Spillovers in International Corporate Taxation

This paper explores the nature, significance and policy implications of spillovers in international corporate taxation—the effects of one country’s rules and practices on others. It complements current initiatives focused on tax avoidance by multinationals, notably the G20-OECD project on Base Erosion and Profit shifting (BEPS). The paper draws on the IMF’s experience on international tax issues with its wide membership, including through technical assistance (TA), and on its previous analytical work, to analyze spillovers and how they might be addressed. In doing so, it goes beyond current initiatives to look at a wide set of possible responses.


This paper explores the nature, significance and policy implications of spillovers in international corporate taxation—the effects of one country’s rules and practices on others. It complements current initiatives focused on tax avoidance by multinationals, notably the G20-OECD project on Base Erosion and Profit shifting (BEPS). The paper draws on the IMF’s experience on international tax issues with its wide membership, including through technical assistance (TA), and on its previous analytical work, to analyze spillovers and how they might be addressed. In doing so, it goes beyond current initiatives to look at a wide set of possible responses.


1. International aspects of corporate taxation, which have long arisen in IMF technical assistance, have come to prominence in public debate. Revenue concerns are now prominent around the world, reflecting the need for consolidation while protecting spending in many advanced economies, and the need to support spending for poverty alleviation and growth in developing1 countries. In this context, the relatively low amounts of tax that, as a result of cross-border tax planning (Box 1), many multinational enterprises (MNEs) pay has given rise to significant public disquiet. 2 The underlying issues themselves are not new. For many years, corporate tax design has had to pay close attention to international issues, and—as documented in IMF (2013a)—these issues have featured prominently in the technical assistance (TA) work of the Fund. These long-standing problems—including the concern sometimes expressed that the system allocates too little tax base to developing countries—have now come to a head.

2. The G20-OECD project on base erosion and profit shifting (BEPS) is an ambitious and constructive effort to strengthen the international corporate tax system. Building on the central role that the OECD has played in developing and maintaining that system, and the unique expertise this has given it, the BEPS Action Plan (OECD, 2013a) aims to make progress on 15 areas by the latter part of 2015. This is an unprecedented effort to address major avoidance opportunities that arise under current international tax arrangements, and one that is already having some effect on national tax policies and, perhaps, the behavior of MNEs. The European Commission is also active on a range of issues bearing on tax spillovers in the context of the single market, through its Code of Conduct on harmful tax practices and proposals on corporate taxation (discussed later), and having, for example, recently issued recommendations on aggressive tax planning.3

3. Drawing on the IMF’s TA expertise, macroeconomic perspective and engagement with its near-universal membership, this paper complements current initiatives4 by identifying and analyzing macro-relevant spillovers arising through international aspects of corporate taxation. It does this not only for the current, but also for possibly reformed, international corporate tax arrangements. A core task of the paper is to begin unpacking these ‘spillovers’ (by which is meant the impact that one country’s international tax practices has on others), both conceptually and empirically, and to assess how they are shaped by actual and potential policy decisions. Two broad and related sets of questions run through the discussion.

4. One set of questions concerns whether international corporate tax spillovers matter for macroeconomic performance. For capital movements, at least, it seems clear that they do. Table 1, showing characteristics of the ten countries with the highest FDI stocks relative to GDP,5 suggests that patterns of FDI are impossible to understand without reference to tax considerations (though these of course are not the only explanation). And the point is significant not only for some individual countries (accounting for a stock of FDI extremely high relative to their GDP) but globally (with relatively small countries accounting for a very large share of global FDI). The potential economic implications of international tax spillovers thus go well beyond tax revenue, with wider implications for the broader level and distribution of welfare across nations. The paper explores these and other dimensions of spillovers.

Table 1.

FDI stocks relative to GDP—The Top Ten (2012)

article image
Source: Calculations from IMF Coordinated Direct Investment Survey (http://cdis.imf.org/).Note: Figure shown is average of outward and inward positions. Singapore and Mongolia (for which outward data are unavailable) would enter the top ten if only inward positions were considered.

5. A second set of questions concerns the impact on developing countries. Since the early 1980s, the stock of inward FDI in developing countries relative to their GDP has roughly tripled, to about 30 percent—making its tax treatment increasingly germane to these countries’ wider fiscal performance. The amounts at stake in a single tax planning case now quite routinely run into tens or hundreds of millions of dollars. These sums may be small relative to total tax revenue in sizable advanced economies, but are large for the developing countries that are increasingly involved in such cases—especially, but not only, in the extractive industries.6 And substantial revenue effects can of course arise from the accumulation of individually modest ones. Moreover, the pervasiveness of the tax incentives that the IMF and others have long stressed as significantly undermining revenue in developing countries7 may to a large extent be a spillover reaction to policies pursued in other countries: a clear instance of tax competition. As a share of all revenue, the corporate income tax (CIT) is actually more important in low and upper middle income countries than in advanced (Figure 1),8 so that their overall fiscal performance is more vulnerable to pressures on these receipts.9 Mindful of these risks, the paper pays particular attention to both the quantification of spillovers on developing countries and specific aspects of current international tax arrangements that IMF TA and surveillance have found to be of particular concern to them.

Figure 1.
Figure 1.

Revenue from the Corporate Income Tax in Percent of Total Revenue

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

Source: IMF staff estimates.Notes: Total tax revenue excluding social contributions (data on which are incomplete); resource-rich counties are excluded due to divergences in the reporting of taxes on natural resources. Figure shows medians with countries ranked by income per capita each year and divided into four equal-size groups.

6. The plan of the paper—which has benefited from extensive consultation10—is as follows. The next section provides background on current international tax arrangements and the nature of associated spillovers. This is followed by an assessment of the quantitative importance and welfare effects of these spillovers, after which the paper takes up selected issues of particular importance to developing countries. The final section considers a variety of proposals, beyond those considered in the current OECD BEPS project, that have attracted interest from government officials, academics and civil society; and assesses the case for and obstacles to enhanced coordination in international tax matters. Each section begins with a short summary of its main messages.

Considering International Corporate Taxation

Key Messages

  • Current international tax arrangements rest on concepts of companies’ ‘residence’ and the ‘source’ of their income, both of which globalization has made increasingly fragile (some would say meaningless).

  • The appropriate balance of taxing rights between residence and source countries, long contentious, has lately become an even more focal question. Traditional tax arrangements that are felt to result in unfair allocation of the tax base may increasingly—if not addressed—give rise to unilateral actions that further undermine the coherence of the international system.

  • There is a distinction between worldwide tax systems, under which a country taxes the income of companies resident there wherever that income arises, but gives a credit for taxes paid abroad, and territorial systems, under which the residence country exempts business income of resident companies arising abroad—but, in practice, there is a spectrum between these extremes.

  • MNEs have many devices—often highly complex, interlocking, and very effective—by which to reduce their total tax bills.

  • The paper focuses on two broad types of fiscal externalities, or ‘spillovers’: ‘base spillovers,’ by which one country’s actions directly affect others’ CIT bases; and ‘strategic spillovers,’ by which they induce changes in other countries’ tax policies.

  • Such spillovers potentially give rise to a collective loss of revenue and welfare—but not all countries are necessarily losers.

A. Current Practice and Key Concepts

7. The present international corporate tax framework is defined by the interplay of domestic laws and tax treaty obligations. There is no comprehensive architecture comparable, for example, to that which regulates international trade. Instead, current arrangements that define and divide the international corporate tax base have evolved over the last century or so with little explicit coordination (other than through bilateral treaties that touch only a subset of relevant matters).11

8. Taxing rights over business profits are based on identifying, on the one hand, the ‘source’ of profits…Source refers—very loosely—to where investment is made and production takes place, and is traditionally determined largely by the physical presence of labor and/or capital. Certain thresholds of contact (proxies for the creation of value)—said to create a permanent establishment (PE)—must be met for a foreign company to become liable to pay tax to the country deemed the ‘source’ of profit. Importantly, the location of ‘sales’ (in the sense of the country into which the goods or services produced are sold) is not, under this long-standing architecture, taken to give rise to a place of ‘source,’ nor, thus, as triggering any liability for income taxation. Under territorial taxation (more often referred to in Europe as the exemption method), tax on business profits is levied only in the source country.

9. …and on the other, on identifying the ‘residence’ of corporate taxpayers, since the right to tax profits is under the traditional architecture retained legally by the residence country unless it is given up by that country through domestic or treaty rules (which establish, for instance, what will be considered to constitute a PE abroad by that country). Residence means the place where the company receiving the income is deemed to have its primary location, with common tests for this being where the company is incorporated (applicable for example in the U.S.) or from where it is effectively managed (in most countries). Assertion in domestic law of the right to tax profits from any geographic source based on a company’s domestic residence is generally referred to as worldwide taxation. The double taxation12—taxation, that is, by both source and residence countries—that would then otherwise arise is typically avoided by the residence country granting a foreign tax credit13 against its own tax on the same profits taxed by the source country. This is done either or both in domestic law and/or in applicable bilateral treaties. The result is that the residence tax is limited to the excess of the residence country’s effective tax rate over that in the source country.

10. In practice, neither worldwide nor territorial taxation is found in pure form. Importantly:

  • Worldwide’ basis countries generally provide for deferral of tax on active business profits earned elsewhere, not levying tax until earnings are repatriated to the residence country which brings the system in effect closer to source taxation for active business income (by which is meant, broadly, income arising directly from some commercial activity entered into directly or controlled by the company in question).

  • Controlled Foreign Corporation (CFC) rules play a prominent role. These vary significantly across countries, but are in essence (complex) provisions that bring immediately into tax passive income—the complement of active business income—arising abroad that has not paid tax there of at least some minimum amount. For worldwide countries, CFC rules in principle provide some protection against tax avoidance through deferral; for territorial countries, they typically simply ensure that only active—not passive—income is exempt in the residence country. CFC rules make the distinction between passive and active income arguably the most critical one for modern tax planning, since they normally apply only to the former.

There is thus a spectrum between worldwide and territorial systems, movements along which—as have recently occurred in some large countries—can have major effects on the tax bases of other countries. This trend is taken up later.

11. Identifying the country that is the source’ of income—critical for applying the current architecture—is increasingly problematic. It has been made more difficult, conceptually and practically, by the increased importance of intra-firm transactions (about 42 percent of the value of U.S. goods trade in 2012,14 for instance) and, more recently, of intangible assets of various kinds-patents, trademarks, other intellectual property (IP)—which can be much more easily relocated than can the bricks-and-mortar facilities of the world for which the current framework was initially built. The digitalization of economic activity also raises issues for which the present system was not designed (discussed in OECD (2014d); see also Appendix II). Nor is the notion of residence entirely clear cut, in the sense that—however defined—companies can change it; and, moreover, the increasing disconnect between a company’s country of residence and that of its shareholders makes even the relevance of the concept less clear. The main issues of this kind, and the planning devices to which they lead—of which Box 1 gives a flavor—are the primary focus of the OECD BEPS project.

12. Once the source of profits and residence of companies have been determined, fundamental issues arise in determining the appropriate balance between source and residence taxation. At its core, a key issue in assessing any international tax arrangement is how it divides the rights to tax between source and residence countries. The existing system worked fairly smoothly when capital movements were essentially a matter of relatively balanced bilateral flows among industrialized countries: ex ante, any country could be assumed to be sometimes ‘source’ and sometimes ‘residence.’ The allocation of rights is especially important for low-income countries, however, as flows are for them commonly very asymmetric—they are essentially ‘source’ countries, the recipients of capital inflows and the site of production, not investors in business activities outside their borders. Since the current architecture allocates taxing rights to the source country through rules regarding PEs, for example, it might seem to favor them. Looking deeper, however, the network of bilateral double taxation treaties based on the OECD model18 significantly constrain the source country’s rights (as discussed later). And, determining ‘source’ itself relies on allocating earnings to particular entities within corporate groups. The core allocation rule for this purpose is the arm’s length principle of valuing transactions within MNEs at the prices that would be agreed by unrelated parties—which leaves considerable scope for manipulation by MNEs to shift their tax base away from high tax (often source) countries.19 At issue here are deeper notions as to the ‘fair’ international allocation of tax revenue and powers across countries (which current initiatives do not address). Arrangements that seem to contradict broad perceptions of fairness, even if those are imperfectly articulated, may increasingly give rise to unilateral domestic measures to change them— with a consequent risk of uncoordinated defensive measures even further undermining the coherence of the international tax system.

International Tax Planning—Tools of the Trade

The essential aim of tax planning schemes is of course to shift taxable income to low tax jurisdictions. Precisely how this is done is driven by specific features of national tax systems and treaty networks,15 but common strategies (some of which are looked at more closely below) include:

  • Abusive transfer pricing (stretching, violating or exploiting weaknesses in the arm’s length principle) is often raised as a concern, ranging from potential mispricing of natural resources to the transfer of IP rights to low tax jurisdictions early in their development, when they are hard to value verifiably.

  • Taking deductions in high-tax countries…—by, for example, borrowing there to lend to affiliates in lower-tax jurisdictions;

  • …and as many times as possible—passing on funds raised by loans through conduit companies (ones, that is, serving solely as intermediaries within a corporate group) may enable double dipping—taking interest deductions twice (or more), without offsetting tax on receipts—leading to thin capitalization (high debt ratios).

  • Risk transfer—conducting operations in high tax jurisdictions on a contractual basis, so limiting the profits that arise there.

  • Exploiting mismatches—tax arbitrage opportunities can arise if different countries classify the same entity, transaction, or financial instrument differently (the U.S. ‘check the box’ rules16 being a prime example of a provision with major unintended consequences in this regard);

  • Treaty shopping—treaty networks can be exploited to route income so as to reduce taxes (discussed further below);

  • Locating asset sales in low jurisdictions—to avoid capital gains taxes (a particular concern in the context of recent resource discoveries in some low tax countries, and also explored further later).

  • Deferral—companies resident in countries operating worldwide systems can defer home taxation of business income earned abroad by delaying paying it to the parent.

  • ‘Inversion’—companies may be able to escape repatriation charges or CFC rules by changing their residence.

Schemes commonly combine several of these devices, turn on quite fine legal distinctions and span several countries and tax systems. They are often extraordinarily complex.17

B. Spillovers

13. By spillover’ is meant here the impact that one jurisdiction’s20 tax rules or practices has on others. ‘Fiscal externalities’ of this kind can arise from many aspects of national tax systems, including purely domestic ones: changes in the excise taxes on the consumption of particular commodities, for instance, can have powerful effects on foreign producers of those commodities. The focus here, however, is on cross-border effects arising through taxation at the corporate level. This perspective sets aside important issues regarding taxation at personal level of the income earned by multinationals, 21 but captures the core concern with the level and cross-country allocation of corporate tax payments.

14. Spillovers can have macro-relevant effects through several channels. One country’s international tax decisions may affect other countries by their impact on:

  • Real and financial flows, with potential impacts, through FDI and corporate financing arrangements, on growth and macroeconomic stability.

  • The corporate tax base, as taxable profits change in reflection of both real responses (through investment and the like) and profit-shifting responses (affecting, loosely speaking, only where profits are booked for tax purposes). This channel of effect, holding constant the tax policy of the affected country, is referred to here as base spillover.

  • Tax-setting incentives, since the best response to tax changes abroad may be to change national tax rules too. These are strategic spillovers—tax competition’ in its broadest sense-most obviously in the potential form of a ‘race to the bottom,’ as countries respond to lower CIT rates elsewhere by reducing their own rates.22 The focus below will be on direct strategic spillovers of this kind—changes in corporate taxation at home in response to changes abroad— but there can also be important indirect strategic effects. In particular, pressures to reduce the CIT rate create pressures to reduce the top personal income tax rate too, to prevent avoidance of the latter through incorporation.

  • World prices, as significant tax-induced changes in investment and saving behavior may generate, for instance, changes in world interest rates and/or, to the extent that corporate taxes ultimately fall on labor, in wages. These are pecuniary externalities (that is, they operate through prices), so will affect global efficiency only to the extent that they interact with other distortions: a tax-induced reduction in the world interest rate, for instance, could amplify the adverse effect of taxes that reduce private incentives to save.

15. These effects are closely related, Decisions on the location of real activities, for instance, may be influenced by the associated opportunities for profit shifting; and awareness of an ability to affect world prices can create its own tax-setting incentives (analogous to the familiar optimal tariff).23 There are other potentially important spillover effects too: on, for instance, companies’ choice of organizational form. The focus here, however, is on the first three spillovers above—and especially on base and strategic spillovers, these being the most directly fiscal.

16. Spillovers can arise from several aspects of national tax policies, and reflect complex interactions between them. Most obviously, spillovers can arise from differences in ‘headline’ statutory rates of corporate taxation, since these create incentives to shift taxable profits between countries. So too do more narrowly-defined preferential regimes that offer special treatment for particular types of income—a topical example being the ‘IP boxes’ that charge reduced rates of tax on income from patents and other forms of intellectual property.24 But spillovers can arise from other than these core parameters of the corporate tax. For example:

  • Network externalities can arise within the system of double tax treaties: if country A, having a treaty with country B, signs a treaty with country C, it may in effect create a treaty between B and C (see Box 3 and related discussion below).

  • Mismatches in national rules can also create spillovers. These can arise, for instance, in terms of how financial instruments are treated: one country may regard an instrument as debt, for example, and so allow an interest deduction, while the country in which payments are received regards it as equity, and so imposes no or little tax. Or the mismatch may be in terms of the characterization of business entities (as corporations or otherwise).

  • Tighter CFC rules, raising the effective tax rate on funds placed in low tax jurisdictions, can indirectly affect other countries by raising the tax cost of investing there.

17. The form and severity of spillovers depend on the structure of the international tax framework. The rate of tax levied in source countries would not matter for corporate decisions, for instance, under a pure system of worldwide taxation in which such income was fully subject to tax, without deferral, by the residence country: the company would then simply pay the residence country rate on all its earnings, wherever they arose. Addressing spillover problems thus inevitably raises issues concerning not just particular tax arbitrage opportunities under current arrangements, but the wider architecture itself.

18. The effects of spillovers are not zero sum, but can create a collective inefficiency—and national interests can diverge sharply. This is clearest in relation to profit shifting: moving taxable income from a high tax jurisdiction to a low tax one so as to reduce total tax payments. Since the company’s purpose in doing so is to reduce its total tax payments, the collective revenue of the countries affected must fall. But revenue in the low tax country can only increase. And similar effects arise more broadly than from such straightforward income shifting techniques. Non-cooperative policy making in the presence of externalities can result in outcomes that, from the collective perspective, are inefficient but from which some countries or jurisdictions nonetheless gain. In a very much second best world it may be, however, that non-cooperative tax policymaking eases other distortions. This complex set of issues is taken up in the final section.

Assessing the Spillovers

Key Messages

  • There is substantial evidence that tax considerations significantly affect FDI and a wide range of other corporate decisions—but estimating aggregate revenue effects remains elusive.

  • New evidence for a large panel of countries assessed here suggests that: • Both base and strategic spillovers are significant and large.

    • Base spillovers arise at least as much from profit shifting as from effects on real activities.

    • These spillover effects are especially strong for developing countries.

  • Assessing the welfare effects of tax spillovers remains problematic:

    • While spillovers are presumptively a source of inefficiency, low tax rates on the most mobile activities ease economic distortions from the corporate tax.

    • A small observed collective revenue loss associated with spillovers (from profit shifting, for instance) does not imply a small efficiency loss, as the overall level of taxation could be too low.

A. Quantification

This section assesses evidence on the nature and importance of the first three types of spillovers identified above.

Real and Financial Flows

19. Aggregate international investment positions and behavior are strongly marked by tax considerations. Tables 2 and 3 show for, respectively, selected G20 and developing countries, where their outward investment immediately goes to and /or their inward investment immediately comes from: 16 percent of outward investment from Brazil, for instance, goes (as least initially) to the Cayman Islands. Such lists—and the countries selected are not in this respect atypical—confirm the impression from Table 1 that taxation plays a key role in shaping the structure of international capital flows: jurisdictions known for attractive tax regimes and extensive treaty networks commonly feature prominently as ‘conduits’ through which investments pass.25 There is some direct evidence too that tax changes can have significant capital account effects.26 Reviewing a large econometric literature on investment effects, the meta analysis of De Mooij and Ederveen (2008) suggests that a 10 percentage-point reduction in a country’s effective average tax rate27 increases its stock of FDI, on average and in the long run, by over 30 percent. Not all of this FDI, it should be noted, represents the development of new productive capacity (so-called ‘greenfield’ investment). Estimates suggest that more than half may reflect mergers and acquisitions. There are some indications though that greenfield investments –which, are presumably more likely to generate the benefits for which inward investment is often sought, such as additions to employment—are a larger proportion of total FDI in lower income countries than in more advanced economies.28

Table 2.

Top 10 Destinations and Sources of FDI, Selected G20 Countries, 2012/1

article image
Source: IMF Coordinated direct investment survey (http://cdis.imf.org).

Main destinations/sources of FDI from/to country in bold; percent of total in parentheses.

Table 3.

Top 5 Sources of Inward FDI, Selected Non-OECD Countries, 2012/1

article image
Source: Coordinated direct investment survey (http://cdis.imf.org).

Percent of total in parentheses.

Base Effects

20. One country’s tax policy can impact others’ CIT bases by affecting either real activities or the shifting of paper profits. Effects on real investment of the kind just noted would be expected to feed through to corporate tax bases.29 Pure (paper) profit shifting might be regarded in a quite different and (even) more negative light, being seen as undermining the traditional notion that profits be taxed in geographic line with the activities that generate them (though, as stressed above, quite where that location is can be very unclear).

21. There is strong evidence of extensive profit shifting… It is highly suggestive, for instance, that more than 42 percent of the net income earned by U.S. majority-owned affiliates is earned in ‘tax havens’,30 while less than 15 percent of their value added is created there;31 and that the presence of an additional ‘tax haven’ subsidiary reduces the consolidated tax liability of a corporate group by 7.4 percent of total assets.32 And more granular evidence increasingly confirms the longstanding but largely anecdotal impression that the opportunities for profit shifting indicated in the previous section are indeed used extensively (Box 2).

22. …but it is very difficult to arrive at ‘bottom line’ numbers for the overall sums at stake. Many attempts have been made to arrive at a single estimate of the extent of profit shifting. For the U.S., one estimate points to a revenue loss of $60 billion—about 25 percent (at that time) of CIT revenue.33 For non-advanced countries, Christian Aid (2008) arrives at an aggregate annual loss from trade mispricing alone of USD 160 billion. Estimates such as the latter, however, have been forcefully criticized.34 In assessing these and other estimates it is also important to remember that they often look at only one side of the story: one country’s revenue loss may be offset, though only partly, by other countries’ revenue gains.

Responses to International Tax Rules—Evidence

There is a large and growing literature, exploiting firm level data (available in usable quantities only for advanced economies) tending to confirm that the planning opportunities outlined in Box 1 do have significant effects on corporate behavior:

  • Transfer pricing abuse. Direct empirical evidence is scarce, and does not unequivocally point to large effects. Clausing (2003) finds signs of significant tax-motivated transfer pricing abuse of intracompany trades by U.S. multinationals; and Heckemeyer and Overesch (2013) attribute about two-thirds of their consensus spillover effect to transfer pricing abuse. Swenson (2001), on the other hand, reports responses of transfer prices with respect to cross-country differences in tax rates that are very small. It seems likely that the potential for abusive transfer pricing in advanced countries occurs not so much for trade in tangible goods—as it may for developing countries, which often lack appropriate information on comparable prices even for these transactions—as for transactions for which even advanced countries may lack comparables, such as intangibles, risk premia or management services.

  • Location of intangible assets. CIT rates appear to have large negative effects on the number of patents filed by a subsidiary (Karkinsky and Riedel, 2012) and on the magnitude of intangible assets reported on a company’s balance sheet (Dischinger and Riedel, 2011). This is consistent with profit shifting and indeed there is evidence that profit shifting activities are larger in MNEs with high IP holdings and R&D intensities (Grubert, 2003).

  • Intra-company debt shifting. There is substantial evidence that taxation induces intracompany borrowing to reduce tax payments in high-tax locations (De Mooij, 2011). Effects are larger for affiliates located in developing economies than for those in developed economies (Fuest and others, 2011); and are found to be important also for multinational banks (Gu and others, 2014).

  • Mismatches and other devices. For the U.S., Altshuler and Gruber (2008) find that ‘check-the-box’ rules in the U.S. (by which such mismatches can be exploited) created a revenue loss for the U.S. treasury of $7 billion between 1997 and 2002.

  • Treaty shopping. Tables 2 and 3 above, and similar, provide quite compelling prima facie evidence for extensive use of conduit countries.35 Using firm-level data, Mintz and Weichenrieder (2010) find strong effects for German MNEs, while Weyzig (2014) documents a significant impact of Dutch Special Purpose entities on the routing of FDI.

  • Inversion. Between 1997 and 2007 about 6 percent of all MNEs relocated their headquarters. Voget (2011) finds that a 10 percentage point higher tax on repatriations increases the probability of such relocation by more than one third. Huizinga and Voget (2009), moreover, estimate that if the U.S. were to eliminate worldwide taxation, the number of parent companies that would choose residence in the U.S. after a cross-border merger would increase by 5 percentage points.

  • Deferral. When the U.S. tax rate on repatriated dividends was reduced from 35 percent to 5.25 percent for one year in 2005, corporations repatriated $312 billion, much of which was distributed as dividends to U.S. shareholders (Marples and Gravelle, 2011; Dharmapala and others, 2011). Studies by the Joint Committee on Taxation and the U.S. Treasury estimate that eliminating deferral would yield an annual revenue gain in the U.S. of between $11 and $14 billion (Gravelle, 2013), allowing a revenue-neutral reduction of the CIT rate to around 28 percent (Altshuler and Grubert, 2008).

23. Panel data analysis points to marked base spillover effects, from both real activities and profit shifting. Much of the literature has focused on how a country’s tax system affects its own tax base (to identify, for instance, revenue-maximizing tax rates). The spillovers that are the focus here, however, have received much less attention. Appendix III develops an empirical approach for assessing base spillovers, relating (proxies of) the corporate tax base in 103 countries, over the period 1980–2013, to both their own statutory CIT rate and the rates of others.36 This approach is in important respects simplistic: many jurisdictions are attractive for tax purposes not because their statutory rate is especially low, for instance, but because of special regimes not captured in this data. The wide country and time coverage, however, provides some offsetting analytical benefits. Several suggestive conclusions emerge:37

  • Spillover base effects through real activities are significant and large. Theory suggests that the base spillover effects felt by any particular country will reflect a weighted average of tax rate changes elsewhere, but with weights that differ depending on the channel of the spillover. If it is only through real activities, then tax rates set by (broadly speaking) larger countries should matter more: intuitively, this is because one would in this case expect that the tax base of a large advanced economy would not be much affected by the statutory rate of a small island. Weighting other countries’ tax rates by GDP to capture this idea, the (short run) semi-elasticity of the implicit corporate tax base38 with respect to statutory CIT rates abroad is around 3.7; that is, a one point reduction in the statutory CIT rate in all other countries reduces the typical country’s corporate tax base by 3.7 percent. With corporate tax rates having fallen, on average, by 5 points or so over the last 10 years, this implies a sizable effect.

  • Spillover base effects through profit shifting are also large—and no less significant. If, on the other hand, effects are through shifting of paper profits, it is simply cross-country differences in statutory tax rates that matter, together with the ease with which they can be exploited: in this case, the tax policy of a small island could well matter for a large economy. To explore this, not having any direct measure of the ease of profit shifting, two weighting structures are considered in the empirical exercise: the simple average of all tax rates elsewhere, and the average only across ‘haven’ jurisdictions. These ‘haven-weighted’ effects prove to be marked; indeed spillovers from this group imply as large an effect as is found using GDP weights—and one estimated with greater confidence.

24. There is also evidence that tax base spillovers are especially pronounced for low-income countries.39 IMF TA has encountered several instances in which single cases account for a significant part of all revenue. Many of the most spectacular relate to the extractive industries: a gold mining sector in which USD 100 billion has been invested over the last decade, but which is almost entirely debt financed; a potential loss from effective elimination of withholding taxes on a single project equivalent to around 15 percent of total revenue;40 and further, especially striking examples related to indirect transfers of ownership interests (Appendix VI). But the issues are not confined to the extractive industries: there are cases, for example, of telecom companies in Africa— often among the most profitable enterprises—being almost entirely debt financed. Civil society has also drawn attention to cases in which substantial sums are at stake for developing countries in the proper application of international tax rules.41 Less anecdotally, the econometric evidence reported in Appendix III points to particular macro-relevance of tax spillovers for these countries, suggesting that:

  • The spillover base effect is largest for developing countries. Compared to OECD countries, the base spillovers from others’ tax rates are two to three times larger, and statistically more significant.

  • The apparent revenue loss from spillovers, relative to a benchmark akin to source taxation, is also largest for developing countries. A separate and very preliminary exercise (outlined in Appendix IV) derives estimates of ‘apparent spillovers’ by comparing actual CIT revenue with that which would be collected were the base simply proportional to gross operating surplus (GOS) —a very crude approximation to a form of source-based taxation. Data limitations mean that this can be done only for a relatively small sample of countries,42 and one that excludes, moreover, the conduit countries that the evidence above suggested to be an especially important source of spillovers. So the exercise cannot capture profit shifting involving such countries. There are other caveats too: as explained in the appendix, apparent gains from spillovers will be higher, for instance, in countries with relatively few exemptions and for those implementing worldwide taxation. For all these reasons, country-specific estimates can be highly misleading. The broad impression, nonetheless, is instructive. And what is notable is that the spillover impacts which emerge from this exercise are more than twice as large in non-OECD as in OECD countries. These losses are in many cases substantial: more than 50 percent of current CIT revenue in several cases. The (unweighted) average revenue loss across all countries in the sample is about 5 percent of current CIT revenue—but almost 13 percent in the non-OECD countries. While there is considerable scope to refine the methodology, the results are, nonetheless, suggestive of broad patterns, and broadly consistent with the direct econometric analysis of base spillovers above.

Strategic Spillovers

25. There is increasingly strong evidence of strategic interactions in tax setting… The dramatic worldwide decline in statutory CIT rates over the last three decades is well-known: most significant in Europe and Central Asia, somewhat less in Sub-Saharan Africa and Latin America—and perhaps now leveling off (Figure 2). The recent spread of IP boxes, noted earlier, is also highly suggestive of strong strategic spillovers, which one would indeed expect to be most marked for the most mobile elements of tax base. And the econometric evidence increasingly confirms that, as policymakers’ rhetoric often suggests, this reflects deliberate competition. For OECD countries, Devereux and others (2008) find that a one percentage point decrease in the statutory CIT rates of others generates, on average, a cut of 0.7 percentage points in response. For developing countries too, there is evidence that their setting of incentives responds to the incentives available in neighboring countries; and that a race to the bottom has become evident among special regimes— most notably in Africa where tax burdens under these regimes have fallen to almost zero.43

Figure 2.
Figure 2.

Corporate Income Tax Rates, 1980–2013

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

Source: IMF staff estimates.Note: Figure shows medians with countries ranked by income per capita each year and divided into four equal sized groups.

26. …including for the panel data used here. Appendix III finds a significant strategic tax response of statutory CIT rates (using either GDP or ‘haven’ weights’) of around unity. Results on strategic responses in terms of effective corporate tax rates, which take account not only of statutory tax rates but also depreciation allowances and other considerations that will matter for real but not for paper transactions, tend on the other hand to find insignificant strategic effects.44 This greater sensitivity to statutory than to effective tax rates thus suggests—tentatively, as for most conclusions in this area—that tax competition is driven at least as much by profit-shifting concerns, including in relation to ‘havens’, as by the desire to attract real investments.

27. Strategic responses diminish the direct revenue effect of base spillovers, but do not eliminate it. The results above imply, for instance, that, holding its own tax rate constant, a one percentage point reduction in statutory rates in all other countries—again, a relatively small cut in terms of historical experience—reduces the typical country’s CIT base by (in the long run) about 6.5 percent (Table 4). But they also imply that the country will not hold its rate constant, but rather cut it by, on average, 0.5 points. This increases its CIT base by 4 percent, leaving a net base loss of 2.5 percent. And the proportional revenue loss will be larger than the proportional base reduction, of course, since the rate cut reduces revenue across the entire base.45

Table 4.

Simulated Impact on a Country’s CIT Base of a Change in Foreign CIT Rates

article image
Source: Staff calculations using estimates reported in Appendix III.

B. Welfare Implications

28. If there are no other distortions, spillovers are presumptively sources of collective inefficiency—with the revenue impact itself, however, saying little about their scale. A fundamental difficulty in assessing the potential inefficiencies which, as noted above, spillovers can create, is that the proximate revenue losses (whether collective or national) are not directly informative of the associated welfare loss. If all tax rates were competed to zero, for instance, there would be no profit-shifting or revenue loss actually observed, but there would be a collective loss from inefficiently low taxation nonetheless. At best, measured revenue losses provide a lower bound indication of the inefficiencies involved.

29. The effects are potentially very different in different countries. Theory suggests,46 for example, that small countries can be winners in tax competition games—because they have relatively little to lose from reducing taxation of their small domestic bases, but much to gain from attracting large bases from abroad.

30. Interactions with other distortions may be important. Two are prominent:

  • Some see tax competition as welcome in counteracting a political bias towards excessive public expenditure.47 Such arguments have become less prominent, however, with the needs for fiscal consolidation now felt in many countries; and perhaps too with recognition that fiscal rules may better address such concerns.

  • Spillovers may also ease distortions from the corporate tax itself, by, for instance, reducing marginal effective tax rates on investment. More subtly, to the extent that avoidance opportunities are more easily exploited by—perhaps even targeted by policymakers to—firms whose assets or activities are more mobile internationally, this can result in a more efficient tax system (consistent with standard tax policy prescriptions) by taxing more elastic tax bases at lower rates. In this respect, low tax and conduit countries can improve the efficiency with which capital is allocated.48

These considerations generate heated dispute, but remain essentially uninformed by empirical knowledge.

31. Tax planning through intra-group borrowing amplifies (unconsolidated) leverage, but may pose few financial stability risks. Such borrowing affects the apparent allocation of risk within the group, but, implicitly if not explicitly, risk is likely borne at the group level, so that lending to affiliates is akin—in all but tax terms—to an equity investment.

32. The potential significance of spillover effects may warrant discussion in Article IV consultations, consistent with the integrated surveillance decision.49 Related, in a report to the Development Working Group of the G20, jointly with the OECD, UN and World Bank, staff have suggested that it would be appropriate for G20 members to assess any spillover effects on developing countries of major tax reforms proposed.50

Selected Key Issues for Developing Countries

Key Messages

  • ‘Treaty shopping’—the use of tax treaty networks to reduce tax payments—is a major issue for many developing countries, which would be well-advised to sign treaties only with considerable caution.

  • Many developing countries also need to be better protected against the avoidance of tax on capital gains on natural resources and some other assets in their jurisdiction by the realization of these gains (i.e., the transfer of ownership) in low tax jurisdictions.

  • Many developing countries also still lack effective provisions to guard against the use of borrowing to shift profits to lower tax jurisdictions.

  • Addressing the challenges of transfer pricing requires not just capacity building, but clearer and appropriately simplified rules and guidance.

  • Important though it is for developing countries to cope better with the challenges of international taxation, this should not distract from wider and more fundamental tax reform objectives.

33. Addressing avoidance by MNEs can discourage investment—but the revenue concern is often paramount in developing countries. To the extent that tax avoidance reduces firms’ cost of capital,51 it tends to make investment more attractive—and, conversely, measures to reduce avoidance opportunities make it less so. This is a real concern: as seen above, taxes can indeed matter for inward FDI. But other factors—infrastructure and labor costs—often matter more. And the impact on investment in any particular country will be less the wider is the set of countries participating in adopted anti-avoidance measures. With that in mind, this section explores some of the issues that IMF TA and other experience suggest are of particular importance for developing countries.

A. Tax Treaties

34. Bilateral tax treaties (BTTs) make more certain, and set limits to, the taxation of cross-border investments. They set out the allocation of taxing rights between source and residence countries. To this end one key set of provisions specifies maximum rates of withholding tax (WHT) on interest, dividends, royalties and other payments from source countries—generally below those otherwise applicable by domestic law. As with the CIT itself, rates of WHT have been trending down, in both treaties and domestic law (Appendix V) but remain a significant concern for investors. BTTs also generally provide: agreed PE definitions; specification of taxes for which foreign tax credits will be provided; 52 non-discrimination rules; and dispute resolution procedures.53 BTTs may also include provisions for exchange of taxpayer information (EOI), though this may also be provided for in stand-alone tax information exchange agreements (TIEAs) or by acceding to the 1988 Council of Europe/OECD Convention of Mutual Administrative Assistance in Tax Matters. Treaties differ, but— building on a framework first established by the League of Nations—are strongly guided by models developed by the OECD and UN, along with associated commentaries.

35. There has been a proliferation of BTTs over the last twenty years—driven by an increasing number that involve developing countries (Figure 3). Initially, almost all BTTs were between advanced economies. The tripling of the number of treaties since the early 1990s, however, almost entirely reflects an increase in the number to which at least one party is a non-OECD country—many of which, of course, will have few capital exports. Overall, there are now around 3,000 BTTs: a large number, but only a small fraction of the number of potential bilateral relationships. Figure 3 also shows the dramatic increase in the number of TIEAs over the last few years, reflecting the renewed focus on EOI following the 2009 Pittsburgh Summit and the work of the Global Forum on Transparency and Information Exchange.

Figure 3.
Figure 3.

Numbers of BTTs and TIEAs, 1975–2013

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

Source: International Bureau of Fiscal Documentation database.

36. Whether a capital importing country benefits from signing a BTT depends on whether it realizes sufficient gains from increased FDI to offset any revenue loss. The primary benefit that developing countries seek from signing a BTT is increased inward investment—an important policy objective for many countries, given the perception of significant associated external benefits—as a result of both increased certainty and lighter taxation. Against this, however, by committing to reduced WHT rates countries that are primarily capital importers forego revenue (at any given level of inward investment); and in the process they also incur far from trivial costs of treaty negotiation—which is heavy on the time of skilled staff.54 (The EOI aspects of BTTs, however, run in the other direction: they may increase source country revenue at unchanged behavior and, by the same token, could discourage inward investment).

37. The empirical evidence on the investment effects of treaties is mixed….(Appendix V). Identifying causality is inherently problematic, since treaties may precede investment not because they spur the latter but because they may be concluded only when there is an expectation of such investment. (This can be a deliberate feature of treaty policy, as it traditionally has been in the U.S.). Studies using macro-level data indeed find a wide range of effects, though perhaps with some signs that a positive effect on FDI is most likely for middle-income countries. Work using firm-level data finds a significant impact on firms’ entry into a particular country, though not on the level of their investment once they are present.

38. …while the potential revenue loss, especially to developing countries—including through ‘treaty shopping’—has caused increasing concern. With a treaty in place, the MNE’s incentive is to extract income in forms that attract a low or zero WHT rate, which may be ones— management fees, for instance, or royalties—that the host authorities find particularly difficult to value. The opportunities for this are amplified by the possibility of treaty shopping: constructing advantageous routing by linking bilateral tax treaties, typically through low tax conduit countries.55 (Box 3). In effect, a treaty with one country can become a treaty with the rest of the world, and the network externalities noted earlier arise. Examples of base erosion arising from treaty structures have frequently arisen in IMF TA, as with the 2004 Mongolia-Netherlands treaty (IMF, 2012b). And the effects can be very sizable. One estimate, for instance, is that treaties with the Netherlands led to foregone revenue for developing countries of at least EUR 770 million in 2011;56 similar, very rough, calculations suggest that U.S. tax treaties cost their non-OECD country counterparts perhaps $1.6 billion in 2010.57 Treaties are rarely cancelled, particularly by developing countries, so it is indicative of the level of current concerns—and the wider pressures on the international tax system—that both Mongolia and Argentina have done so since 2011.

39. ‘Limitation of benefit’ (LOB) provisions can provide important protection… These stipulate that reduced withholding rates and other treaty provisions apply only to companies that meet specific tests of having some genuine presence in the treaty country (such as a minimum share of ownership by its residents or a minimum level of income from conducting an active trade or business there). LOB provisions have not, however, been the norm in treaties other than those of the U.S.58 There are, though, welcome signs that anti-abuse provisions are spreading: India and Japan have been moving in this direction; a recent OECD discussion document59 recommends inclusion of LOB provisions; and the Netherlands has taken the very positive step of indicating an intention to approach all developing countries with which it has (or is negotiating) a tax treaty with a view to strengthening anti-abuse provisions. LOB provisions are, however, often complex and are not self-executing: where capacity is weak and access to information limited, verifying that the pre-requisites for treaty benefits are met can be difficult.

40. …but, more fundamentally, considerable caution is needed in entering into any BTT. A critical decision for any primarily capital-importing country is whether it can achieve more by signing a treaty than it can simply through its own domestic law.60 The reciprocal benefits that a treaty could provide to such a country may actually be of relatively little value, except perhaps for the EOI aspects—but those can in principle be achieved through a TIEA or by signing the Convention on Mutual Administrative Assistance. And key provisions regarding, for instance, WHT rates and the PE definition, can be provided in domestic law. Treaties are, moreover, inherently discriminatory as between partners and others. The main or even only advantage that a BTT can offer may then be one of signaling, acting as a strong commitment device for the tax assurances given to foreign investors. But that in turn may become less needed as countries build up a credibility in tax policy making they may not have had some years back. Some would simply advise developing countries not to sign BTTs, and at a minimum, to include some form of LOB clause if they do, while also providing for LOB in domestic law.61 What is clear is that countries should not enter treaties lightly— all too often this has been done largely as a political gesture—but with close and well-advised attention to the risks that may be created.

Treaty Shopping

Country A has a treaty with B, but not with C, so that residents of C investing directly in A face the relatively high WHT rates specified in A’s domestic law. If C has a treaty with B, however, investors there may be able to benefit from A’s treaty-reduced WHT rates by investing though an affiliate in B. which acts as a conduit. It may, moreover, be that B has a treaty with some country D that taxes such income at a low rate; receipts can then be further passed there, without (unless C applies CFC rules) attracting any further immediate liability in C.

An LOB provision in the A-B treaty would aim to address this by providing the benefits of the A-B treaty only to those undertaking substantial activity in B.

B. Indirect Transfers of Interest

41. The tax treatment of gains on indirect transfers of interests in assets is a controversial issue—and one of particular importance to many developing countries.62 At issue is the possibility that the owners of some asset with respect to which a capital gain arises can avoid tax in the country to which that asset—perhaps a telecom or mineral license—inherently relates, by holding it through a chain of companies and then selling the claim in a low tax jurisdiction (Box 4). Such transactions can involve the host country receiving little or no revenue when substantial gains are realized on assets located there. Over the last few years, including in the context of resource discoveries, this has emerged as a macro-relevant concern in several low income countries (Appendix VI).

Indirect Transfers of Interest

Company A holds an asset located in a relatively high tax source country that increases in value. The ultimate owner of A, company Z, realizes this gain by selling X, an intermediate entity holding A, in a low tax jurisdiction. The result may be that no or little tax is payable anywhere.

42. The laws of many developing countries need strengthening if they are to tax gains on such indirect transfers. Conceptually, there are arguments as to whether or not it is appropriate to tax such gains at all: they presumably reflect accumulated and expected earnings, so it may not be necessary or appropriate to tax them if those earnings have been, or will be, adequately taxed in other ways.63 Nonetheless, most countries do aim to tax such capital gains, and the non-taxation of indirect transfers can then raise serious equity and political concerns. Domestic tax laws need to provide an adequate basis if indirect transfers are to be brought into tax: by ensuring, for instance, that ‘immovable property’ is defined clearly and broadly (including for treaty purposes); and extending taxing rights sufficiently far into multi-tiered corporate structures. Tax treaties can protect and clarify the source country’s rights, but in practice, many existing treaties do not make clear provision on the issue (or define source rights narrowly).64

43. Implementation, however, can be problematic—especially for developing countries. Challenges arise in both discovering the transactions and collecting the tax due. It can be difficult for a tax administration to know about offshore transactions. Possible mechanisms to address this include requiring that relevant authorities be notified of any indirect disposal and the sharing of tax information between countries. And once a gain is identified, it can be difficult to collect the tax from a non-resident: some countries address this by requiring a resident to be liable and/or imposing non-tax penalties (such as withdrawal of a mining license) for non-payment. Appendix VI elaborates on the rules and procedures that can be used to bring indirect transfers into tax.

C. Interest Deductibility

44. Debt shifting through intra-group loans—a common method of profit-shifting—is a significant concern in many developing countries. With interest deductible under the CIT, and low or no withholding taxes, an obvious way to shift profits out of high tax jurisdictions is by lending to them through low tax ones. In itself, this problem is much the same for all countries,65 but its importance for developing countries bears emphasis: examples from IMF TA abound.

45. Restrictions on interest deductibility, which can be relatively straightforward, have considerable potential for addressing avoidance through debt shifting.66 It is reasonable to ask why, in principle, any deduction at all should be given for interest paid to related parties. Certainly, many countries have started to take measures to address the issue. The rules used take different forms: some are based on debt/equity ratios, others on net interest payments; some apply only to intracompany debt, others apply to all debt.67 Restrictions are generally limited to larger companies, and banks are often specially treated. Recently, several countries have adopted comprehensive earnings stripping rules that restrict deductions for interest payments exceeding some specified proportion of a company’s income. These measures are relatively easy to apply, and can be especially attractive for developing countries in protecting their tax base from base erosion. Many countries, however, continue to be made vulnerable by the absence of such provisions.

D. Arm’s Length Pricing

46. The Arm’s Length Principle (ALP)—valuing intra-firm transactions at the prices that unrelated parties would reach68—is at the heart of current international tax arrangements…69 The underlying rationale is that use of ALP to allocate income across members of a corporate group, and hence countries, preserves neutrality between MNEs and independent operations while also defining the tax base on which countries can exercise their primary taxing rights. Following conventional CIT practice, the ALP is self-assessed; taxpayers have to document that their internal transfer prices correspond to arm’s length prices, and tax authorities can challenge this. Close comparability with independent transactions between unrelated parties is crucial for establishing that prices satisfy the ALP. Verifying this requires reviewing the functions performed, assets used and the risks genuinely assumed by the entities within an MNE—all of which depends very heavily on the facts and circumstances of each case.

47. …but is seen by many as having become too complex and permissive, allowing in practice a massive concentration of corporate profits in a few low tax jurisdictions. The empirical evidence on profit shifting through transfer pricing, discussed in the previous section, reaches no consensus on precise magnitudes, but few doubt that the sums at stake in applying the ALP are very considerable. Reflecting this, the ALP has come under increasing pressure—and criticism—both conceptually and in application.

48. Some conceptual criticisms of the ALP are misplaced. The most common theoretical criticism is that, following the classic discussion in Coase (1937), MNEs exist precisely as a more efficient alternative to market transactions, so that market prices cannot provide an appropriate benchmark. But Coase’s is a theory of the size of the firm, a key implication of his argument being that MNEs expand up to the point at which they are no more efficient than the market—in which case market prices are, at the margin, just as relevant for them as for independent parties. In this view, Coase’s argument does not undercut the ALP but, to the contrary, rationalizes it. There may be cases, however, in which MNEs undertake, for non-tax reasons, operations that are very hard to conceive of unrelated parties entering into. They may, for instance, exist to overcome ‘hold up’ problems. MNEs may also enter transactions for which there is no unique price at which one would expect unrelated parties to arrive. Guidelines for applying the ALP implicitly recognize this, by including a methodology—the ‘residual profit split method’ (RPS)—that attributes the total profit associated with a transaction among transacting entities according to their respective contributions to the value created—often in a way that is quite formulaic.70

49. But significant issues of principle arise because intra-group transactions may occur only because of MNEs’ ability to exploit cross-border tax differentials. Prominent examples include:

  • The transfer of intangible assets within a corporate group at an early stage in their development (to low tax jurisdictions, where the subsequent return will then accrue) can raise severe valuation problems given the inherent absence of comparables—and is evidently subject to significant issues of asymmetric information.71

  • Risk transfer among affiliates has become standard practice in MNEs, including for many operations in developing countries. It is achieved by contractual arrangements that, in effect, provide implicitly some (potentially quite complex) degree of insurance between affiliates. Risks can, for instance, be stripped from one (highly taxed) entity and assigned to another (low taxed) affiliate by converting the former into a routine operation with correspondingly low profit margins. The allocation of risk within a group may of course be driven by commercial considerations, and is not in itself problematic. The question for transfer pricing purposes is how to value the transactions by which such risk transfers are achieved. A fundamental issue here is that, since a pure reallocation of risk within the group cannot affect the risk borne by the group as a whole—to the extent, at least, that it is hard for all other group members to step away from the failure of any affiliate—there is inherently no comparable price between unrelated parties on which to draw; and the bundling of such risk transfers with changes in function whose substance can be hard to judge adds further difficulty. Companies may be able to secure reduction in taxation by exploiting these difficulties of pricing.

The current methodological framework is inadequate for fending off such schemes because it asks what independent parties would do in such a situation—but with no comparables to use, as none could exist. The ALP under such circumstances allows re-characterizing transactions so as to ignore for tax purposes those deemed artificial, but this potentially introduces discretion, uncertainty and complexity. 72

50. The practical difficulties in applying the ALP are substantial—as is widely recognized. The burden on companies of justifying, and on the authorities of verifying, that transfer prices used correspond to ALP is widely recognized to be substantial. And they will increase: as businesses become more knowledge-intensive and technology driven, management becomes more geographically diffuse, and intra-group operations become easier to transfer and more difficult to price. This does not mean that other arrangements would be any simpler—but is cause for legitimate concern.

51. Developing countries face particular challenges—with capacity building important…The inherent complexities of applying the ALP are especially challenging where administrative capacity is weak. Strengthening it—including by improving access to information needed73—will be key to progress.

52. …but not all that is needed. The ALP is very sensitive to the facts and circumstances in which controlled transactions take place, and many situations that are more significant to or common in developing countries receive relatively little attention in existing transfer pricing guidance. For example, a large proportion of non-natural resource based multinational businesses located in developing countries are organized as low risk, routine, light manufacturing or commercial ventures, rewarded with accordingly low profit rates. It is common, under the application of transfer pricing methods, to assign these operations a fixed rate of return for tax purposes, under which productivity gains rarely translate themselves into higher local profit margins. A risk in introducing such simplified schemes, despite their attractions for administration, is that they thus may not respond to changing commercial circumstance, and can perpetuate inappropriately low fixed profit rates in developing countries.

53. A specific agenda for developing countries is needed to protect and expand their corporate tax bases in the face of challenges in applying the ALP. This means, first, having appropriate transfer pricing rules in place—including, not least, for domestic transactions as well as cross border:74 Incentives, such as tax holidays, can create significant profit shifting opportunities even among domestic companies. It also means making the best use of the existing tools based on the ALP: much tax planning by MNEs in developing countries appears to be perceived by at least some observers not just to stretch ALP concepts but flout them.75 Countering this aggressiveness would be greatly facilitated by developing concrete guidance where it is lacking and repudiating perverse interpretations of the ALP (commonplace and often tacitly accepted), such as condoning risk stripping and other arrangements that provide no documented productivity gain for the MNE. Carefully designed safe harbors that apply a fixed mark up to certain costs can play a greater role than generally recognized—though subject to the caveats above.76

54. Transitionally, to protect their tax bases developing countries’ may have to rely more heavily on even simpler anti-avoidance measures. These might include, for instance, withholding taxes on payments for services; restricting allowable deductions for some types of expense, such as those related to services provided by the parent company headquarters, and limitations on interest deductions along the lines discussed above, perhaps particularly targeting operations with low tax jurisdictions.

55. A coherent agenda for addressing transfer pricing challenges in developing countries would thus rest on four pillars: introducing appropriate anti-avoidance rules; developing more detailed guidance on how the ALP should be applied in concrete situations that concern them very specifically (Box 5); improving public data availability for comparability studies; and capacity building.77

Challenges in Arm’s Length Pricing for Developing Countries

Geographical adjustments. In the absence of public data on market transactions in developing countries, taxpayers will resort to foreign operations in advanced economies to price their controlled transactions. But market prices (or profit margins) in developed countries may not be directly applicable to the circumstances of developing countries, and there is no guidance on how such geographical adjustments should be made. This would involve adjusting for market structure differences, appropriate interest rates, and country risk differences.

Location savings. The notion that MNEs profit globally from locating their subsidiaries in LDCs has led to the suggestion that this benefit should somehow be shared between the host economy and headquarters. In the absence of generally accepted guidance on this, some countries have devised some quite arbitrary ways to do so.

Non-competitive business arrangements. The preferred reference in the application of the ALP is a comparable transaction in a reasonably competitive market. However, it is common for intra-MNE arrangements rather to mimic a situation in a non-competitive market among independents. Independent parties solve these problems by bargaining, and a range of solutions is possible. No guidance is given as to how the ALP should apply in those circumstances, and developing countries are particularly ill-placed to conduct such bargaining with other competent authorities, or to apply adjustments to taxpayers using knowledge of such market structures.

Asymmetric approaches. Some transfer pricing methods determine the profit associated with the ‘routine’ (thus easier to value) activities performed by one of the parties participating in the transaction, with the residual profit of the controlled transaction allocated by default to the other party—the one with the more complex functions, which are typically located outside developing countries. This risks missing functions in developing countries that contribute value beyond the limited routine activity; without sufficient guidance, it can be applied to their disadvantage.

Risk transfer. Current transfer pricing guidelines do not explain how an MNE could benefit from shifting risks within the group, so cannot help developing countries apply ALP to the risk transfer itself.

56. It is important, however, that addressing ALP and other international tax challenges more effectively in developing countries, not detract from core efforts to strengthen domestic tax systems. The prospective revenue gains to them from better implementation of ALP, for example, should not be overstated. The tough and unglamorous work of strengthening domestic taxation—by, for instance, building the effective personal income tax that many developing countries still lack—must remain a priority.

Dealing Better with Spillovers

Key Messages

  • Current initiatives, which operate within the present international tax architecture, will not eliminate spillovers.

  • Unlike many other areas of tax policy, in international taxation there is little consensus as to the practicable principles that should guide efficient tax design.

  • Among alternative frameworks that have been proposed, which differ in the degree to which they depart from current arrangements:

    • Minimum taxes can in some cases be useful in protecting the revenue derived from inward investments.

    • Strengthening worldwide taxation may help limit tax competition, but would put more pressure on fragile and increasingly arbitrary notions of corporate residence.

    • ‘Formula apportionment,’ which has attracted much attention, would limit conventional transfer pricing problems, but would create new difficulties around the factors used to apportion profits across jurisdictions, and would not necessarily shift tax base towards developing countries.

    • Hybrid schemes that combine straightforward ALP methods and a formulaic allocation of profits not easily allocated by these means merit closer study, as do more radical proposals for destination -based corporate taxes.

  • The obstacles to effective and inclusive action on adverse international tax spillovers are both political and conceptual, and are substantial.

A. Changing the Architecture

57. The nature and extent of spillovers depend on features of international tax arrangements beyond those addressed in current initiatives. Eliminating avoidance would not eliminate spillovers. These arise from the incentives that tax rules imply for both business decisions and tax setting by governments, with the form of those incentives depending on the wider international tax architecture. These incentives will be seen, for instance, to be very different under worldwide and territorial taxation—a distinction that current initiatives do not address. Differing arrangements might mitigate (or worsen) the adverse spillovers that now arise, and recognition of this has led to significant interest, among CSOs and academics, in deeper reforms of that architecture.

58. This section reviews proposals that have been made for alternative international tax arrangements. These differ substantially in how far they diverge from current practice.78 While a significant change in direction may well be impractical for the immediate future, considering the direction in which the international tax architecture could usefully evolve is important for informing choices made now.

59. There are few accepted principles to guide the design of collectively efficient international tax structures. In other areas of tax policy, there are generally accepted guidelines (albeit qualified) for efficient design: there is widespread agreement, for instance, on the desirability of moving away from tariffs, or towards carbon pricing. The literature has not, however, produced guiding principles for international taxation sufficiently compelling to leave a strong mark on the policy debate (Appendix VII). Proposed alternatives to the present international tax framework are commonly framed as responding to perceived difficulties and/or furthering implicit judgments on the fairness of the allocation of tax base79 rather than being grounded in well-articulated efficiency and cross-country equity goals.

Minimum Domestic Taxation

60. One possible approach to bolstering the CIT base in developing countries is through some form of minimum tax (MT). An MT aims to protect revenue by charging tax on something— commonly turnover, book earnings or assets—that is less subject to manipulation than is taxable income, with overall tax payment then being the larger of liability under MT and under the standard CIT. Corporate MTs are already found in over 30 countries. Schemes differ quite widely, and can lead to considerable complexity and significant distortion: a charge on net assets, for instance, can reinforce debt bias, while one on gross assets may introduce distortions between firms with differing capital structures. Nonetheless, MTs have proved both useful and practicable in protecting domestic tax bases, and might also be addressed to combating aggressive international tax planning in relation to inward investment. They could, for example, address in a simplified, aggregate way the need for increased limitations on deductibility of certain cross border payments flowing from developing countries, that is seen by many observers.

Strengthening Elements of Worldwide Taxation

61. Recent years have seen a trend along the spectrum away from worldwide and towards territorial systems. Seventeen OECD countries have moved significantly towards territoriality since the 1990s, and territorial countries now export about twice as much capital as do countries that tax on a worldwide basis.80 The U.S. is now the only large OECD country applying a worldwide system, although many large emerging markets, including the BRICs, do so. Nominally territorial regimes differ widely, however: the hallmark is exemption of dividends from abroad, but some also exempt capital gains arising abroad and there are differences too in the strength of CFC provisions.

62. This trend reflects a variety of concerns, from the perspective of capital exporters, in applying worldwide taxation.81 These include:

  • A perception that firms resident in such countries are at a competitive disadvantage abroad— this in line with notions of capital import neutrality as an efficiency criterion (Appendix VII), but at odds with the view that home country taxes in countries applying worldwide taxation are substantially avoided by deferral.

  • A revenue risk from inversion—essentially, moving the tax residence of a company to another country. Legislation can discourage this,82 though views differ on how effectively.

  • Recognition that foreign tax credit arrangements can create opportunities for base erosion83—though territoriality creates its own different opportunities for profit shifting.

  • The apparent disincentive for earnings repatriation created by deferral (Box 2). There is no doubt that U.S. companies, for instance, retain significant amounts abroad: on one estimate, $1.4 trillion.84 But this may not be an inherent consequence of deferral. So long as repatriation must occur at some point in order for shareholders to attach any value to the underlying income, taxes on repatriation must be faced either now or later; and so, if the tax rate is expected to remain unchanged, they should not influence current repatriation decisions.85 On this view, the accumulation of funds abroad is fundamentally a reflection of the expectation of future holidays on repatriations. Studies have found significant effects of repatriation holidays (in the U.S.) and of shifts toward territoriality (in the U.K. and Japan) on dividend repatriations;86 but this—at least in the holiday instance—is likely due at least partly to the explicitly temporary nature of the policy changes. Nor is it entirely clear why delayed repatriation should be a significant policy concern; undistributed foreign earnings are generally available to the home economy, at least with sufficient tax planning, and indeed are often invested in home-country financial assets.87

  • Considerable practical complexities can arise from, for instance, the operation of foreign tax credits: perhaps 40 percent of the compliance costs of U.S. MNEs relate to international aspects88—though whether these costs would be lower under territoriality is unclear.

63. Territoriality can, however, amplify profit shifting and intensify tax competition—a particular concern for developing countries. Shifting profits to, or investing in, low tax countries becomes more attractive if the income generated there becomes less heavily taxed in the parent’s residence country. And there is significant evidence of such effects in practice. For instance, foreign acquisitions by U.K. and Japanese corporations increased following these countries’ 2009 shift to exempting foreign business profits;89 and U.K. outward FDI became more sensitive to source country CIT rates.90 By the same token, this increased sensitivity to effective tax rates in source countries can lead to more intense competition to attract investment: tax holidays and other tax breaks become more attractive to investors if the tax saved in source countries is no longer offset by increased taxation in their residence country. The point is especially important for developing countries, with CIT bases often substantially weakened already by incentives. But it can reverberate on others too: one explanation that has been given for why large capital exporting countries have, historically, implemented worldwide taxation is precisely their expectation that doing so will bolster the tax rates imposed by others, and hence protect their own tax bases.91

64. Spillovers would likely be less under worldwide taxation without deferral—or, as a close substitute, territoriality with tough CFC rules. Such a structure would not eliminate spillovers: a higher tax rate abroad, for instance, will reduce revenue in the residence country. But it would likely dampen strategic spillovers, mitigating pressures to cut tax rates simply to protect tax bases.

65. Such a system does, however, put increased pressures on notions of residence, since that is what would ultimately define liability. This can be a source of difficulty, both practically given the possibility of inversion and conceptually in that the formal tax residence of a multinational is increasingly removed from the arguably more fundamental realities of its complex, inter-related and global activities and ownership.

Formula Apportionment

66. The use of formula apportionment’ (FA) to allocate tax base across jurisdictions is long-established at subnational level,92 and proposed for the EU by the European Commission as a ‘Common Consolidated Corporate Tax Base’ (CCCTB).93 FA establishes a multijurisdictional enterprise’s tax base on a ‘unitary’ basis—that is, consolidated across the entire corporate group94— and allocates this total across jurisdictions, which then apply their own rate of tax on a formulaic basis, according to varying combinations of the shares of sales, assets, payroll and/or employees located in each.

67. The current debate has intensified interest in FA as a way to ease difficulties with the ALP and perceived misallocations of tax base…The primary appeal of FA is in dispensing with the need to value intra-group transactions, so eliminating direct opportunities to shift profits through transfer pricing and other devices. And by then allocating the base using proxies to substantial activities, it holds the prospect of aligning tax payments more closely with economic fundamentals. That this approach, rather than one based on ALP, is so common at subnational level—indeed there seems to be no subnational CIT that does not use some form of FA—suggests that it has at least some merit in taxing firms operating across highly integrated economies.

68. …and, in the view of some, as especially attractive for developing countries,95 because their capacity limitations leave them especially vulnerable to profit shifting.

69. Significant issues arise, however, in relation to the weights used in the formula:

  • The cross-country allocation of revenue96 is highly sensitive to the choice of apportionment factors… This emerges clearly from early work on the CCCTB, which also showed that the effects could be very large for some countries: up to 50 percent of current CIT revenue. Data limitations prevent similar calculations for developing countries, but rough estimates can be made of the impact for some countries of reallocating the taxable income of U.S. multinationals using the factors commonly discussed: sales, assets, payroll, and employees (Figure 4).97 These are no more than illustrative, but point to large and systematic effects. Advanced economies generally gain tax base whichever factor is used, while substantial tax base moves out of conduit countries; emerging and developing economies clearly gain base only if heavy weight is placed on employment.

  • …with precise definitions critical. It matters a great deal, for instance, whether sales are measured on a destination basis (that is, by residence of the purchaser) or an origin basis (residence of the seller), and whether they include or—more in the spirit of the unitary approach, but further from accounting practice—exclude sales to other members of the corporate group. For reasons of data availability, Figure 4 uses sales by origin and includes intra-group sales. This is likely to be the notion of sales most favorable to lower income countries (where sales for final consumption and to third parties are likely to be relatively modest), so that the gain shown for them for allocation by ‘sales’ in Figure 4 is likely to be a generous upper bound on what would arise under other sales concepts (which, indeed, might plausibly be quite negative). Note too that if sales are defined to exclude those to other members of the group, then where the first sale out of a country is within a corporate group, none of the sales—whether on a destination or an origin basis—will allocate profits to that country. Reflecting this, some resource-rich states within the U.S. include a special factor relating to the production of natural resources.

  • Valuation and avoidance issues remain… The calculations above assume unchanged corporate behavior. But FA would present firms with very different real and profit-shifting effects, now focused on the factors entering the apportionment rule rather than on local profitability and intra-group transactions. If a destination-based measure of sales is used, for instance, there is an incentive to route sales through low-tax jurisdictions (as is easy to do); and if employment is used, the firm has an incentive to hire more low paid workers in low tax jurisdictions then it otherwise would. All this creates evident scope for production inefficiency.

  • …and so do issues of strategic tax-setting. Countries have an incentive under FA to attract whatever factors are given high weight in the formula—and this incentive may be strong, since the revenue gain from attracting such factors is not from a marginal increase in some local tax base—as it is, for instance, under territoriality—but from the greater share of the group’s overall profit that is brought into tax. Tax competition can, for this reason, become even more severe. under FA than even under territorial taxation.98

Figure 4.
Figure 4.

Reallocation of Taxable Income U.S. MNEs, Using Alternative Factors

(percentage change)

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

Source: See Appendix VIII. This figure shows the weighted averages for three country groups, using current taxable income as weights.

70. Practical issues also make FA less simple than it may seem. Definitions of the corporate group are required, for instance; and traditional forms of allocation will remain necessary (with all the problems that now entails) if FA applies only up to some water’s edge. Distinct forms of distortion can also arise, with the potential for artificial incentives to merge or spin off. 99 Whether these difficulties are greater than those under present arrangements is unclear—but nor are they trivial.

71. Whatever its merits in principle, prospects for adoption of international FA seem remote. A substantial legal and institutional infrastructure has been built around current arrangements, so that movement towards international FA would likely involve considerable disruption.100 That might change if a major capital exporter were to move in this direction. But there is little immediate sign of that—with significant resistance within the EU, for instance, to the CCCTB.

Formulary Profit Split

72. There may be scope for more systematic but partial use of formulaic approaches—as with the ‘formulary profit split’ (FPS) method. Current transfer pricing arrangements already envisage, and regard as consistent with the ALP, the splitting of profits on a the transactional RPS basis. One hybrid approach,101 combining both formulaic and ALP-based methods, would be to calculate the tax base on a unitary basis, applying simple ALP methods where their application is evidently straightforward (for transactions with third parties, most obviously) and using some FA-like formula to allocate not aggregate group profits, but the ‘residual’ profit that remains after accounting for arm’s-length transactions.

73. FPS has the merit of building on the strengths of present arrangements while remedying some of its defects. Some of the properties and implications of FPS are familiar—those relating to the choice of apportionment factor(s) for allocating residual profit, for instance. Some are problematic—a group might have an aggregate loss, for instance, but face a strictly positive tax liability. Some may be attractive—preliminary analysis suggests that spillovers may be weaker under FPS than under either separate accounting or FA.

Destination based corporate taxation

74. One radical proposal is to shift corporate taxation to a destination basis102—meaning that exports are excluded from the base, and no deduction provided for imports. In its simplest form, the tax would be levied on a cash flow basis: no deduction for interest or other financial costs, and all investment spending immediately deductible against tax (rather than depreciated over some years). Since value added is cash flow profit plus wages, this would be broadly equivalent to the combination of a VAT and labor subsidy at the same rate. While there are a number of unresolved issues—in the treatment of services, and consistency with the WTO prohibition103 on direct tax equivalents of export subsidies, for instance—this approach has the very strong appeal that, as with the VAT, cross-country spillovers would be expected to be minimal.104 There would be no tax benefit, for instance, in under-pricing sales from a high tax country to a low tax one, since the export attracts no tax and the import attracts no relief. The effect of moving towards a destination-based corporate tax would be to shift the tax base towards countries in which sales are concentrated—so having some of the adverse impact on developing countries noted above for at least some forms of sales-based FA. But the attractions of such an approach may be implicit in the emphasis on increasing reliance on the VAT and reducing labor taxes apparent in recent years. More fundamentally still, this proposal raises in stark form deeper issues concerning the role of the CIT that the current debate has set aside.

B. Challenges of Coordination

75. Substantial coordination to address spillovers in international taxation has proved very difficult to achieve. Some see little if any case for coordination, for political economy reasons touched on above. But even if it were agreed that spillovers cause collective harm, identifying and securing agreement on appropriate measures of coordination is likely to be highly problematic.

76. Countries’ interests diverge widely, most obviously between low and high tax countries, but in other respects too. Countries that are primarily capital exporting, for instance, are likely to be more sensitive to the interests of the residence country, while resource-rich countries may attach more importance to source rights. This diversity of interest need not preclude identifying mutually advantageous forms of coordination. Minimum effective tax rates, for instance, can prove beneficial even for low tax countries initially below the minimum, since an enforced increase in their own tax rates may lead to an induced increase in tax rates elsewhere from which they can benefit. But diversity clearly does make securing agreement on coordination measures harder.

77. …and there is no institutional framework comparable to that regulating international trade. BTTs provide an important structure, but touch only a subset of corporate tax design issues. While some have argued for a ‘World Tax Organization’ to serve as a tax analogue to the World Trade Organization, prospects are very remote—in part reflecting the absence of guiding principles with the force of arguments for free trade.105 In sharp contrast to the trade context, moreover, in tax matters smaller countries can have a major impact, making widespread participation in any agreement especially important. The OECD has come to play a powerful and constructive role in the area, drawing on the technical expertise of its member countries and beyond, and has taken an inclusive approach to its current BEPS project. Drawing too on the long-standing expertise of the UN on treaty and transfer pricing issues in particular, and on the wide membership and experience of other international organizations, as well as on regional organizations, current initiatives may ultimately lead to a more coherent framework for managing international tax relations.

78. In designing specific measures, the notion of ‘harmful tax practices’ remains ill defined. Attempts have been made to identify particular sources of tax spillover as being distinctively ‘harmful.’ These have largely focused on preferential regimes as potentially harmful, as for instance in the European Union’s Code of Conduct for Business Taxation. Even this, however, is problematic: aggressive competition for very mobile parts of the tax base may be preferable to less intense competition over a wider base. The literature points to various margins of behavior on which this issue turns, but in practical terms the issue is unresolved.106

79. Partial coordination can be problematic:

  • Coordination among only a subset of countries, perhaps on a regional basis may be relatively practicable (though experiences in Europe, West Africa107 and elsewhere show that it can be far from easy). There is, however, some risk that such coordination can make participating countries more vulnerable to pressures from lower tax jurisdictions elsewhere.108 This does not mean that participants are harmed by coordination, and the IMF continues to support such regional efforts—but points to the potential benefit from negotiating with leading non-participants, as the EU has in several cases done.

  • Coordinating only on a subset of instruments can worsen distortions by putting more pressure on those not coordinated.109 Making treaty abuse more difficult, for instance, might in itself simply lead to more aggressive use of transfer pricing or intra-group borrowing as profit shifting devices. This makes it important for instance, to view the elements of the BEPS Action Plan as a package; and, more generally, to avoid a piecemeal approach to identifying and addressing spillovers in international taxation.

Appendix I. Consultations

Consultation process

In preparing this paper, staff held: several discussions with civil society organizations, including in both small groups and in an open, large panel discussion at the 2014 IMF-World Bank Spring Meetings; a conference call with international business, including six major corporations, organized through the Confederation of British Industry (CBI); and meetings with academic economists and lawyers expert in the area.

Staff also participated actively in two consultations held with developing countries by the OECD Center for Tax Policy and Administration in connection with the G20-OECD BEPS project (a regional event in Colombia and a global meeting in Paris). Staff have also been providing input into the OECD’s Tax and Development working group report to the G20 development working group on these consultations and related issues.

Many submissions were also received in response to an online invitation for views, posted on the IMF’s external website.

Key issues raised

Civil society expressed the views that….

  • International inequities exist in the corporate tax take, particularly with respect to developing countries who suffer capacity constraints. The misalignment of taxation and value creation is a major flaw in the current architecture. Several highlighted the impact of ‘non-inclusive’ international processes with the development of international standards in fora composed only of developed country/G20 members.

  • Tax treaties, with significant risk to developing countries from treaty shopping and the lowering/abolition of withholding taxes. The benefits of treaties in terms of investment are doubtful; and treaties systematically favor developed countries over developing countries.

  • Tax havens, as vehicles for profit shifting are responsible for significant revenue losses in developed and developing countries. The latter could undertake more action, requiring more transparency and disregarding transactions if there is no economic substance involved.

  • The ALP might have reached its limits due to practical problems and conceptual flaws. A fundamental review of allocation rules is needed to better align profit to economic activity. Simplified approaches (such as fixed margins, profit split methods and cost caps) may be useful for developing countries.

  • Tax competition is generally seen as harmful, and especially problematic in developing countries, where tax competition is often in the form of tax incentives, such as tax holidays, which have doubtful benefits for these countries. Such tax competition raises coordination issues that go beyond simple tax avoidance.

… and proposed steps in the following directions:

  • Systematic analysis of spillovers to help identify the need for and the appropriate type of policy action, and to ensure coherence in policy responses.

  • Limitation of interest deductions in broader more effective ways.

  • Better data, to understand the quantitative importance of spillovers.

  • Further research in areas such as (i) unitary taxation with formula apportionment (which some saw as a logical response to globalization, with smaller risk of spillovers and a fairer distribution of tax revenue); (ii) a multilateral treaty approach; (iii) global harmonization of accounting standards. Some comments went beyond calling for more research, to urge action in these directions.

  • Greater transparency to better identify profit shifting risk and take proper actions. Many called for (publicly disclosed) country-by-country reporting by MNEs; others emphasized automatic exchange of information between tax authorities—although limitations on potential use and capacity constraints are important in this regard, especially for developing countries.

Business made the points that:

  • There is risk of conflict over tax take, with new rules potentially allowing for extensive interpretation by individual tax authorities that could end up in a battle between (advanced) countries in claiming part of the tax base. It could lead to double taxation, increased tax uncertainty and rising litigation.

  • Country-by-country reporting could significantly raise the compliance burden for MNEs.

  • Unitary taxation with formula apportionment has little merit in an international context, as agreement on the formula will be virtually impossible.

  • Residual profit split runs the risk of increased uncertainty, as two governments negotiate about the tax take.

  • Tax incentives are not a major factor in the location choice of businesses; and discretionary incentives can also deter investment by creating uncertainty.

  • Digital economy transactions should be subject to common rules and laws based on the same principles as applied elsewhere, not subject to special treatment.

Developing country representatives drew attention to:

  • The difficulties they have had in tackling specific cases of tax planning.

  • Particular areas of difficulty, including among others, the extensive use of service charges by headquarters, under differing names; and the assignment of low value-added functions to subsidiaries, with no relation to the overall operation of the MNE.

  • A need to revise the allocation of taxing rights across countries, not just tackling avoidance, which was seen by some as much-needed.

  • The harm that developing countries suffer from tax competition, including especially through the proliferation of incentives.

Individual submissions raised many points, including:

  • The development of multi-lateral tax treaties will be necessary in the medium to longer term.

  • Tax competition is a major problem, including among developing countries.

  • Strengthened CFC regimes, restrictions on interest deducibility, and increased and invigorated withholding taxes would in combination help greatly in reducing tax avoidance and base erosion.

  • The arm’s length method must be adjusted to give less deference to contractual arrangements.

  • Formulary apportionment would offer a reasonable way forward, but will be feasible only after considerable policy and administrative design work over a period of years. The IMF should participate in such work.

  • Public distrust and current negative views of the tax architecture are at root based in a belief that the existing system does not fairly split the tax base across countries—this goes beyond solely tax avoidance, the correction of which may well leave this fundamental issue unresolved. Technology has contributed to a shift, under the existing system, of the allocation of taxing rights.

Appendix II. Taxation and the ‘Digital Economy’

Many now prefer to use the phrase ‘the digitization of the economy,’ rather than ‘the digital economy,’ as it is increasingly recognized that the tax issues raised by the growing business use of the internet should be addressed not as specific to sectors in which the internet is integral to delivery of a service (such as the social media) but as generic to much of business activity. A collision of new business models coupled with an increasing proportion of value added arising from intangible assets, such as various forms of intellectual property, with old international rules designed to rely for the basis for taxation on some form of physical presence has created substantial gaps in the taxing model for not only major international technology players, but also for other businesses that are less wholly ‘digital’ in nature.

Until quite recently, most informed opinion viewed the problems created by ‘digitization’ as most important for indirect (that is, consumption) taxation. For example, an OECD Technical Advisory Group as recently as 10 years ago took the view that existing income tax concepts seemed sufficient to ensure tax neutral treatment across e-commerce and physical transactions.110 And in the late 1990’s the problems of applying consumption taxation were seen, in effect as ‘urgent but not fundamental,’ while the problems of corporate income taxation were seen as ‘fundamental but not urgent.’ The problems of digitization have since then been more fully addressed for consumption taxation. These ‘e-commerce’ indirect tax issues have long been the subject of work within the relevant OECD Committee on Fiscal Affairs working group on VAT, for example, and are now in the process of further study there, and in the European Commission. However, issues of corporate income taxation—which are indeed quite fundamental—have now become urgent.111 These are both more complex and more controversial.

Indirect Taxation issues

The most straightforward problem is applying sales tax or VAT in the case of on-line ordering of physical goods for final consumption. Since consumption taxes are normally imposed by the jurisdiction where the goods and services are consumed (on the ‘destination basis’), the issue is to develop a workable system to charge VAT with regard to such sales.112 The original ‘e-commerce’ concern in this context arose as a result of the increased quantity of direct imports by final consumers, resulting in a practical, though not a conceptual, problem of collecting the VAT.113

The place of taxation for VAT is typically defined as the ‘place of supply.’ With regard to non-physical—digital, virtual, or electronic—products that are purchased from sellers abroad, such as movie downloads, the place of supply may be less obvious than for goods but must be defined. New EU rules that are to come into effect on January 1, 2015, explicitly treat the place of supply for all purchased electronic services as being in the country where the customer is located—logical where the basis for taxation is ‘destination,’ the place where the consumer consumes.114 Again, this is relatively straightforward conceptually, and the issue is one of administration and enforcement.115 Unlike the issues in regard to CIT, the core issue is practical, not conceptual.

Direct Taxation Issues

The issues posed by ‘digitization’ for direct taxation are much thornier conceptually, as well as posing practical issues—and have now risen to the fore in the international debate. Fundamentally, the questions relate to the issue of locating the ‘source’ of corporate profits.

There are various problems. First is the location of ‘intangible’ factors of production. Many modern products within the ‘digital’ economy tend to rely more heavily than those of a century ago on intellectual property as important contributors to their value. One aspect of tax planning revolves around strategically ‘locating’ these assets in low tax jurisdictions, and assigning as much value to them as possible in the production process, through the payment of high service fees, royalties, and the like from the actual place where production is occurring to the place where the asset is deemed to ‘be.’116

A second problem arises because ‘sales’ alone do not in the standard architecture give rise to a basis for corporate income taxation in a jurisdiction. The problem is now sometimes re-framed as needing to determine where an enterprise has ‘economic presence,’ but not a physical or traditional tax presence—using the rationale that in some businesses, there is (relatively) little in the way of physical assets anywhere that are contributing to the formation of value and profits. Advances in technology more generally facilitate the carrying on of economic activity with little or no need for employees to be present.117 There is a strong sense, for example, that providers of search engines or online movies are ‘doing business’ or somehow ‘present’ in a country where large numbers of people are using their products, even if there are no hardware or employees there—but the existing tax system does not capture this sense.118 In essence, the question becomes whether the traditional approach to identifying the source of profits should be altered, to move further toward, in these instances, the location of the customer base. This is partly due to the difficulty in determining the ‘real location’ of production activity where that activity is much less reliant upon bricks, mortar, and large numbers of production workers. If it were to be agreed that in such a case the country in question ‘should’ be able to capture a portion of the company’s tax base, the question then arises, what rules would need to be changed to permit this?119 An approach under discussion, for example, is the notion—relying on the existing tax architecture but extending it—of creating a ‘virtual permanent establishment’ in the country where consumers of digital services are located, or, less radically, to lower the threshold regarding what sorts of activities (such as preliminary or preparatory actions) will give rise to a PE.120 In any event, there is now wide agreement that rules should apply to all enterprises, irrespective of their degree of digitalization. To the extent this results in sales themselves creating a presence that could trigger a CIT liability, the change in the current international tax architecture would be profound.121

Appendix III. Estimating Spillovers from International Corporate Taxation

This appendix elaborates on the empirical analysis of spillovers reported in the text.

Empirical Strategy

Base spillovers are explored by estimating equations of the form


where Bi denotes the CIT base in country i = 1,.., N at time t = 1,.., L, τit is the domestic CIT rate in percent, and W-iτ-i indicates some weighted average ΣjiNωjτjt of the statutory tax rates in countries ji(withΣk=1Nωj=1),Xit is a vector of controls, and country and time-specific effects are also included. The lagged dependent variable allows for sluggish response in the tax base to changes in tax rates.

The coefficient β in (1) represents the short run marginal impact of a country’s own CIT rate on its CIT base, with the long run impact given by θ(β) ≡ β(1 – δ); both are expected to be negative. The focus here, however, is on base spillover effects from tax rates set by others. This is captured by the coefficient γ for the short run, and by


for the long run, with both expected to be positive: higher tax rates abroad would be expected to increase the tax base at home. As discussed in the text, such effects can operate by two channels: through tax effects on real investment decisions, and on profit shifting. These imply different structures for the appropriate weighting matrix in (1), which in turn provides a crude but potentially instructive way to assess the importance of each channel.

To see this, suppose first that only real decisions can respond to tax considerations. With capital mobile across borders, after-tax rates of return will be equalized in equilibrium, since otherwise investors will have an incentive to reallocate assets to wherever the after-tax return is higher. This implies that the before-tax rate of return will be higher in countries with higher tax rates, which, assuming decreasing returns to capital, means a lower capital stock than would otherwise be the case. Standard models then imply that how much capital will flow between countries in response to a change in the foreign tax rate—the magnitude of the spillover term in (1)—will depend on such factors as their relative population sizes and quality of their technologies:122 Intuitively, real investment in a large advanced economy is likely to be much more sensitive to the tax policies of other advanced economies than to those of small or poor low-income countries. As a simple way of capturing these considerations, we explore using as weights ωi in (1) each country’s share of global GDP.

Spillovers through profit shifting are different The benefit of shifting a dollar of taxable profit between countries is simply the tax saving from doing so, measured by the absolute difference in their statutory tax rates: the tax base of an advanced economy could indeed then be strongly affected by the rate set by a small, low-tax jurisdiction. More formally, theory suggests that base effects are driven by statutory tax rates in other countries weighted not by size or income but by the ease with which profits can be shifted in or out of them (reflecting other aspects of tax rules or the degree of their enforcement). To explore this possibility, we consider two other possible weighting schemes in (1): an unweighted average of statutory rates in all other countries, and a weighted average only across those jurisdictions included in a commonly-used list of ‘tax havens.’

To explore strategic spillovers—the response of country ;’s tax rate to the tax rates set by other countries—we estimate (following Devereux and others, 2008):


The same reasoning as above points to the use of the same three sets of weights: by GDP, unweighted over all countries, and unweighted across only ‘havens.’

This empirical strategy has significant limitations, largely reflecting the availability of data. Cross-border real investment decisions, for instance, are likely to be driven not by the statutory rate of CIT alone but by an average effective tax rate (AETR) that also reflects depreciation and other allowances123—data on which, however, are not available for as long a period or, more particularly, for the developing countries that, for present purposes, it is important to include in the exercise. There is some comfort, however, in the strong correlation between AETRs and statutory rates, and the use of instruments below should help address the implied problem of measurement error. A further difficulty is that the attractions of tax ‘havens’ do not solely, or even mainly, derive from low statutory CIT rates, but from special regimes and arrangements on which data are again unavailable.124 The identification of haven effects thus depends on a plausible but untested correlation between movements in their rates and special regimes. The results, for these reasons, can be no more than indicative.


The sample is an unbalanced panel comprising 173 countries for the period 1980–2013. The data are taken from the IMF’s Fiscal Affairs Department tax and revenue database. The country coverage of CIT rates is full, though unbalanced in the time dimension. To approximate the CIT base in percent of GDP, we divide the ratio of CIT revenue to GDP by the standard CIT rate; this is referred to as the ‘implicit’ corporate tax base. Unavailability of CIT revenue means, however, that this can be done for only 121 countries. The total number of observations on CIT rates (relevant for estimating strategic spillover effects) is 3,037, while for the CIT base (for base spillovers) it is 2,161. Resource-rich countries, whose CIT base will likely have distinct drivers and reflect a variety of design choices (including the use of state owned enterprises, whose profits are akin to CIT revenue, and of distinct rent-based taxes) are excluded. The countries in the sample (and those classified as ‘havens’), classified by income group, are listed in Appendix Table 4; full descriptive statistics are in Appendix Table 5.

Appendix Table 1.

Base Spillovers with Alternative Weighting Matrices 1/

article image
Note: Dependent variable is CIT base. Full set of year dummies and control variables in all regressions. Robust standard errors, in parentheses; ***, **,* indicate significance at 1,5, 10 percent.

One step, robust, system GMM with instruments based on first lag of differences in the CIT tax base (collapsed to avoid proliferation in the number of instruments) in levels equation, and second lags of their levels in the differenced equation.

Appendix Table 2.

Base Spillovers by Income Level, ‘Haven’-Weighted 1/

article image
Note: Dependent variable is CIT base. Full set of year dummies and control variables in all regressions. Robust standard errors in parentheses; ***, **,* indicate significance at 1,5, 10 percent.

One step, robust system GMM with instruments based on first lag of differences in the CIT tax base (collapsed to avoid proliferation in the number of instruments) in levels equation, and second lags of their levels in the differenced equation.

Appendix Table 3.

Strategic Spillovers with Alternative Weighting Matrices 1/

article image
Note: Dependent variable is statutory CIT rate. Full set of year dummies and control variables in all regressions. Robust standard errors, in parentheses; ***, **,* indicate significance at 1, 5, 10 percent.

One step, robust, with instruments based on first lag of differences in the own CIT rate and weighted CIT rates of other countries (collapsed to avoid proliferation in the number of instruments) in levels equation, and second lags of their levels in the differenced equation.

Appendix Table 4.

Countries in the Sample129

article image
article image
Appendix Table 5.

Descriptive Statistics

article image

As controls, we include in both estimating equations (1) and (3) variables that have been commonly used in modeling tax revenues and rates:125 the log of GDP per capita, the share of agriculture in value-added, trade openness (the sum of non resource exports plus imports, expressed relative to GDP), inflation, and oil rents. Estimation is by system-GMM (Blundell and Bond, 1998) to address endogeneity of both the lagged dependent variable and possibly τ-it.126

Estimation Results

Appendix Table 1 reports the results of estimating equation (1), using in turn the three different weighting matrices discussed above: column (1) uses GDP-weighted rates; column (2) the simple average; and column (3) the simple average over tax havens. The diagnostics are satisfactory, with the Arellano and Bond (1991) test for first- and second-order serial correlation (M1 and M2) suggesting that, consistent with the underlying assumptions, the former is present and the latter is not; the Hansen statistics suggest that instrument proliferation is not a major concern.127

The impact of country i’s CIT rate on its own CIT base is significant and large. The short-run marginal coefficient of 0.17 in column (1), for instance, means that a 1 percentage-point increase in a country’s CIT rate will reduce its CIT base by 0.17 percent of GDP; and the long run effect 6(p) suggests an ultimate reduction of around 0.56 percent of GDP. The own-effects for the other weighting schemes, in columns (2) and (3), are smaller but remain statistically significant.

The main interest here, however, is with the spillover effects, y and 9(y). Column (1) shows a large and significant positive coefficient for the weighted average foreign tax rate, interpreted here as relating to spillovers through real capital flows: a one percentage-point reduction in the GDP-weighted world average CIT rate (excluding its own rate) reduces the typical country’s CIT base in the short run by 0.3 percent of GDP. This effect becomes much smaller and less significant in column (2) where foreign tax rates are not weighted by size. In column (3), however, where only statutory rates in ‘havens’ are included, the spillover is again 0.3, as when using the GDP-weights—and more strongly significant. The long-run spillover of the tax base is between 0.56 and 1.12 percent of GDP, with the higher figure again corresponding to both GDP- and haven-weighted rates, and more significant for the latter. These are potentially quite large effects, as viewed for instance as a share of total tax revenue. What is striking too is that both real and profit-shifting effects seem to be significant, and with a distinct haven effect emerging despite the imperfections, as noted above, with which the distinctive features of havens are captured in the data.

Appendix Table 2 explores whether spillovers vary systematically across countries at different income levels, reporting separate estimates of equation (1), using the haven weights, for: OECD countries (column (1)), non-OECD countries (column (2)), and low and middle-income countries128— (column (3)). The coefficient for the OECD sample is much smaller, and less significant, than that in Appendix Table 1: the long-run marginal impact, for instance, is now 0.27 rather than 1.1. Moreover, the estimated spillovers are substantially larger for the non-OECD and lower income groups (especially the latter) than for OECD countries.

The last column of Appendix Table 2 explores possible changes in spillover effects over time (again for the haven-weighted case) by interacting with the spillover term a dummy variable that takes the value zero for 1980–1995 and unity for 1996–2013. The results suggest that while the own-rate effect has not increased over time, spillovers have. Indeed, spillovers from havens are significant only in the latter period.

Results on strategic spillovers—countries’ responses to the tax rates set elsewhere—are in Appendix Table 3. These point to strategic complementarity in tax-setting: that is, countries respond to tax rate reductions elsewhere by cutting their own tax rate. A one point CIT rate reduction in all other countries, for instance, induces a 0.5 point rate cut, with a slightly larger response to rates in ‘haven’ countries. The coefficient of the GDP-weighted average is both larger and more significant, however, indicating that larger countries’ tax policies have a stronger effect in driving policies elsewhere.

Appendix IV. Using Gross Operating Surplus to Explore Spillovers

This appendix uses panel data on CIT revenue, CIT rates and gross operating surplus to gauge the extent to which cross-border tax effects may affect country’s CIT revenues and dissipate aggregate revenue, as reported in the text

Framework and Data

We define ‘CIT-efficiency’130 in country i (denoted Ei) as the ratio of actual CIT revenue (Ri) to some reference level of CIT revenue, with the latter computed as the standard CIT rate (τi) multiplied by a reference tax base (Gi):


To the extent that the reference CIT base is larger than the actual ‘implicit’ CIT base (Rii), Ei will be less than unity; and the further Et lies below unity, the less effective is the CIT in raising revenue relative to the benchmark. For a given choice of reference base, cross-country variation in Ei can be the result of differences in policies, such as the generosity of tax deductions for depreciation and interest, or of special tax incentives. But Ei might also be affected by behavioral responses, such as profit shifting, which cause the actual CIT base to deviate from its reference. The analysis below explores how variations in Ei correlate with the CIT rate, which may point to such behavioral responses.

For Rt and τi we use the data described in Appendix III. For the reference tax base, Gi we use the Gross Operating Surplus (GOS) of corporations, from the national accounts; this is available for 93 countries, and for up to 33 years, from the UN Statistics Division.131 GOS is an instructive reference base in that provides some proxy to what the base would be if profits were allocated on something broadly similar to a ‘source’ basis.132 It is close to the accounting concept of EBITDA (Earnings Before Interest, Tax, Depreciation and Amortization). Data on GOS are constructed from the national accounts as value added by corporations, minus compensation of employees.133 It is broader than the standard CIT base in that depreciation allowances, interest and other specific provisions are not subtracted, Moreover, loss carry forwards can create differences between the accrued GOS that appears in the national accounts and the CIT base that is relevant for tax purposes.134 For these reasons, one would expect CIT-efficiency relative to this base to be less than unity.

The panel obtained from the UN national account statistics is unbalanced and—significant limitations on the present exercise—sparse in coverage of both developing and conduit countries. Overall, 59 countries appear in both tax and GOS datasets. As with the exercise in Appendix II, resource-rich countries (Norway, Kuwait, Russia, Saudi Arabia, South Africa, Venezuela, Kazakhstan and Trinidad and Tobago) are dropped. The remaining sample contains 840 observations, comprising 51 countries and a maximum of 33 years (1980–2012).135

Mean CIT-efficiency is 43 percent, with a slightly lower median: as expected, the implicit CIT base is typically much smaller than GOS. The variation is large, however, ranging from 7 percent (China in 1995) to 338 percent (Cyprus in 2008) and with a standard deviation of 29. There are large differences between country groups: the averages in the OECD and Eastern Europe and Central Asia exceed 40 percent, while those in Sub-Saharan Africa, the Middle East and Northern Africa and Asia and Pacific are noticeably lower. Notably, average Et exceeds 100 percent for Cyprus, is also high in Ireland and Luxembourg, and is lowest in Africa. Over time, there is an upward trend in Et (Appendix Figure 1), in both OECD and non-OECD countries,136 reflecting the global trend towards lower CIT rates combined with stable or even rising CIT revenue.

Appendix Figure 1.
Appendix Figure 1.

Trends in CIT Efficiency

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

Signs of Profit Shifting?

Variations in Ei might reflect behavioral responses that affect GOS and the implicit CIT base in different ways. One obvious candidate is profit shifting, the incentives for which are determined by differences in statutory CIT rates: if a country has a relatively high CIT rate, outward profit shifting will likely cause an erosion of the tax base, without a corresponding reduction in GOS. Conversely, for a country with a relatively low CIT rate, inward profit shifting will tend to expand the implicit base.137 Thus profit shifting would be expected to induce a negative correlation between Ei and τi.138 To explore this, we regress Ei as the dependent variable on τi controlling for time139 and country fixed effects.140

The results, presented in Appendix Table 6, point to strongly significant effects. Column (1) suggests an intercept at 85.1 percent and a slope coefficient on the tax rate of -1.2: this implies that a country with a CIT rate of 10 percent would have, on average, CIT-efficiency Ei of 73 percent. Column (2) suggests that the relationship is convex: predicted efficiency Ei is now 104 percent in a country with a CIT rate of 10 percent. The last two columns imply that tax rates have a much larger impact on CIT efficiency in the 26 non-OECD countries, with a coefficient more than three times as large as for the 25 OECD countries. Overall, the results point to a strong negative relationship between Ei and τi suggestive of strong profit shifting, and—as with the analysis of spillover effects in Appendix III— again suggesting the effect to be particularly marked for developing countries.

Appendix Table 6.

Fixed Effects Regressions of CIT Efficiency on CIT Rate /1

article image

/ t-values in parentheses, based on heteroskedasticity-robust standard errors. All regressions include time and country fixed effects.

Revenue implications

Profit shifting from high to low-tax countries reduces revenue in the former and increases it in the latter—but with an overall revenue loss, since the shifting occurs precisely to reduce total tax paid. The data on GOS can be used to give a sense of where, and how large, are the underlying gains and losses. These calculations suffer from a number of limitations, stressed later.

Computing for each country the product of its CIT rate and its GOS, multiplied by the average CIT efficiency, Ē—as a crude but practical way of allowing for base erosion (or expansion) from sources other than profit shifting—gives a rough proxy for the CIT revenue it would receive in the absence of income shifting—or, somewhat more precisely, if CIT were effectively levied on a source basis. Revenue without profit shifting (Ri*) is thus approximated by


where Ē is calculated as a weighted average of countries’ CIT-efficiencies, Ē = ∑ωiEi, the weighting being by GOS:bold> ωi=GiΣGs. If—a somewhat heroic assumption—profit shifting is the only source of cross-country variation in CIT-efficiency, the difference between the simulated and the actual CIT revenue for each country could then be interpreted as a rough measure of the revenue gain (or loss, if negative) from profit shifting:


which is positive if a country’s CIT-efficiency exceeds the weighted average, and negative otherwise. This can be written as:


so that Δi is positive if and only if country i’s share of the world’s implicit CIT base exceeds its share of the world’s GOS. Using (4), one would expect aggregate revenue to decline as a result of profit shifting so long as higher tax countries tend to have a larger excess of their share in the sample’s aggregate GOS over that in the sample’s aggregate implicit tax base—which is as the regressions in Appendix Table 6 tend to suggest

For the calculations, we use data between 2001 and 2012 to reflect only the more recent period. Appendix Figure 2 shows the average CIT efficiency per country during this period, ranked from low to high. The GOS-weighted average is 43 percent. Revenue in the absence of apparent profit shifting, Ri*, and the associated revenue effect, Δi, are computed as in (2) and (3), respectively. This gives the results described in the text.

Appendix Figure 2.
Appendix Figure 2.

Mean CIT Efficiency, 2001–2012 /1

Citation: Policy Papers 2014, 071; 10.5089/9781498343367.007.A001

/1 CIT efficiency for Cyprus is 213 percent.


These calculations are, of course, highly speculative as they rely on a number of assumptions. For instance, as is clear from (3), it attributes to profit shifting all cross-country variation in CIT-efficiency in the sample. To the extent that such variation reflects differences in the prevalence of incentives that are themselves a strategic response to the tax policies of others, it can be seen as capturing base erosion from international tax competition. But variations in CIT-efficiency may also reflect such unrelated features as differences in compliance and enforcement. Moreover, the approach can capture only profit shifting between countries in the sample—which in the present case excludes the ‘haven’ countries which the analysis in the preceding appendix suggests are an especially important source of spillovers.

To illustrate the importance of assumptions in the calculations above, let us expand the definition of the tax base by explicitly distinguishing three elements: (i) profit shifting within the group of countries in the sample, Ai; (ii) profit shifting vis-à-vis countries not included in the sample (such as ‘havens’), denoted by Ci; (iii) exemptions, incentives or compliance gaps, denoted by Hi. Actual CIT revenue can then be written as:


Since Ai is pure profit shifting, any reallocation within the group of countries in the sample is a zero sum game: ΣiAi = 0. Using this, we derive:


so that:


Expression (7) shows that Δi represents the true revenue effect of profit shifting for country i shifting only if the last term on the right-hand side between brackets is zero. This is the case if (i) base effects due to exemptions, incentives and compliance gaps are proportional to the GOS share for all countries, (ii) there is no profit shifting from the sample countries to the rest of the world. Conversely, Δi overestimates the gain from profit shifting (or underestimates the loss) if a country has more exemptions or compliance problems relative to its GOS compared to the sample average, or if the group of countries in the sample together lose profits to third countries.

Appendix V. Tax Treaties and Withholding Tax Rates—Evidence

Investment and tax treaties

Studies of the impact of BTTs on FDI using bilateral country-level data show mixed results. Four find either no or a negative impact (Blonigen and Davies, 2004 and 2005; Egger and others, 2006; Louis and Rousslang, 2008). These studies are largely confined to FDI to/from OECD countries. Two studies find some positive impact for subsets of countries: Millimet and Kumas (2007) find a positive impact of treaties for countries with low initial FDI levels (but a negative impact for countries with a high initial level), and Neumayer (2007) finds a positive impact on U.S. outbound FDI only for middle-income countries. Two studies show a positive impact of treaties on FDI: for a broad dataset of 155 countries, Di Giovanni (2005) finds a positive impact on mergers and acquisitions. Barthel and others (2010), analyzing a dataset of 105 countries, find a positive impact of treaties on FDI stocks, particularly for middle-income countries.

Use of microdata alleviates concerns about reverse causality, since firms are likely to take the existence (or not) of a treaty as exogenous. Davies and others (2009), using firm-level data on Swedish corporations, find that treaties raise the probability of entry into a country by 17 percent relative to the sample average, but, conditional on presence, have little impact on investment. Egger and Merlo (2011) find similar results from firm-level data from Germany: controlling for the host country tax rate, the presence of treaty increases the probability of entry by 58 percent relative to the sample mean.

These and other studies that aim to capture the impact of BTTs solely through a dummy variable for their presence risk confounding positive and negative effects. Most BTTs contain both elements that may encourage FDI (such as reduced WHTs and clarity of liability) and elements that could discourage it (such as commitment to information exchange), so either—or neither—effect may dominate for a given sample. A recent study by Blonigen and others (2011) analyzing U.S. firm-level data attempts to separate positive and negative treaty effects by interacting the treaty dummy with industry effects capturing the extent to which firms in a particular sector are likely to experience heavier enforcement due to EOI.141 Controlling in this way for the effect of EOI, they find that the presence of a tax treaty increases average foreign affiliate sales by 45 percent, while the interacted term capturing EOI reduces them by 28 percent.142 Failing to separate the incentive and disincentive effects of BTTs thus biases the coefficient on a treaty dummy toward zero. Netting the two effects against each other, the authors also find that presence of a BTT roughly doubles the entry rate of new foreign affiliates.

There is evidence that ‘tax sparing’ provisions of BTTs have encouraged FDI. Such provisions grant to MNEs that benefit from source country tax incentives a credit against residence country tax equal to the source country tax saved as a result of those incentives: this preserves the value to the investor of host country tax incentives, which would otherwise be negated when the parent country has a worldwide tax regime.143 Both Hines (1998) and Azémar and Delios (2007) find that tax sparing has a positive impact on Japanese outward FDI flows. Similarly, Davies and others (2009), studying Swedish firm-level data, show that, controlling for the presence of a BTT, inclusion of a tax sparing agreement increases affiliate production, sales and exports (but not entry). With the trend towards territorial taxation, discussed in the text, tax sparing becomes less germane—but the evidence is instructive on what that trend might itself mean.

Studies of the impact of WHTs on FDI, which typically do not control for other treaty factors, again find mixed results. Egger and others (2006) find that the outbound FDI of OECD countries is negatively related to source country dividend withholding taxes, controlling separately for CIT rates and depreciation allowances. Similarly, Egger and others (2009) find, for the same dataset, that bilateral cross-border tax rates incorporating WHTs have a negative effect on FDI, controlling for home and host country CIT rates, which have a positive influence.144 Conversely, a study by Barrios and others (2012) using data on European MNEs finds that, while the overall cross-border corporate tax regime influences MNE subsidiary location, dividend WHTs exert no independent influence when separated from home and host country CIT effects. Similarly, Huizinga and others (2008) find for the same database that, while the overall cross-border tax regime influences MNE financing in the expected manner—higher taxation of dividends relative to interest correlates with a greater share of debt finance—interest and dividend WHTs have no independent effect on financing when isolated from CIT effects.145 By contrast, Arena and Roper (2010), examining marginal debt issuance decisions for MNEs based in 29 countries, find that a low ratio of interest to dividend WHTs results in a higher leverage ratio for foreign subsidiaries. The importance of WHTs thus remains unclear, presumably because they often merely shift revenue from source to residence countries without lowering the overall tax rate.

Trends in Withholding Tax Rates

Like CIT rates, both domestic law and treaty WHT rates have trended downward over the past decades (Appendix Table 7), suggesting that they have been subject to a similar process of tax competition. Since the early 1980s, tax treaty WHT rates on portfolio dividends, interest and royalties have on average fallen by about 30 percent, while the average rate on participating dividends has fallen almost 50 percent. Key drivers of this process among OECD countries have been EU directives146 eliminating intra-EU taxes on parent-subsidiary dividends and reducing WHTs on interest and royalty payments to a maximum of 10 percent.

Appendix Table 7.

The Evolution of WHT Rates

article image
Source: International Bureau of Fiscal Documentation database, 2011.

Appendix VI. Gains on Transfers of Interest

Recent Examples

There are several recent cases of significant gains on indirect transfers of interest147 being untaxed (or with tax disputed) where the underlying assets are located. These often relate to mining or petroleum rights, but cases are not restricted to these sectors. Appendix Box 1 provides some examples. Importantly, even though in these examples tax may not have been paid in the country in which the underlying assets are located, it is possible that tax on the transaction could be payable elsewhere (such as in the country in which the share transaction took place).

Examples of Gains on Indirect Transfers of Interests

India: The high profile Vodafone case involved the sale of a 66.9 percent interest in an Indian telecom company in 2010. The interest was held through a chain of offshore entities, but the transaction was effected by the sale of a single share in a Cayman Islands company. The Indian tax authorities sought withholding tax for the gain on the transaction, arguing that the offshore transaction was effectively the sale of rights (i.e., an asset) located in India and the offshore transaction was simply a way of facilitating that transaction. The tax in dispute is $2.2 billion. After a series of court cases the Indian Supreme Court ruled in favor of the taxpayer, but the government responded by proposing a retrospective amendment to clarify that such gains should be taxable. Negotiations are ongoing to settle the dispute.

Mauritania: A Canadian company effectively acquired an interest in a large gold mining project from another Canadian company via a transaction in the Bahamas in 2010, with a potential capital gain of $4 billion. No tax was collected on the transaction in Mauritania.

Mozambique: In 2011 a change in ownership of mining projects in Mozambique was achieved through the sale, on the Australian stock market, of shares in the mining company holding interests in the projects. The value of the transaction was around $4 billion. No tax was collected on the transaction in Mozambique. In the case of the sales of shares in the exploration concessions in the Rovuma basin, the authorities have collected $1.1 billion in capital gains taxes in 2013–14. Changes were made to the tax code, on January 1, 2014, to ensure taxation of capital gains resulting from a direct or indirect transfer between non-residents of assets located in Mozambique.

Possible Responses

An initial step is to ensure that domestic tax laws provide that direct and indirect disposals of immovable property (which should be defined widely enough to cover mining and petroleum rights) are taxable under the capital gains tax regime.148 Such provisions would appropriately cover only significant changes of ownership (of, say, at least 10 percent), to avoid having to identify small transactions. The rules also need to cover any kind of interest in an entity, not just shares. For example, South Africa has a ‘look through’ provision that specifies that if more than 80 percent of a company’s (or trust’s) assets consist of mineral rights (treated as immovable property), then disposals of the company’s shares (trust interests) are treated as disposal of the mineral right itself. If the gain is made by a non-resident, it is taxable as a South African sourced gain. In applying such an approach, the tax treatment for the purchaser needs to be considered (for example, whether the purchaser is entitled to write-off of the purchase price if the transaction is a deemed sale/purchase of a right). Tax treaties can also support the domestic law by ensuring that the country in which the assets are located retains the taxing right for those transactions. Both the UN and the OECD Model Conventions have such a provision, where at least 50 percent of the underlying assets are immovable property. However, the UN Model covers shares and interests in other entities, while the OECD Convention is limited to share transactions.

The challenge of discovering the offshore transactions and collecting the tax can be addressed by requiring that relevant regulatory authorities (such as the mining or petroleum agency, telecommunications authority, foreign investment authority) be advised of significant changes in indirect ownership, with this information shared with the revenue authority. The exchange of tax information under tax treaties and TIEAs can also help in obtaining information about such transactions. Once identified, there are a number of ways of collecting the tax (consideration of tax treaties and avoiding double taxation needing to be taken into account in making the choice):

  • Imposing a withholding tax on the transaction to be withheld by the seller—this is the approach used in India, for example;

  • Treating the resident company as the agent of the non-resident, so that they will be liable if the tax is not paid by the non-resident;

  • Deeming the resident company to have made the transfer (that is, disposed of and reacquired the asset), so that it is liable for the tax; or

  • Making regulatory approval for the transfer conditional on payment of the tax—with the threat, for example, of withdrawing the underlying license.

Appendix VII. Guiding Principles for International Tax Design

One of the most fundamental principles of optimal tax design, set out in Diamond and Mirrlees (1971), is the desirability of production efficiency: the proposition that, given sufficient tax instruments, any Pareto-efficient tax system will have the feature that aggregate production is maximized. There are many qualifications to this even in a purely domestic setting (it presumes, for instance, competitive behavior) and still more in international ones (because of spillover effects between national revenues).149 The principle has proved useful in other contexts, nonetheless, and is the natural starting point for international tax matters. While it can be applied from both national and collective perspectives, we focus here on the latter.

The difficulty is deciding exactly what production efficiency requires. One dimension is the requirement that investors resident in any location earn the same before-tax return wherever they invest—otherwise aggregate output could be increased by reducing investment where it earns the lower return and increasing it where it earns higher. This condition—capital export neutrality (CEN)— points towards residence-based taxation: investors then face the same tax treatment wherever they invest and so, in maximizing their net income by equating after-tax returns on those investments, will be led to equate pre-tax returns (Musgrave, 1969). But there are other dimensions too. Capital import neutrality (CIN) is the requirement that all investments within a country earn the same pretax return—without which aggregate output could again be increased by reshuffling investments. And this condition points, in contrast, to territorial taxation. And capital ownership neutrality (CON) is the condition that the ownership—and hence, presumptively, the productivity—of assets not be affected by taxation;150 a condition that is well-tuned to the importance of mergers and acquisitions within FDI (noted in the text), but to the extent that assets are location-specific, will often have the same implications as CIN.

These neutrality conditions conflict with each other, in the sense that meeting all generally requires fully harmonized tax systems. And very few results are available on efficient design when that cannot be achieved, or there are other constraints on tax-setting.151 Still more fundamentally, however, the concepts of CEN and CIN rest on notions of residence and source whose slipperiness, as discussed in the text, is at the root of the difficulties in current international tax arrangements; and CON, while more sensitive to the realities of modern corporate investment behavior and a step towards a framework that recognizes the importance of intangibles, places a stress on ownership as a determinant of productivity that has not won professional consensus—it is not clear, for instance, that one might not just as plausibly stress management or creativity more generally. Reflecting their limitations, these neutrality conditions—which, it should also be noted, do not in themselves dictate which country should receive the revenues152—provide only limited guidance in navigating the issues addressed in this paper, and indeed are rarely invoked in policy discussions.

Appendix VIII. Formula Apportionment: Details Underlying Figure 4

The calculations begin with the net incomes of U.S. parents and Majority Owned Foreign Affiliates (MOFAs) by country of affiliate (Table 10.2), taken from Bureau of Economic Affairs data (Barefoot, 2012), Tables 8.2 and 10.2. These are adjusted by the average effective corporate income tax rate in the respective country—taken from Hassett and Mathur (2011), PWC (2011), Chen and Mintz (2011) and Markle and Shackelford (2012)—to obtain estimates of taxable income. The average effective tax rate for global taxable income is weighted according to countries’ GDP. Country shares of U.S. MNEs’ sales, assets, compensation of employees and number of employees are obtained from the same tables, adding totals for U.S. parents and MOFAs in all countries. Shares of each apportionment key are applied to global taxable income to derive changes in taxable income.

Appendix Table 8 shows the country-specific estimates underlying Figure 4. Broadly, a country gains from FA on the basis of some factor if its share in the global total of that factor exceeds its share in the net income of US MNEs. That Italy, for instance, gains under all factors reflects the very low share of US MNEs net income reported there: about 0.16 percent. Whether that reflects inherently low profitability or particularly aggressive outward profit shifting cannot be determined from these data.

Appendix Table 8 shows the country-specific estimates underlying Figure 4.

Appendix Table 8.

Reallocation of Taxable Income from Alternative Factors, U.S. MNEs Percent of change

article image
Source: Staff calculations based on Bureau of Economic Analysis.


  • Abbas, S. M. Ali, and Alexander Klemm, 2013, “A partial race to the bottom: corporate tax developments in emerging and developing economies,” International Tax and Public Finance, Vol. 20 (August), pp. 596617.

    • Search Google Scholar
    • Export Citation
  • Action Aid, 2010, “Calling Time: Why SABMiller should stop dodging taxes in Africa,” November 28, 2010. Available via the Internet: https://www.actionaid.org.uk/sites/default/files/doc_lib/calling_time_on_tax_avoidance.pdf

    • Search Google Scholar
    • Export Citation
  • Altshuler, Rosanne, and Harry Grubert, 2009, “Formula Apportionment? Is it better than the current system and are there better alternatives?National Tax Journal, Vol. 63 (December), pp. 114584.

    • Search Google Scholar
    • Export Citation
  • Altshuler, Rosanne, and Harry Grubert, 2008, “Corporate Taxes in the World Economy,” in Fundamental Tax Reform: Issues, Choices, and Implications, ed. by John W. Diamond and George R. Zodrow (Cambridge, MIT Press).

    • Search Google Scholar
    • Export Citation
  • Altshuler, Rosanne, and Harry Grubert, 2006, “Governments and Multinational Corporations in the Race to the Bottom,” Tax Notes International, Vol. 41 (February), pp. 45974.

    • Search Google Scholar
    • Export Citation
  • Arena, Matteo, and Andrew Roper, 2010, “The Effect of Taxes on Multinational Debt Location,” Journal of Corporate Finance, Vol. 16 (December), pp. 63754.

    • Search Google Scholar
    • Export Citation
  • Arellano, Manuel and Stephen Bond, 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations,” The Review of Economic Studies, Vol. 58 (April), pp. 27797.

    • Search Google Scholar
    • Export Citation
  • Auerbach, Alan, and Michael Devereux, 2013, “Consumption and Cash Flow Taxes in an International Setting,” NBER Working Paper 19579 (Cambridge: National Bureau of Economic Research, Inc.).

    • Search Google Scholar
    • Export Citation
  • Auerbach, Alan, and Michael Devereux, and Helen Simpson, 2010, “Taxing Corporate Income” in Dimensions of Tax Design: the Mirrlees Review, ed. by James Mirrlees and other, (Oxford: Oxford University Press).

    • Search Google Scholar
    • Export Citation
  • Avi-Yonah, Reuven, and Ilan Benshalom, 2010, “Formulary Apportionment – Myths and Prospects,” Law & Economics Working Paper 28 (University of Michigan Law School).

    • Search Google Scholar
    • Export Citation
  • Avi-Yonah, Reuven, Kimberly A. Clausing, and Michael C. Durst, 2009, “Allocating Business Profits for Tax Purposes: A Proposal to Adopt a Formulary Profit Split,” Florida Tax Review, Vol. 9 (5), pp. 497553.

    • Search Google Scholar
    • Export Citation
  • Azémar, Céline and Andrew Delios, 2007, “The Tax Sparing Provision Influence: A Credit versus Exempt Investors Analysis” (Glasgow: University of Glasgow).

    • Search Google Scholar
    • Export Citation
  • Barefoot, B. Kevin, 2012, “U.S. Multinational Companies, Operations of U.S. Parents and their Foreign Affiliates in 2010,” Survey of Current Business, (November) pp. 5174.

    • Search Google Scholar
    • Export Citation
  • Barrios, Salvador, Harry Huizinga, Luc Laeven, and Gaetan Nicodème, 2012, “International Taxation and Multinational Firm Location Decisions,” Journal of Public Economics, Vol. 96 (December), pp. 94658.

    • Search Google Scholar
    • Export Citation
  • Barthel, Fabian, Matthias Busse, and Eric Neumayer, 2010, “The Impact of Double Taxation Treaties on Foreign Direct Investment: Evidence from large Dyadic Panel Data,” Contemporary Economic Policy, Vol. 28 (July), pp. 36677.

    • Search Google Scholar
    • Export Citation
  • Benedek, Dora, Ernesto Crivelli, Sanjeev Gupta, and Priscilla Muthoora, 2014, “Foreign aid and revenue: Still a crowding-out effect?FinanzArchiv: Public Finance Analysis, Vol. 70 (March), pp. 6796.

    • Search Google Scholar
    • Export Citation
  • Bennett, Mary, Michael Devereux, Judith Freedman, Martin Hearson, Chris Lenon, Glen Loutzenhiser, David McNair, William Morris, Richard Parry, Douglas A. Shackelford, Roberto Schatan, and Susan Symons, 2011, “Transparency in Reporting Financial Data by Multinational Corporations” (Oxford: Oxford University Center for Business Taxation). Available via the Internet: http://www.sbs.ox.ac.uk/sites/default/files/Business_Taxation/Docs/Publications/Reports/transparency-in-reporting-financial-data-by-multinational-corporations-july-2011.pdf

    • Search Google Scholar
    • Export Citation
  • Bettendorf, Leon, Michael Devereux, Albert van der Horst, Simon Loretz and Ruud A. de Mooij, Corporate Tax Harmonization in the EU, 2010, Economic Policy, Vol. 25, pp. 53790.

    • Search Google Scholar
    • Export Citation
  • Blonigen, Bruce, and Ronald Davies, 2005, “Do Bilateral Tax Treaties Promote Foreign Direct Investment?” in Handbook of International Trade, Economic and Legal Analyses of Trade Policy and Institutions, Volume II, ed. by Eun Kwan Choi and James Hartigan, (London, Blackwell).

    • Search Google Scholar
    • Export Citation
  • Blonigen, Bruce, and Ronald Davies, 2004, “The Effects of Bilateral Tax Treaties on U.S. FDI Activity,” International Tax and Public Finance, Vol. 11 (September), pp. 60122.

    • Search Google Scholar
    • Export Citation
  • Blonigen, Bruce, Lindsay Oldenski, and Nicholas Sly, 2011, “Separating the Opposing Effects of Bilateral Tax Treaties,” NBER Working Paper 17480 (Cambridge: National Bureau of Economic Research, Inc.).

    • Search Google Scholar
    • Export Citation
  • Blouin, Jennifer, Harry Huizinga, Luc Laeven, and Gaetan Nicodeme, 2014, “Thin Capitalization Rules and Multinational Firm Capital Structure”, IMF Working Paper no. 14/12 (Washington: International Monetary Fund). Available via the Internet: http://www.imf.org/external/pubs/ft/wp/2014/wp1412.pdf

    • Search Google Scholar
    • Export Citation
  • Blumenthal, Marsha, and Joel Slemrod, 1995, “The compliance cost of taxing foreign source income: its magnitude, determinants and policy implications,” International Tax and Public Finance, Vol. 2 (February), pp. 3753.

    • Search Google Scholar
    • Export Citation
  • Blundell, Richard, and Stephen Bond, 1998, “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models,” Journal of Econometrics, Vol. 87 (August), pp. 11543.

    • Search Google Scholar
    • Export Citation
  • Brennan, Geoffrey, and James M. Buchanan, 1980, The Power to Tax: Analytical Foundations of a Fiscal Constitution (Cambridge: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Burns, Lee, Honoré Le Leuch, and Emil Sunley, (Forthcoming in 2014), “Transfer of an interest in a mining or petroleum right,” in Resources without Borders, ed. by Philip Daniel, Michael Keen, Artur Swistak, and Victor Thuronyi (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chen, Duanjie and Jack Mintz, 2011, “New estimates of Effective Corporate Tax Rates on Business Investment,” CATO Institute Tax & Budget Bulletin, No. 64. Available via the Internet: http://www.cato.org/pubs/tbb/tbb_64.pdf

    • Search Google Scholar
    • Export Citation
  • Christian Aid,Death and Taxes: The True Toll of Tax Dodging, ” May 2008. Available via the Internet: http://christianaid.org.uk/images/deathandtaxes.pdf

    • Search Google Scholar
    • Export Citation
  • Clausing, Kimberly A., 2009, “Multinational Firm Tax Avoidance and Tax Policy,” National Tax Journal, Vol. 62 (December), pp. 70325.

    • Search Google Scholar
    • Export Citation
  • Clausing, Kimberly A., 2003, “Tax-Motivated Transfer Pricing and U.S. Intrafirm Trade Prices,” Journal of Public Economics, Vol. 87 (September), pp. 220723.

    • Search Google Scholar
    • Export Citation
  • Coase, Ronald H., 1937, “The Nature of the Firm,” Economica, Vol. 4 (November), pp. 386405.

  • Crivelli, Ernesto, and Sanjeev Gupta, 2014, “Resource Blessing, Revenue Curse? Domestic Revenue Effort in Resource-Rich Countries,” IMF Working Paper 14/5 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Davies, Ronald, Pehr-Johan Norback, and Ayca Tekin-Koru, 2009, “The Effect of Tax Treaties on Multinational Firms: New Evidence from Microdata,” The World Economy, Vol. 32 (January), pp. 77110.

    • Search Google Scholar
    • Export Citation
  • De Meza, David, Ben Lockwood, and Gareth Myles, 1994, “When are origin and destination regimes equivalent?International Tax and Public Finance, Vol. 1 (1), pp. 524.

    • Search Google Scholar
    • Export Citation
  • De Mooij, Ruud A., 2011, “The Tax Elasticity of Corporate Debt: A Synthesis of Size and Variations,” IMF Working Paper 11/95 (Washington: International Monetary Fund). Available via the Internet: http://www.imf.org/external/pubs/ft/wp/2011/wp1195.pdf

    • Search Google Scholar
    • Export Citation
  • De Mooij, Ruud A., 2012, “Tax Biases to Debt Finance: Assessing the Problem, Finding Solutions,” Fiscal Studies, Vol. 33 (December), pp 489512.

    • Search Google Scholar
    • Export Citation
  • De Mooij, Ruud A., and Sjef Ederveen, 2008, “Corporate Tax Elasticities: A Reader’s Guide to Empirical Findings,” Oxford Review of Economic Policy, Vol. 24 (4), pp. 68097.

    • Search Google Scholar
    • Export Citation
  • De Mooij, Ruud A., 2003, “Taxation and Foreign Direct Investment: A Synthesis of Empirical Research,” International Tax and Public Finance, Vol. 10 (November), pp. 67393.

    • Search Google Scholar
    • Export Citation
  • De Mooij, Ruud A., and Hendrik Vrijburg, 2010, “Enhanced Cooperation in an Asymmetric Model of Tax Competition,” Working Papers 102, Oxford University Centre for Business Taxation.

    • Search Google Scholar
    • Export Citation
  • Desai, Mihir, C. Fritz Foley, and James R. Hines Jr., 2006, “The Demand for Tax Haven Operations,” Journal of Public Economics, Vol. 90 (February), pp. 51331.

    • Search Google Scholar
    • Export Citation
  • Desai, Mihir A., and James R. Hines Jr., 2003, “Evaluating International Tax Reform,” National Tax Journal, Vol. 56 (September), pp. 40940.

    • Search Google Scholar
    • Export Citation
  • Dethier, Jean-Jacques, and Thierry Madies, 2010, “Fiscal Competition in Developing Countries: A Survey of the Theoretical and Empirical Literature,” World Bank Policy Research Working Paper 5311 (Washington: World Bank).

    • Search Google Scholar
    • Export Citation
  • Devereux, Michael and Rachel Griffith, 1998, “Taxes and the Location of Production: Evidence from a Panel of U.S. Multinationals,” Journal of Public Economics, Vol. 68 (June), pp. 33567.

    • Search Google Scholar
    • Export Citation
  • Devereux, Michael, Ben Lockwood, and Michaela Redoano, 2008, “Do Countries Compete Over Corporate Tax Rates?Journal of Public Economics, Vol. 92 (June), pp. 121035.

    • Search Google Scholar
    • Export Citation
  • Devereux, Michael, and Simon Loretz, 2008, “The Effects of EU Formula Apportionment on Corporate Tax Revenues,” Fiscal Studies, Vol. 29, pp. 133.

    • Search Google Scholar
    • Export Citation
  • Dharmapala, Dhammika