'any other method' that looks like the fourth method under IRC section 482). Transactions value is defined as: the actual price of the goods, adjusted for certain costs incurred by the buyer with respect to packaging, selling, commissions, related royalty or license fees, plus any proceeds payable to the seller by buyer on resale of the goods, and an apportioned part of any 'assists' supplied or paid for by the buyer. (Singer and Karlin 1983, 230)

Transactions value is acceptable to U.S. customs authorities only if (1) the parties are unrelated, or (2) the parties are related but the relationship did not influence the price, or (3) the transactions value is approximately the same as that charged for comparable goods sold to unrelated buyers. Where transactions value cannot be used, one of the other three methods must be substituted. In 1986, section 1059A of the Tax Reform Act was passed which requires the IRS commissioner to take the customs valuation into account in determining the transfer price for tax purposes. In the legislation, the customs valuation now acts as an effective ceiling price since the costs taken into account in computing value for tax purposes may be no higher than those used for tariff purposes. In this manner the option to set one low price for tariff purposes and another higher one for tax purposes is eliminated - but only for imported goods, not services or intangibles, by related persons into the United States. Section 1059A therefore is a one-way constraint, since the price of inbound transfers for income tax purposes cannot be higher than the price for customs duty purposes, but it can be lower.

The U.S. Tax Transfer Pricing Regulations: Rules 397 Section 1059A has been somewhat problematic in practice since the customs valuation and tax valuation are computed differently (Lindsay 1987, 52). Take the following example.16 Suppose the US price paid for good X is $100, freight to the U.S.-Canada border is $15, and freight to the Canadian warehouse is $10. Suppose the Canadian import duty is $10 (10 per cent of $ 100), and that the agent is paid a buying commission of $10. Therefore the cost of goods sold, the basis for section 482, is $145, whereas the valuation for customs duty is $100. A recent IRS ruling on section 1059A, reviewed in Fuller (1994a), deals with this issue. Where the customs valuation does not include amounts that should be included in the cost basis or inventory cost for income tax purposes, but are not appropriate. for customs valuation purposes, the taxpayer may increase the customs valuation by these amounts to determine the transfer price for tax purposes. Applying the Arm's Length Standard to Intangibles One of the most contentious issues related to intercompany pricing in U.S. tax law has been the pricing of intangibles. Historically, if a U.S. parent transferred technology through a sale or a licence to its foreign affiliates, the intangible was subject to the arm's length pricing standard of section 482 at the time of transfer. If, however, the transfer did not take the form of a sale or licence, U.S. multinationals could use IRC section 351 to transfer the intangibles without recognizing taxable income at the time of the transfer. There was no general rule within the U.S. Internal Revenue Code requiring that all transfers of intangibles be treated as a sale or disposition at fair market value or as a contribution to capital. U.S. MNEs could therefore avoid the tax consequences of technology transfers to their foreign affiliates. The MNEs could transfer technology, developed in the United States and written off for tax purposes against the U.S. corporate income tax, to foreign subsidiaries acting as contract manufacturers in low-tax locations.17 The subsidiaries acted as the manufacturing division, providing contract services without having to incur any of the risks, R&D costs, or proprietary interest in the products being produced since these costs were incurred by their U.S. parents. Profits would be declared in the low-tax subsidiary while deductible R&D costs were declared in the United States. Thus, U.S. parents could effectively reduce their U.S. income taxes by not recognizing taxable income at the time of the transfer and then subsequently by undercharging for royalties based on the foreign income generated with the transferred assets.18 Since the early 1980s, the U.S. Congress has attempted to plug this loophole through various pieces of legislation together with Treasury regulations. In this section, we provide a brief history of the changes in U.S. tax law

398

The Rules of the Game in North America

affecting intercompany transfer pricing of intangibles. We begin with legislative changes relating to Puerto Rican affiliates and to imputed royalty payments in general and then turn to the 1986 commensurate with income standard and the 1988 Treasury White Paper. Section 936: Puerto RicoTransfers19 In the United States, tax incentives to foreign direct investment in U.S. possessions have been around for a long time. The 1921 Revenue Act exempted from U.S. taxation the foreign-source income of U.S. firms that received at least 80 per cent of their income from U.S. possessions, if at least 50 per cent of the income came from an active business or trade. Dividends paid to the U.S. parent were taxable on repatriation, while liquidated distributions were tax free. The purpose behind the exemption was threefold: (1) to help U.S. multinationals compete with other foreign firms doing business in the Philippines, then a U.S. possession; (2) to give U.S. investments in U.S. possessions the same tax deferral treatment received by American investments in foreign countries; and (3) to encourage economic development in the possessions. IRC section 351 provided an additional tax incentive because U.S. MNEs could transfer intangible assets to their possessions corporations, as contributions to capital, without recognizing U.S. taxable income at the time of the transfer. Operation Bootstrap In 1948, the Puerto Rican (P.R.) government set up Operation Bootstrap to encourage economic development in the U.S. possession. Under the P.R. Industrial Incentives Act of 1948, the government provided tax holidays to profits earned by new foreign businesses setting up on the island. The tax holiday was expanded under the 1963 Industrial Incentive Act so that qualifying corporations received a 10- to 25-year, 100 per cent tax-free holiday from P.R. corporate income, property, and local taxes. Operation Bootstrap tax incentives induced many U.S. multinationals to set up foreign affiliates (called possessions corporations) in Puerto Rico in the 1950s. A U.S. parent company could transfer the ownership of technology developed in the United States to a possessions corporation; the affiliate could manufacture products using the technology for sale to the parent; and the parent then could market the final product in the United States. Since possession income was exempt from U.S. taxation, as long as profits were kept on the island, no taxes were payable to either government for the period of the tax holiday. Once the holiday expired, the parent could liquidate the corporation and bring the capital home tax free.

The U.S. Tax Transfer Pricing Regulations: Rules

399

BOX 8.2 The Pre-1982 Possessions Tax Credit

Income and expenses of PRCORP, a possessions corporation Direct material costs 45.00 Direct and indirect labour costs 5.00 Total manufacturing cost (= material + labour costs) 50.00 Manufacturer's mark-up (30 per-cent of mfg. cost) 15.00 Price sold to parent firm 65.00 Income and expenses of USCORP, the U.S. parent Price paid to possessions corporation Worldwide R&D costs Distribution costs Distributor's mark-up (5 per cent of final price) Final selling price

65.00 10.00 20.00 5.00 100.00

Section 936 (pre-1982) Profits earned by PRCORP are not taxed by the Puerto Rican government if retained. If profit is remitted to USCORP, PRCORP must pay P.R. tax of 10 per cent. Tax is 10 per cent x (PRCORP's mark-up) = 10 percent x 15 = 1.50. USCORP profits are taxable at U.S. rate of 46 per cent. If PRCORP's profits are retained offshore, or if PRCORP is liquidated when profits are returned, then USCORP's tax is 46 per cent x 5 = 2.30. If PRCORP does remit dividends, they are brought into USCORP's consolidated income, but P.R. tax is 100 per cent creditable. USCORP pays U.S. tax of 46 per cent x (5 + 15) - 1.50 = 7.70. The U.S. tax break is 100 per cent for profits retained offshore or only returned when PRCORP is liquidated. The tax break is worth 46 per cent x 15 = 6.90.

A numerical example, based on a hypothetical U.S. multinational (USCORP) and its P.R. affiliate (PRCORP) is provided in Box 8.2. USCORP develops and transfers an intangible to PRCORP. PRCORP manufactures a product, using its own materials and labour, and sells the output to USCORP for final sale in the United States. Before 1976, PRCORP's profits initially are not taxed by either government. Between 1976 and 1982, the U.S. tax exemption is converted to a tax credit and the P.R. government adds a withholding tax. The tax break

400

The Rules of the Game in North America

remains 100 per cent for retained profits or liquidations; dividends face the full U.S. tax and credit program. The U.S. Treasury Protests The IRS was not happy with this situation, even though tax-free transfer of intangibles was legal under section 351, because of income tax loss implications. Under section 482, the commissioner could allocate income among related parties so as to prevent tax evasion or clearly reflect the income. Individual section 482 audits started piling up across the United States in the mid1960s. Partly as a result of these audits and the lack of a common IRS policy for dealing with them, the U.S. Treasury developed the 1968 transfer pricing regulations. Several cases involving audits of U.S. MNEs with P.R. income in the 1970s also went to tax court.20 The IRS was concerned with two issues: the separation of income and expenses, and the appropriate valuation of intangibles. First, U.S. law allowed American MNEs to develop intangibles, write off current R&D expenses against the U.S. tax, and then transfer the ownership of the intangibles tax free to P.R. affiliates where the subsequent income earned with the intangibles would not be taxed by either government. So the expenses were declared in the United States and the income in Puerto Rico. Second, the ownership of very profitable intangibles was being transferred offshore, with inadequate compensation to the U.S. developer of the intangibles, with the result that huge profits were going untaxed. In the P.R. cases that went to tax court, the Service argued that the possessions corporations were contract manufacturers, performing no more than routine functions, and should only be allowed a small mark-up over costs. In addition, the IRS argued that transfers of intangibles to the affiliates were invalid because they would not have occurred between unrelated parties. The tax courts generally disagreed with the IRS's view on both issues. The Eli Lilly Case The issues can clearly be seen in the Eli Lilly and Company v. Commissioner case.21 Eli Lilly, a U.S. pharmaceutical MNE, owned an extremely profitable patent on the drug Darvon, the largest-selling prescription drug in the United States at the time. The parent firm developed the drug and deducted the R&D costs from its income. Then, under section 351, the parent transferred ownership of the patent, tax free, to Lilly's P.R. affiliate, Lilly P.R. Subsequent income on Darvon was thus declared, and faced a minimal tax bill, in Puerto Rico. The IRS argued that the income from the intangibles belonged to the U.S. parent, despite their tax-free transfer under section 351. Eli Lilly argued that it

The U.S. Tax Transfer Pricing Regulations: Rules 401 had transferred the manufacturing intangibles to its affiliate, and had been paid for them in the form of paid-up capital in the affiliate. The Tax Court held that the transfer of the Darvon intangibles to Lilly P.R. was legitimate, and the subsidiary was the owner of the manufacturing intangibles. Even so, the IRS had the right to allocate income among related parties so as to clearly reflect their income. The court allocated a per cent of the parent's worldwide R&D costs to the possessions corporation, and then used a profit split methodology to divide the remaining profits between the two parties.22 The Court of Appeal rejected the R&D allocation, on the grounds that Lilly P.R. owned the manufacturing intangibles, but affirmed the profit split methodology. Congress Adds Section 936 (the Possession Tax Credit) Concern over the tax losses led the U.S. government, in the Tax Reform Act of 1976, to tighten this generous tax treatment. Congress added IRC section 936, which converted the foreign tax exemption to a foreign tax credit for 'qualified possession source investment income (QPSII),' defined as income from FDI in an active business or trade. The Senate Finance Committee Report accompanying the 1976 act stated that the purpose of section 936 was to encourage employment-producing investments by U.S. MNEs in Puerto Rico. The P.R. government also reduced the tax holiday from 100 to 90 per cent for new investments, and added a 10 per cent 'toll charge' on repatriations of 10 per cent (Cole 1987, 17). The question as to how many jobs were created and at what cost has been an important part of the ongoing controversy surrounding section 936 ever since its passage. In 1982, a U.S. Treasury report concluded that the tax loss to the U.S. government per job in a possessions corporation was $22,000, whereas the average compensation per job paid by possessions corporations was only $14,210 (Turrol993e, 1417). This controversy, and the IRS's lack of success in the tax courts, stimulated the U.S. Congress to add section 936(h) as part of the 1982 Tax Equity and Fiscal Responsibility Act (TEFRA '82).23 To qualify for possessions status under section 936, the per cent of active business income was raised from 50 to 65 per cent of gross income. All intangible income earned by a possessions corporation was allocated to the U.S. parent, under the so-called 'Dole rule.' Thus possessions corporations were to be effectively treated as contract manufacturers, which the IRS wanted. Two escape clauses from the Dole rule were permitted through which a U.S. MNE could continue to earn tax-free intangible income: cost sharing and a profit split. All members of the MNE family involved in the same product line or services had to make an irrevocable election to use one of these two methods.

402

The Rules of the Game in North America

The first escape clause or safe harbour was based on cost sharing. The subsidiary was given ownership of the manufacturing intangibles and the parent ownership of the marketing intangibles. The subsidiary had to make an R&D costsharing payment to the parent. The payment, based on a specified formula, forced the P.R. affiliate to pay a share of the parent's worldwide R&D costs.24 Any manufacturing intangibles were then the property of the possessions corporation. Finally, section 482 was used to determine the transfer price for the products manufactured by the possessions corporation and sold to the U.S. parent. The second safe harbour was a 50-50 profit split, on a product-by-product basis, of the combined taxable income of the U.S. parent and its P.R. affiliate. Since the owner of the intangible was entitled, under U.S. law, to the profit from the use of the intangible, the second safe harbour effectively split profits, after all costs were covered, between the marketing and manufacturing processes. The ease of this split, of course, depended on identifying which returns come from marketing and which from manufacturing, and on the assumption that they could be separated. In practice, more than 50 per cent of the profit was allocated to the U.S. parent. Box 8.3, which is based on the information in Box 8.2, provides a numerical example of section 936(h). Under the Dole rule, PRCORP is faced with full, annual U.S. taxation. In this example, the cost-sharing and profit split safe harbours each generate only half the U.S. tax of the Dole rule. It is not surprising that U.S. MNEs have opted for the safe harbours, in order to preserve at least part of their tax holiday. Who Benefits from 936? Even with the 1982 tax changes, section 936 still sheltered income earned in Puerto Rico from U.S. tax, in effect, keeping Puerto Rico as a U.S. tax haven. Firms that shifted manufacturing operations to Puerto Rico paid lower labour costs, had duty-free access to U.S. markets for their products, and made minimal tax payments to either government. William Cole (1987, 17) estimated the effective tax rate on possessions corporations to be less than 5 per cent on retained earnings and less than 12 per cent on remitted dividends. The 1982 IRS study of transfer pricing adjustments involving section 482 found that P.R. adjustments totalled $508 million, or 11 per cent of total adjustments for all countries. Of the P.R. audits, pricing adjustments were 83 per cent of the total, followed by expense allocations (7.7 per cent) and net income allocations (5 per cent). By 1991, there were about $13 billion in tax-exempt investments under section 936, implying a potential tax loss to the U.S. Treasury of about $3 billion dollars (Richardson 1992, 171-2). While pharmaceutical MNEs received more than half the tax credits, they

The U.S. Tax Transfer Pricing Regulations: Rules

403

BOX 8.3 The 1982 Revisions to the Section 936 Possessions Tax Credit

1982 Section 936(h): The Dole rule All of PRCORP's intangibles are allocated to USCORP. This is equivalent to eliminating the tax holiday. Thus USCORP pays 46% x (PRCORP mark-up + USCORP mark-up) = 46% (5 + 15) = 9.20 in U.S. tax. The U.S. tax break is zero compared with the pre-1982 situation. 1982 Section 936(h), Election 1: Cost sharing PRCORP pays portion of USCORP's worldwide R&D costs. If PRCORP's share is 50%, 50% of R&D cost is disallowed as an expense of USCORP and allocated to PRCORP. USCORP's new profit is 5 + (50% x USCORP R&D costs) = 5 + 5 = 10. In this example, 5/15 or 33.3% of PRCORP's profits are allocated back to USCORP. U.S. tax paid by USCORP = 46% x 10 = 4.60. This tax, as a per cent of the Dole rule tax = 4.60/9.20 = 50%. 1982 Section 936(h), Election 2: Profit split Total MNE Profit = PRCORP's mark-up + USCORP's mark-up = 15 + 5 = 20. Profits allocated to USCORP = 50% x 20 = 10. U.S. tax paid by USCORP = 46% x 10 = 4.60. This tax, as a per cent of the Dole rule tax = 50%.

provided only 18 per cent of the jobs created under section 936. A 1992 General Accounting Office (GAO) report concluded that, in 1987, drug companies with manufacturing operations in Puerto Rico received an average tax benefit worth $70,788 for each job paying $26,471 (Tax Notes International1993e, 519). Thus, each dollar in wages paid to P.R. workers cost the U.S. Treasury $2.67 in forgone tax revenue. For a handful of firms, the 936 credit stood not just as a break, but as a bonanza. The GAO study found that Pfizer Inc.'s tax savings amounted to about $156,400 per worker, or six times the average compensation at its P.R. operations. Merck and Co.'s tax savings amounted to $110,493 per employee, or more than four times the average compensation (Wartzman and Calmes 1993, Al). By 1991, there were about $13 billion in tax-exempt investments under section 936, implying a potential tax loss to the U.S. Treasury of about $3 billion (Richardson 1992, 171-2). The top ten U.S. companies manufacturing in Puerto Rico under section 936 are listed in Table 8.2. Four of the ten are pharmaceuticals.

404

The Rules of the Game in North America

TABLE 8.2 The Top Ten 936 Multinationals

Company

Industry

Number of employees in P.R. possessions corporations

Baxter Intl. Sara Lee General Electric H.J. Heinz Johnson & Johnson Westinghouse Electric Abbott Laboratories Bristol-Myers Squibb Warner-Lambert United States Surgical Average of top ten 936 multinationals

Pharmaceuticals Underwear, hosiery Electric components Processed tuna Health care Electric components Pharmaceuticals Pharmaceuticals Pharmaceuticals Surgical/medical

5,547 5,037 3,555 3,550 3,354 3,281 2,633 2,163 1,649 1,594 3,236

SOURCE: Based on data in 'A Hurricane Heads for Puerto Rico,' Business Week (14 June 1993): 52, 54

The 1992-1993 Attack on Section 936 Congress, in the summer of 1992, discussed cutting the tax credit for new investments in Puerto Rico to 85 per cent, but there was strong opposition from MNEs already located in Puerto Rico and from local government officials. In November 1992, President Bill Clinton singled out the pharmaceutical multinationals, promising 'to protect American consumers and bring down prescription drug prices, I will eliminate tax breaks for drug companies that raise their prices faster than Americans' incomes rise' (quoted in Turro 1993e, 1418). Clinton's economic advisers called for a radical revision of the P.R. tax credit that would reduce the tax credit by half ($7 billion) over five years and replace it with a tax credit related to wages. The Possessions Wage Credit Act of 1993, s. 362, was introduced by Senator David Pry or (Democratic-Arkansas) on 16 February 1993. Pry or characterized 936 as a 'gigantic tax windfall for the pharmaceutical industry' (quoted in Tax Notes International, 1993e, 518). The bill called for a phasing out of section 936 over five years, and a simultaneous phasing in of a wage-based credit, effective 1 January 1993. The end result would have been, by 1998, a 40 per cent nonrefundable tax credit for the first $20,000 of qualified possessions wages, the same incentive available to enterprise zones in the United States. During its phase-in period, the bill would permit 936 companies the lesser of

The U.S. Tax Transfer Pricing Regulations: Rules 405 100 percent of the section 936 credit or 100 percent of qualified wages in 1993 and 1994, dropping in increments so that, by 1998, section 936 companies would be entitled only to the 40 per cent wage credit. The Puerto Rican government was very concerned with the possible loss of 936. With 11 per cent of the economy's labour force employed by about 400 U.S. possessions corporations, $14 billion held by these firms in P.R. bank deposits and a $34 billion economy at stake, the fears were real. Manufacturing accounted for 40 per cent of the island's GDP; when taken together with the banking and finance industry servicing the 936 possessions corporations, the dependence of Puerto Rico on section 936 was evident (Business Week 1993). P.R. Governor Pedro Rossello entered the debate by proposing two alternatives to the administration's plan, either of which would have raised significantly less revenue for the U.S. government: a total compensation-based cap on the section 936 credit or a revised income-based incentive.25 Rosello estimated that the Clinton plan would raise $7.2 billion over the 1994-8 period while his plan would raise about $2.8 billion in tax revenue. The administration's proposals, somewhat diluted, passed the House in April 1993. However, continued lobbying by the P.R. community and the U.S. drug MNEs led to further watering down of the proposals. Drug multinationals like Pfizer Inc. and Merck & Co. did not want a wage-based credit since their possessions corporations were capital intensive; labour-intensive MNEs such as Westinghouse Electric Corp. and General Electric Co. were prepared to settle for the wage credit (Wartzman and Calmes 1993, A4). An additional concern, voiced by Senator Patrick Moynihan (DemocraticNew York), chair of the Senate Finance Committee, was the upcoming non binding plebiscite on whether Puerto Rico should remain a U.S. possession, become a U.S. state, or go it alone. The fear was that dropping 936 could push the islanders into demanding statehood, a status that would have generated substantial additional costs for the U.S. government (Krauss 1993).26 Coupled with the NAFTA (which would make Mexico a competitive location vis-a-vis Puerto Rico), the situation was politically precarious. The Outcome: Take Your Pick President Clinton eventually committed himself to a compromise backed by Moynihan and Senator Bill Bradley of New Jersey (where many of the drug MNEs are located) that would restore more than $2.6 billion in tax breaks. The Omnibus Budget Reconciliation Act (OBRA) was passed on 6 August 1993, with a much-watered-down version of the initial administration proposal. The reduction in the possessions tax credit is now expected to raise $3.8 billion over the 1994-8 period.

406

The Rules of the Game in North America

Under OBRA '93, taxpayers must make a choice between two types of limitations on the section 936 credit. The first option is an income tax credit, which amounts to a percentage limitation of the old 936(h) credit. Under this method the current 100 per cent income tax credit was to fall to 60 per cent in 1994 and an additional 5 per cent each year to 40 per cent in 1998. Taxpayers may deduct a portion of the possessions taxes paid. The second option is a nonrefundable activity-based credit (Krauss 1993, 28; Turro 1993g, 435-6). The credit is the sum of 60 per cent of qualified labour compensation plus a certain percentage (varying between 15 and 65 per cent) of depreciation deductions for qualified tangible property plus a percentage of possession income taxes (if the profit split method is not used). If the profit split method is used, taxpayers may deduct a portion of the possessions taxes paid. Box 8.4 provides a numerical example of the 1993 tax changes to section 936. Both the income tax credit and the activity-based credit raise more U.S. tax revenues than were generated under section 936(h); neither is as punitive as the Dole rule. In this example, where labour costs are a small percentage of PRCORP's manufacturing costs, the activity-based credit is less generous than the income tax credit. Based on this example, it looks likely that most MNEs, and certainly all the capital-intensive firms such as pharmaceuticals, will opt for the income tax credit. Even with these changes, U.S. MNEs with possessions corporations managed to escape with more of their tax breaks intact than one might have expected since the final bill was expected to raise less than half the revenue of President Clinton's first proposal. Moreover, the MNEs managed to preserve, at least in part, the very form of credit that Clinton's advisers had insisted was indefensible: a credit that rewarded U.S. companies in Puerto Rico for the profits they earned rather than for the jobs they created. As one real-world example, Salzarulo (1995), in his study of Schering-Plough Corporation, a pharmaceutical MNE with operations in Puerto Rico, concludes that the firm had used the income tax credit method under section 936(h), claimed a larger possession tax credit, and paid a lower effective U.S. tax rate in 1994 compared with 1993. Conclusion: The Debate Is Not Over Yet The 1993 legislation was not the end of the assault on the possessions tax credit nor on the pharmaceutical multinationals. Senator Pryor, who introduced the bill to repeal 936, had vowed to take the drug MNEs on again in 1994, saying: The drug companies still have an extremely lucrative benefit not enjoyed by companies elsewhere' (quoted in Wartzman and Calmes 1993, A4). As a result, the opportunities for tax avoidance through possessions corporations continued

The U.S. Tax Transfer Pricing Regulations: Rules

407

BOX 8.4 The 1993 Revisions to the Section 936 Possessions Tax Credit

1993 section 936(h) option 1: Income tax credit U.S. credit drops to 60 per cent of the section 936 credit in 1994, falling to 40 percent in 1998. Election 1: Cost-sharing method If USCORP has elected cost sharing, the tax credit in 1998 is 40% of (PRCORP's mark-up - PRCORP's share of USCORP's R&D costs) = 40% x (15 - 5) = 4. The effective tax break is 4/15 = 27%. USCORP must include in its income 60% of (PRCORP's mark-up - PRCORP's share of R&D costs) = 60% x 10 = 6. USCORP's tax is 46% (5 + 6) = 5.06. This tax, as a per cent of the Dole rule tax = 5.06/9.20 = 55%. Election 2: Profit split method If USCORP has elected the profit split, the tax credit for PRCORP is 40% x (PRCORP mark-up - 50% (USCORP mark-up + PRCORP markup) = 40% x (15 - 50% (5 + 15)) = 40% x 5 = 2. The effective tax break is 2/15= 13%. In this case, USCORP's tax is 46% x ( 50% x (5 + 15)) + 60% x 5) = 46% x 13 = 5.98. This tax, as a per cent of the Dole rule tax = 5.98/9.20 = 65%.

1993 section 936(h) option 2: Activity-based credit Under this method all of PRCORP's profit would be added to USCORP's income, but a nonrefundable tax credit equal to 60% of the wage bill plus a certain per cent of depreciation deductions plus a per cent of the P.R. tax would be given. Puerto Rican taxes would be creditable under the cost-sharing method, deductible under the profit split method. USCORP would pay 46% (USCORP mark-up + PRCORP mark-up) minus the tax credit = 46% x (5 + 15) - 60% x 5 = 9.20 - 3 = 6.20. This tax, as a per cent of the Dole rule tax is 6.20/9.20 = 67%.

to be reduced by the U.S. government between 1993 and 1996; in 1996, section 936 was repealed. The 1993 temporary regulations for section 482 (482-T93) provided new methods for valuing intrafirm transfers of intangibles, in accordance with the commensurate-with-income principle adopted by the U.S. Congress in 1986.27

408

The Rules of the Game in North America

One subsection of the new regulations dealt with U.S. possessions corporations.28 If a possessions corporation made a cost-sharing election under section 936, the temporary regulations required that the payment be no less than that computed for transfers of intangibles under section 482. This meant that possessions corporations now had to make annual payments to their U.S. parents commensurate with the income earned from the use of the intangible. Section 936(h) was thus made subordinate to section 482. The final transfer pricing regulations, issued in July 1994, strengthened the 'back stop' role of section 482 (Hexner and Jenkins 1995, 248). First, the regulations confirmed that section 936(h) was subordinate to section 482; that is, the amount of the cost-sharing payment required under section 936(h) could be no less than that required by section 482. Second, a possessions corporation had to apply the section 482 pricing methods for intangibles before applying the 936(h) rules on intangibles. In May 1996, the House passed a resolution to repeal section 936, which the Senate quickly followed in June. In August 1996, as part of the Small Business Job Protection Act, section 30A became part of the Internal Revenue Code (Muniz 1996). The new law repeals section 936 and establishes a ten-year period for phasing out its benefits. During this period, qualified domestic corporations that were in existence before October 1995 continue to receive tax breaks tied to the MNE's contribution to the Puerto Rican economy (i.e., their benefits are grandfathered but are now determined by section 30A rather than 936). After 2001, a ceiling on the amount of net income that can be sheltered from U.S. tax takes effect, further reducing the tax benefits of section 30A. New firms, on the other hand, are not entitled to any U.S. benefits under either section 936 or 30A. In 2005, section 30A is to be phased out and the long history of the possessions tax credit is to end. The government of Puerto Rico, however, continues to lobby for reinstatement of the credit. Given the likely trade diversion impacts on Puerto Rico of Mexico's joining the North American Free Trade Agreement, and the U.S. government's long-standing commitment to the island, we conclude that section 30A is not the end of the story. The debate over the possessions tax credit is not over yet. Section 861: Allocation ofR&D Expenses29 Two other measures besides section 936 were introduced to response to the Treasury's twin concerns that U.S. MNEs were developing intangibles, writing off current R&D expenses against the U.S. tax, and then transferring the ownership of the intangibles tax free to low-taxed foreign affiliates. The first was section 861, dealing with allocation of R&D expenses incurred by the U.S. parent.

The U.S. Tax Transfer Pricing Regulations: Rules 409 The second was section 367, denying tax-free transfers of intangibles where tax avoidance was the reason for the transfer. We look at 861 in this section, and 367 in the next section. Before 1977, U.S. MNEs had wide latitude in allocating R&D costs between parent firms and their foreign affiliates. Since R&D costs were deductible against the corporate income tax, MNEs could reduce their overall tax liability by declaring costs in the high-tax jurisdiction. In the 1960s and 1970s, U.S. rates were higher than those in most developed countries; this encouraged the firms to take the deductions at home, thus reducing the U.S. tax base. In 1977, the U.S. Treasury issued regulations for allocating these expenses designed to shift more of these expenses offshore. Under the 1977 regulations, U.S. MNEs are required to allocate governmentmandated R&D costs against gross income arising in the country where the R&D is performed. Thirty per cent of R&D costs can first be apportioned exclusively to the source where more than 50 per cent of the taxpayer's research occurs, with the rest normally allocated between domestic and foreign income depending on the ratio of domestic to foreign sales.30 American multinationals did not like these regulations because some host governments disallowed the allocated R&D costs, generating double taxation for the parent firm. In addition, allocating these expenses to foreign subsidiaries reduced the size of their foreign income and foreign taxes, thus reducing the size of the foreign tax credit against U.S. tax. Lobbying by the firms led to a series of modifications in the rules between 1981 and the present. At one extreme was full deductibility of all U.S.-incurred R&D costs against U.S. source income (the 1981 Economic Recovery Act). At the other extreme, the 1986 Tax Reform Act reduced the allocation to 50 per cent, with the remainder apportioned on the basis of gross sales or income. In between, the 1988 Technical and Miscellaneous Revenue Act allocated 64 per cent of R&D costs to the jurisdiction where they were incurred, with the remainder allocated on the basis of gross income or sales. The 1992 IRS Revenue Procedure 92-56 gave taxpayers the option of the 1977 or 1988 rules. In 1996, the IRS issued new regulations on the allocation and apportionment of R&D expenditures (Fuller 1996). Section 1.861-17 allows 50 per cent of U.S.-based R&D expenses to be allocated to U.S.-source income, but only for taxpayers using the sales method to apportion the remaining amount of R&D expenses. If the MNE elects to use the optional gross income methods, the election is binding, and a maximum 25 per cent of R&D expenses can be allocated to the United States. In allocating expenses among product categories, the MNE must use three-digit rather than two-digit SIC codes, implying less flexibility for the firm in allocating costs among products. In most other countries, 100 per cent of local R&D costs can be written off

410 The Rules of the Game in North America against domestic income. Hufbauer (1992, 136-8) recommended that the United States adopt a similar rule, allowing full deductibility for all overhead costs, including R&D expenses. At the same time, he recommended that all royalties generated from the use of the intangibles generated from U.S.-sourced R&D be allocated to the U.S. income base and therefore taxable in the United States. Hufbauer's proposal, which would have been much simpler than the current 861 regulations, was clearly ignored by the IRS, probably because the proposal required the cooperation of all U.S. treaty partners. Such cooperation was unlikely to be forthcoming, given the implication of Hufbauer's proposal for their tax revenues. Section 367(D): Imputed Royalty for IntangibleTransfers31 Section 351 has been an important component of the possessions corporation strategy for U.S. multinationals. Tax-free transfer of the ownership of U.S. intangibles, coupled with a P.R. tax holiday and U.S. tax sparing, made for a very generous package encouraging foreign direct investment in Puerto Rico. Even though section 351 allowed tax-free transfer of the ownership of U.S. intangibles to offshore affiliates as well as to U.S. businesses, the U.S. government has restricted this transfer where tax avoidance was seen to be the main motive for the transfer. In 1976, at the same time that section 936 was added to the tax code, section 367 was also added as a general rule, denying tax-free status to transfers to foreign corporations, not involving a sale or a licence, where a tax-avoidance motive was evident. MNEs could apply for a ruling from the Service under this section that would grant tax-free status when an intangible asset was transferred offshore if the purpose was not tax avoidance. (Eli Lilly applied for, and got, this exemption.) Section 367 required transfers that were not tax free to be treated as contributions to capital and levied with a toll charge. Section 367, however, did not apply to intangibles transfers to possessions corporations because they were considered to be American, not foreign, affiliatesGiven the concern in the early 1980s with intangible income, again along with the 1982 addition of section 936(h), in 1984 Congress added section 367(d). Section 367(d) ensured that if intangible property was transferred abroad, the transferor was treated as receiving an imputed royalty over the life of the intangible. The U.S. developer and transferor of the intangible was required to include arm's length payments, treated as being received annually over the life of the intangible, in its income whether received or not. Effectively, by 1984, the United States had plugged the intangibles loophole through which U.S. MNEs with foreign affiliates had historically been able to avoid U.S. income tax (Boidman 1988a, 44:10).32 Any MNE transferring intan-

The U.S. Tax Transfer Pricing Regulations: Rules

411

gible assets abroad is now treated as receiving an imputed royalty payment over the life of the intangible and the royalty counted as transferor's U.S. source income, regardless of the method of transfer or the type of intangible property. Section 1231(E): The Commensurate with Income Standard33 Even with sections 367(d) and 936(h) in place, Congress was still concerned about underpayment of royalties and licence fees for intangibles transferred offshore. In the Tax Reform Act of 1986, Congress changed the rules with respect to taxing the income from intangibles. Congress passed Section 123l(e), which added a second sentence to section 482, and to section 367(d)(2)(A), requiring that payments to related parties for licenced or transferred intangibles be commensurate with the income (CWI) from these intangibles. Congress then mandated the U.S. Treasury to investigate ways to incorporate the CWI standard into section 482, which led to the U.S. Treasury's 1988 White Paper (see below). Section 1231(e) reads as follows: In the case of any transfer (or licence) of intangible property (within the meaning of section 936(h)(3)(B)), the income with respect to such transfer or licence shall be commensurate with the income attributable to the intangible. (U.S. Federal Tax Guide-2, Tax Tactics - Codes and Regulations, section 482, 5221)

The Committee Report accompanying the House version of section 1231(e) clearly states that the purpose of the legislation was to eliminate the tax loss on transfers of intangibles: There is a strong incentive for taxpayers to transfer intangibles to related foreign corporations or possessions corporations in a low tax jurisdiction, particularly when the intangible has a high value relative to manufacturing or assembly costs. Such transfers can result in indefinite deferral or effective tax exemptions on the earnings, while retaining the value of the earnings in the related group ... [S]ections 482, 367(d), and section 936 ... may not be operating to assure adequate allocations to the U.S. taxable entity of income attributable to intangibles in these situations (U.S. House of Representatives 1985,423).

The Committee Report stresses that the purpose of the CWI legislation is to ensure that the division of income between related parties reflects the economic activities undertaken by the parties. If we read between the lines, this means that the intent of the legislation is to ensure that U.S. multinationals receive a fair, and taxable, return on their outbound transfers of intangible assets to their

412 The Rules of the Game in North America foreign affiliates. A 'fair return' is achieved when the income paid to the developer of the intangible (the U.S. MNE) is commensurate with the income earned by the firm using the intangible (the foreign subsidiary). To see what difference the CWI standard makes to section 482, recall that the general rule for pricing intangibles, section 1.482-2(d), which has been part of the Treasury regulations since 1968, requires that an arm's length return be charged when intangibles are sold or licenced to foreign related parties. The valuation of the intangible therefore depends on the parties' expectations, at the time of the transfer, of the income that will be generated using the intangible. Where normal profit intangibles are involved, outside comparables are more likely to exist that can be used to establish an arm's length charge under section 482. Industry average royalty rates, for example, can be used as a safe-harbour estimate. These arm's length prices should be reasonably accurate reflections of the lifetime income earned from the intangible. On the other hand, where high-profit, crown jewel intangibles are involved, royalty rates are harder to establish.34 These intangibles are generally unique with highly uncertain returns so that MNEs are less likely to transfer them outside the corporate entity. And, given uncertainty about the future, it is possible that using a section 482 valuation may substantially underestimate the income that would eventually be generated, for example, if a pharmaceutical company transfers a technology that leads to a cure for a major life-threatening disease. Thus, for normal profit intangibles, the CWI method and the old 482 method should yield the same results, but a super royalty charge would be needed to capture a fair share of the returns from crown jewel intangibles such as wonder drugs. Since the actual amount of income from the intangible is likely to vary over time, the CWI standard demands periodic adjustments to the transfer price to take into account changes in income as well as in the economic circumstances of the related parties. Congress did not want U.S. firms to transfer potentially high-profit intangibles abroad, unless the price reflected the future stream of income from the intangible. It was concerned that using industry average rates of return and setting the return at the time of transfer would allocate most of the eventual income stream from high-profit intangibles to the foreign subsidiary. The CWI standard was designed to get around these problems by insisting that the actual amount of income derived from the intangible over the long run should be the primary factor in determining the royalty payments. The 1988 Treasury White Paper35 The 1986 legislation did not set out the specific methods by which the CWI standard was to actually apply. Instead, Congress mandated the Treasury and

The U.S. Tax Transfer Pricing Regulations: Rules

413

the IRS to prepare a comprehensive study of intercompany pricing rules and suggest modifications to section 482 to incorporate the CWI standard. In response to the request from the U.S. Congress, the Treasury published a White Paper called A Study of Intracorporate Pricing (U.S. Department of the Treasury 1988). The methods for valuing intrafirm transactions proposed in the White Paper are summarized as part of the comparisons provided in Table 8.1. Bringing CWI into the 482 Regulations The White Paper first reviews the history of U.S. transfer pricing regulations and recent IRS experience with section 482. The Treasury affirms its commitment to the arm's length standard and the CUP method in spite of the difficulties in finding suitable comparables. The report then turns to the Congressional request to address section 482 in the light of the CWI standard. The Treasury argues that CWI is an arm's length pricing method since it 'ensure[s] that each party earns the income or return from the intangible that an unrelated party would earn in an arm's length transfer of the intangible' (U.S. Department of the Treasury 1988, 47). The paper suggests two different approaches to valuing intangibles according to CWI: a pricing approach and an income approach. The pricing approach is based on two methods: the exact comparable method, which can be used where true comparables to the intangible exist, and the inexact comparable method, in which true comparables do not exist and adjustments have to be made. The pricing methods are intended to apply the arm's length standard of section 482, that is, that of an uncontrolled taxpayer dealing at arm's length with another uncontrolled taxpayer. Where exact or inexact comparables are unavailable, the U.S. Treasury suggests that the income approach should be used. The Basic Arm's Length Return Method (BALRM) The income approach includes two methods: the basic arm's length return method (BALRM) and BALRM with a profit split.36 BALRM acknowledges that MNE transactions consist of a bundle of tangibles, intangibles, and services, and that each asset needs to be valued at arm's length. The analysis assumes that all the firm's assets, tangible and intangible, can be identified and measured. All costs of developing the intangible assets should be charged according to the functions performed, risks undertaken, and capital invested by the related parties. Both income methods use a functional analysis to allocate income between the two parties based on their shares of intangible assets measured in terms of industry-average rates of return. The functional analysis identifies all the valueadding activities undertaken within the MNE and puts a value on those functions, as if they had been performed at arm's length.37 Income is then split between the parties according to their relative economic contributions.

414

The Rules of the Game in North America

To apply BALRM, the MNE's value chain is split into four functions: R&D, manufacturing, distribution, and marketing, and into two types of intangible assets: manufacturing intangibles and marketing intangibles.38 In BALRM, assets are then assigned to the MNE's various functions and external weighted average rates of return calculated for each asset. The rates of return are supposed to measure the long-run opportunity cost of the assets in terms of what they could earn in their next-best alternative use in external markets, that is, an economist's definition of return, not the industry average. Generally there will be an unexplained 'residual' term after all the functions have been valued and BALRM allocates the residual to the parent firm. The White Paper suggests that if both parties contribute intangibles then BALRM with a profit split based on the relative shares in the intangibles may be more appropriate. Cost-Sharing Arrangements The last chapter of the White Paper discusses how cost-sharing arrangements can be used to allocate the costs and benefits of intangibles among related parties. A cost-sharing arrangement, according to the White Paper, is an alternative method by which related parties can develop and exploit intangibles... [It is] an agreement between two or more persons to share the costs and risks of research and development as they are incurred in exchange for a specified interest in any property that is developed. (U.S. Department of the Treasury 1988, 109)

The Treasury argues that the costs borne by each of the participants should be proportionate to the benefits each party expects to receive over time from exploiting the intangibles developed under the arrangement. Each participant is expected to bear an appropriate share of R&D costs on successful and unsuccessful products within a particular product area. The cost share should be proportionate to the profits, adjusted for differences in assumed levels of risk, before R&D deductions. The report goes on to discuss problem areas in defining bona fide cost-sharing arrangements. It concludes that such arrangements should be permitted if they produce results consistent with the CWI standard, follow the benefit-cost principle, and meet certain other tests.39 Summary and Criticisms Historically, the Internal Revenue Code has not had a broad rule enforcing arm's length pricing of intangibles when transferred by a U.S. corporation to its foreign affiliates or P.R. possessions corporations, unless the transfer involved a

The U.S. Tax Transfer Pricing Regulations: Rules 415 licence or sale. Section 367 was added to the act to require that tax-free transfers of intangibles be treated as transfers to capital and levied with a toll charge. In 1982, Congress reduced the tax loophole for possessions corporations and, in 1984, further tightened section 367. In 1986, Congress added the commensurate with income standard to section 482 and required periodic revaluations of the royalty charges, the so-called super royalty. In 1988, the U.S. Treasury suggested four methods for implementing the CWI standard and outlined possible rules for cost-sharing arrangements. General Criticisms of the White Paper Section 123 l(e) and the 1988 Treasury White Paper caused a 'furore, undiminished to date, in international tax circles' (Boidman 1988a, 44:1). The U.S. government's attempts to plug the intangible transfers loophole through the super royalty method generated concerns among taxing authorities outside the United States for several reasons. First, the section applied both to tax haven and to high-tax countries, even though the stated purpose was to lessen tax avoidance. Other OECD tax authorities did not see why the legislation should apply to high-tax countries. Second, other governments feared that the CWI standard would lead to excessive licensing or other charges being levied on foreign affiliates of U.S. multinationals. The super royalty was seen as equivalent to a tax grab by the IRS. Foreign tax authorities would then be forced to decide whether to allow the subsidiaries of U.S. MNEs to deduct the (excessive) royalties from their foreign income taxes or to deny the deductions, leaving the affiliates with double taxation burdens. Third, probably the most controversial feature of the super royalty has been the requirement for periodic reassessment of the income from the intangibles, regardless of the terms negotiated between the two parties. The U.S. Treasury argued that parties negotiating a long-run contract at arm's length generally do allow for adjustment of the royalty terms, particularly if profits rise significantly over time but were unanticipated at the time of the contract. Treasury therefore concluded that periodic adjustments were consistent with the arm's length principle. Other governments argued that the periodic assessment of the actual income of the group required to fulfil the CWI standard violated the arm's length standard. There was (and is) a strong conviction among OECD revenue authorities, including Revenue Canada, that the appropriate transfer price for any transaction should be based on the information the two parties could have reasonably been expected to know at the time of the licence or sale, and not on hindsight. Thus periodic adjustments were unacceptable to most tax authorities.

416

The Rules of the Game in North America

At least one transfer pricing expert, however, has been sanguine about the effects of CWI. Nathan Boidman contends that the super royalty does not break international standards (Boidman 1988a). He argues that all countries, the United States included, use comparables where available to determine arm's length prices. Where comparables are not available, there are no specific rules and the transfer price is determined by the facts and circumstances of the case. The super royalty merely provides some specific rules for cases in which there are no comparables. Boidman concludes: The super royalty raises at most an issue of form, and not substance, in the international context; it does not give rise to conflict with the tax law in Canada or other countries governing international transactions' (Boidman 1988a, 44:14). Criticisms of BALRM BALRM, as a method for valuing intangibles, has some benefits. First, the method is based on a functional analysis of the MNE's value chain, which, as we have outlined in Chapter 3, is an important step in understanding the nature of an integrated business. The method is based on the opportunity cost principle, that is, what each factor or asset could earn in its next best alternative use. Again, we have argued in Chapter 5 that the underlying principle that should drive the choice of a transfer price is its opportunity cost. The basic problem with BALRM is putting the theory into practice. Commentators criticized the prominence given to BALRM in the White Paper, arguing it would be difficult to apply because the information would not be generally available, would be unfair to corporations with rates of return varying from the average industry, and would allocate too much income to the U.S. entities.40 We outline a few of these criticisms below; the interested reader is also directed to the large number of articles on this subject. First, BALRM works best if all markets are 'plain vanilla markets'; that is, products are homogeneous commodities, firms are price takers, all firms earn normal rates of return on all their assets, technology is mature, and markets are certain. The presumption behind BALRM is that product markets are perfectly competitive and in long-run equilibrium. Assets are assumed to be perfectly mobile, so they earn an equilibrium normal rate of return. Assets are homogeneous within class groups and not firm specific; as a result, comparables exist and can be measured. However, in the real world of internationalized business none of these is likely to occur. As a result, the average valuations generated by BALRM are not good measures of the true opportunity costs of the firm's intangible assets. Second, BALRM assumes the MNE can be decomposed into discrete, independent units with identifiable, routine functions that can be easily valued.

The U.S. Tax Transfer Pricing Regulations: Rules 417 BALRM ignores the interdependencies among the units of the MNE due to economies of scale and scope, savings in transactions costs, and benefit externalities that are the motivation behind internalization in the first place. As we know from the Hirshleifer rule, the opportunity cost principle breaks down when there are interdependencies among the demand and/or cost functions of the divisions of the multinational. Third, BALRM ignores the management intangible. While BALRM incorporates two intangibles, it ignores a third intangible or, at best, assumes it belongs to the parent firm. Strategic management (the so-called 'marginal productivity of the entrepreneur' [Langbein 1986] or 'returns to the organization') is one of the most complex activities for the firm in the globalized world of the 1990s, and a function each affiliate performs, not just the parent firm. While it is true that the parent performs a long-run overall management function on behalf of the MNE's shareholders, each affiliate engages in strategic planning at the local level, the degree of sophistication depending on the role the affiliate plays in the MNE's value chain (e.g., varying from a simple distributor to a full-fledged lead plant). BALRM assumes either that all management expertise comes from the parent firm or that the subsidiaries 'run themselves.' This is illustrated in Figure 8.1, where we assume the parent firm supplies a product to a foreign affiliate for marketing and sale in the host market. Total output Q0 is sold at price P (see point a in Figure 8.1) generating total sales revenue of P times Q0. The minimum amount the parent firm must receive to cover its variable production costs is area A, below its marginal cost or supply curve. The maximum excess profit is therefore areas B + C + D + E. To whom do these rents belong? Area B normally belongs to the supplier if one price is set for all units of the supplied good; area B represents the producer surplus that accrues to the manufacturer because price exceeds marginal costs for inframarginal units. We assume area C represents the return to manufacturing intangibles involved in the product; if the technology is owned by the parent firm and nothing is contributed by the foreign affiliate, then area C should belong to the parent, according to BALRM. We also assume area D is the return to the marketing intangible. If the foreign affiliate holds the brand name and franchise rights to the product, and does all the distributing, then BALRM would assign area D to the foreign affiliate. This leaves area E, the residual profit, unassigned. Since management intangibles are ignored, BALRM assigns all excess profits to the parent firm. Even if the management function is recognized, the assumption of BALRM is that the parent firm is the overall manager of the global enterprise and excess profits belong to the parent. An additional complication is that the home country tax authority may argue

418

The Rules of the Game in North America

FIGURE 8.1 The Basic Arm's Length Return Method (BALRM)

that the foreign affiliate is a simple contract distributor, and/or that the brandname intangibles reside with the parent. In this case, area D is also returned to the parent firm. The taxing authority in a capital-exporting country like the United States has an incentive to do this because returning areas D and C to the parent firm increases the home country tax base. Fourth, BALRM ignores the product life cycle and therefore has difficulty coping with high-tech MNEs.41 High-tech industries are unlikely to ever be in long-run equilibrium necessary for BALRM to work effectively. Their volatility is due to the short product life cycles and very high overhead costs of technology development characteristic of high-tech firms. The shorter the product life cycle, the more important global market share is for covering costs and generating the returns to cover the costs of succeeding innovation. As a result, the earnings pattern for a high-tech MNE will vary over the life cycle of the product, being high at the beginning and declining over time. In terms of accounting profits, R&D expenses are expensed as incurred, even though the returns (if they develop) occur later. So expenses are out of sync with revenues and hightech MNEs have volatile earnings patterns. The distribution of firm returns will

The U.S. Tax Transfer Pricing Regulations: Rules

419

vary widely, and the average return, which BALRM would calculate, is unlikely to be a good proxy for the real returns to intangible assets. Creating One Set of Rules: The New 482 Regulations Responding to direction from Congress to integrate the CWI standard into the section 482 regulations, the U.S. Treasury developed and issued, in January 1993, draft proposals for revising the regulations. The release of the proposals was 'greeted by generalized criticisms and expressions of confusion and dismay from the legal community' (McLennan 1993, 1201). The OECD was very critical on the grounds that the Treasury had moved away from the arm's length standard, and, in late 1992, issued its own negative analysis of the proposals.42 In February 1993 the Treasury responded with new temporary regulations that went into effect in April 1993. Final regulations were released on 1 July 1994. In this section we examine the three stages in the development of the new section 482 regulations. We identify them as Mark I: 482-P92 or IRS (1992), the proposed version of the regulations; Mark II: 482-T93 or IRS (1993b), the temporary version of the regulations; and Mark HI: 482-F94 or IRS (1994b), the final version of the regulations. Mark I: The 1992 Proposed Regulations43 Outline of Section 482 Mark I The first version, the January 1992 draft proposals (Mark I or 482-P92), attempts to integrate the Congressional directive on the commensurate with income standard into the 482 regulations. 482-P92 contains several major changes from the 1968 regulations. The new rules for pricing intrafirm transactions are summarized in Table 8.1 above. The first major change is the proposed revision to part of section 1.482l(b)(l) (the definition of the arm's length standard) as follows:44 In determining whether controlled taxpayers have dealt with each other at arm's length, the general principle to be followed is whether uncontrolled taxpayers exercising sound business judgement on the basis of reasonable levels of experience (or, if greater, the actual level of experience of the controlled taxpayer) within the relevant industry and with full knowledge of the relevant facts, would have agreed to the same contractual terms under the same economic conditions and other circumstances under which controlled taxpayers dealt. (482-P92, 42). The key change is the addition of the phrase 'sound business judgement on

420

The Rules of the Game in North America

the basis of reasonable levels of experience.' In interpreting this statement, closely related transfers of tangible and intangible property (the so-called 'round trip transactions') are to be considered together. The second major change is the proposed replacement of section 1.482-2(d) on transfers of intangible property.45 In the proposal, the new definition of the arm's length standard is to be applied to these transfers. An 'arm's length consideration' for an intangible is defined as the amount an uncontrolled taxpayer would have paid for the intangible under the same circumstances. The amount has to be commensurate with the income generated by the party using the intangible. The form of the consideration is normally to be a royalty payment. The IRS can make periodic (annual) adjustments to the arm's length consideration to ensure that the royalty is commensurate with the income attributable to the intangible.46 The third major change is the introduction of a hierarchy of three new methods for pricing intangibles: the matching transaction method (MTM), the comparable adjustable transaction method (CATM), and the comparable profit interval (CPI). CPI, defined in a new section 1.482-2(d)(2), is now referred to as the comparable profits method (CPM) in the final 482 regulations, so we use the term CPM here. Where matching transactions47 exist, the MTM is to take priority because it requires the fewest adjustments and uses the most accurate and complete data. Where matching transactions do not exist, CATM should be used. To use CATM, there must be 'an uncontrolled transfer involving the same or similar intangible under adjustable economic conditions and contractual terms' (482P92, 58-9). The taxpayer then makes appropriate adjustments for differences in intangibles, economic conditions, and contractual terms. If neither MTM nor CATM can be used, CPM must be used. Clearly, the first two methods follow the pricing approach advocated in the Treasury White Paper (U.S. Department of the Treasury 1988). MTM is very similar to the exact comparable method and CATM to the inexact comparable method. However CPM, while it is an income-based method, is not similar to the basic arm's length return method. CPM is an income-based method that uses average industry rates of return to value the taxpayer's income. Since finding matching transactions for intangibles is highly unlikely, due to the MNE's unwillingness to sell or licence non routine intangibles outside the corporate group, in practice the proposed regulations mean that all transfer pricing cases are subject to CPM. CPM requires the operating income from the transaction under review to be compared with the operating income of similar uncontrolled taxpayers. In order to make this comparison, a computed profit interval (CPI) is constructed in which the MNE's transaction must fit. Box 8.5 outlines the evolution of the comparable profits method, from the 1992 proposed regulations through the 1994 final regulations.

BOX 8.5 The Comparable Profits Method The 1992 proposed comparable profits method (CPM-1992)

The 1993 temporary comparable profits method (CPM-1993)

The 1994 final comparable profits method (CPM-1994)

Select the tested party: Select the taxpayer whose operating income can be verified with the most reliable data and fewest adjustments. For an intangible transfer this is normally the party that uses the intangible.

Select the tested party: The tested party is the one not using any valuable nonroutine intangibles. It is the party with the simplest and most easily compared activities. Where intangibles are licensed, the tested party is normally the licensee.

Select the tested party: The tested party is the one whose operating profit attributable to the transaction can be verified using the most reliable data and requiring the fewest and most reliable adjustmemnts, and for whom the most reliable data on uncontrolled comparables can be found. Generally, the tested party is the least complex party and does not own any valuable intangible property or unique assets.

Determine the business classification: Identify the tested operations (type of product, functions), match these to similar operations of uncontrolled parties, and then determine the applicable business classification as the broadest category of tested operations that corresponds to the matched ones.

Select the industry segments: Divide the tested activities into industry segments. CPM measures the total return on the business activities of a tested party, applied separately to each segment.

Select the relevant business activity: Find the most narrowly identifiable business activity for which data incorporating the controlled transaction are available; this is the relevant relevant business activity. CPM measures the total return on the relevant business activities of a tested party, applied separately to each activity. Where a taxpayer has many transactions, CPM can be applied to product lines or other groupings; in addition, sampling and other statistical techniques may be used to evaluate the arm's length results.

Comparability: Select financial data for the matched uncontrolled parties that provide a reliable basis for comparing profits.

Comparability: The tested and uncontrolled parties need only be broadly similar. Significant product diversity and some functional diversity are acceptable. Reliable data for comparable parties in the same industry segment are preferred.

Comparability: Comparability of the parties depends on the facts and circumstances, including functional risk, resource comparability, and other material differences. Adjustments for material differences should be made.

BOX 8.5 (continued) The 1992 proposed comparable profits method (CPM-1992)

The 1993 temporary comparable profits method (CPM-1993)

The 1994 final comparable profits method (CPM-1994)

Select a profit level indicator: Calculate the profit level indicators.

Select a profit level indicator: Select a single profit level indicator, derived from the comparable parties. Adjustments should be made to the indicator to improve consistency and achieve greater similarity between the tested and uncontrolled parties. Data from at least the year under review and the previous two years should be used.

Select the profit level indicator: Select a single profit level indicator (generally based on operating profit), derived from the comparable parties. Adjustments should be made to the indicator to improve consistency and achieve greater similarity between the tested and uncontrolled parties. A sufficient number of years to reasonably measure returns to the uncontrolled comparables should be used; in general, this includes at least the year under review and the previous two years.

Types of profit level indicators: Possible indicators include the rate of return on assets, operating income to sales, gross income to operating expenses.

Types of profit level indicators: There are two types of indicators: the rate of return on capital employed, and financial ratios (operating profit to sales, and gross profit to operating expenses).

Types of profit level indicators: There are at least two possible types of indicators: the rate of return on capital employed, and financial ratios (operating profit to sales, and gross profit to operating expenses). The indicator should be chosen based on nature of the activities of the tested parties, reliability of the data, and extent to which the indicator is likely to produce a reliable result.

Calculate the comparable profit interval (CPI): Apply the profit level indicators to compute the tested party's constructive operating income (COI). Adjust for significant differences. Look for convergence of two types of COIs: those derived from (1) several profit indicators of one uncontrolled party and (2) one or more profit level indicators from multiple uncontrolled parties. This sets the range of the CPI.

Calculate the arm's length range: Use the profit level indicator to construct operating incomes for the test party. If adjustments were made to the profit level indicators, the range is all observations; if not, the range is limited to the 25-75 per cent range.

Calculate the arm's length range: Apply the CPM to two or more uncontrolled transactions of similar comparability and reliability. Use the profit level indicator to construct operating incomes for the test party. The arm's length range consists of all estimates if exact comparables are used; if inexact comparables are used, the interquartile range (25-75 per cent) applies.

BOX 8.5 (concluded) The 1992 proposed comparable profits method (CPM-1992)

The 1993 temporary comparable profits method (CPM-1993)

The 1994 final comparable profits method (CPM-1994)

Select the appropriate point with the CPI: Select the most appropriate point based on analysis of the data. This point determines the constructed operating income for the tested party. Ordinarily this will be the midpoint.

Determine the arm's length result: The result is arm's length if the tested party's operating profits are within the arm's length range. If it lies outside the range, the IRS can allocate to any point in the range; ordinarily this will be the midpoint. Where multiple year data are used, the tested party's profits in the multiple-year period are compared with the range over the whole period.

Determine the arm's length result: The result is arm's length if the tested party's operating profits are within the arm's length range. If it lies outside the range, the IRS can allocate to any point in the range; ordinarily this will be the median of all results if the interquartile range is used; in other cases adjustment is to the arithmetic mean. Where multiple-year data are used, the tested party's profit for the tax year in question is compared with the arm's length range using the multiple-year data; the amount of adjustment depends on the comparables' profits from the year under examination.

Determine the transfer price: Calculate the transfer price that would have resulted in the tested party earning the appropriate point within the CPI range. This is the arm's length price.

Determine the transfer price: Calculate the the transfer price that would have resulted in the tested party earning the arm's length result. This is the arm's length price.

Determine the transfer price: Calculate the transfer price that would have resulted in the tested party earning the arm's length result. This is the arm's length price.

SOURCE: Based on Carlson et al. (1994), Horst (1993), U.S. Internal Revenue Service (1992, 1993b, 1994b)

424

The Rules of the Game in North America

Where the controlled parties have 'significant self-developed intangibles that are not reflected on their financial statements' (482-P92, 95), a fourth major change is to allow a profit split to be used to determine the appropriate allocation of income. Either an overall, or a residual, profit split can be used. For a total profit split, the following steps are performed: (1) determine the combined operating income of uncontrolled taxpayers that have transactions and functions similar to the controlled parties; (2) calculate each uncontrolled party's percentage of the combined operating income; (3) use these percentages to split the combined operating income of the controlled firms. Alternatively, the taxpayer can use a residual profit split, a form of the White Paper's BALRM-with-profitsplit method (482-P92, 93). Another major change is that under all three methods - MTM, CATM, and CPM - data must be used from three years: the taxable year under review, the previous year, and the year after the taxable year. This means that the taxpayer must calculate the acceptable transfer pricing using data from a year after the transfer took place.48 The 1992 proposals also introduce detailed rules for qualified cost-sharing arrangements. These proposals are closely related to those outlined in the 1988 White Paper. The general principles are that: (1) each participant must expect to use the developed intangibles in the active conduct of its business or trade; (2) the costs of all related intangible development, successful and unsuccessful, must be shared, with each party's share proportionate to its share of the income due to the intangibles; (3) where costs are not appropriately shared, participants must pay an arm's length amount of the use of the property to the developers (482-P92, 25-6). The last major change is the pricing of tangibles under section 1.482-2(e). The 1968 regulations specify a hierarchy of methods: comparable uncontrolled price (CUP), resale price (RP), cost plus (C+), and other methods. The Treasury argued that changing the rules for intangibles, but not for tangibles, would have created an artificial distinction between the two and led to transfer pricing disputes. In terms of priorizing the methods, CUP remains the highest priority, but RP and C+ are put on an equal footing. The proposed 482 regulations require that RP, C+, and other methods be double checked to see if CPM is also met (i.e., these methods result in an operating income for the taxpayer that is within the comparable profit interval). If this test is not satisfied, then none of the three methods is acceptable and the transfer price must be calculated using CPM. Criticisms of Section 482 Mark I The proposed regulations met with a storm of criticism, focused on the comparable profits method. The 46th annual Congress of the International Fiscal

The U.S. Tax Transfer Pricing Regulations: Rules 425 Association (IFA) adopted a resolution stating that CPM deviated from the arm's length standard (Turro 1992a). The Congress was also critical of the tax authorities using information that could not reasonably have been available to taxpayers at the time the transfer price was established. The IFA argued that CPM was a shift from arm's length pricing to an arm's length profit method. Comparison of profits 'reflects none of the basic features of the arm's length principle, particularly because it is not transactional, it does not take into account the contractual arrangements between the related parties and can hardly reflect special circumstances' (quoted in Turro 1992a, 867). The IFA stressed the importance of adhering to arm's length pricing based on a transactional approach, recommended a broader interpretation of CUP to allow substitute comparables, and recommended that fourth methods be used only to verify established methods. In 1992, the OECD's Committee on Fiscal Affairs established a special task force, made up of representatives from the tax authorities of its 24 member countries, to evaluate the Treasury proposals. The OECD report, issued in January 1993, was highly critical (OECD 1993b; Tax Notes International 1993d). Two major criticisms were made: the transfer pricing methodology was not consistent with the arm's length standard, and it was not clear how the proposed regulations would work under the mutual agreement procedures in bilateral tax treaties. The major deviations from the arm's length standard, as seen by the report, were the adoption of CPM and periodic adjustments. The OECD argued that CPM was a problem because the incomes of uncontrolled taxpayers in the same or similar activities varied not only because of differences in prices, but also because of other factors including differences in cost, location, degree of business experience, goals of the firm, and so on. Unless these differences were held constant (i.e., the analysis is conducted ceteris paribus), the results from CPM could be fundamentally inconsistent with arm's length pricing. Also, since firmand industry-level data were required for the CPM, there would be a built-in bias towards using U.S. data since access to data from other countries would be limited. In addition, the OECD argued, the arm's length standard is based on evaluating bargains at the time they take place, without the benefit of foresight. Events that are not known and could not be reasonably predicted are not taken into account in practice in arm's length negotiations between unrelated parties. The internationally accepted rule follows actual practice: payments are judged by the facts and circumstances known to the parties at the time, not by subsequent events. This follows the court decision in R.T. French v. Commissioner (60 T.C. 836 [1973]). Therefore, the OECD concluded, the Treasury proposals for

426

The Rules of the Game in North America

annual reappraisals of the pricing of intangibles based on the CWI standard were inconsistent with the arm's length standard. The OECD stopped short of demanding that the U.S. Treasury withdraw the CPM and the periodic adjustments clause. Instead, the report recommended that the proposed regulations be 'substantially amended.' The report requested that the CPM only be used for tax abusive cases (e.g., tax havens) and that CPM primarily be used as a cross-check or complementary method, a method of last resort, rather than a replacement for transactions-based methods (CUP, RP, C+) or profit split methods, and that CPM be rebuttable by the taxpayer. The OECD report also argued that a profit split was preferable to the computed profit method because profit splits were based on analysis of the MNE's actual commercial activity. The report asked that the 'sound business judgement' wording that had been added to section 1.482(1) be redefined as judgment based on the facts known or that could reasonably have been known by the taxpayer at the time of the transaction. The proposed sound-business-judgment test required that the taxpayer have full knowledge of the relevant facts, which the OECD viewed as unlikely to be satisfied in practice. We return to this point about sound business judgment in Chapter 12. The OECD's second major criticism had to do with tax treaties and arbitration of transfer pricing disputes. The report predicted the U.S. proposals would result in transfer pricing adjustments that would not be acceptable to other tax authorities under the mutual agreement procedure (MAP), thus increasing the probability of double taxation of MNEs. The OECD report requested that the U.S. Treasury agree that agreements reached under the MAP would take precedence over the proposed regulations. Mark II: The 1993 Temporary Regulations49 In response to the barrage of criticism, in January 1993 temporary regulations (T.D.8470, referred to herein as 482-T93) were introduced that confirmed the IRS's commitment to the arm's length pricing standard. On the same date that 482-T93 was released, Treasury International Tax Counsel James Mogle discussed the regulations at a California State Bar conference (Ekman 1993). Mogle admitted that criticisms of the 482-T93 regulations were valid. The proposals did not meet the arm's length standard and looked like a long laundry list of rules. The temporary regulations were designed to address both issues: to return to the arm's length standard and avoid laundry lists. The goal of the temporary regulations, according to Mogle, was flexibility (provided for by removal of the priority of methods and by adoption of a new 'best method'

The U.S. Tax Transfer Pricing Regulations: Rules

427

rule), but in exchange for flexibility much greater written documentation and disclosure would be required of the taxpayer. Mogle stated that three concepts stood out among the major revisions in 482-T93: the importance of comparables, the focus on identification and evaluation of functions and risk, and the need to provide contemporaneous documentation (Ekman 1993, 244). The temporary regulations, which went into effect in April 1993, replaced all provisions of the existing 482 regulations except for loans and advances, services, and the use of tangible property. In addition to these regulations, the Treasury issued new proposed rules dealing with foreign legal restrictions and profit splits, and with accuracy-related penalties. All the proposals in 482-P92 were withdrawn with the exception of the rules for cost-sharing arrangements. We outline the changes in the pricing regulations in this section (see Table 8.1 for a summary).50 While 482-T93 contains many other changes, the ones we identify below are the major innovations,51 with one exception. The temporary regulations also introduced proposals for significant new penalties for TPM, accompanied by requirements for contemporaneous documentation of the firm's transfer pricing policies. We deal with these in the next chapter. Purpose and Scope The arm's length standard is defined in the regulations as the consideration charged in comparable transactions under comparable circumstances between uncontrolled parties. The commitment to the arm's length standard can be seen in the stated purpose of section 482: to ensure that taxpayers clearly reflect income attributable to controlled transactions, and to prevent the avoidance of taxes with respect to such transactions. Section 482 places a controlled taxpayer on a tax parity with an uncontrolled taxpayer by determining the true taxable income of the controlled taxpayer in a manner that reasonably reflects the relative economic activity undertaken by each taxpayer. (482-T93, 47)

The IRS can therefore make an allocation either to clearly reflect income or to prevent tax avoidance. Realization of income is not a prerequisite; the IRS can make an allocation even if the income anticipated from a transaction has not or never is realized. The Best-Method Rule Acknowledging that transfer prices depend on the facts and circumstances of the case, the 1993 temporary regulations are less mechanical and more judgmental than the 1992 proposals. The regulations adopt a new best-method rule;

428

The Rules of the Game in North America

that is, the best method is the one 'that provides the most accurate determination of an arm's length result.' Since this method may vary with the facts and circumstances of the transaction, it is both flexible and informationally demanding. In choosing the best method, three factors must be taken into account: 'completeness and accuracy of the available data, the degree of comparability between controlled and uncontrolled transactions, and the extent of adjustments necessary to apply the method' (482-T93, 13). In terms of comparability between transactions, the standard is weaker in 482-T93 than in the 1968 or 1992 proposed 482 regulations. For two transactions to be comparable, they do not need to be identical, but only 'sufficiently similar that the uncontrolled transaction provides a reasonable and reliable benchmark for estimating an arm's length result (482-T93, 15). The regulations provide new definitions of types of comparables. Product comparables focus on similar products, functional comparables on similar functions, with similar resources and levels of risk. Where uncontrolled transactions are similar in product and function, other methods (e.g., CPM) look for comparable returns. Five factors must be examined for comparability: the taxpayer must look for comparable functions, risks, contractual terms, economic conditions,52 and properties or services. Comparability of functions must be ascertained through a functional analysis of the controlled and uncontrolled taxpayers. Several different types of risk are identified (market, R&D, financial, credit and collection, product liability, and general business risks). Returns should be commensurate with the level of risk (482-92T, 58-61). Since contractual terms can affect the price, they must be investigated and adjustments made where necessary. Significant economic factors that could affect the price must also be considered.53 The last factor requires a comparison of the property or services transferred. Two special circumstances are considered that may affect the comparability analysis: a market penetration strategy and different geographic markets, including the issue of location savings. The first (market penetration) is designed primarily for Japanese MNEs, either distributors or manufacturers, in the United States that have shown low taxable income or losses over several years (see Chapter 7 for more details). A transfer price based on a market share strategy is acceptable only if (1) an uncontrolled firm with the same strategy would have set the same price; (2) the business strategy is likely to achieve its intended results, lead to future profits, and is pursued for a reasonable time; and (3) the strategy, related costs, and expected returns are documented before the strategy is implemented. Given these restrictions, few firms are likely to be able to meet this special circumstances test. The second special circumstance is different geographic markets. Generally,

The U.S. Tax Transfer Pricing Regulations: Rules

429

data from the same market are preferred. Where the MNE has activities in lowcost locations, the location savings can only accrue to the foreign affiliate if the MNE can prove that the profits have not been competed away through the 'competitive effects attributable to the other producers in similar geographic markets capable of performing the same functions at the same low costs' (482T93, 66-7). The assumption of competition among producers in developing countries eliminating the rents for the foreign affiliate means that any location savings accrue to the U.S. parent - a view the IRS has argued for several years (see Chapter 5 for more details). Calculation of a method is to be based on data from the year under review, but the IRS is permitted to use multiple-year data, both before and after the year in question, even though some information (e.g., from the year under review and after) could not have been available to the parties at the time of the transaction. All the alternatives facing the taxpayer are to be considered, including whether or not the firm could have obtained the item itself rather than from an affiliated party (the 'make or buy' decision). Aggregation of interrelated (round trip) transactions is permissible, as in the 1992 proposed regulations. In addition, the temporary regulations explicitly state that there can be more than one arm's length acceptable price; in fact, there can be a range of such prices. Two or more uncontrolled comparable transactions, using the same pricing method, each of which independently establishes an arm's length result, determine an arm's length range. An arm's length range cannot be determined by applying two or more methods to the same uncontrolled transaction. If the taxpayer's transfer price is within the range, it is considered to be an arm's length price. If the transfer price is outside the range, the IRS can move the price to any point within the range, which ordinarily will be the midpoint. There is therefore a clear incentive for the MNE to ensure that its prices fall within the arm's length range for the best method. Methods for Pricing Tangibles54 For transfers of tangibles there are five methods: CUP, RP, C+, other methods, and the comparable profits method (CPM). In certain cases, a profit split method may also be used (see below). CUP is formally defined as a product comparable method; RP and C+ are functional comparable methods. CPM can only be used for pricing tangibles in certain cases.55 Payments for tangibles are not subject to periodic adjustment, whereas payments for intangibles are (see below). The taxpayer can elect to use another method only if none of the enumerated ones apply, if the method is disclosed on the tax return, and if the firm prepares contemporaneous documentation supporting the method, which documentation is to be given to the IRS within 30 days of a written request.56

430

The Rules of the Game in North America

Methods for Pricing Intangibles For transfers of intangible property through a licence or royalty,57 the temporary regulations require the taxpayer to distinguish between routine and nonroutine intangibles. The regulations define a nonroutine intangible as an intangible central to the conduct of a business activity and without which the business activity could not be conducted. It would normally be expected that such property is unique or nearly unique, that its use or application is very valuable, and that there subsequently would not be examples of substantially similar transactions between unrelated parties ... The term would not include intangible property that is a normal result of conducting a business... or ... for which there may be acceptable substitutes available in the marketplace at a comparable price. (482-T93, 28)

Where neither party to the transaction owns nonroutine intangibles, three methods are outlined for pricing routine intangibles: the comparable uncontrolled transaction method (CUT), the comparable profits method (CPM), and other methods.58 The transfer is normally to be in the form of a royalty payment that is subject to annual adjustment to ensure that the payment is commensurate with earned income. CUT, the equivalent to CUP for tangibles, is based on an amalgamation of MTM and CATM in the 1992 proposals. CUT is not subject to a mandatory check by the comparable profits method, but periodic adjustments are required for its use. The firm must take eight factors into account to ensure comparability with transactions by unrelated parties, including a calculation of the net present value expected from using the intangible. Given the difficulty of estimating net present values for one's own intangibles, much less for those of unrelated parties, CUT may not be used very often in practice. The Comparable Profits Method CPM carries over from section 482-P92, but with much simpler requirements. The general principle behind CPM is that 'similarly situated taxpayers will tend to earn similar returns over a reasonable period of time' (482-T93, 36). The new CPM constructs a range of operating incomes for the taxpayer based on industry-average financial returns for third-party firms performing similar functions. CPM is generally assumed to provide an accurate measure of the arm's length standard. A summary of the 1993 version of CPM is provided in Box 8.1, for comparison with the 1992 proposal. CPM starts by identifying the 'tested party' as the party which is not using any valuable nonroutine intangibles, in order to focus on the party with the simplest and most easily compared activities. Where intangibles have been

The U.S. Tax Transfer Pricing Regulations: Rules

431

licenced, the tested party is normally the licensee. The method then measures the total return on the firm's activities, applied separately to each industry segment59 of the tested party. A range of operating incomes is constructed for the tested party based on the financial results of third-party firms performing similar functions. A single profit level indicator can be used to construct the arm's length range. Two types of profit level indicators are recommended: rate of return on capital employed, and financial ratios such as the ratio of operating income to sales and the ratio of gross income to operating expenses. Once an income is established for the tested party, all remaining profits are allocated to the other party. When the 1992 and 1993 versions of CPM are compared, the 1992 version had much stricter requirements in terms of information gathering and statistical manipulation. The 1993 CPM can be calculated by simply looking at industrywide rates of return on assets for firms performing similar functions. The Proposed Profit Split Method In addition to the temporary regulations, the Treasury issued proposed regulations for profit splits, for discussion purposes only. Where both parties to the transaction own valuable nonroutine intangibles, the proposals allow firms to use the profit split (PS) method, but provide several hurdles that must be cleared before the firm is free to use this method. The PS method can only be used if it provides the most accurate measure of the arm's length result. The firm formally elects the method; once elected, the method is binding on all subsequent tax years. The firm must document the combined profit and loss of the controlled parties, prove there are significant transactions between them, and that each party owns valuable, nonroutine, self-developed intangibles that contribute significantly to earning the combined income from the activity. One of four profit split rules can apply; three (the fourth is 'other') are described below: The residual allocation rule: This method is similar to the basic arm's length return method (BALRM-with-profit-split). Market returns to each party for all activities other than nonroutine intangibles are first calculated, using either CUP, C+, RP, or CPM. The residual profit is divided into manufacturing and marketing intangibles. The value of these intangibles is estimated by looking at factors such as R&D expenditures (for manufacturing intangibles), and advertising costs (for marketing intangibles), and then the residual is split between the two parties based on their shares of these intangibles. The values may be current or capitalized expenditures. No third-party data are necessary for these calculations. The capital employed allocation rule: If the two parties bear equal levels of

432

The Rules of the Game in North America

risk (measured as the probability of success or failure) with respect to their investments in the activity, an equal return to capital employed is assumed to apply to each firm, and the combined profit or loss is allocated based on the relative percentages of capital each party employs. Capital is defined as the book or fair market value of all operating assets plus the value of any employed intangible property. The value of intangibles is estimated in the same way as in the residual allocation rule. This is a type of formulary approach, but it is based on the parties to the transaction, not on the worldwide MNE family. Again, no outside data are required.60 The comparable profit split rule: The firm must find outside data on unrelated parties performing comparable functions, see how they have divided their combined operating profits, and then use their allocation. The method was first outlined in 482-P92, but the requirements are so demanding, both in terms of finding similar firms and having access to information on costs and revenues, that the method is unlikely ever to be used. Criticisms of Section 482 Mark II When the temporary regulations were issued, the initial reaction was one of relief - that the rules were less draconian than the 1992 proposals. Several detailed analyses were published which, as time passed, became more critical of the regulations (Fuller and Aud 1993; Granfield 1993a; Hannes 1993; Higinbotham et al. 1993; Nolan et al. 1993). These criticisms were echoed in the OECD's special task force, which reviewed the temporary regulations and issued a report in the fall of 1993 (OECD 1993a). The OECD's report welcomed the direction taken by the Treasury, particularly in reaffirming its commitment to the arm's length standard, in adopting a more flexible approach to the transfer pricing methods, and for placing greater emphasis on the facts and circumstances of individual cases. The report, however, was still critical of the comparable profits method and periodic adjustments, and made several recommendations for change. The major OECD recommendations were: 1. The U.S. government should confirm that it will not be bound by the temporary regulations when dealing with a case through the mutual agreement procedure or arbitration under a tax convention. 2. CPM should be used only in abusive cases and as a method of last resort, and should be based on publicly available data that should be shared with the taxpayer. 3. The preference for transactional methods should be clearly reflected in the process for method selection.

The U.S. Tax Transfer Pricing Regulations: Rules

433

4. Periodic adjustments shoud be confined to truly abusive cases, and once a transaction has been determined to be at arm's length, adjustments should not be made for later years for factors that could not have been foreseen or were outside the control of the taxpayer. 5. The standards of comparability for CUP and CUT should be loosened so they are easier to meet. 6. The advance reporting and documentation requirements for using a profit split should be dropped to make the method more widely available. 7. The compliance burden imposed by contemporaneous documentation should be reduced. As we show below, most of these recommendations were at least partly implemented. Mark HI: The 1994 Final Regulations61 The final section 482 regulations (U.S. Internal Revenue Service [1994b], referred to here as 482-F94), 262 pages in length, were issued on 1 July 1994. The rules deal with many of the criticisms of the temporary regulations, but, in general, follow the same approach.62 According to the preamble, the 'format and substance of the final regulations are generally consistent with the 1993 regulations,' and any changes are intended to 'clarify and refine' the temporary rules 'without fundamentally altering the basic policies' (482-F94, 117). Purpose and Scope The purpose and scope of the regulations are unchanged. The IRS can intervene, not only in tax abusive situations, but whenever the income allocation does not reflect the arm's length standard, and even if the income is not realized. The key criterion is that the taxpayer reports its 'true taxable income' (482-F94, 117).63 Section 1.482-l(b) states that the key principle behind section 482 is the arm's length standard, but the wording has changed. Whereas the temporary regulations refer to the results of comparable transactions between uncontrolled taxpayers in comparable circumstances, the final wording is: A controlled transaction meets the arm's length standard if the results of the transaction are consistent with the results that would have been realized if uncontrolled taxpayers had engaged in the same transaction under the same circumstances (arm's length result). However, because identical transactions can rarely be located, whether a transaction produces an arm's length result generally will be determined by reference to the results of comparable transactions under comparable circumstances. (482-T94, 133)

434

The Rules of the Game in North America

The argument here is that taxpayers should be placed 'on the same (rather than a merely similar) footing' (482-F94, 118). The Best-Method Rule The best-method rule has been kept and broadened. Whereas in the 1993 regulations, the best method was that which provided the 'most accurate measure' of an arm's length result, the final wording is 'most reliable measure.' There is no strict ordering of pricing methods, and no one method is to be considered more reliable than any other (482-F94, 133). While the rule does not require the taxpayer to evaluate all methods in order to determine which is the best method, the 1994 revised penalty regulations (issued to parallel the final 482 regulations) do require the taxpayer to evaluate the potential applicability of the other specified methods. Thus the penalty regulations, not the 482 rules, may require the taxpayer to invest in substantial administrative and information search costs to find the best method. Reliability of a method depends on three factors: the degree of comparability between the controlled and uncontrolled transactions, the completeness and accuracy of the data, and the reliability of the assumptions. For two transactions to be comparable they do not have to be identical but must be sufficiently similar that the uncontrolled transaction is a 'reliable measure' of the controlled transaction. If there are 'material differences' - that is, differences that would materially affect price or profit - adjustments must be made to account for these differences if they improve the reliability of the result (482-F94,119). Completeness and accuracy of the data affect the taxpayer's ability to identify and quantify material differences between the controlled and uncontrolled transactions. The same five factors determining comparability in the temporary regulations (functions, contractual terms, risks, economic conditions, and property or services involved) are developed in more detail in the final version.64 Risk is treated separately, and the decision on who bears the risk is different from the temporary regulations. In general, to the extent that taxpayers allocate risk by contract and their conduct is consistent with the contract, the IRS will respect the allocation of risk between the parties. Where the contractual basis is unclear, risk is allocated according to the economic substance of the transaction, each party's ability to fund losses, and the extent of operational control over the business activities. The final regulations also include the two special conditions: market share strategy and geographic markets. The market share strategy (running short-run losses in order to gain entry to a market or to increase market share) is only acceptable under certain strict conditions, including contemporaneous documentation, the costs being borne by the party that expects to derive the benefits,

The U.S. Tax Transfer Pricing Regulations: Rules

435

and a reasonable likelihood that the strategy will succeed. Location savings from operating in a low-cost jurisdiction must be allocated among the parties consistent with the competitive conditions in the foreign location; thus the savings will only be left with the subsidiary under strict conditions. The arm's length range is redefined. In the temporary regulations all valid applications of a particular method are included; in the new regulations there are two ways to determine the range: where comparability is high (exact cornparables), all results are included, but where the reliability of the comparables is questionable (inexact comparables), only the interquartile range (the 25th to 75th percentile) is used. The Treasury argument is that broadening the definition of acceptable comparables requires the narrowing of the statistical range (i.e., throwing out the outliers) in order to reduce the risk of spurious results. Methods for Pricing Tangibles The list of pricing methods for tangibles now includes six methods: CUP, RP, C+, CPM, profit split, and unspecified methods. The method chosen must follow the best-method rule: the more direct and reliable measure of an arm's length result. The CUP method may now be used even if there are more than minor differences between the controlled and uncontrolled transactions, or where adjustments for minor differences cannot be made. This is an important change from the temporary regulations. However, the use of inexact comparables lessens the reliability of this method. In addition, indirect evidence from public exchanges or quotation media can be used where such pricing data are widely available and used routinely in business to establish prices for commodities traded in large volumes (e.g., Chicago Exchange prices for pork and beef). Where CUP is used, the most important of the five comparability factors is product comparability; for RP and C+ it is comparability in terms of functions, risks, and contractual terms. Unadjusted industry average returns (e.g., gross profit margins) cannot be used to independently establish an arm's length result for the RP method. If an unspecified method is used, the method should reflect the principle that uncontrolled taxpayers would negotiate an uncontrolled price by considering all realistic alternatives to the transaction, such as, for example, a bona fide offer from another party. The method chosen should, however, be based on actual transactions. Where an intangible in embedded in the transfer of a tangible (e.g., sales of a brand-name product to a distributor for resale), in most cases only the price of the tangible must be evaluated, unless the purchaser of the product also acquires the right to exploit the intangible, in which case both the tangible and intangible must be evaluated.

436

The Rules of the Game in North America

Methods for Pricing Intangibles The final regulations prescribe two methods for pricing intangibles. The first method involves the licensing or sale of the intangible where a price (royalty) must be determined. Under 482-F94, both lump-sum and per-unit royalty payments are subject to periodic adjustment in order to satisfy the CWI standard. The second method is a cost-sharing arrangement in which costs and benefits are allocated among the participants. We look at each below. Licensing or Sale of Intangibles. The final regulations list four methods for pricing intangibles: CUT, CPM, profit split, and unspecified methods.65 There is no hierarchy of methods; again, the best-method rule must be employed to choose the method. The CUT method determines an arm's length royalty for an intangible by looking at 'comparable transfers of comparable intangible property under comparable circumstances' (482-F94, 125). In the 1993 regulations, this sentence meant that for two transactions to be comparable, the intangibles must be used with similar products or processes within the same general industry, and the intangibles must have 'substantially the same' profit potential. The key difference in the 1994 regulations is that 'substantially the same' is broadened to 'similar' profit potential. Profit potential is to be measured where possible through 'direct calculations, based on reliable projections, of the net present value of the benefits to be realized through the use of the intangible' (482-F94, 125). The 1994 rules expand on the definition of who owns intangible assets. Ownership of intangibles is determined in one of two ways: legal ownership of the right to exploit the intangible, or, in the absence of legal ownership, the controlled taxpayer that bore the greatest share of the development cost of the intangible is given ownership (thus size of firm and the place where expenses are located should influence ownership). Where a nonowner enhances the value of an intangible, that party must be compensated for effectively performing a service on behalf of the owner of the intangible. Where an unspecified method is used, the method should provide information on prices or profits that the controlled parties could have realized by choosing a 'realistic alternative' to the controlled transaction. The regulations suggest a general principle whereby uncontrolled parties examine the realistic alternatives to a transaction and only enter into the transaction if no better alternative exists. The example provided in the 482-F94 regulations to support this argument is a disguised version of the recent Bausch & Lomb case (see U.S. Internal Revenue Service 1994b, 13-4154). In the example (shortened here), USCO licences

The U.S. Tax Transfer Pricing Regulations: Rules

437

a process technology to its European subsidiary EURCO; the process substantially reduces the cost of manufacturing an adhesive. Using the technology, EURCO produces and sells the adhesive to related and unrelated parties at a market price of $550 per ton. In addition, EURCO pays a royalty of $100 per ton to USCO. The 482 regulations argue that in determining the arm's length price for the intangible, the IRS should look at USCO's realistic alternatives, which would include producing the product itself instead of having EURCO do it. If USCO could realistically produce and sell the adhesive itself for $300 per ton, then having EURCO produce it instead costs USCO $550 - $300 or $250 per ton in forgone revenues. Since the royalty is $100 per ton, the current arrangement leaves USCO with a net loss of $150 per ton. Thus the $100 royalty charge is not arm's length because it is not commensurate with the income earned from the intangible. The example does not say, but presumably the royalty charge would be adjusted upwards to $250 per ton. Periodic Adjustments. The regulations continue to generally presume that payment for transfers of intangibles is in the form of a royalty subject to periodic adjustment according to the CWI standard. Lump-sum payments are also potentially subject to periodic adjustments by treating the payment as an advance equivalent to a stream of royalties over the life of the contract. The 'equivalent royalty amount' is to be based on a net present value calculation using the lumpsum payment, an appropriate discount rate, and projected sales over the relevant period. This equivalent royalty serves as the basis for determining the arm's length price, and is used for purposes of calculating the periodic adjustment. Five limited exceptions from periodic adjustments are provided in the regulations under section 1.482-4(f)(2)(ii): (A) Exact Comparables: Periodic adjustments will not be made if the same tangible was transferred under substantially the same circumstances to an unrelated party (i.e. an exact comparable) and the price of this unrelated transaction is used to generate an comparable uncontrolled price for the related party transaction (i.e. a CUT) in the first year of the transfer, then no subsequent adjustment is required. In short, if an exact comparable exists and was used to determine the first year's price there is no need for an adjustment. (B) Inexact Comparables: Where a comparable arm's length transaction in a comparable intangible exists and is used to provide an inexact CUT, no adjustments are required if each of the following occurs: (a) The controlled parties signed a written agreement providing that the price would be an arm's length price for the first year and outlining a price for each subsequent year, and that agreement remained in force during the tax year under review; and

438

The Rules of the Game in North America

(b) The uncontrolled parties, in similar circumstances to the controlled parties in the year under review, had also signed a written agreement which contained no provisions allowing any change in the price and not allowing any renegotiation or termination of the agreement. (c) The controlled and uncontrolled agreements are substantially similar in terms of their time periods and provisions. (d) The controlled agreement limits use of the intangible in a manner that is consistent with industry practice, and any such limitation is also in the uncontrolled agreement. (e) No substantial changes in functions performed by the controlled transferee occurred after the agreement was executed, except changes required by unforeseeable events. (f) The aggregate profits actually earned, or cost savings realized, by the controlled party from exploiting the intangible in the tax year under review, and in all previous tax years, were not less than 80 per cent nor more than 120 per cent of the prospective, foreseeable profits or cost savings when comparability with the outside transaction was first established. (C) Other Methods: When a method other than an exact or inexact CUT was used to establish the arm's length price, no allocation is required if each of the following holds: (a) The controlled parties had signed a written agreement that established a price for each taxable year under the agreement, and the agreement was in force during the year under review. (b) The price for the intangible was an arm's length amount in the first taxable year during which a substantial periodic consideration was required to be paid, and relevant supporting documentation on the pricing policy was prepared at the same time as the controlled agreement was executed. (c) There were no substantial changes in the functions performed by the transferee since the agreement was executed, except those changes required by unforeseeable events. (d) The aggregate profits actually earned, or cost savings realized, by the controlled transferee from exploiting the intangible in the tax year under review, and in all previous tax years, were not less than 80 per cent nor more than 120 per cent of the prospective, foreseeable profits or cost savings when comparability with the outside transaction was first established. (D) Extraordinary Events: No allocation will be made if the following requirements are met: (a) Due to extraordinary events that were beyond the control of the related parties and which could not have been reasonably foreseen at the time the written contract was signed, the total actual profits or the total cost savings to the controlled taxpayer are less than 80 per cent or more than 120 per cent of the prospective profits or cost savings, and (b) All the requirements of either section (B) Inexact Comparables or section (C) Other Methods are otherwise satisfied.

The U.S. Tax Transfer Pricing Regulations: Rules

439

(E) Five-Year Period: If all the requirements of either section (B) Inexact Comparables or section (C) Other Methods are met for each year of the five-year period beginning with the first year in which a substantial periodic consideration was required to be paid, then no adjustment is required for any subsequent year.

The problem, of course, with these exemptions is the very difficulty taxpayers will have in meeting these tests. The first exemption (the exact CUT) will almost never be met since there are few examples of arm's length transfers of exactly comparable intangibles. The ability of the taxpayers to satisfy the six tests included in the inexact comparables is also very problematic. Where other methods are used, the MNE must have contemporaneously documented its transfer pricing policy in addition to all the other requirements. The extraordinary events exemption apparently provides a breathing space, but the additional restrictions prove that this is more apparent than real. Lastly, if transfers of intangibles meet an arm's length test for five years (i.e., the transfer was arm's length in the first year and for the next four years the results lie within 80-120 per cent of what was reasonably foreseeable), periodic adjustments will not be made after the fifth year. Thus the likelihood of periodic adjustments has somewhat been reduced relative to the proposed or temporary section 482 regulations, but periodic adjustments have certainly not been eliminated for most taxpayers. If the taxpayer fails to meet these tests, then section 1.482-4(f)(2)(i) applies; that is, the transfer price will be adjusted by the IRS to ensure that it is commensurate with the profits earned or costs saved by the transferee when using the intangible. The adjustments must be consistent with the arm's length standard using the best-method rule. Cost-Sharing Arrangements, Although final transfer pricing regulations for section 482 were issued in 1994, section 1.482-7 on cost-sharing arrangements was omitted. In December 1995, the IRS filled this gap when final cost-sharing regulations were issued (see Dodge et al. 1996; Dolan 1996; Fuller 1996; Knee, Shapiro, and Joy 1996; and Wood 1996). Cost sharing provides an alternative to licensing or selling intangibles to related parties. Each party to a cost-sharing arrangement is treated as the owner of any intangibles that are developed, and the income from those intangibles accrues to the owner. Costs of developing the intangibles are to be shared among the participants in the same proportion as each party's expected share of the benefits from the intangibles. In addition, the final regulations require contemporaneous documentation of the arrangements and periodic adjustments to reflect changed circumstances.

440

The Rules of the Game in North America

Costs for a participant include the firm's own incurred costs of intangible development, plus cost-sharing payments made to other participants, minus all cost-sharing payments received from other participants. Reasonably anticipated benefits are defined as all additional income that the firm can reasonably expect to derive from the use of the intangibles. Benefits can be measured on a direct basis (revenues generated and/or costs saved) or indirect basis (units produced or sold, operating profit, sales). Net present values may be used where parties anticipate different timing streams of benefits and/or costs. In effect, the new cost-sharing arrangements treat the participants as a nonequity joint venture, splitting costs in proportion to the benefits the partners expect to derive from the joint venture. This method has some clear advantages over the pricing methods for intangibles under the final section 482 regulations. As Wood (1996, 421) notes: the 'commensurate with income' requirement of section 482, with all its complexity, uncertainty, and audit exposure, is effectively avoided. The ultimate result is essentially a profit split between participants (and tax jurisdictions) on income attributable to the intangibles developed.

Wood concludes that the new regulations are 'consistent with the principles of the transfer pricing "trinity" - i.e. flexibility, comparability, and documentation' (1996, 432). We therefore expect to see cost-sharing arrangements used much more frequently now that final regulations have been issued. The Comparable Profits Method The comparable profits method (CPM) can be used for either tangibles or intangibles, but the method is significantly weakened in the final regulations, in response to repeated complaints about inconsistency of CPM with the arm's length principle (compare the first and last columns in Box 8.5). The sentence that CPM 'ordinarily will provide an accurate measure of an arm's length result' is deleted from the final regulations. Given adequate data, the Treasury expects that CUP or RP is likely to achieve more comparability than CPM, and therefore CUP or RP is more likely to be the method chosen under the best-method rule. Thus, CPM will generally be a method of last resort where comparable data from uncontrolled parties are unavailable. Under the 1993 regulations, CPM could not be used if either party owned valuable nonroutine intangibles; this condition is now weaker so that intangible property is one of several factors to consider in using this method. The profit level indicators to be used to calculate CPM include the rate of return on capital employed and financial ratios; other indicators can be used if

The U.S. Tax Transfer Pricing Regulations: Rules 441 they are objective measures of profitability derived from uncontrolled taxpayers. Internal measures of profitability are not acceptable. In determining an arm's length result under CPM, the taxpayer's 'average reported operating profit' for the year under review and the two previous years is used for comparison (482-F94, 127). A number of comparability factors must be taken into account, particularly resources employed, risks assumed, and functional comparability. If the taxpayer's result falls outside the arm's length range, ordinarily the IRS will adjust the result to the median of the range. As Thomas Horst notes, 'the CPM is destined to be widely used because it is so simple to apply' (Horst 1993, 1443). How easy it is to use can be seen from the example of CPM using the 1994 regulations, which is shown in Box 8.6 (parts A through C). The box assumes that a U.S. MNE (USCO) with a Canadian affiliate (CANCO) manufactures and sells a product X to CANCO, which CANCO then sells to unrelated parties. Part A of the box shows the financial statements of the two parties. USCO sells ten units of X to CANCO at a transfer price of $200 per unit, for total sales of $2,000. USCO's cost of goods sold is $1,400, leaving a gross profit of $600. After operating costs are subtracted, USCO has an operating profit of $100. CANCO imports X from USCO, incurs its own operating expenses of $400, and sells the final product to consumers for $280, creating total sales revenue of $2,800 and an operating profit of $400. The U.S. tax rate is assumed to be 45 per cent, whereas the Canadian rate is 35 per cent. Thus the after-tax profit of USCO is $55 and for CANCO is $260. Initially CANCO is much more profitable than USCO in terms of the comparable profits method indicators. We calculate the rate of return on capital employed for USCO at 0.05, and for CANCO as 0.20. CANCO also has an operating profit-to-sales ratio significantly higher than USCO's (0.20 versus 0.05). Lastly, the ratio of gross profit to operating expenses (the Berry ratio; see Chapter 13) is 1.20 for USCO and 2.00 for CANCO. If we compare these indicators with the mean and range for similar unrelated U.S. manufacturing firms (shown in Part B of Box 8.6), USCO's profit rates are well below the relevant industry averages. If the comparable profits method is used by the IRS and USCO is the tested party, the parent's profit indicators can be brought into line with the industry range by having CANCO make a 5 per cent royalty payment to USCO. This is illustrated in Part C of the table. The royalty payment from CANCO to USCO raises the effective transfer price from $200 to $214. This increases the U.S. firm's operating profit, while the affiliate's falls. The U.S. tax revenues rise from $45 to $108, while Canadian taxes, assuming the Canadian government provides a full compensating adjustment, fall significantly from $140 to $91. However, CPM is successful at

442

The Rules of the Game in North America

BOX 8.6 The Comparable Profits Method: An Example Part A: Financial statements of the controlled parties MNE's financial statements (in US$ millions) Sales (USCO to CANCO, CANCO to unrelated parties) Selling price Cost of goods sold Gross profit (equals sales minus cost of goods sold) Operating expenses Operating profit (equals gross profit minus operating expenses) Taxes paid (per cent of operating profit) (US = 45%, CAN = 35%) Operating assets (fixed, current) Rate of return on capital employed (operating profit to operating assets) Ratio of operating profit to sales Ratio of gross profit to operating expenses

Parent (USCO)

Subsidiary (CANCO)

2,000 200

1,400

2,800 280 2,000

600 500

800 400

100

400

45

140

2,200

2,000

0.05 0.05 1.20

0.20 0.20 2.00

Part B: Profit level indicators

Rate of return on capital employed Operating profit to sales Gross profit to operating expenses

Range of similar uncontrolled parties in the United States

USCO

CANCO

Mean of similar uncontrolled parties in the United States

0.05

0.20

0.11

0.06-0.20

0.05

0.20

0.11

0.06-0.24

1.20

2.00

1.39

1.25-1.90

bringing the returns to USCO more in line with U.S. industry averages, as the last three rows in Part (C) of the box show. We return to our discussion of CPM in Chapter 13. The Profit Split Method The 482-F94 regulations finalize and liberalize the profit split (PS) rules first proposed in the 1993 regulations. The Treasury has historically been reluctant to allow use of profit splits since they are not based on outside, objective data. However, given industry demands for this method, the final regulations now allow the PS method in cases where it is impossible to find reliable outside data.

The U.S. Tax Transfer Pricing Regulations: Rules

443

BOX 8.6 (concluded) Part C: The revised financial statements after CPM is applied MNE's financial statements (in US$ millions) Royalty (5% royalty on sales paid by CANCO to USCO) Sales (USCO to CANCO, CANCO to unrelated parties, includes royalty income) Selling price, including per-unit royalty Cost of goods sold Gross Profit (equals sales minus cost of goods sold) Operating expenses Operating profit (equals gross profit minus operating expenses) Taxes paid (per cent of operating profit) (US = 45%, CAN = 35%) Change in tax revenues (assuming the foreign government provides a correlative adjustment) Total assets Rate of return on capital employed (industry mean = 0.11) Operating profit to sales (industry mean = 0.11) Ratio of gross profit to operating expenses (industry mean = 1 .39)

Parent (USCO)

140

Subsidiary (CANCO) -140

2,140 214 1,400 740 500

2,800 280 2,140 660 400

240

260

108

91

+63

-49

2,200

2,000

0.09

0.13

0.11

0.09

1.48

1.65

SOURCE: Hypothetical example; data on profit ratios of uncontrolled manufacturers from McLennan (1993, 1207)

The PS method can be used for either tangibles or intangibles; taxpayers do not have to make a binding election in order to use the method; and the requirement that both parties must have valuable nonroutine intangibles is deleted. However, since the best-method rule assumes that results based on comparisons with uncontrolled results are generally more reliable, the PS method, which uses internal data, will generally be a method of last resort. To calculate a PS, the taxpayer must 'estimate an arm's length return by comparing the relative economic contributions that the parties make to the success of a venture, and dividing the returns from that venture between them on the basis of the relative value of such contributions' (482-F94, 128). Only two of the four methods proposed in the 1993 regulations are kept: comparable profit splits and residual profit splits.66 Since the first requires knowledge of how comparable unrelated taxpayers split their profits (information the taxpayer is unlikely to have), the first method will probably be little used. The second method is equivalent to the BALRM with profit split method proposed in the 1988 Treasury White Paper: a method such as CPM is used to estimate market

444

The Rules of the Game in North America

returns for routine intangibles and to allocate the returns to the two parties; the residual is allocated between them according to the relative value of the parties' contribution to the activity. Comparability with the New OECD Guidelines The final version of the section 482 regulations was issued in July 1994, almost at the same time as Part I of the OECD's draft transfer pricing guidelines were issued. Since then the OECD has issued all parts of the report (OECD 1994b, 1995a,b, 1996) so that a comparison between the U.S. rules and the OECD guidelines for the pricing of goods, services, and intangibles can be made. Cheng (1995) provides such a survey, together with helpful tables showing the U.S. regulations on the left-hand side and the equivalent OECD guidelines on the right (there may be an implicit comment here). Cheng argues that both sets of rules embrace the arm's length standard and emphasize comparability on a transactional basis as the key to meeting the ALS. There are still some real differences, however. For example, the CPM and transactional net margin method (TNMM), while they may look alike on paper (as Culbertson [1995a] argues), may be treated very differently in practice by U.S. and European tax authorities. CPM is expected to be used much more frequently than TNMM also. Second, the U.S. rules insist on periodic adjustments as necessary for the commensurate with income standard, where as the OECD frowns on such ex-post adjustments. This implies that royalty revaluations by the IRS may not be acceptable at the competent authority stage, leading to double taxation of MNE income. Levy and Wright (1995) also compare the U.S. and OECD Part I transfer pricing rules, concluding that the two sets of regulations have become more similar in that they have both moved to broader, more flexible rules. However, the U.S. regulations are much more technical, rule-based and formalized than the OECD guidelines, and, like Cheng (1995), the authors see areas of disagreement, in particular regarding CPM and periodic adjustments. Because both the U.S. and OECD rules now give the taxpayer more freedom to choose the transfer pricing method, but also more responsibility in documenting and applying it, Levy and Wright offer the following practical advice for taxpayers: (a) know the characteristics of your intercompany transactions; (b) know what makes companies successful in your industry; (c) understand how the market works and how unrelated parties deal with each other in the open market; and (d) put all of these together to develop the 'business case' to develop arm's length transfer prices and to explain those prices to the relevant tax authorities. (Levy and Wright 1995, 32)

The U.S. Tax Transfer Pricing Regulations: Rules

445

Taxpayers need more expertise (particularly economic expertise in functional analysis) so that they can take advantage of the greater flexibility in the new regulations, while avoiding the potential pitfalls (e.g., penalty rules). Summary: Section 482 Mark III The final regulations, according to James Mogle, formerly of the Treasury and head of the team that drafted the 1993 temporary regulations, provide 'a tremendous amount of flexibility ... [that has] advantages and costs, and will create greater controversy' (quoted in Turro 1994f, 81). Steven Hannes, previously chief of international rulings at the IRS, concludes that the final regulations are closer to international norms since the role of CPM has been reduced while that of CUP, RP, and C+ methods expanded (quoted in Turro 1994f, 81). The new regulations appear to offer more flexibility and room for judgment, and therefore more room for error and IRS-taxpayer dispute. The 'facts and circumstances' of each case are now critical factors determining the best-method rule and the appropriate transfer price, or range of prices. The Evolution of Section 482 The commensurate with income legislation, passed by Congress in 1986, has had a major impact on the evolution of U.S. transfer pricing regulation over the 1986-94 period. The U.S. Treasury, and the Internal Revenue Service, have had to reappraise the regulations for pricing of tangibles as well as intangibles. The simple three-method list (CUP, RP, and C+) was inadequate to handle the complexities of CWI, with its valuation of income potential and periodic adjustments. The reforms have gone in three directions: The rules have been broadened and made more flexible (e.g., new methods such as CPM and PS, inexact comparables used for CUP, a range of acceptable arm's length prices, the best-method rule, some relief from periodic adjustments). Reasonableness and reliability are key concepts in determining comparability. The emphasis is on the facts and circumstances of the particular case. More guidance is provided to the taxpayer (e.g., detailed explanations of comparability, discussion of special circumstances such as market share strategies and location savings). More documentation (contemporaneous documentation, functional and risk analyses) is required and harsher penalties can be levied (under section 6662). This is the stick that ensures taxpayers comply with the new transfer pricing regulations.

446

The Rules of the Game in North America

Overall, the final regulations provide more flexibility in terms of finding comparable transactions. An uncontrolled transaction must be 'sufficiently similar' that it provides a 'reliable measure' of an arm's length price. Adjustments are made to the uncontrolled transaction only if they improve reliability of the results. There can be a range of arm's length results. While the regulations implicitly favour CUP and CUT, and suggest CPM and PS as methods of last resort, in general, there is no hierarchy of methods; the method selected by the taxpayer is to be the best (most reliable) method. CPM uses external information on rates of return and financial ratios to benchmark the results of uncontrolled taxpayers; PS uses internal information on profitability (effectively, BALRM) to allocate profits over and above normal returns. Conclusions Internal Revenue Code section 482 is not long, but the 482 Treasury regulations that have developed around it since 1968 are voluminous. As one commentator on 482-T93 recently complained: One wonders whether the United States needs over 250 pages of regulations - almost equal in amount to all the national income tax regulations in Canada - to enforce a twosentence statute ... when the essence of the regulatory message is, appropriately, that transfer pricing is a fact-sensitive issue. (Hannes 1993, 418)

As we have seen in this chapter, up until the mid-1960s, the principal thrust of the U.S Internal Revenue Service in applying section 482 was to offset attempts by U.S. taxpayers to shift income to tax havens where tax rates are low or nonexistent. The IRS was concerned that U.S. multinationals were underpricing their outbound transfers to their affiliates (for example, parts exports and technology transfers) in order to shift income abroad to countries with lower tax rates. Since the U.S. tax system allows deferral of foreign-source income until repatriated, both the taxes received by the U.S. government and the total tax burden of the group of related taxpayers could be significantly reduced in this manner. The Treasury and the IRS were concerned about tax avoidance and evasion in the 1960s, particularly in two types of multinationals: U.S. petroleum multinationals with foreign branches in the Middle East, and U.S. pharmaceutical MNEs with foreign subsidiaries in Puerto Rico. In the 1970s, this concern was extended to U.S. MNEs with offshore processing operations in Ireland and Southeast Asia (Guttentag and Miyatake 1994, 376). The issue of transfers of intangibles offshore was important for many of these cases. In most cases, for-

The U.S. Tax Transfer Pricing Regulations: Rules

447

eign tax rates were negligible and the U.S. government feared that substantial profits were being diverted offshore. More recently, the IRS has begun to change its focus, for two reasons. First, with the rise in the 1980s in inward foreign direct investment into the United States, the IRS is now concerned with the income taxes paid by foreign corporations in the United States. The fear is that foreign MNEs are overcharging for transfers to their affiliates in the United States with the purpose of reducing their U.S. incomes and thus their U.S. taxes. The IRS has been especially attentive to Japanese transplants in the auto and consumer electronics industries where foreign affiliate returns and U.S. tax payments have been relatively low.67 Second, as the U.S. statutory corporate income tax rate has fallen, particularly since tax reform in 1986, while the U.S. budget deficit has simultaneously risen, the need to earn more tax revenue per dollar of tax base has become more acute. The budget deficit has led to stricter enforcement by the IRS and the casting of a wider net for transfer price violators.68 The history of U.S. regulation in the area over the past ten years reflects this desire by Congress and the IRS to increase the United State's share of taxes collected by all tax authorities worldwide. Most 482 adjustments now involve the United States and another developed market economy where corporate tax bases and tax rates are similar or higher than U.S. ones. Shifting taxable income from one high-taxed country to another does not reduce the MNE's total tax burden, only the location where the tax is paid. The location, presumably, does not matter to the multinational, but it clearly does to the two tax authorities. The nature of transfer pricing disputes has therefore changed from disputes over allocations between U.S. multinationals and their tax haven affiliates, the perceived problem in the 1960s and 1970s, to disputes between the IRS and either (1) U.S. firms with majority-owned foreign affiliates (MOFAs) in deyeloped market economies such as Canada and the European Community, or (2) foreign MNEs, particularly Japanese firms, with subsidiaries in the United States.69 Thus the legislation is now used less often as a method of preventing tax abuse through tax avoidance or evasion, the first prong of section 482. The focus of the Internal Revenue Service has shifted to the second prong, ensuring that the true taxable income of the group is reported for tax purposes. The United State's desire to protect its tax base is generating 'turf wars' as the U.S. Treasury attempts to take a larger share of the total tax base worldwide. In order for the American share to rise, either other countries must reduce their tax share or more taxes must be levied on the MNEs' global income. For multinationals

448

The Rules of the Game in North America

caught between warring tax regulators, higher U.S. tax collections mean either lower collections for other tax authorities if these authorities provide correlative adjustments, or double taxation if they do not. As a result, the second prong of section 482 is creating some conflict between the IRS and tax authorities in other developed market economies where their focus has been on the more traditional first prong of tax avoidance (Guttentag and Miyatake 1994). We will return to this topic in chapters 12 and 13, but now we turn to the second part of our history: the U.S. administrative procedures for dealing with transfer pricing disputes.

APPENDIX 8.1 A HISTORY OF U.S. TAX TRANSFER PRICING REGULATION Date

Section

Topic

Descriptions of legislation or document

1917

41

Allocation of income and deductions

Commissioner is authorized to allocate income and deductions among affiliated corporations, can require consolidated returns be filed.

Exemption from U.S. taxation for possessions

Exempts foreign-source income of U.S. MNEs that receive at least 80 per cent of their income from U.S. possessions, where at least corporations 50 per cent of the income comes from an active trade or business, from U.S. taxation. Dividends paid to the U.S. parent are taxable on repatriation, while liquidated distributions are tax free.

War Revenue

Act 1921 Revenue Act

1928 Revenue Act

351

Tax-free transfers of intangibles

Non-recognition of income in extraordinary transactions. U.S. MNEs can transfer intangible assets tax free to foreign affiliates in return for shares in the foreign subsidiary.

1928 Revenue Act

45

Rationale for allocating income and deductions

Commissioner is authorized to allocate income and deductions among related corporations so as to prevent tax avoidance and determine the true tax liability of the related parties.

1935 Treasury Regulations

45-l(b)

Defines arm's length standard

Arm's length standard - i.e., that of an uncontrolled taxpayer dealing at arm's length with another uncontrolled taxpayer - is to be used by the commissioner in making a clear reflection of income and preventing tax evasion among related parties.

The U.S. Tax Transfer Pricing Regulations: Rules 449 APPENDIX 8.1 (continued) Descriptions of legislation or document

Date

Section

Topic

1954 Revenue

482-1

Renumbering 45 as 482

Section 45 becomes section 482.

1968 Treasury Regulations

482-2

Intercompany transfer pricing regulations

Defines arm's length standard for five types of transactions: loans or advances, business services, rental of tangible property, use or transfer of intangibles, sales of tangibles. Outlines four methods for tangibles: comparable uncontrolled price, resale price, cost plus, fourth methods.

1976

367

Transfers of intangibles to foreign affiliates

Denies tax-free status for intangibles transfers, not involving a sale or licence, to foreign affiliates where a tax avoidance motive is evident. Recharacterizes transfers as contribution to capital and levies them with a toll charge.

936

Tax credit for possessions corporations

Converts the tax exemption to a foreign tax credit for 'qualified posession source investment income (QPSII),' defined as income from FDI in an active business or trade.

1977 Treasury Regulations

861

Allocation of overhead, R&D and interest expenses

Treasury regulations for allocating and apportioning R&D costs, general and administrative expenses, interest costs, etc. among related parties. Suspended by statute in 1981.

1981 Economic Recovery Tax Act

861

Allocation of R&D costs

100% of R&D costs incurred in U.S. could be allocated to U.S. source income. Extended in 1984 and 1985.

Formal document request

Admissibility of documents maintained in foreign countries. If IRS issues a formal document request and the documents are not produced, subject to reasonable cause, the documents cannot subsequently be submitted by the taxpayer in a court of law to challenge an IRS tax allocation.

Act

Tax Reform Act

1976

Tax Reform Act

1982 Tax 1982 Equity and Fiscal Responsibility

Act (TEFRA)

450

The Rules of the Game in North America

APPENDIX 8.1 (continued) Date

Section

Topic

Descriptions of legislation or document

1982 TEFRA

6038A

Information reporting by foreign corporations

Required all U.S. corporations 50% or more owned by a foreign person to report annually on intrafirm transactions using Form 5472. Minimal penalties for noncompliance.

1982 TEFRA

936(h)

Safe harbours for possessions corporations

Raises the per cent of active business income necessary to qualify as a possessions corporation from 50 to 65 per cent of gross income. All intangible income earned by a possessions corporation must be allocated to the U.S. parent (the Dole rule), unless one of two safe harbours is chosen: a 50-50 profit split or cost sharing.

1984 Tax Reform Act

367(d)

Deemed royalty payments for outbound intangibles

Annual arm's length payments for outbound intangibles must be included in the U.S. parent's income whether received or not. Treated as stream of royalty payments over life of intangible.

1986 Tax Reform Act

1231(e)

CWI standard

Commensurate with income (CWI) standard to apply to section 367(d) on deemed receipts from intangible transfers, so-called super-royalty.

1986 Tax Reform Act

861

Allocation of overhead, R&D and interest expenses

Specific allocation rules for R&D and interest expense between U.S. and foreign-source income in an affiliated group. Interest apportioned on asset values. Headquarters expenses apportioned on gross income. 50% of U.S. R&D costs allocated to U.S. income, with remainder split between U.S. and foreignsource income based on sales or gross income.

1986 Tax Reform Act

1059A

Customs valuation as ceiling on transfer price

Commissioner is required to take the customs valuation into account in determining the arm's length price; transfer price for tax can be no higher than for customs purposes.

Allocation of R&D expenses

64% of U.S.-incurred R&D could be allocated to U.S. income; 64% of foreign R&D to foreign income; remainder apportioned on sales or gross income basis. Continued in 1989, 1990, and 1991.

1988 861 Technical & Miscellaneous Revenue Act

The U.S. Tax Transfer Pricing Regulations: Rules 451 APPENDIX 8.1 (continued) Date

Section

Topic

Descriptions of legislation or document

1988 Treasury White Paper

Treasury White Paper on intercompany pricing of intangibles

White Paper focuses on pricing of intangibles for CWI standard. Recommends either pricebased method (exact and inexact comparables) or profit-based method (basic arm's length return method [BARLM] or BARLM with profit split). Rules for cost sharing proposed.

1989 6038A Omnibus Budget Reconciliation Act (OBRA)

Information reporting by foreign corporations

All corporations with minimum of 25% foreign ownership must report annually intrafirm transactions. Penalties increased substantially. Corporations required to keep on hand for IRS perusal upon request any documents relevant to transfer pricing examination.

1990 6038A Revenue Reconciliation Act (RRA)

Information reporting by foreign corporations

Extended to all open tax years.

1990

603 8C

Information reporting by foreign branches

Same information filing and record keeping required of U.S. branches of foreign corporations as for U.S. corporations under 6038A.

6662

Accuracy penalties extended to 482 violations

Penalties for substantial and gross valuation misstatements. 20% penalty for pricing 200% above or 50% below IRS's arm's length price, or for adjustments over $10 million. Penalty doubles if inaccuracies double.

1991 IRS Revenue Procedures

91-22, 9123, 91-24

Advance Pricing Agreement (APA)

Voluntary procedure whereby taxpayers can get an (up to three-year) advance ruling on their transfer pricing methodology from the IRS. Taxpayer provides detailed information on transfer pricing methodology (TPM).

1992 Treasury Regulations

482-92P

Intercompany transfer pricing preliminary regulations

Draft proposals for 482 revisions based on CWI. Computed profit interval (CPI) method based on rates of return introduced. Negative reactions from OECD and tax groups.

RRA

1990

RRA

452

The Rules of the Game in North America

APPENDIX 8.1 (continued) Date

Section

Topic

Descriptions of legislation or document

1992

92-56

Allocation of R&D expenses

Gives taxpayers option of 64% rule or 1977 regulations.

1993 Treasury Regulations

482-93T

Intercompany transfer pricing temporary regulations

New 482 temporary regulations introduce bestmethod rule, computed profit method (CPM). Possible use of profit splits where both parties hold nonroutine intangibles. Reaffirms arm's length standard and CUP for goods (CUT for intangibles), admits range of prices exists. Periodic adjustments for intangibles. Includes proposals for accuracy-related penalties. 936 corporations using cost sharing must follow 482 principles.

1993 OBRA

6662

Accuracy-related penalties

'Final' rules for implementing 6662. OBRA passes modified version of 482-93T penalty regulation proposals. Defines 'reasonable cause and good faith' exemption. Must use 482 pricing method. Contemporaneous documentation required.

1993 OBRA

936

Possessions corporations

Section 936 tightened by reducing the income tax credit for cost-sharing or profit split exemptions. Activity-based credit added as alternative.

1993 OBRA

936 with 482

Possessions corporations

Section 482 temporary regulations require possessions corporations using the costsharing safe harbour to make annual payments to their U.S. parents commensurate with the income earned by the affiliates from the use of the intangible.

1994 Treasury Regulations

482-94F

Intercompany transfer pricing final regulations

Final section 482 regulations focus on reasonableness and comparability. Best-method rule replaces hierarchy of methods. Profit split and CPM are specific methods, normally of last resort.

1995 Treasury Regulations

6662

Accuracy-related penalties

IRS issues 'final' penalty regulations.

IRS Revenue Procedures

The U.S. Tax Transfer Pricing Regulations: Rules

453

APPENDIX 8.1 (concluded) Descriptions of legislation or document

Date

Section

Topic

1995 Treasury Regulations

861

Allocation of R&D expenses

IRS issues final rules on the allocation and apportionment of research and experimental expenditures.

1995 Treasury Regulations

475

Final cost-sharing regulations

Final cost-sharing regulations are issued as section 1.482-7.

Possessions corporations

Section 30A is added to Income Tax Code. Repeals section 936 and phases out the possessions tax credit over a 10-year period.

1996 30A Small Business Job Protection

Act 1996 Treasury Regulations

6662(e)

Accuracy-related penalty

IRS issues final and temporary regulations for 6662(e) that reduce the probability of a penalty.

1996 IRS Revenue Procedures

96-53

Advance Pricing Agreement

IRS issues revised APA procedures that increase fees, set up APA teams, and encourage multilateral APAs and rollbacks to open years.

9

The U.S. Tax Transfer Pricing Regulations Part II: Procedures

Introduction The previous chapter dealt with the U.S. rules or methods for pricing intrafirm transactions. We turn now to the U.S. administrative procedures which supplement the rules. In the first part of the chapter, we outline the general procedures used by the Internal Revenue Service in auditing taxpayers under the corporate income tax at the examination, appeals, litigation, and competent authority stages, and discuss some problems with the general process. We then turn to recent solutions. Since the early 1980s, the Internal Revenue Service has introduced two types of procedural changes designed to improve taxpayer compliance in the transfer pricing area. First, the IRS has introduced specific administrative procedures that require more, and now contemporaneous, documentation from taxpayers and impose penalties where evidence exists that transfer prices deviate substantially from the arm's length price. Second, the IRS has introduced two new dispute resolution procedures: the Advance Pricing Agreement (APA) and binding arbitration. The APA process allows taxpayers to negotiate an acceptable transfer pricing method prior to the tax audit. Binding arbitration has recently been used to settle one transfer pricing dispute (with Apple Computers). We turn first to a discussion of the traditional audit procedures followed by the IRS. General Audit Procedures It is a broadly held view among U.S. Treasury officials and the U.S. Congress that the Internal Revenue Service has a compliance problem in the section 482 area. As we saw in Chapter 7, foreign-controlled corporations report low levels of taxable income in the United States, and transfer pricing manipulation is gen-

The U.S. Tax Transfer Pricing Regulations: Procedures

455

erally believed to be one cause of these low taxes. Bill Clinton, in his campaign for president, promised to substantially strengthen the Service's compliance procedures, and indeed the number of IRS examination agents has grown dramatically over the last few years. What are the problems with the current auditing procedures? We look first at each stage and then draw some general conclusions. The Examinations Stage Most transfer pricing cases begin with an audit of a firm's revenues and expenditures for a particular tax year, done by the Examinations Division of the Internal Revenue Service (Brunori 1994). An audit generally takes place within three years of filing the tax return. The examinations or field agent is responsible for identifying an audit issue and for developing the supporting facts and the appropriate transfer pricing method. Agents are provided with IRS audit guidelines, updated as the regulations change, that identify problem areas and outline procedures.1 For several years, U.S. tax auditors have been using functional analysis to determine which party has the responsibility for which functions, and to allocate costs and revenues on this basis. For example, the IRS manual directs the IRS examiner to 'go behind' the books and records of a taxpayer and discover the underlying realities of particular transactions. The auditor establishes the functions performed by the related parties and determines the ownership and value of any intangibles involved in the transactions, using a prepared list of questions (see Box 9.1 below).2 The advantage of a functional analysis is that it is based on solid microeconomic foundations. It looks at the value chain of the firm, determines the valueadding functions at each stage, and documents the intrafirm flows of goods, services, and intangibles. In other words, functional analysis attempts to get inside the 'black box' of the integrated business, as we explained in Chapter 3. This is both the advantage and disadvantage of the technique since the MNE as an integrated business by definition has interdependencies that cannot be decomposed in this manner. At best a functional analysis can provide a road map to the integrated business, outlining the responsibilities and risks of the various entities; sources and uses of common resources; and intrafirm flows and prices. At worst, it can impose the world of perfect competition, perfect certainty, and long-run equilibrium on real transactions that are anything but perfectly competitive, certain, or long-run. Functional analysis is a powerful tool of microeconomics, but like most powerful tools it needs to be wielded with care. Taxpayers are now expected to perform a functional analysis as a required

456

The Rules of the Game in North America

BOX 9.1 List of Questions in a Functional Analysis

What was done? What economically significant functions were involved in doing it? Who performed each function? How was the function accomplished? Are there any valuable intangibles used in performing the given function? Why were the transactions structured the way they were? Where and when did the transactions occur and which entities were involved? What economic risks were assumed? SOURCE: Fuller (1989, 34-5; 1994b, 243)

step in meeting the new section 482 regulations. The regulations require contemporaneous documentation by the MNE of its transfer pricing policy. Such documentation is to include a functional analysis (see Chapter 8 for details). A functional analysis is also required as part of the U.S. Advance Pricing Agreement process (see below). In Canada, Information Circular 87-2 recommends the use of functional analysis by the MNE and the tax auditor. The Canadian APA may include, but does not require, a functional analysis (see Chapter 10). The Appeals and Litigation Stages If the initial audit leads to a reallocation of income, with assessed taxes and possible penalties, the taxpayer may dispute the assessment. If this happens, the dispute proceeds to the IRS Appeals Division. The administrative appeals process is the traditional dispute resolution technique used by the Service. The purpose of Appeals, according to the IRS, is 'to resolve tax controversies, without litigation, on a basis that is fair and impartial to both the government and the taxpayer and in a manner that will enhance voluntary compliance and public confidence in the integrity and efficiency of the Service' (quoted in Wrappe 1994, 1583, n. 21). Appeals officers are responsible for exploring the range of a

The U.S. Tax Transfer Pricing Regulations: Procedures

457

reasonable settlement with the taxpayer, given the strengths and weaknesses of the case and the assessed probability of winning or losing if the case goes to tax court. Face-to-face settlement conferences between the IRS and Appeals officers settle about 80-90 per cent of the 65,000 disputed cases coming out of the examination process.3 However, only about 20 per cent of all section 482 income adjustments developed at the examination stage are sustained through the appeals process (Wrappe 1994, 1581). At each level of the process, the taxpayer provides information, develops transfer pricing methodologies, and negotiates with the Service. If the Service demands foreign documents, the process takes even longer. If Appeals cannot resolve the issue, the case goes to Litigation and may end up in tax court. It may take up to ten years for the whole process to be resolved; transfer pricing cases are particularly notorious for their length and expense.4 The Competent Authority Stage The bilateral tax treaty has been the traditional vehicle for international tax cooperation. Traditionally, a tax treaty's main purposes have been to avoid double taxation, reduce discriminatory taxation, and deter tax evasion.5 The United States had 42 bilateral income tax treaties (Hufbauer 1992, 21619) in 1993. Generally, these treaties define residency, taxation of business profits, attribution of profits, allocation of expenses, and treatment of sources of income. As a home country, a major goal of the United States has been to convince treaty partners to reduce their withholding taxes on remitted dividends, royalties, and head office fees.6 The U.S. competent authority is responsible for administering and implementing these tax treaties. The competent authority function is the responsibility of the IRS assistant commissioner (international) and his or her staff in the Tax Treaty Division. IRS personnel assigned to competent authority cases include international examiners, auditors, lawyers, and economists (Matthews 1994a). Problems with the Audit Process Disputes over transfer pricing issues are frequent because of (1) the absence of a well-defined, objective standard, (2) the large dollar amounts involved, and (3) the failure of the taxpayer and the IRS (the two parties) to narrow the issues prior to the trial (Wrappe 1994, 1584). The various transfer pricing methods laid down in the Treasury regulations do not provide much guidance for pricing individual transactions in specific facts and circumstances. The negotiations are

458

The Rules of the Game in North America

plagued with principal-agent problems since the taxpayer has all the internal information (or at least its own) while the IRS has much less information. The taxpayer faces a large tax assessment and is likely to be defensive and to refuse to provide information, while the large amounts of money are likely to attract increased scrutiny by the Service. The problem with the process is therefore at least partly due to its adversarial nature. The competent authority stage has been criticized as being too slow and for not allowing taxpayers to participate directly in the process. In the 1990 fiscal year, the average time for processing a competent authority case was 3.5 years, but this had fallen to about 18 months in fiscal year 1993 (Halphen and Bordeaux 1994, 437). The process has also been criticized for cases where full relief was not attained, since negotiations between the two authorities may lead to 'splitting the difference' (i.e., each party compromises by providing some relief) or 'horse trading' (i.e., trading winning on one case for losing on another, where several cases are discussed at the same time). In addition, many foreign governments are unwilling to provide full relief, particularly where they disagree with the IRS's tax assessment (e.g., periodic adjustments and CPM are not recognized by many governments as acceptable methods under the arm's length standard). The U.S. competent authority estimates, however, that full relief was provided in over 90 per cent of all cases over the 1991-3 period (Halphen and Bordeaux 1994, 437). Because of problems with the traditional audit process, the Internal Revenue Service has moved in two directions to improve compliance by taxpayers and reduce the IRS's administrative overload. The first approach was to increase its enforcement regulations and penalties for noncompliance; the second was to introduce new dispute resolution procedures. We examine enforcement and penalties first, and then turn to dispute resolution. Enforcement and Penalties Three basic categories of enforcement and penalty procedures relate to section 482: (1) formal document requests made by the IRS to the taxpayer, (2) reporting requirements for domestic and foreign-owned firms, and (3) accuracyrelated penalties for significant under- or overinvoicing of transfer prices. We deal with each in turn. Section 982: Formal Document Request1 Section 982, 'Admissibility of Documents Maintained in Foreign Countries,' passed in 1982 as part of the Tax Equity and Fiscal Responsibility Act (TEFRA

The U.S. Tax Transfer Pricing Regulations: Procedures

459

'82), added a formal document request to the Internal Revenue Code. The IRS can request any documents relevant to a tax audit, including tax data from foreign parents, and the taxpayer is only allowed to withhold the documents if it can demonstrate 'substantial compliance' and 'reasonable cause' for failing to provide the document(s). The IRS can prohibit the later use of the missing documents by the taxpayer as a penalty for noncompliance. Foreign nondisclosure rules or lack of control over the corporation holding the documents are not sufficient justification for noncompliance. Section 6038(A, C): Reporting and Record Keeping* Until the 1980s, U.S. and foreign MNEs were generally not required to report their intrafirm transactions to the U.S. government, nor to keep on hand copies of their books and financial statements. The U.S. Congress was concerned with the difficulty the Internal Revenue Service had in accessing information about offshore activities of multinationals. Regulatory changes were made to increase reporting requirements and to ensure that documents would be available for IRS scrutiny upon request. In terms of reporting requirements, before 1982, a U.S. parent corporation had to report transactions with and between its foreign affiliates on Form 5471. Foreign MNEs with affiliates in the United States did not have to report their intrafirm transactions, however. Congress saw two problems with this: first, that U.S. parents with foreign affiliates and foreign parents with U.S. affiliates were not on an equal footing, and, second, that transfer pricing manipulations would be less visible in the latter case because the foreign MNEs were not required to report on intrafirm transactions to the Internal Revenue Service. In TEFRA '82, in addition to the Formal Document Request, the Congress therefore added a clause, Section 6038A, requiring 'all U.S. corporations (and foreign corporations engaged in a U.S. trade or business) that were controlled (owned 50 per cent or more by voting power or value of all classes of stock) by a single foreign person at any time during the taxable year, to provide to the IRS information reports ... relate[d] to any transactions between the reporting corporation and any related parties' (McCawley 1991, 6). The information was to be supplied on an annual information return, Form 5472, with a penalty (minimum US$1,000, maximum US$24,000) for failing to file. The reporting corporation had to report transactions with related parties involving: sales and purchases of tangibles, rents and royalties, sales and purchases of intangibles, payments and receipts for services, commissions, amounts loaned and borrowed and interest paid, and insurance premiums (Cotton 1990,445). The purpose of this information was to help the IRS look for 'outliers,' cor-

460 The Rules of the Game in North America porations with out-of-range average receipts or expenditures, as a flag for more intensive audits down the line. The initial information returns 'were of little use to the Service,' according to Charles Triplett, IRS deputy associate chief counsel (International), so section 6038 was broadened in the 1989 Omnibus Budget Reconciliation Act (OBRA '89) to lower the threshold test to all corporations with a minimum of 25 per cent foreign ownership, and raise the minimum penalty for noncompliance to US$10,000 (Lowell et al. 1994, 351; Triplett 1990, 9). In OBRA '89 new record-keeping requirements were added for foreign multinationals. Reporting corporations are now required to keep on hand and available for IRS examination the books and records necessary to document the transfer prices for intrafirm transactions. Thus, for example, a Canadian corporation selling goods to a related U.S. firm must keep a set of books in the United States or file the relevant information with the IRS. The categories of documents to be maintained include: original entry books and transactions records, profit and loss statements, pricing documents, foreign tax and regulatory filings, ownership and capital records, and nonsales transactions records (Goldberg 1991, 427). Noncompliance may result, not only in fines, but also in loss of deductions and credits. In the Revenue Reconciliation Act of 1990 (RRA '90), section 603 8A was made applicable to all open tax years. Section 603 8C was added to impose similar information reporting and record-keeping requirements on U.S. branches of foreign corporations. The rules now cover all foreign corporations doing business in the United States regardless of the degree of foreign ownership. As we discuss in Chapter 10, Canada has reporting requirements that are similar to those under section 6038(a,c), but does not have any requirement that MNEs keep copies of all records onshore and available for scrutiny by Revenue Canada. To the extent that the two authorities exchange information, presumably the Canadian government can access these documents if they are not confidential. We return to this issue in Chapter 14. In addition to reporting requirements, Congress has substantially increased the penalties that can be levied for noncompliance with the Internal Revenue Code. The transfer pricing penalties are discussed below. Section 6662: Accuracy-Related Penalties9 In 1990, Congress added a penalty clause to the Internal Revenue Code, section 6662. This section was designed to provide the stick that would ensure that taxpayers complied with the arm's length pricing standard, section 482. In 1993, the IRS issued proposed penalty regulations, implementing section 6662; these

The U.S. Tax Transfer Pricing Regulations: Procedures

461

were updated in 1993 and again in 1994 at the same time as the section 482 regulations were updated. Final regulations were issued in 1996 (see U.S. Internal Revenue Service 1993a, 1995, 1996). Under section 6662, valuations either substantially above or below the arm's length price face severe penalties. In addition to staying within a limited range of the IRS valuation, taxpayers must carefully document their transfer pricing policies either before or at the time the transactions take place, and be prepared to provide this documentation to the IRS on request. In Chapter 6 we outlined the impact penalties could have on the MNE's incentive to manipulate transfer prices. We showed that a penalty could be an effective deterrent to transfer price manipulation. Now let us look at the evolution of the U.S. penalty regulations over the 1990-6 period. We focus first on the legislation and then on the Treasury regulations. Adding Section 6662 to the Internal Revenue Code In OBRA '89 the penalty provisions under the Internal Revenue Code were consolidated, but valuation matters were not discussed. The 1990 hearings on tax compliance by foreign-controlled corporations (the Pickle hearings, see Chapter 7 for details) brought the issue of transfer pricing to the attention of the Congress, and in RRA '90, a penalty clause for section 482 violations was added to the Internal Revenue Code. The purpose of section 6662 is to induce the taxpayer, through threat of an imposed penalty, to provide contemporaneous documentation of the firm's transfer pricing policies. The rationale provided for the section 482 penalty was IRS experience in transfer pricing audits. The Service found that most taxpayers were unable to provide an explanation of their transfer prices at the time of an audit (often some years after the transaction had occurred). This increased the time spent and costs of an IRS audit for both the Service and the taxpayer. The penalty provision for under- or overvaluation would provide a stick encouraging taxpayer documentation and compliance with the 482 regulations. In the final penalty legislation, section 6662 differentiates between a substantial valuation mis statement (SVM) and a gross valuation misstatement (GVM). A substantial valuation misstatement occurs if 1. the price for any property or services, or use of property, charged in connection with a transaction between related parties is 200 per cent or more, or 50 per cent or less, of the price as determined under section 482; or 2. the net effect in one taxable year of all adjustments under section 482 is the lesser of an increase in taxable income of more than $5 million or 10 per cent of gross receipts.

462

The Rules of the Game in North America

The latter provides a de minimus limitation, preventing the imposition of a penalty for amounts below US$10 million. A gross valuation misstatement occurs when these thresholds are doubled (i.e., over 400 per cent or under 25 per cent of the transfer price, or the lesser of an adjustment of more than US$20 million or 20 per cent of gross receipts). The penalty charged is a 20 per cent tax levied on the portion of the underpayment of tax attributable to an SVM (section 6662[a]), or 40 per cent for a GVM (section 6662 [g]). The penalty is in addition to the tax assessment, bears deficiency interest, and is not tax deductible. With a U.S. federal tax rate of 34 per cent, the penalty raises the effective rate of tax to 41 per cent for an SVM (1.2 x 34) or to 48 per cent for a GVM (1.4 x 34). Relief from the penalty is provided, in section 6662(c), only if there was reasonable cause for the underpayment and the taxpayer acted in good faith. This 'reasonable cause and good faith' exception is not defined in the legislation, and has been the subject of some controversy (e.g., what is necessary to prove that the MNE acted in good faith?). It is interesting to note that in the July 1990 public hearings on section 6662, no witness recommended adopting section 6662 (Wickham 1991, 17-19). Spokespersons for both the Treasury and the IRS recommended against transfer pricing penalties on the grounds that (1) general penalties also existed and could be used instead; (2) penalties would have little effect in improving taxpayer compliance because of the great difficulty the IRS had in proving transfer price manipulation; (3) other compliance methods, in particular, information reporting requirements, more audit activity, and new section 482 regulations, were preferable to penalties. Congress, however, went ahead and passed the legislation and directed the U.S. Treasury to develop regulations for section 6662. Below, we provide an example to illustrate how the regulations might work in practice, and then discuss the history of the Treasury regulations. The. Transfer Pricing Penalty: An Example An example might help clarify the regulations. In Figure 9.1 we assume the regulated transfer price W is $10.00. We show the various ranges created by section 6662 for the taxpayer as choices along a line of possible transfer prices. If the MNE's transfer price falls within a range which lies above a floor of 50 per cent less than the regulated price ($5.00) and lies below a ceiling of 200 per cent more than the regulated price ($20.00), no penalty is levied. Thus the nopenalty range includes all transfer prices higher than $5.00 but lower than $20.00, that is, all prices between $5.01 and $19.99. A substantial valuation misstatement occurs if the transfer price falls between 25 and 50 per cent below W (that is, in the $2.51-$5.00 range) or between 200 and 400 per cent above

The U.S. Tax Transfer Pricing Regulations: Procedures

463

FIGURE 9.1 Section 6662 Penalties for Transfer Price Manipulation

the regulated price ($20.00-$39.99). Thus the SVM range consists of two segments, a small one below the regulated price and a large one for transfer prices in excess of W. The gross valuation misstatement occurs for transfer prices of $2.50 or below, or prices of $40.00 or higher; again the upper range is more open than the lower range (assuming a floor of zero). Now let us look briefly at the history of the IRS penalty regulations. The 1993 Proposed Penalty Regulations In January 1993, as part of 482-T93, the IRS issued proposed regulations, 1.6662-5(e), for interpreting section 6662(e, h). The proposals follow the outline provided above in more detail.10 The penalties are intended to pressure MNEs to gather data on comparables at the same time intrafirm transactions occur, develop a transfer price methodology based on the best-method rule, and

464

The Rules of the Game in North America

to document, use, and monitor that methodology. The penalties are also high enough to discourage adventuresome transfer pricing methods. The key part of the proposals is the definition of the 'reasonable cause and good faith' exemption. In order to qualify for the exemption, a taxpayer must make a reasonable effort to accurately determine its tax liability, and have a reasonable belief that its transfer price meets the arm's length standard. In terms of the 'reasonable effort' test, the taxpayer must attempt to accurately determine its transfer price and its tax liability. This determination must be made no later than the time when the tax return is filed, and must be documented at the time of filing. The documentation must contain a written analysis of the transfer pricing methodology showing why the result meets the arm's length test of the section 482 temporary regulations. No specific guidance is provided on the type of written analysis required. If the taxpayer has not documented its transfer pricing methodology, or does not provide it to the IRS within 30 days of a request, the exemption is voided and the IRS assumes the taxpayer did not make a reasonable effort to accurately determine the transfer price. In terms of the 'reasonable belief test, the taxpayer must show that it reasonably believed that its transfer pricing result, more likely than not, would be sustained on its merits in a transfer pricing audit. In determining reasonable belief, the regulations suggest the following factors are important: the experience and knowledge of the taxpayer; the use of a listed section 482 pricing method (e.g., CUP, RP, CPM, C+, CUT, but not a fourth method); the taxpayer's reasonable belief in the method and that the critical assumptions underlying the method would not change before the tax return was filed; and the use of a qualified professional (tax lawyer, accountant, economist) in preparing the transfer pricing methodology. The new penalty regulations were strongly criticized by tax practitioners. Morrison (1993a, 859), for example, argued that the penalty 'places a premium on getting one's transfer price right and gives the IRS a potent weapon to encourage less aggressive reporting positions and to obtain more favourable negotiated results to the disputes that do develop.' As a result, he concluded, contemporaneous documentation is, in effect, mandated in all crossborder intrafirm transactions since that is the only way to avoid the 6662 penalties. The 1993 Revised Penalty Regulations Hearings were held on the 6662 regulations in May 1993 and the Treasury proposals were modified and signed into law in August 1993 as part of OBRA '93. In the 1993 version, there are two penalties: transactional and net adjustment (Lowell et al. 1994). The transactional penalty applies to substantial and gross valuation misstatements. For an SVM, a 20 per cent penalty applies if the trans-

The U.S. Tax Transfer Pricing Regulations: Procedures

465

fer price is 200 per cent or more than, or 50 per cent or less than, the arm's length price under section 482. For a GVM, a 40 per cent penalty is levied if the misstatement is 400 per cent or more than, or 25 per cent or less than, the section 482 valuation. The transactional penalty is not imposed on underpayments where the requirements for reasonable documentation are satisfied. The net adjustment penalty is triggered by the size of the adjustment. An SVM occurs if the net section 482 adjustment exceeds the lesser of US$5 million or 10 per cent of gross receipts. A GVM occurs if the net adjustment exceeds the lesser of US$20 million or 20 per cent of gross receipts. These are the de minimus thresholds for the tax penalty (Fuller and Aud 1993, 528). The term 'net section 482 adjustment' refers to the net increase in taxable income, not to the net increase in assessed tax, resulting from any kind of section 482 assessment. Thus, not only is the range of issues under which a penalty can be assessed very large, but the penalty provisions relate to the total amount of net section 482 adjustments, taking all issues into account simultaneously. Section 6664, 'Definitions and Special Rules,' which accompanies section 6662 provides that no penalty shall be imposed for an underpayment of tax if the taxpayer can show that it acted in good faith with respect to the underpayment. Section 6662, defines reasonable cause and good faith, however, by differentiating between those taxpayers using specific 482 methods and those that do not. If a specific method (other than a fourth method) is used, the firm must prove that its use of the method was reasonable, contemporaneously document the method, and provide it within a 30-day window to the IRS upon request. If an unspecified method was used, the taxpayer must prove that none of the other 482 methods was sustainable and that its method was the best method. The method must be documented and also available. Some light on these penalty regulations was provided by IRS associate chief counsel (International) Robert Culbertson, at a February 1994 National Foreign Trade Council seminar on the penalty regulations. Culbertson said that taxpayers are expected to make 'a reasonable effort to collect reasonable data and to choose the best method,' that is, 'given what is in front of you, did you do the best you could?' (quoted in Turro 1994b, 501). The penalty regulations are designed to encourage self-assessment by the taxpayer and to make that information available to the Service. In addition, large MNEs with large volume transactions are expected to provide more data and analysis than small taxpayers or firms with small volumes of transactions. Two types of documents - principal and background - are to be kept. Principal documents consist of the basic transfer pricing analysis done by the taxpayer while background documents support the principal ones. The taxpayer must keep sufficient documentation, in existence at the time of filing, to establish that the firm selected the best transfer pricing method.

466

The Rules of the Game in North America

The 1994 Temporary Penalty Regulations In July 1994, the penalty regulations were updated to conform with the revised best-method and arm's length range rules under the final section 482 regulations. Under the 1994 temporary regulations, the taxpayer must select and apply a specific method in a reasonable manner, given all the facts and circumstances. Five factors are relevant in this determination: 1. the experience and knowledge of the taxpayer; 2. whether accurate data were available and were used in a reasonable manner; 3. whether the taxpayer followed the best-method rule requirements under section 482; 4. whether the taxpayer reasonably relied on an objective analysis done by a qualified professional who had access to all the relevant information concerning the transactions in question; and 5. if more than one uncontrolled comparable is used to establish the arm's length range whether the taxpayer arbitrarily selected an extreme point in the range. (Lowell et al. 1994, 364-5) Reasonableness is a key criterion in the 1994 penalty regulations. It is defined in section 6662 as in the section 482 regulations - that is, given the available data and pricing methods, the one chosen by the taxpayer must provide the most reliable measure of an arm's length result under the best-method rule. The difference from the section 482 regulations is that the taxpayer, in order to avoid a possible penalty, must have made a reasonable effort to evaluate the potential applicability of the other specified methods (Fuller 1994b, 242). The cost of searching for data is a factor that can be considered in determining reasonableness (Lowell et al. 1994, 369). In effect, the penalty regulations are more onerous than the section 482 regulations in terms of the best-method rule. Under the 482 rules, the taxpayer must use the best method, but does not have to work through all the other listed methods to prove that the one chosen was the best. In the penalty regulations, however, the best-method rule requires the taxpayer to prove that the method chosen was superior to all others, and this means documenting that none of the other methods was preferred. Thus, if the 1994 proposed regulations became final rules, the workload for multinationals would be significantly increased. The required documentation in the 1994 regulations is divided into three categories: principal documents, background documents, and tax return documentation. In addition to the 1993 documentation requirements, specific rules were added for profit splits and lump-sum royalty payments. If a profit split methodology is selected, specific information on income, assets, and costs for each rel-

The U.S. Tax Transfer Pricing Regulations: Procedures

467

evant related party and each relevant business activity is required; documentation must be attached to the tax return also. Similarly, if an unspecified method is used, documentation must accompany the tax return. Even if there is no business purpose to the transfer price (e.g., casual intrafirm transfers that would not normally be priced and for which no funds would be exchanged between affiliates), the penalty rules require that a transfer price be set and be defended against all other prices. The 1996 Final Penalty Regulations11 In early 1996, the IRS issued its final penalty regulations under section 6662(e). The IRS (1996b) regulations are quite similar to the temporary ones, with a few key changes. The first set of changes reduce the probability of the penalty being applied. One change is that the size of the net transfer pricing adjustment can be taken into account in determining whether a taxpayer applied a specific method in a reasonable manner. Because the threshold for the net section 482 adjustment penalty is $5 million, MNEs with a small price adjustment but large volumes of intrafirm trade could easily exceed this floor. Similarly, the final regulations delete the requirement that taxpayers must disclose the use of the profit split, lump-sum payments, and unspecified methods on tax returns, in order to avoid the 20 and 40 per cent penalties. Both changes should make it easier to avoid triggering the penalty. The second set of changes affects the best-method rule. One change allows the use of the transfer pricing methodology developed in an Advance Pricing Agreement (APA) to be used as a factor in choosing a specified method. Additionally, not all specified methods need to be tested in all cases if the taxpayer has made a 'reasonable effort to evaluate the potential applicability of other specified methods in a manner consistent with the principles of the best method rule' (section 1.6662-6[d][2][ii]). Also, the data that must be searched and collected in order to apply the best-method rule has been reduced from the data available prior to filing the tax return, to the data available by the end of the taxation year. An Assessment of Section 6662 Wickham (1991, 20-3) provides a list of criteria (given below, modified slightly) against which the penalty regulations can be compared. A good penalty system should be: Efficient: A penalty should reinforce and be rationally related to a legal standard of behaviour that is stated clearly in the law in terms comprehensible to the taxpayer; that is, there should be clear written rules in the law, known to

468

The Rules of the Game in North America

the taxpayer, against which the taxpayer can self-assess and voluntarily pay the tax. Equitable: A penalty should be fair and be seen to be fair. Appropriate: The size of the penalty should be seen to be fair and reasonable and not disproportionate to the culpability of the conduct in question. Effective: A penalty should be effective. It should improve taxpayer compliance, and should not be so severe as to undercut compliance (e.g., cause more litigation). The more equitable and appropriate the penalty, the more effective it can be. Easy to administer: The penalty should be easy to administer. It should be imposed only when necessary, not be so severe that it is difficult to impose, and not impose undue administrative demands on the regulator. Imposed for the right reasons: A penalty should not be levied for reasons other than improving taxpayer compliance - for example, to raise revenue, engage in tit-for-tat retaliation against another country, or punish foreigncontrolled corporations on grounds of political expediency. Nonretroactive: A penalty should not be retroactive; that is, it should not apply to transactions before the enactment of the penalty. Retroactivity is punitive. Compared with these criteria, the section 6662 regulations do not perform very well. First, the penalty regulations are only as efficient as the self-assessment system behind section 482. Without clear and easily followed transfer pricing regulations, the probability of differences between the taxpayer's and IRS's determination of an arm's length transfer price is high, increasing the probability of a tax penalty. Since transfer pricing cases are so sensitive to the facts and circumstances, they lend themselves to disputes, litigation, and penalties. The tax was also introduced to raise revenue and punish FCCs in the United States, and applied initially on a retroactive basis. Because the tax applies to all section 482 adjustments, and calculates the minimum penalty relative to aggregate adjustments, the likelihood of a penalty assessment is high; thus the administrability of the tax is in question. Lastly, given the punitive size of the penalty amounts, section 6662 cannot be viewed as either equitable or appropriate. The burden of proof also falls on the taxpayer to prove it acted reasonably, meeting all five criteria outlined in the 1994 temporary regulations. Thus the effectiveness of the penalty regulations may be undermined. For example, Lowell et al. (1994) assess the effective tax cost of the 1994 temporary regulations as very high, arguing that the elements of this higher effective tax cost (section 482 penalty and penalty interest

The U.S. Tax Transfer Pricing Regulations: Procedures

469

rates)... create a negative arbitrage for multinational taxpayers ... The draconian spectre of this penalty will cause tax executives of any U.S.-connected company ... to develop a strategy for contemporaneous documentation and to otherwise satisfy the conditions of the penalty. The issues that will be faced in even straightforward situations in this regard will be significant, which suggests this is an area where significant evolution can be expected in the future. (Lowell et al. 1994, 374)

It is clear that the U.S. Treasury expects to see a major change in taxpayer compliance as a result of the new penalty regulations. A letter to Congressman Paul Kanjorski from Treasury officials states that: the Administration has launched a vigorous enforcement program focusing in particular on transfer pricing ... Before these changes [the new 482 and 6662 regulations] were adopted, multinationals were able to set their transfer prices in any way they wished with little or no fear of penalty ... [Now] taxpayers that fail to set their prices in accordance with the IRS guidelines, and that fail to document such compliance, will be subject to harsh penalties ... [This should lead to] a marked change in taxpayer behavior in transfer pricing that will improve compliance in this very significant area of international taxation. (Quoted in Tax Notes International1994d)

This concludes our discussion of the U.S. penalty regulations. We come back to penalties for taxpayer noncompliance in Chapter 13. Now we turn to the last of the three sections of this chapter: new dispute-settlement mechanisms. Updating the Dispute-Settlement Process In this section we look at two new dispute-settlement mechanisms: the Advance Pricing Agreement (APA), a domestic mechanism for settling disputes outside the court process, and binding arbitration, a new method which substitutes for the court process (the first domestic example being the 1994 Apple arbitration case). The Advance Pricing AgreementProcess12 An Outline of the APA Process In March 1991, the IRS published revenue procedures introducing an Advance Pricing Agreement (APA), which were modified in 1996 (IRS 1991, 1996a). The APA is an alternative dispute-resolution process whereby a firm meets voluntarily with the Service to negotiate a transfer pricing method that will be used for specific transactions over a specified period.

470

The Rules of the Game in North America

The stated purpose of the APA is to 'lessen uncertainties and enhance predictability of the tax consequences of international transactions by relieving the costly and time consuming examination and resolution of major transfer pricing issues for both the taxpayer and the involved government(s)' (U.S. Internal Revenue Service 1991, 3). An APA is designed to serve as a 'safe harbour' over the period in which the APA was in force, increasing certainty for the taxpayer. An APA generally consists of three elements: (1) agreement between the taxpayer and the Service about the relevant facts and circumstances under negotiation, (2) the development of an acceptable transfer pricing method, and (3) the application of the methodology to determine an arm's length range of results. The firm's transfer prices over the life of the APA are expected to fall within that range.13 The process for obtaining an APA is outlined in Box 9.2. Firms are invited to pre-filing conferences where the IRS and the taxpayer can discuss informally the suitability of an APA. If the taxpayer goes forward with an APA request, the MNE must supply a list of information to the Internal Revenue Service, along with a US$5,000 user fee (US$25,000 in 1996). All requested materials are part of the IRS file and are not returned. When the request is filed, the taxpayer provides a proposed transfer pricing method, together with all supporting documents. The IRS then assembles a multifunctional team drawn from all relevant parts of the Service. The members of the IRS's APA evaluation and negotiation committee then evaluate the proposal. The IRS committee meets 30 to 60 days later to negotiate one opinion, and then meets with the taxpayer. The committee either accepts the proposed method as is or with revisions, or rejects the method. If the taxpayer's proposal is acceptable, the taxpayer enters the APA. The IRS issues a ruling which takes force for up to three years, and can be renewed if the economic facts and circumstances have not changed in the meantime. The APA confirms the method for determining the transfer price and the range of acceptable transfer prices (the range), but not the actual price itself. The agreement includes (1) a transfer pricing methodology, (2) a possible range of expected arm's length results, and (3) the critical assumptions underlying the economic analysis of the method. The taxpayer is expected to file a similar APA with the appropriate foreign competent authority, but a competent authority agreement is not mandatory. An APA can be issued without one - for example, with a nontreaty country or if the taxpayer shows 'good and sufficient' reasons for requesting a unilateral APA. If, subsequent to a unilateral APA, double taxation does develop with a treaty country, the normal competent authority rules follow. Rollbacks to include open tax years are encouraged. In addition to spelling out the transfer pricing methodology, the APA stipulates the annual report - that is, the information the taxpayer must produce each year

The U.S. Tax Transfer Pricing Regulations: Procedures

471

to ensure that it has complied with the terms of the agreement. If the transfer price varies within the range (as defined in the APA ruling) or if profits or their division fall within acceptable ranges (however defined), the APA will not be revoked or revised. The firm is also subject to the regular examination process. During the term of the APA, the taxpayer is still audited by the Service, but the audit is limited to five areas. The taxpayer must prove that: (1) it complied in good faith with APA's terms and conditions; (2) the material representations made in the APA negotiations and annual reports remain valid; (3) the supporting data and analyses used in applying the transfer price methodology were correct; (4) the critical assumptions remain valid; and (5) the firm consistently applied the transfer pricing method and the critical assumptions (Schwartz et al. 1994, 1521). There can be a renewal of the APA for a US$7,500 fee if all parties consent. A new or updated study with analysis and supporting data must accompany the renewal request. The ruling can be revoked by the IRS if the taxpayer fails to comply with the terms and conditions of the ruling, or if the taxpayer misrepresented or left out material facts in the APA request. If the ruling is 'revoked for cause,' additional taxes, interest, and penalties - including the foreign tax credit - may be assessed. The ruling may also be revised or revoked if there is a change in the 'critical assumptions (specific triggering events') underlying the APA that renders the determination unfair or unworkable. Having outlined the process, it might be useful to review one of the few APAs that have been discussed publicly. Generally, the results of the process are not public information, but a few agreements (e.g., Apple Computer, Sumitomo Bank Capital Markets, Barclays Bank, and Matsushita Electric) have been made public with the taxpayer's permission.14 We look at the Matsushita Electric APA below. An Example: The Matsushita Electric APA Matsushita Electric Corporation of America (Matsushita) is a U.S. subsidiary of Matsushita Electric Industrial Co. Ltd., the world's largest consumer electronics multinational. The parent firm, headquartered in Japan, has 12 subsidiaries around the world with a total of 354 plants. Matsushita's total annual revenues in fiscal year 1993^ were US$64 billion (Pollack 1994, Cl). The MNE produces everything from batteries and washing machines to robots and video equipment. Its products are sold under brand names such as Panasonic, Technics, and Quasar, and the MNE owns such movie and entertainment firms as Universal Pictures and MCA Music. Matsushita USA was regularly audited by the Service in the 1980s (Turro 1992b, 1096). The IRS was concerned about the prices Matsushita paid its

472

The Rules of the Game in North America

BOX 9.2 The U.S. Advance Pricing Agreement (APA) Process

Background information names, addresses, etc. for all parties to the requested APA a statement as to whether the taxpayer or any related party has requested an identical or similar APA ruling a general history of the taxpayer's business operations a description of the taxpayer's worldwide structure, principal businesses and their location, major transactions flows, significant transfer pricing practices Measures of profitability for the taxpayer detailed information on the tax base - i.e., the parties and the transactions that will be subject to the APA summarized financial and tax data of the parties for the past three years, including sales, cost of goods sold, operating expenses, profit before taxes, assets, liabilities, number of employees, Form 5471 and 5472, statutory and effective tax rates for each involved foreign country, currency information, description of current (GAAP) accounting methods, differences between U.S. and foreign accounting methods that might influence the APA tax returns, financial statements, annual reports, other U.S. and foreign government filings, etc. which may affect the APA to be cited and supplied to the IRS on request Information on the proposed transfer pricing methodology a detailed explanation and analysis of the proposed transfer pricing method, showing it to be in accordance with Section 482 and consistent with the arm's length standard of tax treaties a complete discussion of the current methods that have a bearing on the APA request, including information on: allocation of gross income, deductions, credits or allowances; principal transactions flows; transfer pricing practices; all tangibles and intangibles subject to TP methods relevant to the proposed method; separation of developed from acquired intangibles and documentation for acquired intangibles a discussion of any relevant legal provisions, court rulings, relating to the current or proposed TP method an explanation of the taxpayer's and government's positions on any previous or current issues at the examination, appeals, court, or competent authority levels relevant to this case; any information involving foreign tax authorities; notification if any of the above start while the APA is in process

The U.S. Tax Transfer Pricing Regulations: Procedures 473

BOX 9.2 (concluded) Economic analysis of industry competitors relevant measures of profitability and return on investment (e.g., gross and net profit margins, the Berry ratio, return on assets) which can be used for competitor comparability a functional analysis of each party: economic activities, assets employed, economic costs, risks assumed relevant measures of profitability for the general industry a list of comparable competitors, or if these do not exist, a list of types of businesses similar to the taxpayer's; the comparability is in terms of economic activities, assets employed (tangible and intangible), economic costs and risks specific economic measurements of selected independent competitors that can be used to establish comparability such as: segregating lines of business; differences related to activities performed, assets employed, risks and costs assumed; volume or scale differences; different economic assumptions on cost of capital or inflation; market penetration rates; maturity of product; terms of sale (FOB, GIF), debt-equity ratios development of profitability measurements for independent comparable competitors After the APA is approved The taxpayer must file an annual report showing yearly results and demonstrating compliance with the APA The IRS can conduct a complete examination as to how the pricing mechanism in the APA is being operated by the taxpayer. The IRS will look at whether the taxpayer is properly complying with the terms and conditions of the ruling, whether the taxpayer's submissions at the time of the APA were based on an accurate statement of the material facts, whether the transactions were carried out as proposed, and whether any critical assumptions have changed. The taxpayer must keep all records used in reaching the APA and all supporting data in any annual report as long as such information is relevant. The taxpayer may be required to provide an independent expert to review the annual reports, compare actual operating results with the APA ruling, and render an opinion as to whether the ruling was complied with and whether the APA represents the economic interests of the parties. The taxpayer must waive disclosure for the expert. SOURCE: Based on U.S. Internal Revenue Service (1991, 1996a)

474

The Rules of the Game in North America

parent for consumer electronic products that were then resold in the U.S. market under such brand names as Panasonic, Quasar and Technics. One transfer pricing investigation of the firm's 1981 and 1982 tax years, started in 1984, was not settled until 1991. The IRS disputed the price that the foreign affiliate had paid its Japanese parent for videocassette recorders. In the end