eonstor www.eonstor.eu Der Open-Aess-Publikationsserver der ZBW Leibniz-Informationszentrum Wirtshaft The Open Aess Publiation Server of the ZBW Leibniz Information Centre for Eonomis Gérard, Marel; Prinen, Savina Working Paper Investment and finaning strategy of a multinational enterprise under alternative tax designs CESifo Working Paper: Publi Finane, No. 3838 Provided in Cooperation with: Ifo Institute Leibniz Institute for Eonomi Researh at the University of Munih Suggested Citation: Gérard, Marel; Prinen, Savina (2012) : Investment and finaning strategy of a multinational enterprise under alternative tax designs, CESifo Working Paper: Publi Finane, No. 3838 This Version is available at: http://hdl.handle.net/10419/61052 Standard-Nutzungsbedingungen: Die Dokumente auf EonStor dürfen zu eigenen wissenshaftlihen Zweken und zum Privatgebrauh gespeihert und kopiert werden. Sie dürfen die Dokumente niht für öffentlihe oder kommerzielle Zweke vervielfältigen, öffentlih ausstellen, öffentlih zugänglih mahen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweihend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrehte. Terms of use: Douments in EonStor may be saved and opied for your personal and sholarly purposes. You are not to opy douments for publi or ommerial purposes, to exhibit the douments publily, to make them publily available on the internet, or to distribute or otherwise use the douments in publi. If the douments have been made available under an Open Content Liene (espeially Creative Commons Lienes), you may exerise further usage rights as speified in the indiated liene. zbw Leibniz-Informationszentrum Wirtshaft Leibniz Information Centre for Eonomis
Investment and Finaning Strategy of a Multinational Enterprise under Alternative Tax Designs Marel Gérard Savina Prinen CESIFO WORKING PAPER NO. 3838 CATEGORY 1: PUBLIC FINANCE JUNE 2012 PRESENTED AT CESIFO AREA CONFERENCE ON PUBLIC SECTOR ECONOMICS, APRIL 2012 An eletroni version of the paper may be downloaded from the SSRN website: www.ssrn.om from the RePE website: www.repe.org from the CESifo website: Twww.CESifo-group.org/wpT
CESifo Working Paper No. 3838 Investment and Finaning Strategy of a Multinational Enterprise under Alternative Tax Designs Abstrat This paper investigates the onsequenes of a series of alternative international tax designs on the strategy of a multinational enterprise regarding the ross border distribution of its investment and the hoie of its finaning behavior. We start with a world where no international tax rules are at work. Then we suessively introdue (i) the rules provided by the OECD Model Tax Convention, (ii) the EU Parent-Subsidiary Diretive of July 23, 1990; and (iii) a ombination of Allowane for Corporate Equity (ACE) and Comprehensive Business Inome Tax (CBIT). Finally, we leave systems based on Separate Aounting (SA) aside and turn to Consolidation and Formulary Apportionment (C&FA) adopted either by all the jurisditions at work in the model, or by a sole subset of them within the framework of an Enhaned Cooperation Agreement (ECA). JEL-Code: F230, H250, K340. Keywords: orporate tax, multinational firms, MNE. Marel Gérard Louvain Shool of Management / UCL Louvain-la-Neuve / Belgium marel.gerard@ulouvain.be Savina Prinen Louvain Shool of Management / UCL Louvain-la-Neuve / Belgium savina.prinen@ulouvain.be This paper is part of the PhD dissertation of Savina Prinen written under the supervision of Marel Gérard. We are indebted to Gaetan Niodeme, Joann Weiner, Edoardo Traversa and other members of Savina s PhD ommittee, as well partiipants to various onferenes where we presented hat paper, for valuable omments and suggestions. Savina Prinen is espeially indebted to the Interuniversity College of Management, ICM, for its ontinuous support during the preparation of her PhD.
1 Introdution and Motivation As shown by the orporate finane literature, orporate taxation distorts a ompany s finanial deision making in a twofold way. First, the unequal tax treatment of debt and equity, applied by most tax systems, influenes a ompany s apital struture. Based on Modigliani and Miller (1958,1963) s work, Stiglitz (1973) and King (1974) formalized the inidene of the tax disrimination between debt and equity on the ost of apital and the value of a single firm. Their work triggered an important number of empirial studies quantifying the impat of the distortions due to this tax disrimination (a.o. Kaplan (1989), Fama and Frenh (1998), Desai et (2004)). Seond, the orporate tax differenes between ountries impat the funtioning of multinational enterprises (MNEs). A ompany ative in a multinational setting faes a wide range of tax regulations, haraterized by diversity in the definition of the tax base and in tax rates. The existene of as many tax odes as ountries, therefore, is at the root of many strategi behaviors of MNEs. International tax divergenes enhane ross-border tax arbitrage and as a result impat the level of foreign diret investment (a.o. Hartman (1985), Weihenrieder (1996)), the hoie of legal form (a.o. De Mooij and Niodème (2008)) and the loation deision (a.o. Hines (1996), Devereux and Griffi th (1998)) of ompanies regarding real investment as well as taxable inome. A MNE, however, is also onfronted with a lot of barriers impeding the development of ross-border ativities. One onsiderable hurdle hampering international business is the administrative burden related to the diversity of national tax odes. Another major obstale is double (or even multiple) taxation, i.e. the double (or multiple) taxation of the same multinational regarding the same revenue during the same time period. In order to takle these distortions, several institutional bodies, aiming at the development of international trade, tend to eliminate barriers to ross-border ativities by providing a set of rules, guidelines and suggestions. For MNEs taxed in the European Union, two institutional bodies have played a major role in shaping international rules for tax purposes. A first institutional body, is the Organisation for Eonomi Cooperation and Development (OECD), regrouping the governments of around thirty ountries. It suggested in 1958 a "Model Tax Convention" as framework for the negotiation of international tax treaties. Those tax treaties, onluded between two ountries, aim at avoiding double taxation of inome or apital by providing a tax relief system. A seond institutional body is the European Commission, aiming at the ahievement of the Single Market. In the past it issued two diret tax Diretives, abolishing withholding taxes on dividend, interest and royalty payments between assoiated ompanies of different Member States (Parent-Subsidiary Diretive (90/435/EEC) and Interests and Royalties Diretive (2003/49/EC)). In order to take hold of the administrative burden related to international taxation, the European Commission introdued in 2011 a Diretive proposal to adopt a Common Consolidated Corporate Tax Base (CCCTB), taxing a MNE on its onsolidated tax base instead of taxing eah entity separately (EU Commission (2011)). The topi of profit-shifting and the strategi behavior of MNEs was addressed 2
in several studies (Allingham (1972), Grubert (1991), Hines (1994), Mintz and Smart (2004)). Moreover, various authors theoretially studied how the tax environment determines the behavior of the firm. In partiular, the shift from separate taxation to onsolidated taxation has been analyzed extensively (a.o. MLure (1980), Weiner (1994), Mintz (2000)). Gérard (2007) and Gérard (2010) showed how the use of anti-abuse measures an ounter the adverse results obtained if only a subgroup of ountries shift to onsolidated taxation. Three features differentiate this paper from previous researh. First, we simultaneously investigate a real variable, i.e. the international distribution of investment, and a finanial variable, i.e. the amount of internal debt. We determine how eah tax design alters the investment and finaning deisions of a MNE and espeially fous on how it reates inentives to use internal debt. This is the main originality of this paper. Another feature that differentiates this study from the previous ones is that we follow the development of MNE taxation from the issuing of the OECD Model Tax Convention (1958) to the EU Diretive proposal regarding a Common Consolidated Corporate Tax Base (2011). Hene, we fous on those tax designs related to the evolution of MNE taxation in the European Union. Finally, we analyze alternative traks for MNE taxation and determine how they would alter a MNE s tax strategies. As suh, we ontribute to the evaluation of the existing tax environments, by omparing them with alternative tax environments serving the same objetives. At a moment where a proposal for a European ommon tax approah issued, this study is partiularly relevant. It would not be surprising to see the proposal rejeted by some EU Member States. Our analysis leads to several results. First, the differenes between the initial and optimal levels of investment and internal debt, illustrate the large set of profit-shifting opportunities, whih various tax environments offer to MNEs. Seond, we find that alternative environments, like a ombined ACE-CBIT system, provide relevant results as it redues profit-shifting strategies at best. In setion 2, we present the framework of our study and we model a world where no international tax rules are at work. We onsider both the ase where the profits are reported in the ountry in whih they are generated and the ase where a lurative detour is used. In setion 3, we analyze a MNE s profitshifting behavior under separate taxation. We start with the rules provided by the OECD Model Tax Convention and we then suessively introdue the rules provided by the EU diret tax Diretives related to MNE taxation. In setion 4, we leave systems based on separate taxation aside and turn to MNE taxation based on a onsolidated tax base. We analyze the setting where this type of taxation is adopted by all EU Member States, as well as the setting where it is adopted by a sole subset of them (Enhaned Cooperation Agreement). Setion 5 onludes. 3
2 The Model In order to analyze the impat of the tax environment on a MNE s profit-shifting behavior, a theoretial model is developed. In that model, we progressively inorporate the international tax rules proper to eah of the tax environments. First, we present the framework against whih the theoretial model is developed. Then, we model the tax situation of a MNE in an institutional environment, free of any international tax rules and subjet to double (or multiple) taxation of the same inome. 2.1 Framework In the model, we onsider a MNE whih is present in three ountries p, i and j. Countries p and i have the resoures to host a prodution ativity and they have the onsumers to host an eonomi market on whih the produts of the MNE an be distributed; ountry j does not. Country p hosts the parent ompany of the MNE (whih is a prodution and ommerial ompany), ountry i a fully-owned ommerial subsidiary and ountry j a servie subsidiary. Additionally, we suppose that the size of the ountry is expressed aording to the size of the eonomi market. As we assume that the fration of sales of the MNE in ountry p, denoted by q, exeeds that in ountry i, denoted by (1 q) (i.e. q > 1 q), ountry p is larger than ountry i, whih in turn is larger than ountry j. We assume that this relation also holds for the fration of real investment, i.e. real investment in ountry p (α) exeeds that in ountry i (1 α) (i.e. α > 1 α). Total sales and total investment amount eah to unity. Moreover, we onsider that the distribution of investment α is ontrolled by the MNE, but that the distribution of sales q is given. In other words, apital is mobile but final demand is not. Fousing on those variables we deliberately seem to leave aside other important aspets like transfer priing. Transfer priing is, however, as muh as the internal debt, an illustration of profit-shifting ativities and paper profit generation. Finally, we require that both q and α are between 0 and 1. For the purposes of the model, we assume that p r is the retail prie (disounted on an infinite horizon), whih is exogenously determined by the final market and obtained by selling the produt on that market. Moreover, p w is the wholesale prie, also exogenously determined by the wholesale market and paid by one entity of the MNE to another entity for aquiring its prodution. Suh intra-mne trade ours beause we assume sales to be performed by the loal entity, either the subsidiary loated in i or the parent ompany loated in p. Regarding the funding of the MNE, suppose the subsidiary in ountry j is entirely finaned through shares. The subsidiary in ountry i is funded partially through a loan granted by the parent ompany and partially through shares. Hene, an amount x represents the present value of interest payments (disounted over an infinite horizon) 1 made by the subsidiary to its parent om- 1 In this model, nothing opposes x to be negative; that orresponds to a situation where 4
pany and orresponds to the value of the internal debt. Interest payments are a dedutible expense for the paying entity but a taxable item for the reeiving one. Two types of taxes are at work in the model, the orporate inome tax (τ) and withholding taxes (w). Regarding the first type of taxes, i.e. the orporate inome tax, we assume that the profits generated by a subsidiary are subjet to the orporate tax of its host ountry i or j, and possibly to the orporate tax of the parent s ountry (MNE s home ountry). The seond type of taxes, withholding taxes, are levied at soure on interest and dividend payments. Sine we assume that eah subsidiary distributes its entire after-tax profits as dividend to its parent, the host ountry of the subsidiary levies a withholding tax w on this dividend. In our model, we designate those withholding taxes as wi d and wj d, where the supersripts refer to the type of inome (interest i or dividend d) and the subsripts to the ountries (i or j). Furthermore, if one of the entities needs to pay interest to another entity, the ountry of the paying entity levies a withholding tax wi i or wi j on the interest paid. We do not expliitly investigate the determination of the tax rates by the respetive ountries at stake. The relative values of the tax rates stem from the assumptions that we issue regarding the relative size of the ountries. Consistent with the theory that large ountries are less subjet to tax ompetition than small ountries, we assume that orporate tax rates τ are given and that they inrease with the size of the ountry (0 τ j < τ i < τ p < 1) 2. Finally, we suppose that the orporate tax rates τ are onsiderably higher than the withholding tax rates w, an assumption in line with most frequent observations. 2.2 Absene of International Tax Rules In a first stage, we onsider the taxation of a MNE in an institutional environment without any international tax rule. As suh, the MNE is subjet to the tax rules of all ountries in whih it has eonomi ativities. Hene, both the host ountry of the subsidiary and the home ountry of the MNE laim the right to tax the profits generated by the subsidiary and paid out to the parent ompany as dividend. Consequently, we assume maximal taxation and suppose that eah ross-border inome is taxed three times: τ host, w host and τ home. We first onsider a ase without a lurative detour, then we introdue the profit-shifting strategy of making a lurative detour through ountry j. 2.2.1 Without lurative detour Suppose that the MNE is not engaged in profit-shifting ativities and reports the profits where they are generated. Hene, only ountries p and i ollet taxes the entity makes a loan to its parent for profit-shifting purposes. 2 This is in line with the literature on tax ompetition and might be shown with a simple model. That result has been established a.o. by Kanbur (1993) for ommodity taxation and extensively used sine then. 5
from the MNE, as we assumed that ountry j does not have the resoures to host an eonomi ativity. Assume that B p is the pre-tax profit of the MNE generated in ountry p. It onsists of the revenues of selling q to onsumers in ountry p, the revenues of selling prodution (α q) to the group entity in ountry i, and the amount of interest x reeived from the entity in ountry i. B p = p r q + p w (α q) + x (1) Similarly, assume that B i is the pre-tax profit of the MNE generated in ountry i. It onsists of the sales (1 q) to onsumers in ountry i from whih the ost of aquiring a fration of the parent s prodution, as well as the interest payments x are subtrated. B i = p r (1 q) p w (α q) x (2) As no international tax relief rules exist, we assume that ross-border inome is subjet to triple taxation. First, the host ountry i levies orporate inome tax (τ i ) on the subsidiary s profits. Then, as the subsidiary in ountry i pays out interests to its parent ompany, withholding tax wi i is levied by ountry i on those interest payments x. The subsidiary in ountry i distributing its entire after-tax profit as dividend to its parent, a seond tax applies, i.e. a withholding tax wi d levied by ountry i on this dividend. From his side, the home ountry of the MNE laims the right to tax the profit generated by the subsidiary and levies orporate inome tax τ p on both benefits, B i and B p. Hene, B i is taxed for a third time. Moreover, adjusting its investment and internal debt levels triggers some additional osts for the MNE. Aording to the existing literature, we assume a quadrati ost funtion. Let 2 (x x 0) 2 be the ost of adjusting the amount of interest from its initial amount x 0 to optimal amount x and 2 (α α 0) 2 the ost of modifying investment from its initial distribution α 0 to optimal distribution α. We request 0 α 1. Assume the present value V (α, x) of the MNE is omputed as its after-tax profits. Sine the MNE seeks to maximize its present value V (α, x) with respet to investment α and interest payments x, its objetive funtion beomes max α,x V (α, x) = (1 τ p)b p + (1 w d i )[(1 τ i )B i xw i i] τ p B i 2 (α α 0) 2 2 (x x 0) 2 (3) where the first term is the after-tax profits of the parent ompany, the seond term the profit of the subsidiary after being taxed by its host ountry, the third term the tax levied by the home ountry and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the MNE s objetive funtion with respet to the fration of real investment α and to interest payments x, we obtain the following first order onditions: = [ 1 (1 wi d )(1 τ i ) ] p w (α α 0 ) = 0 (4) dα 6
= 1 (1 wi d )( τ i + w i dx i) (x x 0 ) = 0 (5) and seond order onditions: d 2 V (α, x) dα 2 = < 0 (6) d 2 V (α, x) dx 2 = < 0 (7) Based on the first order onditions, the equilibrium values of α and x in ountry p an be expressed as α NIR = α 0 + τ i (1 τ i )w d i p w (8) = α 0 + ET Rd i,nir p w (9) x NIR = x 0 + τ i + (1 τ i )w d i (1 wd i )wi i = x 0 + ET Rd i,nir ET Ri i,nir (10) (11) In the equations above ET Ri,NIR d stands for the effetive tax rate on dividends paid out in ountry i in an environment without international tax rules and without lurative detour. Similarly, ET Ri,NIR i stands for the effetive tax rate on interests paid in ountry i. Given a tax environment without lurative detour, it turns out that the optimal level of real investment α NIR inreases when the effetive tax rate on dividends paid by the subsidiary inreases. Hene, the investment in ountry p inreases with taxation of dividends in ountry i. As total investment amounts to unity, investment in ountry i is disouraged when the effetive tax rate on dividends raises. Note that the orporate tax rate in ountry p does not influene the effetive tax rate on dividends. Indeed profits are taxed similarly in ountry p wherever they have been generated. Regarding the optimal amount of internal debt, it inreases with the effetive tax rate differential between dividends and interests. In partiular, if the effetive tax rate on dividends exeeds that on interests, the MNE has an inentive to irulate the profits as interest payments rather than as dividends, and vie versa. Also for interests, the orporate tax rate in ountry p does not influene the effetive tax rate. In addition, when ET Ri d = ET Ri i, x = x 0 and finanial neutrality or the famous "irrelevane of orporate finane" is at work. 2.2.2 With lurative detour We now onsider the ase where the MNE, in order to optimize its after-tax profit, will try to loate part of its profits in the ountry with the lowest tax 7
rate. Country j orresponds to this riterion. However, ountry j only hosts a servie subsidiary of the MNE, sine it offers no opportunity for prodution or onsumption. Hene, the MNE will not make real investments in ountry j. One option for the MNE to benefit from ountry j s favorable tax rate, is to make a detour through ountry j to invest indiretly in ountry i. Instead of the parent ompany diretly granting a loan to the subsidiary in ountry i, it now makes a lurative detour through the entity in ountry j, whih then grants a loan to the subsidiary in ountry i. As a result, the pre-tax profit B p onsists of the revenues of selling q in ountry p and of the revenues of selling (α q) to the group entity in ountry i. B p = p r q + p w (α q) (12) Pre-tax profit B i onsists of the revenues of selling (1 q) in ountry i from whih the interest payments x, as well as the ost of aquiring a fration of p s prodution is subtrated. B i = p r (1 q) p w (α q) x (13) Hosting a servie subsidiary granting loans to other group members, the pre-tax profit B j onsists of the interests paid by the subsidiary in ountry i. B j = x (14) As no tax relief rules exist, ross-border dividends and interests are subjet to triple taxation. First, the profits B i and B j are subjet to their host ountry s orporate tax (τ i and τ j respetively). Seond, sine both subsidiaries distribute their entire after-tax inome as dividend to their parent and sine the subsidiary in ountry i pays out interests to the subsidiary in ountry j, these outgoing dividends and interest are subjet to a withholding tax (wi d, wd j and wi i respetively). Finally, the profits of all group entities are subjet to the home ountry s orporate inome tax τ p. This leads to expressing the MNE s objetive funtion as max α,x V (α, x) = (1 τ p)b p + (1 w d i )[(1 τ i )B i xw i i] +(1 w d j )(1 τ j )B j τ p (B i + B j ) 2 (α α 0) 2 2 (x x 0) 2 (15) where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its host ountry, the fourth term the tax levied by the home ountry and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the MNE s objetive funtion with respet to the fration of real investment α and to the amount of interest payments x, we obtain the following first order onditions: dα = [1 (1 w d i )(1 τ i )]p w (α α 0 ) = 0 (16) 8
= (1 wi d )(1 τ i + w i dx i) + (1 wj d )(1 τ j ) (x x 0 ) = 0 (17) and the same seond order onditions as under setion 2.2.2.1. The equilibrium values of α and x in ountry p an be expressed as α NIRD = α 0 + τ i + (1 τ i ) w d i = α 0 + ET Rd i,nird p w (18) = α NIR x NIRD = x 0 + [τ i + (1 τ i )w d i ] [(1 wd i )wi i ] [τ j + w d j (1 τ j)] p w = x 0 + ET Rd i,nird ET Ri i,nird ET Rd j,nird (19) = x NIR ET Rd j,nird where the effetive tax rate on dividends and interests in ountry i are unhanged. When omparing these equilibrium values with those related to an environment without lurative detour, we observe that a detour through ountry j does not alter the optimal investment level. The optimal amount of internal debt, however, is smaller when shifting inome using a lurative detour, what might seem surprising. Sine ountry p taxes x in any ase, making a detour through j instead of going diretly from i to p simply introdues an extra tax burden. Finally, in order to determine whether using a detour is really lurative, we need to ompute the additional value of the MNE when moving through ountry j. Substituting for the variables α and x their equilibrium values in equations (3) and (15), enables to generate the value of the MNE under both environments. So far, we have onsidered a world without international tax rules. Therefore, ompanies are entirely free to organize their transations aording to their eonomi needs. As shown by the model, some of their profit is, however, subjet to multiple taxation. When relaxing the assumption that no international tax rules exist, the model hanges onsiderably. Starting from the last tax environment (setion 2.2.2), several tax environments are ompared. For eah of them, we ompute the optimal investment level and interest level, and we ompare them with the results found before. 3 EU Taxation under Separate Aounting In this setion, we analyze the tax situation of a MNE ative in the urrent European Union (EU) setting. Several tax relief systems are available, both preventing double orporate taxation and withholding taxes. We first extend the model 9
to an environment omprising the rules provided by the OECD Model Tax Convention. Then, we suppose the EU Parent-Subsidiary Diretive (90/435/EEC) and Interests and Royalties Diretive (2003/49/EC) at work. Finally, we model in that framework the introdution of a ombined ACE-CBIT system. 3.1 OECD Model Tax Convention A first institutional body, whih has shaped international double tax relief rules, is the Organisation for Eonomi Cooperation (OECD). This organization, regrouping the governments of around thirty ountries, suggested in 1958 a "Model Tax Convention" as framework for the negotiation of international tax treaties. Those tax treaties, onluded between two ountries, aim at avoiding double taxation of inome or apital by providing a tax relief system. When entering into a double tax treaty aording to the OECD Model Tax Convention, ountries need to hoose between two methods of double tax relief, the redit method and the exemption system. Under a redit method, the ountry of the benefiiary may tax ross-border inome provided that the taxes paid abroad are dedutible at home. Under an exemption system, the ountry of the benefiiary may tax at most a fration δ of ross-border inome. In the model below, we assume again that the MNE organizes its finanial struture in order to benefit from the lowest tax rate, whih ountry j offers to orporate profits. Hene, a lurative detour through ountry j is used and the profits are defined as under setion 2.2.2. 3.1.1 Credit method for dividends One method to avoid the double taxation of dividends is to request from the taxing ountry to redit foreign withholding tax. Under a redit method as defined by the OECD Model Tax Convention (also alled diret redit ), the ountry of the benefiiary may only tax ross-border inome provided that withholding taxes paid abroad are dedutible at home up to the amount of taxes owed to the ountry of the benefiiary. This means that the host ountry, on top of orporate inome tax, will levy a withholding tax on ross-border dividends and that the home ountry p redits this tax on its orporate inome tax (up to the amount that ountry p would have olleted). Hene, the ross-border dividends will be taxed at the highest of following tax rates: the orporate inome tax levied by ountry p and the withholding tax levied by ountry i. Note that this definition of the redit method differs from the one given by the EU Parent-Subsidiary Diretive (90/435/EEC), whih will be modeled in the following subsetion. For interests, we assume that the ountry of the ompany paying the interests (ountry i) levies a withholding tax on that amount. Knowing that pretax profits are defined as under setion 2.2.2.2, the MNE will 10
define its objetive funtion as follows, max α,x V (α, x) = (1 τ p)b p + (1 max { τ p, w d i +(1 max { τ p, w d j } )(1 τ j )B j } )[(1 τ i )B i xw i i] 2 (α α 0) 2 2 (x x 0) 2 (20) where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its host ountry, and the last two terms the adjusting osts of moving to optimal values of α and x. Sine we assumed that orporate tax rates are higher than withholding tax rates, the home ountry will only levy that fration of its orporate inome tax whih exeeds the withholding tax levied by the host ountry. Hene, max(τ p, w d i ) = max(τ p, w d j ) = τ p. When maximizing the value of the firm with respet to the fration of real investment α and to the amount of interest x, we obtain the following first order onditions, dα = [(1 τ p ) (1 τ p )(1 τ i )]p w (α α 0 ) = 0 (21) = (1 τ p )(1 τ i + w i dx i) + (1 τ p )(1 τ j + wi) i (x x 0 ) = 0 (22) and the same seond order onditions as under setion 2.2.2.1. It turns out that the equilibrium values of α and x in ountry p are, and α OCRE = α 0 + τ i(1 τ p ) p w = α 0 + ET Rd i,ocre p w (23) x OCRE = x 0 + τ i(1 τ p ) w i i (1 τ p) τ j (1 τ p ) = x 0 + ET Rd i,ocre ET Ri i,ocre ET Rd j,ocre (24) We observe that the effetive tax rates on dividends and interests are now influened by the orporate tax rate in ountry p. Moreover, omparing these optimal values with the values found in the previous setion reveals that investment in ountry p inreases when double taxation is eliminated. The optimal amount of internal debt depends of the orporate tax rate of the three ountries involved, as well as on the withholding tax rate of ountry i. The higher the effetive tax rate of dividends with respet to interests, the higher the inentive to shift inome by using internal debt. 11
3.1.2 Exemption method for dividends Another method to avoid the double taxation of dividends is to exempt the ross-border dividend from orporate taxation in the ountry of the benefiiary. The OECD Model Tax Convention stipulates that the ountry of the benefiiary may tax at most a fration δ of the ross-border dividends. Hene, the host ountries will levy orporate tax and withholding tax on the dividends distributed by the subsidiaries to the parent ompany. Only a fration δ of those dividends is subjet to the orporate tax of ountry p. For ross-border interests, we assume that only the ountry of the paying ompany may levy a withholding tax on that amount. Hene, the interests paid out by the subsidiary in ountry i to the subsidiary in ountry j is subjet to a withholding tax wi i. As a result, the MNE s objetive funtion beomes max α,x V (α, x) = (1 τ p)b p + (1 δτ p )(1 w d i )[(1 τ i )B i xw i i] +(1 δτ p )(1 w d j )(1 τ j )B j 2 (α α 0) 2 2 (x x 0) 2 (25) where the pretax profits are defined as under setion 2.2.2 and where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its respetive host ountry, and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the MNE s objetive funtion with respet to α and x, we obtain the following first order onditions: dα = [(1 τ p ) (1 δτ p )(1 w d i )(1 τ i )]p w (α α 0 ) = 0 (26) dx = (1 δτ p )(1 w d i )(1 τ i + w i i) +(1 δτ p )(1 w d j )(1 τ j ) (x x 0 ) = 0 (27) and the same seond order onditions as under setion 2.2.1. The equilibrium values of α and x in ountry p are given by and α OEXE = α 0 + (1 τ p) (1 δτ p )(1 wi d)(1 τ i) p w (28) = α 0 + ET Rd i,oexe τ p p w x OEXE = x 0 + (1 δτ p)[(1 wi d)(1 τ i) (1 wi d)wi i (1 wd j )(1 τ j)] (29) = x 0 + ET Rd i,oexe ET Ri i,oexe + ET Rd j,oexe 12
One more, we observe that real investment in ountry p inreases with the effetive tax rate on dividends paid in ountry i. The tax relief system provided by the OECD Model Tax Convention, therefore, enourages foreign investment. Moreover, as under the redit method, the orporate tax rate of ountry p influenes the fration of investment, as well as the amount of interest x OEXE shifted. The higher the orporate tax rate of ountry p, the lower the optimal amount of interest shifted to ountry j. Hene, the OECD tax relief system disourages interest shifting. 3.2 EU Diret Tax Diretives In a seond setting, the rules of the OECD Model Tax Convention are supplemented by the EU treaties, regulations, and diretives. They set out the priniples and rules for the reation of the Single Market, ensuring the free movement of goods, servies, apital, and labor among the 27 EU Member States. Although tax sovereignty still applies in the EU, Member States an unanimously deide to give up part of their national sovereignty to enhane the development of ommon tax measures. With respet to diret taxation, two Diretives are of major importane, i.e. the Parent-Subsidiary Diretive (90/435/EEC) and the Interests and Royalties Diretive (2003/49/EC), eliminating withholding taxes on dividend, interest and royalty payments between related ompanies. In order for those Diretives to apply, ompanies need to be subjet to orporate tax in the EU, be tax resident in an EU Member State, and be of a type listed in the Diretives. We assume that the three ountries of our model are Member States of the European Union and that the ompanies in those ountries may apply the mentioned Diretives. Furthermore, in order to benefit from the withholding tax exemption for dividends, the EU parent ompany should hold at least 10% of the shares in its foreign EU subsidiary. In our model, we assume that those onditions are verified for all ompanies, inluding the entity in ountry j. Consequently, withholding taxes will no longer appear in our model. 3.2.1 Credit method for dividends As the OECD Model Tax Convention, the EU Parent-Subsidiary Diretive provides for two methods to avoid the double taxation of dividends. One of those is the redit method (full redit method), whih slightly differs, however, from the one used in the OECD Model Tax Convention (diret redit method). Under full rediting, the ountry of the benefiiary may only tax ross-border inome provided that all taxes paid abroad are dedutible at home up to the amount of taxes owed to the ountry of the benefiiary. This means that the host ountry will levy orporate tax and that the home ountry redits this tax on its orporate tax up to the amount that it would have olleted. As the EU diret tax Diretives apply, all withholding taxes are eliminated and the ross-border dividends will be taxed at the highest of both orporate tax rates. This leads 13
the MNE to define its objetive funtion as follows, max α,x V (α, x) = (1 τ p)b p + (1 max {τ p, τ i })B i + (1 max {τ p, τ j })B j 2 (α α 0) 2 2 (x x 0) 2 (30) As we assumed that τ j < τ i < τ p, this objetive funtion an be rewritten as: max α,x V (α, x) = (1 τ p)b p + (1 τ p )B i + (1 τ p )B j 2 (α α 0) 2 2 (x x 0) 2 (31) where the profits are defined as under setion 2.2.2 and where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its respetive host ountry, and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the value of the firm with respet to α and x, we obtain the following first order onditions: dα = (α α 0 ) = 0 (32) = (x x 0 ) = 0 (33) dx and the same seond order onditions as under setion 2.2.1. The equilibrium values of α and x in ountry p an be written α ECRE = α 0 (34) x ECRE = x 0 (35) We observe that the optimal values are independent of tax parameters, meaning that the tax environment is neutral with respet to both the investment and finane deision of the MNE. The redit method under the EU Parent-Subsidiary Diretive an, therefore, be onsidered as an eonomially effi ient tax environment. Gérard and Traversa (2010) suggest to move to the redit system. A move to rediting, though, should imply - see setion 3.2.1 above - that dividends from ountry j be taxed in a similar way as profits from ountries p and i, thus in a similar way as profits not subjet to a lurative detour. Suh a move, however, seems to be in ontradition with the trend among ountries. Indeed, a ountry like the UK whih was for long haraterized by rediting has moved to exemption. The main reason, presumably, is that the redit method may reate disrimination among domesti and other European resident shareholders, sine redits hardly ross the national borders. Moreover, moving to a redit system does not prevent lurative detours when the profit remains in the ountry of the subsidiary and is from there used to finane further investments of the MNE. 14
3.2.2 Exemption method for dividends A seond tax relief method provided for by the EU Parent-Subsidiary Diretive, is to apply an exemption method. As under the OECD Model Tax Convention, the exemption method onsists of exempting all but a fration δ of the ross-border dividends from orporate taxation in the ountry of the benefiiary. Hene, after being subjet to the orporate tax of their host ountry, a fration δ of the dividends paid out by the subsidiaries is taxed aording to the home ountry s orporate tax rate (τ p ). As the EU tax Diretives apply, withholding taxes are eliminated both on dividends as on interests. This leads to the following objetive funtion: max α,x V (α, x) = (1 τ p)b p + (1 δτ p )(1 τ i )B i + (1 δτ p )(1 τ j )B j 2 (α α 0) 2 2 (x x 0) 2 (36) where the benefits are defined as under setion 2.2.2 and where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its respetive host ountry, and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the MNE s objetive funtion with respet to α and x, we obtain the following first order onditions: dα = [(1 τ p ) (1 δτ p )(1 τ i )]p w (α α 0 ) = 0 (37) = (1 δτ p )(1 τ i ) + (1 δτ p )(1 τ j ) (x x 0 ) = 0 (38) dx and the same seond order onditions as under setion 2.2.1. The equilibrium values of α and x in ountry p beome α EEXE = α 0 + (1 τ p) (1 δτ p )(1 τ i ) p w (39) and x EEXE = x 0 + (τ i τ j )(1 δτ p ) Notie that in many ountries, δ = 0. In that latter ase, (40) α EEXE = α 0 + τ i τ p p w (41) and x EEXE = x 0 + τ i τ j (42) We observe that neutrality of the tax system with respet to the investment and finane deision of the MNE further requires the equality among orporate tax rates. 15
3.2.3 Combination of ACE and CBIT A suggestion to redue the orporate tax distortion between soures of finaning is the introdution of a system ombining an Allowane for Corporate Equity (ACE) and a Comprehensive Business Inome Tax (CBIT). Suh proposition takles the unequal tax treatment of debt and equity (and also retained earnings), by giving partial, but equal tax relief to both finaning modes. Hene, the dedutibility of interests is partially abolished and the dedutibility of dividends is partially established. Suppose therefore that a fration θ of the interests an no longer be deduted by the paying ompany, but that a tax shield is granted for a fration 1 θ of the dividend payments. The ACE-CBIT system leaves the benefits before tax unhanged and the pretax profits B p, B i and B j, therefore, have their usual definition. The tax base of the subsidiary in ountry i, however, is altered sine a fration θ of interests is no longer tax dedutible. Consequently, for orporate tax purposes, the fration θx is added to the pretax profits of the subsidiary in ountry i. Moreover, sine a tax shield is granted for a fration 1 θ of dividend payments, both subsidiaries an benefit from a tax advantage amounting to τ host (1 θ) of their dividends. As we assumed that the subsidiaries distribute their entire after-tax inome as dividend to their parent ompany, the amount of dividends equals the subsidiary s after-tax profits. Assuming an exemption system for the taxation of dividends in line with the EU Parent-Subsidiary Diretive, the ompany s objetive funtion now is, max α,x V (α, x) = (1 τ p)b p + (1 δτ p )[B i τ i (B i + θx) + τ i (1 θ)(b i + θx)] +(1 δτ p )[(B j τ j B j + τ j (1 θ)b j ] 2 (α α 0) 2 2 (x x 0) 2 (43) where the first term is the after-tax profits of the parent ompany, the two following terms the profits of the subsidiary after being taxed by its respetive host ountry, and the last two terms the adjusting osts of moving to optimal values of α and x. When maximizing the MNE s objetive funtion with respet to α and x, we obtain the following first order onditions: dα = [(1 τ p ) (1 δτ p )]p w (1 θτ i ) (α α 0 ) = 0 (44) = (1 δτ p )θτ i (1 θ) (1 δτ p )θτ j (x x 0 ) = 0 (45) dx and the same seond order onditions as under setion 2.2.1. The equilibrium values of α and x in ountry p an be expressed as, α ACBIT = α 0 + (1 τ p) (1 δτ p )(1 θτ i ) p w (46) x ACBIT = x 0 + (1 δτ p)θτ i (1 θ) (1 δτ p )θτ j (47) 16
Espeially if δ = 0 and θ = 1/2, α ACBIT = α 0 + τ i 2 τ p p w (48) x ACBIT = x 0 + τ i 2τ j (49) 4 We observe that the ACE-CBIT system has lowered the impat of the orporate tax rate in ountry i in the distribution of investment ompared to the exemption system (setion 3.2.2). Moreover, it sharply dereased the optimal amount of internal debt and thus the importane of a profit-shifting strategy. Hene, the ACE-CBIT system provides for an interesting alternative, as it brings the amount of interest lose to its effi ient level. 4 EU Tax Environment under Consolidation The analysis of an ACE-CBIT system shows that tax effi ieny an be approahed by other tax environments than the redit method. This system, however, does not entirely eliminate profit-shifting and does not takle the ompliane ost issue. This is mainly due to the fat that eah entity is taxed separately, based on its individual aounts, without onsidering the group as a whole (Separate Aounting (SA) approah). In order to address this issue, ountries like the United States and Canada deided, for state taxation purposes, to move to a system of onsolidated taxation (alled Consolidation and Formulary Apportionment (C&FA)). In this system, one onsolidated tax base is omputed, whih is distributed amongst the affeted ountries aording to a given apportionment formula. Also the European Union onsiders to move to suh a system, suggesting a Common Consolidated Corporate Tax Base (CC- CTB). Beause the unanimity priniple for tax issues makes multinational deisionmaking diffi ult in the EU, the mehanism of an Enhaned Cooperation Agreement (ECA) may well be used for this purpose. This alternative deision-making method, adopted in the Treaty of Nie (2002), allows a minimum of eight EU Member States to integrate more or faster than other Member States. It was introdued as a means of takling the problem of the growing diversity in the European Union and allowing the further integration and development of the European projet. In this setion, we onsider two situations. In the first one, all EU Member States unanimously deide to introdue the CCCTB taxation system; under the seond one, only the ountries p and i adopt the reform (under an Enhaned Cooperation Agreement) and ountry j stays out of the onsolidation area. 4.1 Unanimity Under a Common Consolidated Corporate Tax Base (CCCTB), one onsolidated tax base is omputed in whih the intra-group payments of dividends, 17
interests, and royalties are ignored. The pretax profits of all group ompanies are onsolidated regardless of whether these ompanies are residents or non-residents of the MNE s home ountry. The onsolidated tax base is then distributed amongst the ountries using a formula. We assume that this formula is a weighted linear ombination of real investment (with weight λ) and final sales (with weight 1 λ). Eah ountry taxes its tax base fration aording to its own tax rate. Given that all intra-group inome is ignored when onsolidating, the onsolidated tax base B onsists solely of the revenues of selling to onsumers. Sine total sales amount to unity, B is expressed as: B = p r (50) The fration of the onsolidated profits Bp F A, attributed to ountry p, onsists of a fration of B, proportional to investment and sales. As a result, the apportioned profits for ountry p and ountry i are: B F A p = [λα + (1 λ)q]b (51) B F A i = [λ(1 α) + (1 λ)(1 q)]b (52) Hene, the objetive funtion of the MNE beomes: max α,x V (α, x) = (1 τ p)b F A p + (1 τ i )B F A i 2 (α α 0) 2 2 (x x 0) 2 (53) where the first term is the profit attributed to ountry p after being taxed by the home ountry, the seond term the profits attributed to ountry i after being taxed by the host ountry and the last two terms the adjusting osts of moving to optimal values of α and x. There is no longer room for interest-shifting under this tax environment, sine intra-mne movements vanish. Aordingly no tax base is alloated to ountry j. We then maximize the MNE s objetive funtion with respet to the sole fration of real investment α, keeping x possibly equal to x 0 in order to avoid the extra ost to set x equal to any other value. We obtain the first order ondition dα and the equilibrium values of α and x are, = (1 τ p )λp r (1 τ i )λp r (α α 0 ) = 0 (54) α F A = α 0 + τ i τ p λp r (55) x F A = x 0 (56) Two observations deserve attention at this point. First, as the irrelevane of the determination of x shows, there is no longer room for profit-shifting strategies with respet to the soure of finane, the transfer prie or any other instrument. 18
Seond, the move from separate aounting to onsolidation might redue tax ompetition and allow orporate tax rates to go up when determined as the outome of a non-ooperative game between ountries. The ondition therefore is that λp r < p w (57) Sine we know that p r > p w - the retail prie exeeds the wholesale prie - the ondition requires that the weight of the formula be rather on the distribution of sales, the variable not or less under ontrol of the MNE, or on the variable with respet to whih the MNE is less elasti, than on the distribution of investment. That property has been demonstrated by Riedl and Runkel (2007) as well as by Gérard (2007). Although that system exhibits interesting properties, it does not guarantee that every partiipating ountry will gain tax revenues. The ase of j above is emblemati, sine ountry j no longer has revenues to tax. It may therefore be diffi ult to onvine all EU Member States to join the reform, and justifies that the adoption of the reform by a sole subset of Member States, through an Enhaned Cooperation Agreement, is investigated. 4.2 Enhaned Cooperation The mehanism of an Enhaned Cooperation Agreement (ECA) allows a number of EU Member States to integrate more or faster than other Member States. Applied to the C&FA issue, this alternative deision-making method would luster the EU Member States in two groups. One group of Member States would maintain their urrent separate aounting system in order for them to further attrat orporations through their ompetitive tax system. The other group of Member States would implement the CCCTB, allowing them to lower orporate transation osts. Consider that only the ative ountries p and i adopt a ommon onsolidated tax base and that ountry j deides to stay out of the onsolidation area, maintaining its urrent environment. Hene, the use of profit-shifting strategies is relevant again and interests x will be shifted from the onsolidation area tot the servie subsidiary outside of the area in ountry j. The ommon tax base inludes the tax bases of the ative ountries from whih the flow of interest x shifted to ountry j is subtrated. B = p r x (58) A separate tax base of the entity in ountry j oexists: The objetive funtion of the MNE remains: B j = x (59) max V (α, x) = (1 τ p)bp F A + (1 τ i )Bi F A + (1 τ j )B j α,x 2 (α α 0) 2 2 (x x 0) 2 (60) 19
The apportioned tax bases are defined as in the former setion 2.4.1. When maximizing the MNE s objetive funtion with respet to α and x, we obtain the following first order onditions: dα = (1 τ p )λ(p r x) (1 τ i )λ(p r x) (α α 0 ) = 0 (61) dx = (1 τ p )[q(1 λ) + λα] (1 τ i )[(1 q)(1 λ) + λ(1 α)] +(1 τ j ) (x x 0 ) = 0 (62) The equilibrium values of α and x in ountry p an be expressed as: α ECA = α 0 λ(τ i τ p )(x 0 p r ) (τ i τ j ) + (τ i τ p ) 2 λq(1 λ) (τ i τ p ) 2 λ 2 (63) + x ECA = x 0 + (τ p τ i )[q(1 λ) + λα ECA ] + (τ i τ j ) (64) Comparing those equilibrium values with the ones obtained under a unanimous introdution of the C&FA reform, we observe that optimal investment in the parent ompany is redued and the optimal debt fration is inreased under enhaned ooperation. Hene, a detour through a non-onsolidating ountry is still profitable. Not the ooperating ountries, but the ountry staying out of the onsolidation area benefits from the enhaned ooperation agreement, as an important number of tax planning strategies persist. We an therefore reasonably onsider that the onsolidating ountries will attempt to ounter this. 5 Poliy Impliations Given the tax ompetition between Member States, it may be expeted that some of them may not be willing to introdue a ommon onsolidated tax base. Remote ountries with attrative tax regimes like Ireland and Estonia may not be eager to give up their favorable tax features without having the guarantee of at least maintaining their urrent tax revenues. This is also the ase for Member States whih offer speial depreiation shemes, R&D tax redits or other nondebt tax shields (like the Allowane for Corporate Equity in Belgium). As a result, it is highly probable that the introdution of a ommon onsolidated tax base in the EU will only be possible through the use of an Enhaned Cooperation Agreement. The analysis above, however, has shown that the introdution of suh a tax base by a limited number of Member States does not lead to effi ient taxation and leaves room for inome-shifting strategies. In line with the existing theoretial literature (a.o. Gérard and Traversa (2010)), we find that the use of a redit method as tax relief for ross-border dividends leads to a finanially effi ient solution and the elimination of profitshifting inentives. Nonetheless, as notied earlier, the politial trend is to 20