CESifo Economic Studies Advance Access originally published online on June 18, 2007
CESifo Economic Studies 2007 53(2):329-361; doi:10.1093/cesifo/ifm012
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Reforming the Taxation of Multijurisdictional Enterprises in Europe*
ARPEGE/FUCaM, Catholic University of Mons, Belgium and Cesifo, e-mail: gerard{at}fucam.ac.be
 |
Abstract |
|---|
 Top Abstract 1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
In 2001, the European Commission proposed replacing the current system of taxation of multinational companies by the taxation of a consolidated base, computed at the level of all the European entities of a multinational enterprise, and then distributed for taxation purposes among the various jurisdictions in which these entities operate, according to pre-established criteria. In this article, we propose a discussion, especially focusing on two related issues, the choice of the formula and the composition of the consolidating area—either the entire European Union (EU) or some Member States within an Enhanced Cooperation Agreement—, as well as on their impact on the size and distribution of tax revenue and economic activity, and on the intensity of tax competition. Our tentative policy conclusion is that the reform deserves support provided that (i) the formula puts emphasis on criteria that the firm may not too easily manipulate, (ii) the activities of the multijurisdictional enterprise are enough mobile, (iii) the consolidation is made compulsory within the consolidating area and (iv) the consolidating area protects its capacity to actually levy tax by adopting a crediting system vis-à-vis the rest of the world. (JEL code: H32, H73, H87)
Key Words: multinational enterprises • multinational companies • European taxation • tax consolidation • tax competition
|
1 Introduction |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
In our increasingly global economy, the taxation of multinational enterprises, in short MNE's, has become a key challenge for tax designers. They are confronted with a two-fold reality. On one hand, organization of the tax system has to secure the capacity for sovereign jurisdictions to levy taxes at rates and on bases determined by them, though possibly coordinated through a network of bilateral tax treaties or multilateral tax arrangements. On the other hand, more and more mobile firms, willing to settle branches or subsidiaries all over the world, complain against the complexity and diversity of tax systems, and simultaneously develop a capacity to reduce their tax liabilities through the more and more extensive use of sophisticated tax planning strategies. And indeed the present organization of international taxation, primarily based on a model proposed by the OECD (OECD, 1996) and thereafter called Separate Accounting or SA, provides the various jurisdictions with, apparently, the power to make sovereign decision on taxation, but, however, it has at least two undesirable outcomes: first, it forces the companies willing to operate in many jurisdictions to learn as many tax codes, and second, it allows those companies to undertake various tax shifting strategies in order to minimize their tax liabilities.
Those undesirable outcomes are especially present in the European Union (EU). The first one is even considered as a main tax obstacle to the operation of the Single Market (European Commission, 2001). The second one is, in the EU, often exacerbated by the principle that no EU Member State may be considered as a tax haven by other European jurisdictions provided its tax system applies in a non-discriminatory way to every EU taxpayer. The issue for EU tax designers is then, more than elsewhere, to find out a system that simultaneously removes the tax obstacle mentioned earlier and is compatible with the principle of subsidiarity—leave as much power as possible to national authorities—and the tax sovereignty of national parliaments.
Therefore, in the Fall of 2001, the European Commission (2001, 2003) suggested replacing the current system of taxation of MNE's, based on separate taxation of different national entities in a group, by the taxation of a consolidated base calculated at the level of all the European entities in a group according to a single set of rules, and then distributed for taxation purposes between the different jurisdictions in which these entities operate, according to pre-established criteria. In so doing, it proposed replacing a typical system of tax relations between sovereign states with a mechanism that is more characteristic of tax relations within a federation; such a system is, e.g. applied in the United States to tax companies operating in several States, and in Canada to tax a given company operating in more than a single province, thus without consolidation across companies—for lessons for Europe from the US and Canadian experiences see Hellerstein and McLure (2004), Weiner (2005) and Martens-Weiner (2006); on the US application of the system, see also Goolsbee and Maydew (2000).
This system, which has in the meantime been examined and discussed by experts and by the parties concerned,1 certainly has the great advantage, providing it is sufficiently widespread, of putting an end to a certain number of tax strategies which MNE's find it in their interest to practice. As shown in the seminal work of Gordon and Wilson (1986) and the studies motivated by the planned reform in Europe—see e.g. Sorensen (2004)—that reform could, however, and under some conditions, increase tax competition between States.
More specifically, one can show that the effect of this change on tax competition is ambiguous, with the intensification of tax competition being all the less (viz. more) probable if the formula adopted for the distribution of the consolidated taxable base between the jurisdictions concerned gives less (viz. more) emphasis to a criterion over which MNE's have control, such as the geographic distribution of investment, production or payroll, and more (viz. less) emphasis to a criterion over which those firms have no or little control, such as the distribution of demand, and thus of sales destinations—see e.g. Gérard (2005a, 2005b, 2006). Therefore, the selection of the formula is a key political decision.
The EU Commission also proposed, as an intermediate step in the way to consolidation and formulary apportionment, in short C&FA in the sequel of this article, to allow for international compensation of losses among companies operating in the EU and belonging to the same MNE—for an analysis see Gérard and Weiner (2003, 2006) and Weiner and Gérard (2004).
In this article, we propose a discussion of the move from separate accounting based taxation, SA, to consolidation and formulary apportionment, C&FA. We especially focus on two related issues, the choice of the formula and the composition of the consolidating area—either the entire EU or some Member States within an Enhanced Cooperation Agreement—, and on their impact on the size and distribution of tax revenue and economic activity, and on the intensity of tax competition.
Our tentative policy conclusion is that the reform deserves support provided that (i) the formula puts emphasis on criteria that the firm may not too easily manipulates, (ii) the activities of the MNE are enough mobile, (iii) the consolidation is made compulsory within the consolidating area, and (iv) the consolidating area protects its capacity to actually levy tax by adopting a crediting system, possibly extended to accrued capital gains through anti-CFC rules, vis-à-vis the rest of the world.
For the ease of the exposition, we will follow the behaviour of a multinational enterprise and three jurisdictions. All those players are involved in a six-step game which is formally presented in Appendix. Steps 1–3 characterize the part of the game played under the SA system, steps 4 and further, the part of the game played under the C&FA hypothesis—see Figure 1. The reader can easily relate those steps to the history and development of the EU.
|
The multinational firm plays at steps 3 and 6. It aims at maximizing its value. Therefore, it distributes a fixed amount of investment between the only two jurisdictions able to host real activity, like production, in order to satisfy a final demand, exclusively located in those two jurisdictions and whose size and distribution is fixed. The authorities of the jurisdictions play at steps 1, 2, 4 and 5. Those of the jurisdictions able to host real activity aim at maximizing the welfare of their residents deemed to depend on the amount of public goods, financed by tax revenue, on one hand, and on the amount of labour income generated by real economic activity produced locally by the MNE, on the other hand. The third jurisdiction hosts a financial centre and offers a relatively low tax rate; on its territory, there is neither any demand for the good produced by the MNE nor any possibility to conduct real economic activity. The three jurisdictions are located within the European Union and none can be considered by another as a tax haven. Depending on the economic environment, the MNE may decide to manage various forms of intra-firm trade and to channel financing and profits through the third jurisdiction, possibly using intermodal finance.
Stated otherwise, in step 1, governments observe the initial distribution of the activities of the MNE and that of the demand for its product, as well as the economic environment and the institutional arrangements, like the tax system. In step 2, they play non-cooperatively in order to determine tax rates which maximize the welfare of their own jurisdiction, anticipating correctly the behaviour of the MNE. In step 3, the MNE adapts to the new set of tax rates. Then, in step 4, all or some jurisdictions may decide for adopting C&FA and those who adopt that reform decide cooperatively on the formula to be used to apportion the common tax base—in that respect our hypothesis differs from Wellisch (2004); that difference is justified by the decision process in tax matters within the EU: unanimity prevails except when an Enhanced Cooperation Agreement is set up by some Member States. Next, in step 5, governments revise their tax rates, again non-cooperatively; we call that stage—maybe improperly—tax competition: actually it is a stage where jurisdictions revise their tax rates egoistically, in the sole best interest of their own residents, and for the sake of simplicity we assume that steps 4 and 5 occur simultaneously. In step 6, the MNE adapts its behaviour to the new setting.
Economic environment is primarily characterized by the degree of mobility of the MNE. We consider two such environments. In the first one, the MNE is deemed to be "one-degree mobile", by which is meant that it only decides on the distribution of its investment, which implies that of production and employment, taking what we call a "real" decision. Then only two jurisdictions are concerned. In that setting, if the distribution of investment chosen by the MNE is such that production in one jurisdiction exceeds demand in that jurisdiction, intra-firm trade is conducted using an exogenous at arm's length transfer price. This setting is investigated in section 2 subsequently.
In the second environment, the firm becomes "two-degree mobile" or even "n-degree mobile". A two-degree mobile MNE decides on two variables, the distribution of its real activities on one hand, and another variable, called a "paper" variable, on the other hand. That other variable may be e.g. the transfer price or the fraction of the investment and repatriated profit which is channelled through a third jurisdiction—on that last issue see also Mintz and Smart (2004). In any case, that second variable might be a source of "paper profits". Combining the distribution of real investment with more than one such additional variable, we produce an "n-degree mobile firm" and actually we will investigate up to a "three-degree mobile" MNE. That will be done in section 3. Then we will stress, in particular, that tax shifting strategies at the root of paper profits redistribute tax revenue between the jurisdictions though the distribution of real activity may remain unchanged.
The move from SA to C&FA is investigated in sections 4 and 5. In the former we assume that all the jurisdictions—either the two of section 2 or the three of section 3—join the consolidating area. Unlike that, in the latter, the third country remains outside that area. Then one can cope with more advanced issues. Let us mention the political economy issues related to decision taking mechanism in Europe, especially such questions like the adoption of the reform at unanimity by all the EU Member States versus its adoption by some within an Enhance Cooperation Agreement, are on the agenda—on that latter topic see also Bordignon and Busco (2006)—and the relation between the consolidating area and jurisdictions outside that area, either within or outside de EU.
Though C&FA does not make taxation neutral as regards decisions by a MNE (only complete harmonization of effective tax rates and systems could achieve this), it does, however, form part of the solution of eliminating tax obstacles to economic activity, notably because of its implications in terms of common rules on constituting the tax base and, upstream, on accounting (see, for instance, Jacobs et al., 2005). Moreover, it can easily be combined with subsidiarity, a principle that is at the heart of the whole organization of the EU. Also the introduction of a third country, possibly remaining outside the consolidation area, paves the way for thinking about the most efficient geographic area to consider MNE tax coordination. Those elements will be reviewed in concluding section 6.
In addition to the contributions already mentioned, papers on related topics are numerous. Let us mention Eggert and Schjelderup (2003), Nielsen, Raimondos-Moller and Schjelderup (2003), Pethig and Wagener (2003), Eichner and Runkel (2006) and Riedl and Runkel (2007), and in French, Gérard (2003).
|
2 "Real" decisions under separate accounting |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
In this section, the MNE is "one-degree mobile", it operates in a situation of separate accounting, adopting only "real" strategies, by which is meant strategies that imply changes in the location of real investment or activities, as opposed to decisions aimed at getting "paper profits". In that setting only two governments are at work, they first observe (step 1) the initial distribution of investment and the distribution of demand—all the investment and production is initially located in the home jurisdiction and demand is unevenly distributed between the two jurisdictions with a larger part in the home jurisdiction—, they know the economic environment and the institutional arrangement, SA. Then they choose the tax rates (step 2) and the MNE adjusts the distribution of its investment and production activities in such a way that provides it with the largest possible value (step 3).
In order to focus on the impact of taxation, let us assume that the two countries only differ in tax rates,2 initial endowment in production capacity and size of the market. And as it is standard in economics, let us resolve the game backward.
2.1 Decision of the multinational enterprise
Following a pattern made popular by Devereux and Griffith (1998), when the company realizes that it has a foreign market, two options are open to it: exporting to this market and establishing a local facility there. In line with that approach we assume that initially all the investment and production is located at home.
The facility located abroad at step 3, if any, will often first be a permanent establishment, and will then become a subsidiary. In the latter case, the parent company aiming at keeping the largest possible control over its affiliate, will finance it either by buying new shares issued by the subsidiary or making a loan to that company. In that facility, the MNE will either produce for one or for both markets, depending on its own best interest. In that latter case, it will have a distribution entity in the other jurisdiction, then conducting intra-firm trade between the production foreign subsidiarity and that entity, at regular transfer price.
We can decompose the decision to be taken by the MNE into four strategic decisions. The first one is, given the tax rates and tax systems applied in the two jurisdictions, to decide of a possible departure from the initial situation where all is produced at home. If the first decision is to depart, the second one is to decide what to set up abroad, a branch or a subsidiary. In case the MNE decides for a subsidiary, the next question is how to finance that subsidiary—either through buying new shares issued by the subsidiary or making a loan to that company—and, in the former case, how to repatriate the income—through dividends or through capital gains. Finally, a last issue is to decide on the size of the investment and production abroad: partly or totally, for the sole local market or partly or totally for both markets of the MNE.
Let us review how those decisions may be influenced by taxation, under SA. Therefore, three aspects of the tax system are relevant, the domestic tax system applying in each jurisdiction, the international tax arrangement regarding the taxation of income and the tax treatment of transfer prices. After that review we will come back to the MNE decision and the comparison between the tax gain and the cost incurred by relocating real economic activity.
The domestic tax systems
Each domestic tax system consists of a set of rules determining how to compute the domestic tax base, and a set of tax rates. The former includes a.o. the mechanism of depreciation allowances, the latter the corporate income tax rate. Combining both sets allows us to compute domestic effective tax rates. Comparison of those rates may for sure highlight the decision. However, that sole information is not sufficient and might even be misleading. Actually it is only sufficient if the choice is limited to producing at home versus in a permanent establishment abroad.3 If instead a subsidiary is set up abroad, international tax arrangements among the countries deserve further consideration.4
The international arrangements on the taxation of income
Those arrangements are ruled by the provisions of the tax treaty between the two countries, in line with the OECD model tax convention, and by EU directives as well. Two ways for the parent company financing its subsidiary abroad deserve consideration, buying new shares issued by the subsidiary or making a loan to that company. In the first case dividends are repatriated, an operation which can be possibly differed through the accumulation of reserves in the subsidiary, translated into accrued capital gains at the level of the parent. In the second one, interests are repatriated and, possibly, dividends too—for the excess of profits over the interest payments for those payments are limited by the obligation to be in line with the world interest rate.
The Directive of 23 July 1990 governing the circulation of dividends between parent companies in the EU first states that, under conditions that we assume to be satisfied, dividends cannot be subjected to a withholding tax in the country in which they are paid.5 Additionally the Directive provides Member States with two options. One is exemption: at most 5 percent of the cross border dividends can be taxed in the country of residence of the company receiving them (the parent company).6 This system shares with that applicable to foreign permanent establishments, an economic property, known as capital import neutrality: if the rate of additional taxation is zero, the value of the subsidiary is independent of the origin of the capital financing it. An immediate corollary is that in such a system, the location of subsidiaries is what is important, not that of the parent company. Consequently, tax competition between countries will focus on attracting subsidiaries.
The other option provided by the Directive (and outside the EU this is the practice in countries such as the US but also Australia, Canada, UK, New Zealand and many other) is crediting: the parent company shall be taxed on the group's global profits, but taxes levied outside the borders, within the EU, shall be credited to its tax liability up to the amount owed to its country of residence. As far as economic properties are concerned, this system may be capital export neutral: if the foreign tax rate does not exceed the rate in the parent company's country of residence, the group's value does not depend on the geographical distribution of its subsidiaries. An immediate corollary of this observation is that in such a system, the location of the parent company in the country of lower taxation is, all things being equal, likely to raise the value of the MNE. Consequently, tax competition will also aim at attracting parent companies or, where appropriate, intermediate holding companies. On the contrary—if the tax rate in the subsidiary's country exceeds that of the parent company—there may be capital import neutrality according to the definition in the preceding point.
If the investment in the subsidiary was financed by a loan from the parent company, the latter may receive interest that, in most tax systems, is deductible by the company that pays it and taxed in the case of the company receiving it. Two comments must be made, however. First, most countries apply a withholding tax on the payment of interest. As it is generally lower than the corporate tax rate and tax treaties’ provisions provide for its crediting, it can be ignored. In fact it will be ignored all the more readily since EU legislative developments provide for its disappearance within multinational groups. Second, financing by loans is limited by measures aimed at averting thin capitalization of companies and use of none at arm's length interest rate.
The tax treatment of transfer prices
When production in one country does not fit exactly local demand part of the local sales needs to be provided by the plant in the other jurisdiction. Assuming that the sales to local customers are operated by the local facility that completes the product before the final sale, such a situation implies intra-firm wholesale trade. Such a trade needs to be conducted at an arm's length price,7 the one that the firm should use for transactions with facilities outside the MNE. Practically, the transfer price divides the profit between the part to be taxed in the jurisdiction of the production facility and the part to be taxed in the jurisdiction of the distribution facility. For that purposes let us denote it by pw while c stand for the cost of production and p for the retail price – see Figure 2.
|
We will come back later on that issue since it is a key one for the MNE making "paper profits" possible. However, such strategic opportunities are ruled out by a move to C&FA.
The decision of the MNE again
Eventually the size of the transfer of real activity will be dictated by the equalization between its marginal gain, in terms of reduced tax liabilities, and its marginal cost, in terms of expenses related to moving the activity from one country to another; that latter includes costs related to dismantling part of the plant in the home country, building a new factory in the other jurisdiction, firing workers in one country and hiring workers in the other.
2.2 Decisions of the governments and consequences for the MNE
Observing at step 1, the governments decide, at step 2, on their respective corporate tax rates, the single variable they are permitted to alter in this section of the article. In taking their decisions, they are guided by the maximization of the welfare of their residents deemed to depend on the amount of public goods, financed by tax revenue and on the amount of labour income generated by real economic activity produced locally by the MNE.
Since the entire production is performed in the home jurisdiction and most of the demand is also located there, the foreign government is potentially poorer in terms of taxable income. Therefore, unsurprisingly, it will be more aggressive when engaging in tax competition, and it will set its tax rate at a smaller value than its competitor.
Consequently, the multinational enterprise will move its real activity to the foreign country up to the point where the marginal gain in terms of reduced tax liabilities will equal the marginal cost of moving. Moreover, the parent company will finance its subsidiary by buying new shares issued by that company to have its profits taxed abroad. Additionally if the home country is a crediting jurisdiction, in order to benefit from the reduced rate abroad, the MNE also needs to move the parent company to the foreign jurisdiction; then the decision as to finance the subsidiary by shares or by a loan becomes irrelevant.
In the sequel of the article however, since this is the case for most European countries, we assume that the home country is an exemption jurisdiction.
The foreign jurisdiction gets a double dividend: on one hand it obtains more real economic activity, a source of real private income for its population, possibly amplified by a multiplier effect, and on the other hand, it benefits from extra revenue enabling it to finance extra public services or infrastructure.
To sum up, the decision by the MNE of relocating the real activity abroad increases its value and impacts on the distribution of both tax revenue and private income between the jurisdictions. An immediate corollary emerges then: only the equalization of effective tax rates can ensure neutrality of taxation in terms of the location of subsidiaries and of parent companies, and hence in terms of the distribution of tax revenue among the countries concerned. Various combinations of tax parameters may obtain that equalization, but the simplest way of doing this is to use identical methods of composing the tax base and to equalize the statutory tax rates.
Most studies in the literature stop at this lesson—see Bénassy-Quéré, Fontagné and Lahrèche-Révil (2000, 2005), Grubert and Mutti (1991, 2000), de Mooij and Ederveen (2003)—, forgetting that MNE's often pursue more complex strategies and are established simultaneously in more than two countries—a contrario, Grubert (2004) and Gérard and Gillard (2004) explicitly consider tax planning strategies.8 In the next section, we will try to go beyond this limit by examining transfer pricing strategies, and then financial detour strategies.
At this stage, the reader interested by an analytical investigation will read section A.1.1 of the Appendix, and then come back to this point.
|
3 From "real" to "paper" profits in a system of separate accounting |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
Two ways at least are available to add "papers profits" to the value of the MNE obtained so far. In the first one, the MNE engages in internal transactions involving its entities and implying either internal trade at a price chosen by it for strategic reasons, which must then be justified, or management fees and related instruments, which are often easier to use. In terms of the game described in the introduction, we are again in step 3 but with a "two-degree mobile" firm.
In the second situation, the MNE replaces the direct investment and revenue flow circulation with an indirect circulation involving a third, low-tax, jurisdiction, but located in the territory of the EU, and intermodal finance, by which is meant e.g. that a flow of interests is turned into a flow of dividends within a passive entity located in the third jurisdiction.
3.1 Manipulation of transfer prices
The company—and let it supposed to produce abroad including for the home market—could then take the risk of distancing its internal transfer price from the arm's length price in order to boost its taxable profit in the jurisdiction with the lowest tax rate, in this case the foreign country. Suppose that it raises this price from an exogenous value pw to an endogenous one p* which minimizes the overall MNE tax liabilities though it keeps the tax bases positive in both jurisdictions—see Figure 3. In this case, actually the MNE takes two decisions in order to maximize its value: on one hand it decides on the distribution of its real activity, and on the other hand, it selects an optimal transfer price. The opportunity to take that second strategic decision allows the firm to save as much tax liabilities as previously but at a reduced cost: manipulating transfer price costs less than moving real activity. Consequently the MNE will presumably decide for a strategy combining such a manipulation with less change is the distribution of real investment and production.
|
The extra profit the MNE obtains in this way—and this will be the case through all this section—is pure "paper profit" since it arises from a change in the sole company's financial strategy without real investment behind it.
However, for the MNE, this gain is not without risk9 but that risk is limited for, instead of explicitly manipulating the transfer price, the MNE may substitute the use of various kinds of royalties and management or expertise fees.
For the foreign jurisdiction engaged in tax competition, the risk is to have a larger fraction of the tax base of the MNE located on its territory, but only a smaller fraction of the real activity of the enterprise. Undoubtedly that may impact on the intensity of the interjurisdictional competition.
3.2 A lucrative detour
We can now suppose that the MNE discovers that there is a jurisdiction within the EU that taxes corporate profits at a very modest rate, or which provides on its territory attractive tax holidays for financial entities. Let us say that this country does not constitute a market for the company's product or a place where it could produce it, but it will certainly host a passive facility owned by this group, say a financial centre—see Figure 4.10
|
The detour which the MNE then comes up with is as follows: rather than subscribe to new shares in its foreign production subsidiary, it will only subscribe in a limited amount of such shares—the minimum required for the subsidiary to be set up—but for the largest part of the investment, subscribe to shares in a passive subsidiary in the third country which, in turn, will lend the amount collected to the production subsidiary. As already mentioned, financing by loans is limited by measures aimed at averting thin capitalization of companies and we assume that we are not in this situation here. The profit in excess over interest payment can be directly channelled to the parent company as dividends compensating the very limited number of shares issued. Interest paid to the passive subsidiary is taxed there and the after-tax amount is either paid out as dividends to the parent or accumulated in the financial centre and possibly turned to loans to other subsidiaries of the MNE. A similar detour can be adopted for financing investment at home.
It turns out that this setting will impact on the intensity of tax competition between the two active jurisdictions: since the return on that competition is now largely deprived of its revenue part, those countries have less incentive to engage in such competition.
Two observations conclude that section. First, without making any new real investment, both the MNE and the passive facility's country obtain a substantial gain. Second, the geographic (re)distribution of the private and public components of the social welfare is changed differently: the country of the active facility is deprived of a large part of its tax revenue but not at all of its private income. Consequently, a distinction must be made between competition between jurisdictions to attract real investments (active facilities) and competition to only attract tax bases, including "paper profits".
That latter observation has a corollary of prime significance: the tax sacrifice to which a country consents in order to attract a real investment can be infinitely expanded simply by the existence of a jurisdiction that is attractive for tax bases.
Of course, these tax strategies would be irrelevant if the effective tax rates were identical across jurisdictions.
What happens to that outcome if SA is replaced by C&FA is the topic of the next two sections. Prior to reading them, the model-oriented reader can go to section A.2.1 of the Appendix.
|
4 Moving to consolidation and formulary apportionment |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
The changeover to consolidation and formulary apportionment discussed subsequently corresponds to step 4 of the game. Then governments of the active jurisdictions decide on tax rates at step 5 and MNE adapts at step 6. In this section, we assume that all the three jurisdictions consider entering the consolidation area, and we revise in that new setting the situations encountered in sections 2 and 3. In section 5, we will reconsider the case where only the two active countries join that area while the third jurisdiction remains outside, either because it decides not to participate or because the MNE has the option to keep the entity located in that jurisdiction outside the consolidation perimeter. The C&FA implies first moving from three to a single tax base, then apportioning that single tax base.
4.1 Consolidation
When consolidation is conducted, not only tax bases are summed up, but also all intra-firm flows are cancelled out.11 As a consequence, neither the value of the transfer price used for intra-firm trade nor the detour still have any effect on the consolidated tax base, and consolidation erases the interest of adopting strategies relating to the manipulation of transfer prices and detours via third units. However, as we will see later, this important argument in favour of consolidation does not hold unless consolidation is compulsory and adopted, at least, by all the Member States of the EU. If a facility, for instance the third country facility, remains outside the consolidation perimeter, then all the strategies remain possible. That remark is especially important for the case where some EU Member States should want to adopt Consolidation and Formulary Apportionment within the framework of an Enhanced Cooperation Agreement—this is not a reason per se to reject such an Agreement, but it shows the need to be careful about its consequences.
4.2 Formulary Apportionment
The formulary apportionment method is based on criteria that the company will find more or less difficult to control or manipulate. Let us consider and discuss four such criteria thereafter.
First, the investment criterion. Since investment decisions are in the company's power, it can be said that this is a criterion controlled by the MNE. If this criterion prevails, the company will continue to decide on the location of its investments on the basis of taxation. Not only will tax competition continue, but also, maybe, it will increase since the attraction of investments will again mean the attraction of a taxable base. The same occurs if the criterion is payroll, especially here where payroll is strictly proportional to investment; therefore, we will disregard that criterion in this article.
Next, the criterion of final sales—understood as sales to outside the company and in line with the principle of destination. It can be supposed that since the distribution of sales is partly or totally driven by that of demand, the company does not control this criterion or in any case controls it less. Then, for a jurisdiction, attracting the company to its territory has no implications for tax revenue provided that sales are not affected—of course, it is quite a different story if attracting investment stimulates local sales—and the effect on tax competition might be reduced.
A third rule consists of taking an average of investment and sales. Since payroll is strictly proportional to investment here, that criterion is equivalent to the average of payroll and sales used e.g. in Canada—on the properties of the Canadian formula, see Weiner (2005) and Martens-Weiner (2006).
Lastly, we can take a quite arbitrary criterion: since the company is established in three jurisdictions, each one of them has the right to tax one-third of the consolidated base. Alternatively we can use the distribution of the population or that of GDP within the area of operation of the MNE.
Following the model developed in the Appendix, we limit our discussion to two criteria, sales—not under the control of the enterprise—and investment—under the control of the MNE.
4.3 Back to the game
At step 6 of the game, the MNE will now only decide on the distribution of its real activity since transfer price and lucrative detour no longer matter. Then, at given tax rates the fraction of investment and production moved to the foreign country—remember that the foreign active country offers a lower tax rate than the home jurisdiction—will depend on the formula. If the formula emphasizes the distribution of sales, then the MNE is likely to have less incentive for moving than under SA. On the contrary, if the formula gives a high weight to the distribution of investment, the MNE is strongly encouraged to move to the active foreign country. Especially, since moving real activity to that country is now the only way for the MNE to reduce its tax liabilities, the fraction of investment located in that country will be larger than under section 3.
The attitude of the governments of the two active jurisdictions when determining non-cooperatively their tax rates at step 5, will be dictated by the formula too. If the formula is mainly based on the distribution of sales, the variable not controlled by the firm, the governments have less incentive to engage in tax competition: the distribution of the tax base, and that of tax revenue, is not influenced by the relative attraction of the territory for the real activity of the MNE—incidentally attracting real activity, if desirable, needs other instruments in that context. Unlike that, if the formula puts its emphasis on the distribution of investment, production or payroll, tax competition might be boosted: attracting investment will increase the local tax base, and there is no longer risk of the tax base to be shifted to the third jurisdiction. Therefore, it is possible to determine a no-race-to-the-bottom condition, i.e. a design of the formula such that tax competition will not be boosted; in the present case, it turns out to be a minimum value of the relative weight given to the variable not controlled by the firm.
We have assumed that the jurisdictions needed to decide cooperatively on the reform, and thus on the formula. This seems to us in line with the decision process in tax matters in the EU. When taking that decision, the three jurisdictions will have in view the following two issues at least. On one hand, if they adopt the reform, they will seek for the formula that maximizes their best join welfare. On the other hand, they will only adopt the reform if they are not worse off, together, after the reform.
Given the discussion above, and their objectives, the governments are most likely to decide for a formula that puts emphasis on the variable not controlled by the MNE, thus on sales rather than on investment or payroll. Notice incidentally, that the best interest of the MNE's is, on the contrary, to have a formula that puts emphasis on criteria that they control; therefore, presumably, they will lobby the governments for the adoption of a formula more in line with their best interest.
After selecting the formula the governments still have to check that the reform make them not worse off, together. Since the new tax system prevents tax shifting opportunities and gathers the tax bases in the two jurisdictions with the highest relative tax rates, the aggregate level of welfare will not be reduced by the reform, and actually it will be increased. Obviously, the distribution of the tax bases, and thus of tax revenue, will be altered: especially if the emphasis is on sales, the home jurisdiction will not only get back the base shifted to the third jurisdiction, but also will gain base and revenue w.r.t. the active foreign jurisdiction, while that latter will get back the base shifted to the third country but will lose base and revenue w.r.t. the home country; finally the third jurisdiction will have no longer base and revenue. The problem with the EU decision process in tax matters is to ensure that no government will veto the decision. Happily, if the formula is designed in such a way that the joint welfare of the three countries is maximized, there is enough gain, in terms of tax revenue, to allow for side payments such that not only the three countries are better off together, but also none is worse off individually.
To sum up, in relation to the MNE, the more weight given by the formula to a lower rate of taxation, the lower the tax liabilities to which the enterprise are subjected. Thus, recourse to the sole criterion of investment, giving precedence to the active country where the tax rate is low (the attractive country for real investment), leads to lower effective taxation of the MNE than the sole criterion of sales.
From the point of view of the jurisdictions, (i) the tax revenue is affected considerably by the formulary apportionment criteria used, while private consumption is not (as long as distribution of real activity is not revised)—generally speaking, looking at the situation with the detour via a third country, the changeover to C&FA produces two tax revenue winners and one tax revenue loser among the States concerned; (ii) but however, global tax revenue is higher so that it may be supposed that compensation between jurisdictions could be organized and consist of side payments likely to induce the third country to become involved if unanimity is required.
Sections A.1.2 and A.2.2 of the Appendix provide the interested reader with the equations supporting the arguments developed in that section.
|
5 Move to C&FA when the lowest tax jurisdiction remains outside |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
So far we have assumed that C&FA applied to the whole MNE. However, a facility, for instance the third country facility, may remain outside the consolidation perimeter. That is the case if consolidation is optional for the MNE, or if only some EU Member States adopt C&FA within the framework of an Enhanced Cooperation Agreement; this is also an important case for the relationship between the EU and the rest of the world.
In that case, the benefit from the reform, for the jurisdictions joining the consolidating area, is lost: tax shifting to the third countries remains possible, and is even more valuable for the MNE.
5.1 Active jurisdictions are worse off, the MNE is better off
Indeed that partial adoption of C&FA turns out to imply that (i) the third country has the same tax revenue as before the partial introduction of C&FA—since that amount is presumably the side payment it should have requested to adhere to a compulsory C&FA system, that country is deemed to be as well off—; (ii) aggregate tax revenue is smaller, so that tax revenue obtained by the countries where the active entities are located is small and in some case, even smaller than under SA combined with lucrative detour; therefore those two countries are worse off; (iii) since its aggregate tax liabilities are smaller, the MNE firm is better off.
5.2 Tax revenue protection is possible
However, there are replies that those two jurisdictions can set up in order to protect their tax revenue. First of all, the jurisdictions participating the C&FA area could decide to apply crediting instead of exemption vis-à-vis the other EU Member States and other foreign countries as well—notice that the US simultaneously apply C&FA within the federation and crediting vis-à-vis the rest of the world. The main consequence of that decision is that now the profit behind dividends paid out by the passive entity in the third country will be effectively taxed at the consolidation area tax rate. Thus any euro channelled through that jurisdiction and then repatriated to the home country is compelled to pay to the consolidating area tax authorities a larger amount of tax, what deters the MNE to use the detour.
Consequently, in case of adoption of C&FA within an Enhanced Cooperation Agreement, EU Member States concerned should be advised to simultaneously decide for applying crediting instead of exemption vis-à-vis other Member States and other foreign countries as well. Therefore, that case provides a useful guideline for an application of the reform by the entire EU as long as the EU is not unconnected with the rest of the world.
Moreover, that system could be, or should be, extended to profits not repatriated but, instead, accumulated into the third country passive entity, through a similar taxation of capital gains accrued in the parent company. That system looks like, but is larger than, anti-CFC rules.12
The reader will find the equations sustaining that argument in section A.2.3 of the Appendix.
|
6 Illustration and conclusion |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
The numerical illustration thereafter both illustrates and summarizes the discussion conducted in the article. It is based on the model developed in the Appendix and used to sustain the discussion. For the sake of simplicity, however, we assume that governments are only interested by revenue; otherwise the share of the home market is fixed to 0.6, the retail and wholesale prices are respectively 1 and 0.8, the discounting rate is 0.05 and the parameter
of the cost of moving function is set to 10. When permitted the detour concerns half of the profit. Based on those parameters, the tax rates computed by the model at step 2 amount to 0.5 in the home jurisdiction and 0.28 in the foreign jurisdiction; since the no-race-to-the-bottom condition is deemed to hold those rates are not revised in step 5. At step 4, the chosen formula is only based on the distribution of sales; however, using the same tax rates, the key parameters are computed for two alternative formulas: one solely based on the distribution of real investment, the other one based on the Canadian formula—equal weight for payroll (investment here) and sales. We both consider the case where all the jurisdictions join the consolidating area (lines C&FA-all) and that where only the entities in the active jurisdictions are consolidated (lines C&FA-active). The table especially shows the sensitivity of the size and distribution of tax revenue to the formula—compare lines 3–5—, the interest to have all the entities and jurisdictions in the consolidating area—compare those lines with the last three—and the divergence between the best interest of the governments and that of the MNE. Notice that if a base protection measure is introduced, the detour is no longer profitable, and thus deemed to be no longer used.
To sum up, in this article we have investigated a move from Separate Accounting to Consolidation and Formulary apportionment, with especially in mind the reform suggested by the EU Commission in 2001. In that exercise we have focused our investigation on two related issues, the choice of the formula and the composition of the consolidating area—either the entire EU or some Member States within an Enhanced Cooperation Agreement—, and on their impact on the size and interjurisdictional distribution of tax revenue and social welfare, and on the intensity of tax competition.
The case of an Enhanced Cooperation Agreement especially deserves interest since it not only enables to investigate a possible device within the EU but also what can happen for the relationship between an EU-wide consolidating area and the rest of the world.
Our tentative policy conclusion is that the reform might be supported provided that (i) the formula puts emphasis on criteria that the firm may not too easily manipulate, (ii) the real activities of the MNE are enough mobile, (iii) the consolidation is made compulsory within the consolidating area and (iv) the consolidating area protects its capacity to actually levy tax by adopting a crediting system, possibly extended to accrued capital gains, vis-à-vis the rest of the world. That recommendation is valid even if the entire EU adopts the reform, as long as the EU is not unconnected with the rest of the world.
This study also paves the way for further investigation. Three directions for further research seem to be especially relevant, one is the analysis of the decision process within a bottom up Federation like the EU, including that of alternative decision rules and coalition formations, another is the determination of the optimal area for tax policy purposes, and a third one is certainly the building process of a EU-like Federation per se.
|
Appendix: Formal analysis |
|---|
|
TopAbstract1 Introduction2 "Real" decisions under...3 From "real" to...4 Moving to consolidation...5 Move to C&FA...6 Illustration and conclusionAppendix: Formal analysisReferences |
|---|
In this appendix, we provide the formal analysis which supports the argument developed along sections 2–5. That appendix is a simplification of Gérard (2006) where a more general model is proposed and a detailed presentation and discussion of the results can be found. We consider first a one-degree mobile MNE, then a more mobile one.
A.1 A one-degree mobile MNE
The MNE distributes a given real unitary investment as a fraction 
in its active entity in jurisdiction h and a fraction 1 – in its active entity in jurisdiction f; h and f designate both the jurisdiction and the entity of the MNE located in the jurisdiction, and we assume that jurisdiction h is that of the parent company where initially the entire investment is located and the whole production performed. One unit of investment produces annually one unit of a consumption good. The MNE has to serve a fraction q 
0.5 of that unit to final consumers of market h and a fraction 1-q to the other market, in both cases at net retail price p. The difference between and q is traded within the firm at arm's length net wholesale price pw < p. The two jurisdictions are identical except for the demand for goods, the initial distribution of investment and production –0 = 1 > q– and possibly for tax rates.
We disregard the case where the affiliate is a branch so that f is a subsidiary; its profits are first taxed locally, then they are repatriated to the parent company under the form of dividends and the exemption principle fully applies. Accordingly, under SA, we define profit in a given jurisdiction as the value of sales to final consumers (net price multiplied by the amount of goods) minus the cost of acquiring goods from the other entity or plus the value of the goods sold to the other entity. This means that we require that sales to final consumers be made through the local entity.
Under C&FA, a single consolidated tax base is computed for the two entities together, which is then distributed between the two jurisdictions in line with a predetermined formula, here a linear combination of the distribution of investments and sales, measured on a destination basis.
For the ease of the presentation, we assume that there are no depreciation allowances and that the time span can be approximated by infinity. Also, there is no room for change in the total amount of investment, but only for a substitution effect between alternative locations of fractions of a total amount of investment deemed to be equal to unity. Finally we assume that shareholders non resident of either h or f own the MNE.
A.1.1 Separate accounting
Tax authorities of the two jurisdictions have observed the initial distribution of investment and demand (sales)—step 1 of the game. Then they set their respective tax rates, 
h and f, in order to maximize their own individual welfare—step 2; since then h > f, we call jurisdictions h and f the high tax and low tax jurisdictions, respectively. Finally the MNE revises the distribution of its investment in line with the incentive provided by the jurisdictions—step 3. Let us now solve that game backward.
The MNE under SA
The MNE maximizes its long run value with respect to the variable under its control,
|
| (1) |
hBh and fBf are tax liabilities in jurisdictions h and f, respectively, and c() the cost of departing the distribution of real investment from its initial distribution.
Furthermore, we define the tax bases as
|
| (2) |
0 = 1
|
| (3) |
that the equilibrium value of that variable is
|
| (4) |
The jurisdictions under SA
Each government maximizes a Social Welfare Function—step 2—defined on the welfare of its own residents deemed to depend on the effect of the investment on local employment, say w/r in jurisdiction h where w is the shadow price of hiring a worker (Boadway and Bruce, 1984), and on the amount of public goods available to the residents, uhBh in jurisdiction h, u being the shadow price of public goods - u > 1 -, thus, for jurisdiction h,
|
| (5) |
h.
The first order condition of that maximization implies a reaction function
|
| (6) |
h < 1.
A.1.2 Consolidation and formulary apportionment
If we move to C&FA, a consolidated tax base B is first calculated, using Equation (2) and cancelling out all intra MNE flows, which is then distributed, or apportioned, between the two jurisdictions using a given formula, cooperatively determined by the two jurisdictions at step 4. Simultaneously jurisdictions determine non-cooperatively their tax rates under the new setting—step 5—possibly entering a new round of tax competition. MNE responds in step 6.
We suppose here that the formula is a linear combination of real investment and receipts from final sales, with weights 
and 1 – , respectively. Since wage cost is strictly proportional to investment in this model, we disregard that criterion but the reader can also see our formula as a combination of sales (destination) and wage costs. What is important for the purposes of this investigation is that the distribution of one criterion, here real investment, is under control of the firm, and the distribution of the other is not. It turns out than we now have a consolidated tax base as well as two local tax bases
|
| (7) |
The MNE under C&FA
Now, substituting the last two lines of (7) for (2) into (1), (4) is replaced by
|
| (8) |
p and pw will command the relative effects of the tax change on the distribution of real investment: at unchanged tax rates, investment in jurisdiction h is smaller (viz. larger) under C&FA if p > pw (viz p < pw).
Let us add that, at given tax rates, and beyond the argument of saving in terms of knowledge of (too) many tax systems, the MNE can gain from the tax change if 
V = VFA – Vq 0. More precisely, it can be shown that, a one degree mobile MNE is more likely to lobby for apportionment be based exclusively on the distribution of investment, the variable under its control.
The jurisdictions under C&FA, tax competition again?
Does the reform attenuate or boost tax competition, or, more precisely, if the governments revise their tax rates in the sole best interest of their residents, will we observe higher or lower rates than under SA? To answer that question, imagine that, though they have selected the apportionment formula cooperatively, jurisdictions keep the right to simultaneously engage in competition on the tax rates—step 5. In that case, we derive new reaction functions, like
|
| (9) |
|
| (10) |
p.
Therefore, inequality (10), and more specifically pw > p, defines a no race-to-the-bottom condition. Especially, if the formula is such that the distribution of the consolidated tax base does not depend at all on that of investment and production, = 0, the reform will generate no incentive for tax competition. On the contrary, if = 1 then tax competition is boosted for sure: the distribution of the tax base depends entirely on a decision variable of the MNE. We will see that, fortunately, the no-race-to-the-bottom condition is compatible with the optimal expression for the apportionment formula.
The jurisdictions under C&FA, deciding on the reform and selecting the formula
The determination of the formula is thus of primary importance and we assume that it is decided cooperatively at the same time as the jurisdictions decide whether or not to adopt the reform—step 4. Therefore, the two jurisdictions jointly maximize W = Wh + Wf with respect to . From the first order condition of that maximization, and using Equation (9) as well as its counterpart for the other jurisdiction, the equilibrium value of the formula is such that
|
| (11) |
1/2), the equilibrium value of the formula is characterized by = 0: the formula only depends on the distribution of demand and sales, a parameter which escapes the control of the MNE. Therefore, the equilibrium value of the formula is compatible with the no-race-to-the-bottom condition.
A last issue is still unresolved however. The decision to undertake the reform needs that the jurisdictions together are made better off or at least not worse off by the reform. The two jurisdictions together are not worse off if WFA WSA, or, for = 0
|
| (12) |
|
| (13) |
0 = 1 > q > 1/2 and that h – f > 0. Then condition (13) will be satisfied and the two jurisdictions will gain together from the reform if the real investment of the MNE is "enough mobile" by which we mean that in response to the observation that f < h the MNE is able to move from 0 > q to < q. Since q can be as small as 1/2, that involves that, under SA, at least one half of the investment and production is located in jurisdiction f. That the two jurisdictions are not worse off together does not mean that no jurisdiction, individually, is worse off. However, the overall welfare gain enables the jurisdiction which gains to compensate that which loses through a side payment. It turns out that the reform can be adopted by the two jurisdictions unanimously provided that a side payment is possible from the higher taxing jurisdiction to the lower taxing one.
Now we see if those results hold when more sophisticated strategies are permitted to the MNE.
A.2 A two-degree mobile MNE
This sub-section completes sections 3 and 5. However, we limit our analysis to the profitable detour described in section 3.2. That strategy will also allow us to compare the adoption of the reform by the three concerned jurisdictions versus by only two of them.
A.2.1 Separate accounting
We denote by k the third and passive jurisdiction, where k < min(h, f) and which is only used for the purposes of the taxation of income, through a lucrative detour and intermodal financing. Profits of both active entities are channelled to jurisdiction k to be taxed there, up to a fraction 
due to the upper limit imposed by the law and the necessity of avoiding thin capitalization. Using intermodal financing and lucrative detour has a cost; however, to avoid unnecessary complication in the exposition, we suppose that cost being either zero, for 
, or infinity, for 
; indeed, beyond the threshold 
, interests are considered as hidden dividends and no longer allowed to be deductible against the tax base in the paying jurisdiction so that the cost of the detour becomes infinite (the detour is no longer interesting).
The MNE under SA
The tax bases in the three jurisdictions now respectively are
|
| (14) |
|
| (15) |
The jurisdictions under SA
At step 2, individual jurisdictions decide on tax rates in order to maximize the social welfare of their residents. Since the selected tax rates must be relevant for the setting investigated, an incentive compatibility constraint needs to be introduced, i.e. that the value of the MNE is not reduced by the opportunity to use a detour. Let µ be the Lagrange multiplier related to that constraint, then the reaction functions are of the type,
|
| (16) |
Two cases deserve a particular interest. First, the unconstrained case, µ = 0. Then, however, the incentive compatibility constraint is not satisfied and there is no guarantee that 
. Second, the constrained case: the tax rates satisfy those constraints—the value of the MNE is not reduced, the tax rates are smaller than one, the Lagrange multiplier is positive—and might be justified on economic grounds—see Gérard (2006) for details. Therefore, we keep the tax rates unchanged with respect to the case investigated in section A.1.1, which means not higher.
A.2.2 C&FA is applied by the three jurisdictions—FA3
The first and main implication of the adoption of C&FA by all the jurisdictions is that Bk vanishes since the inflow in the entity located in that jurisdiction is cancelled by the corresponding outflow. Then c = 0 under C&FA and the distribution of investment between the two active jurisdictions is again given by (8). This implies that the extension of the consolidation perimeter makes the firm more sensitive to the tax differential between the two jurisdictions that can host real investment, h and f.
Three questions then arise. First, is the no-race-to-the-bottom condition affected by the new setting? In the unconstrained case (µ = 0), the answer is positive. Compared with its counterpart in A.1.2, the condition for that result to hold is more demanding but still compatible with = 0. In the constrained case 
, that condition is unchanged, remaining < pw/p satisfied when = 0. Second, is the equilibrium value of the apportionment formula still = 0? The answer is positive too so that the no-race-to-the-bottom is compatible with the equilibrium value of the apportionment formula. Third, does the reform involve a welfare gain for the three jurisdictions together, and thus does it generate the capacity to compensate the jurisdiction k for the disappearance of its tax revenue? In the constrained case—the most relevant one—, that condition—cfr (12) above—holds a fortiori being now
|
| (17) |
c > 0.
Notice that the tax incentive for the MNE to move its investment under SA is smaller than in the case of a one-degree mobile firm so that q – c is less likely to be negative—in other words "enough mobile" is a stronger assumption; but the last term in parentheses in the right hand side of the inequality is positive since the average tax rates over the active jurisdictions exceeds that of the