Corporate Tax Competition and Differentiated Public Goods

Corporate Tax Competition and Differentiated Public Goods∗

Fran¸cois Pouget†

´ ıse St´eclebout-Orseau Elo¨

‡

December 16, 2010

Abstract We present a theoretical model of corporate tax competition that combines two public finance issues: the level of corporate tax rates and the composition of public spending. As firms can shift profits for tax avoidance purposes, national decision makers are led to compete in terms of corporate tax rates. But they can also attract capital inflows by allocating a larger share of public expenditure to the provision of a factor-augmenting public good. We analyse the strategy of countries seeking to attract the mobile tax base and we show that tighter tax competition may actually reduce the share of productive public spending. We then discuss the consequences of tax rate coordination: such a policy does not eliminate competition but it changes its focus, as each country needs to rely more on the public spending instrument. We finally assess the issue of a common corporate tax base. We find that firms may manipulate the apportionment formula and that a race to the bottom in tax rates cannot be excluded. JEL Classification: H25, H40, F20 Keywords: Tax competition; Public good provision; Corporate income tax.

∗ We would like to thank Hubert Kempf, Ad van Riet and Jean-Pierre Vidal for their comments. The opinions expressed herein are those of the authors and do not necessarily reflect those of the ECB. † EURIsCO, University of Paris Dauphine, 1 Place du Mar´ echal de Lattre de Tassigny, F-75016 Paris, France; e-mail: [email protected] ‡ European Central Bank, Directorate General Economics, Kaiserstraße 29, D-60311 Frankfurt am Main, Germany; e-mail: [email protected]

1

Non-technical summary In the last decades, the literature has pointed to the potentially harmful effects of tax competition. Downward pressure on taxation may lead governments to engage in a race to the bottom in tax rates, resulting in lower tax revenues and finally insufficient provision of public goods. This phenomenon is all the more likely to take place as the tax base is mobile, which is particularly the case for corporate income. Given the high mobility of capital and profits in the European Union, various arrangements to prevent a race to the bottom in corporate taxation have been discussed since the 1960s, but their implementation has remained limited. This paper investigates the strategic interaction at work in corporate tax competition and assesses alternative proposals. We set up a game in which a single representative firm operates in two competing countries and we make three major assumptions. First, the mobility of the tax base has two components: capital movements and profit shifting. Capital is perfectly mobile, whereas it is costly for the firm to shift profits from one country to the other because of operational and accounting constraints. Profit mobility as measured by an inverted index of the cost of profit shifting is used as an index of tax competition. Second, tax competition is not one-dimensional. What makes jurisdictions attractive is not only the level of corporate tax rates, but also the provision of a productivityenhancing public input. Governments combine these two policy instruments to attract the mobile corporate tax base. Third, society derives its utility from the consumption of two public goods: the public input and a general public good. Starting with a decentralised framework, we assume that in the first stage of the game each government decides on its domestic tax rate and the composition of public expenditure. In the second stage, the multinational firm allocates capital and shifts profits. As expected, we find that higher profit mobility triggers a race to the bottom in tax rates. In addition, governments seek to attract capital by systematically providing more public input than what society’s relative preferences would suggest. This bias, however, is reduced when profit is highly mobile, as profit shifting outweighs capital movements. Moving on to coordination, we consider the harmonisation of tax rates. We show that when countries are symmetric, the harmonised tax rate is at its highest possible value, i.e. the tax rate that would maximise tax revenue if profits could not be shifted. Correspondingly, the spending bias in favour of the public input is also at its maximum because in this case, attracting capital inflows is of crucial importance. When countries are asymmetric, the harmonised tax rate lies between the revenue-maximising domestic tax rates of the two countries. This implies that the harmonised rate is on the downward slope of the Laffer curve in the formerly low-tax country, which consequently loses the ability to attract the mobile tax base by means of low taxation. Finally, we assess the impact of adopting a common corporate tax base, whereby profits are consolidated and then apportioned among countries according to a formula, which we assume 2

depends on capital stocks (as in existing systems). We show that although the firm is de facto unable to shift profits to the low-tax country, it may arbitrage on the location of its activity to manipulate the apportionment formula and minimise its overall tax burden. Therefore, a common tax base system may paradoxically increase the strategic use of capital movements, which in turn could result in a race to the bottom in tax rates. This finding advocates a formula based on factors that cannot be easily manipulated by the firm.

3

1

Introduction

Although tax policy is traditionally the responsibility of national Member States in the European Union (EU), corporate tax coordination prominently features on the European Commission’s agenda. As economic integration increases the mobility of the corporate tax base, thereby putting pressure on corporate tax rates, fears of a potential race to the bottom have motivated public intervention at the supranational level. This has been a cause for concern over several decades: the first proposal for corporate tax harmonisation, the Neumark Report, dates back to 1963 and in 1992, the Ruding Report advocated a minimum statutory corporate tax rate of 30% in the EU. However, all the forms of tax harmonisation proposed at the European level have failed so far, partly because of the reluctance of Member States to give away their fiscal sovereignty. Yet the debate is not closed and may re-emerge through the intergovernmental political game. Figure 1 gives us a clear picture of the current situation: although corporate tax rates have converged over the last twenty years, the current discrepancy between old and new Member States is quite striking and can be a source of conflict.1 For this reason, the high-tax countries may insist on coordinating corporate tax policies. [Figure 1] Recently, the European Commission appears to have shifted its strategy in favour of less binding and constraining methods by proposing the adoption of a common corporate tax base. The rationale behind this new approach is that corporate taxation in the EU comes under a patchwork of national corporate tax systems. Although the current separate accounting (hereafter SA) system is organised by tax treaties to avoid double taxation (based on OECD and EU guidelines), a multinational firm still needs to cope with each particular domestic tax code in the countries where it operates. These ”compliance costs” constitute an obstacle to the adoption of a well-functioning single market of capital and therefore justify the intervention of the European Commission (see EC 2001[?]). Moreover, the current SA regime provides incentives for firms to set strategies to shift their profits between the different tax systems to reduce their overall tax burden. Our purpose is to build a theoretical model of corporate tax competition and evaluate the alternative options for coordination. Most of the proposed options draw on the well-known conclusions of the tax competition literature initiated by Oates (1972[?]) and developed more formally by Zodrow & Mieszkowski (1986[?]) and Wilson (1986[?]). The traditional theory states that the need to attract a mobile tax base would force countries to lower their statutory tax rates, which would reduce tax revenue and eventually lead to an under-provision of public goods (see Wilson, 1999[?] for a survey of the numerous contributions in this field and Dhillon, Wooders 1 A good illustration is that in 2004, in a television interview, the French Finance Minister strongly criticised the level of corporate tax rates in the new Member States and threatened to propose that countries below the European average should not be eligible for structural funds.

4

& Zissimos, 2007[?] for a recent contribution). Following this conventional wisdom, there is a need to control the strategic interaction initiating a race-to-the-bottom mechanism, especially through coordination measures. A large number of empirical studies confirm that multinational firms react to corporate tax policy (see for instance B´enassy-Qu´er´e, Lahr`eche-R´evil & Fontagn´e, 2005[?] and De Mooij & Ederveen, 2003[?]), which makes a race to the bottom in corporate tax rates likely. Indeed, we observe such a phenomenon in Europe: in 2008, the EU-27 average corporate tax rate amounted to 23.2% while it was 10 points higher a decade earlier.2 However, the harmful consequences of corporate tax competition on public goods provision are still in question. Figure 2 is informative for two reasons. First, the share of corporate taxation relative to other sources of tax revenue has remained broadly stable and even recently increased in spite of cuts in corporate tax rates, meaning that governments may have used other instruments to attract larger tax bases. Second, corporate taxation hovers around 8% of total tax revenues, therefore one cannot expect corporate tax competition to have a dramatic effect on the total level of public spending. [Figure 2] In this paper, we consider that the theoretical framework provided by the traditional theory of tax competition may prove insufficient if it is applied to corporate taxation. We suggest that corporate tax competition could be a more complex phenomenon than a race to the bottom in corporate tax rates and its direct negative consequence for public goods provision . This idea partly draws on a study by B´enassy-Qu´er´e, Gobalraja & Trannoy (2005[?]) on the determinants of foreign direct investment (FDI) flows. They show that a so-called “public input”, defined roughly as a stock of public infrastructure that enhances capital productivity, has a significant impact on FDI flows. They conclude that competition for FDI is at least two-dimensional. Similarly, Hauptmeier, Mittermaier & Rincke (2009[?]) find that local governments react to cuts in tax rates in neighbouring jurisdictions by both lowering their own taxes and increasing spending on public inputs. Assuming in turn that certain categories of public spending can be used by policy makers to attract a mobile corporate tax base, our intuition is that corporate tax competition may produce significant side-effects on the composition of public spending. On this issue, Keen & Marchand (1997[?], thereafter KM) demonstrate that tax competition leads to a systematic distortion in the pattern of public spending, with a tendency to under-provide public goods that are available for direct consumption (in contrast with public goods that are also used for productive purposes). We build a theoretical model of corporate tax competition based on the framework of Zodrow & Mieszkowski (1986[?]). Countries can attract the corporate tax base through both taxation and 2 Nevertheless, as this fall has been accompanied by a base broadening in all Member States, effective tax rates do not follow exactly the same pattern.

5

the provision of a public input. As it enhances the marginal productivity of capital, the public input is targeted to firms, but it is also consumed by households (roads and public infrastructure at large usually have this dual purpose). In each country, the decision maker provides this public input as well as a “non-productive” public good that is only aimed at households (think of redistribution, health or cultural programs). In this paper, public spending can only be allocated to the provision of these two public goods. Public decision is twofold: it involves solving an issue of collection of public resources (through the choice of a statutory tax rate) as well as an issue of composition of public spending. In addition, we assume that society has well defined relative preferences for the two types of public goods. Finally, contrary to KM who consider all sources of tax revenue, we focus our analysis on corporate taxation. This simplification enables us to analyse corporate tax competition in two different environments: first, under an SA system where firms can easily shift profits from one country to another and, second, under a common tax base. In each case, we investigate the nature and outcome of tax competition as well as its consequences on the composition of public spending. The outline of the paper is as follows. We present our model of corporate taxation in Section 2. Following Kolmar & Wagener, 2006[?] and Sørensen, 2004[?], the mobility of the corporate tax base depends on a single representative firm’s decision on capital and profit allocation. Section 3 analyses tax competition. When accounting systems are separated, competing decision makers are led to lower their statutory tax rates. In this context, it is the threat of profit shifting that triggers strategic interaction between national tax authorities. Looking at the expenditure side and how public spending is allocated, we show that tax competition always entails a distortion in favour of public input provision compared to society’s preferences. We compare the magnitude of this distortion under various degrees of tax competition. In Section 4, we discuss corporate tax coordination. We find that harmonising tax rates does not eliminate the issue of the composition of public expenditure and, on the contrary, leads countries to compete more by means of public spending. Furthermore, coordination also raises the issue of heterogeneity. In Section 5, we study the case of corporate tax competition under a common tax base regime where profit shifting can no longer occur and we show that a new type of strategic interaction is likely to emerge. Section 6 concludes.

2

The model

We develop a model of corporate tax competition where two countries (or jurisdictions) denoted A and B rely on the taxable profit of a representative multinational firm to finance their public policy.

6

2.1

Production technology

The firm operates in the two countries. Ki represents the amount of capital that the firm decides to allocate to country i. There is no exogenously given world supply of capital and the firm can choose to purchase it or simply invest on financial markets. The world rate of return is equal to r. Output in country i also depends on a publicly provided amount of human capital denoted Hi ¯ i + hi gi . H ¯ i is the initial human capital stock in country i and is exogenously given where Hi = H in our model. gi is the composite productive public good provided by the government of country i. It includes all types of business-friendly infrastructures such as highways, bridges or various business facilities. Following KM, we assume that it is not firm-specific and therefore it is also provided to households. The specificity of gi is that it can increase the amount of human capital in society (for this reason, education and research can also fit into this category). Nevertheless, countries can be more or less efficient in transforming gi into an input for the firm so that hi  0 can be interpreted as a transformation rate (therefore gi will also be called a ”public input”). The multinational firm’s output in country i is described by a Cobb-Douglas production function: F (Ki ; Hi ) = Kiβi Hi1−βi . The public input enhances the marginal productivity of capital so that FKg (Ki ; gi ) > 0 if hi > 0. This productive nature of gi is in line with Zodrow & Mieszkowski (1986[?]), KM, and Dhillon, Wooders & Zissimos (2007[?]). We also assume that countries can differ in terms of production technologies, so that β is indexed by i.

2.2

Corporate taxation and tax base mobility

The allocation of capital between countries is the first explanatory element for tax base mobility. Nevertheless, the geographical distribution of the multinational firm’s activity only partly defines each country’s tax base because profit can then easily be shifted ”on paper” from one country to another. Under an SA system, the multinational firm is required to establish separate accounts for its activity in every country where it operates. As a result, the parent firm and its foreign subsidiary are separated. Transfer prices for a given service or product between the two entities must be set at arm’s length (i.e. the price that an independent party would pay), but in reality they can be easily manipulated. Management fees or royalties between the parent and the subsidiary are other ways to shift profit for tax evasion purposes. Our model is designed to capture these two components of tax base mobility (capital and profit). As our primary objective is to focus on public policy decisions, we follow Kolmar & Wagener (2006[?]) and Sørensen (2004[?]) by choosing a simple but comprehensive description of tax base mobility. Each tax authority levies a tax on the firm’s gross profit that remains in the country after profit has been shifted. The corporate tax base in a given country i is therefore: Bi (Ki ; S) = F (Ki ; Hi ) − δS

7

(1)

The parameter δ is a dummy variable equal to 1 if the considered country is A and −1 otherwise. S is the amount of profit shifted, such that when S is positive (resp. negative), taxable corporate income is shifted from country A to country B (resp. from B to A). The firm’s total after-tax net profit under an SA system is: ¯ − bS 2 ΠSA = (1 − τA ) BA (KA ; S) + (1 − τB )BB (KB ; S) − r(KA + KB ) − W

(2)

where τi is the corporate tax rate in country i. Note that the cost of capital, as well as the ¯ , are non-deductible for corporate tax purposes. aggregate cost of fixed factors denoted by W This assumption is necessary because we want to restrict the analysis to competition by means of tax rates; we leave aside the issue of differences in the legal definition of tax bases. Moreover, profit shifting manipulations are assumed to be costly to the firm. To prevent tax evasion, tax authorities require explanations in case of distortions in transfer prices and this involves some operational (or ”concealment”) costs, reflected by the cost parameter b. We also assume that profit shifting may have significant side effects in terms of reputation for the firm (a high level of corporate tax evasion could turn out to be extremely unpopular among domestic taxpayers who could put pressure on the government to restrain it). One can therefore intuitively think of this cost as a convex function of the volume of profit shifted. Under these assumptions, profit is imperfectly mobile between the two countries, except when b = 0. In contrast, capital is assumed to be perfectly mobile (as expected in a single market).

2.3

Differentiated public goods

We consider that society derives its utility from the consumption of two types of public goods. We assume that the public input gi is not firm-specific and that households also benefit from it. Public infrastructures primarily built as amenities for the private sector are also largely used by households (think of roads or IT infrastructures). Other good examples are education and research policies which enhance productivity in the private sector but also serve the public interest. Furthermore, in line with Zodrow & Mieszkowski (1986[?]) and KM, we model a second category of public spending specifically aimed at households by introducing a general composite public good denoted Gi . It covers all public spending with no direct productive purposes, such as redistribution.3 Countries tend to differ in their relative preferences for the two categories of public spending.4 To capture this heterogeneity, we assume that the government’s objective function reflects the 3 Cultural policy also fits into this category. Note that the distinction between the two types of public goods is obviously far from being clear-cut in reality. An obvious example is health expenditure, which is assumed to be classified under G although it is also likely to improve productivity indirectly. 4 For instance, European countries tend to spend more on redistribution than the United States while public investment is lower (see Alesina & Glaeser, 2005[?]).

8

preferences of society. The two types of public goods are assumed to be imperfect substitutes and country i ’s relative preference for the public input is measured by αi (∈ [0; +∞[). The utility derived from the consumption of public goods is therefore: Ui (gi ; Gi ) = αi ln (gi ) + ln (Gi ) ; for i = A; B

3

(3)

Corporate tax competition

The public finance issue addressed in our model is twofold. On the one hand, the decision maker of country i needs to collect public revenues through taxation. This involves setting a statutory corporate tax rate denoted by τi . On the other hand, these resources are spent to provide the two different public goods at the national level. This public allocation problem involves determining a ratio gi /Gi where gi and Gi respectively represent the amount of public spending allocated to the public input and the general public good. We study corporate tax competition under a traditional SA system. The timing of the game is as follows: in t1 , public finance decisions are made simultaneously and non-cooperatively in countries A and B ; in t2 , the firm decides where to allocate capital and its taxable profit.

3.1

The firm’s decision

Given the public decisions in A and B, the firm maximises (??) with respect to Ki and S in t2 . As capital is perfectly mobile, the firm will equalise its after-tax marginal productivity in the two countries: ∗ ∗ (1 − τA ) Fk (KA ; HA ) = (1 − τB ) Fk (KB ; HB ) = r

(4)

Cuts in corporate tax rates reduce the net cost of capital r/ (1 − τi ) and therefore attract more capital. As Figure 3 shows, however, they are not the only way for a country to increase its tax base. Indeed, as long as the public input can be transformed into human capital (i.e. hi 6= 0), a country can increase the marginal productivity of capital by providing more gi :  βi /(1−βi ) ∂Ki∗ βi = hi >0 ∂gi Fk (Ki∗ ; Hi )

(5)

[Figure 3] The final allocation of taxable profit, however, does not only depend on the allocation of capital but it also responds to the tax rate differential. Profit will be shifted from A to B if 9

τA − τB > 0. Profit-shifting flows are a decreasing function of the cost parameter associated with these operations: S∗ =

τA − τB 2b

(6)

The mobility of the tax base therefore derives from a typical free-riding problem: the multinational firm seeks to benefit from the public input provided in a given country without paying for it, by shifting its taxable profit to the other (low-tax) country.

3.2

Public policy decision

Each decision maker anticipates the firm’s behaviour and maximises (??) in t1 subject to the following budget constraint: (

¯ i + Ri (τi ; K ∗ ; S ∗ ) Gi + gi = R i with: Ri (τi ; Ki∗ ; S ∗ ) = τi Bi (Ki∗ ; S ∗ )

(7)

Ri (τi ; Ki∗ ; S ∗ ) represents corporate tax revenues in country i. For simplicity, we assume all ¯ i is collected ex ante and can be viewed as other categories of public revenues to be exogenous. R receipts from the taxation of non-mobile factors (for instance a lump sum personal income tax). ¯ i reacts to neither corporate tax rates nor public input provision, it will not interfere with As R the decision maker’s choices. Furthermore, we do not allow budget imbalances in our model, so that all public resources will be used to provide the two types of public goods. Given (??) and (??), we now analyse how decision makers solve the two public finance issues addressed in this model.

3.2.1

Level of the corporate tax rate

The optimisation program leads to choosing a statutory tax rate characterised by (??). As expected, when setting the tax rate, the decision maker acts as a Leviathan and maximises domestic corporate tax revenue. This is not inconsistent with (??), which indicates that social utility increases with the total level of public spending dedicated to the provision of public goods. ∂Ri (τi ; Ki∗ ; S ∗ ) = 0 ; for i = {A; B} ∂τi

(8)

With each country acting simultaneously and non-cooperatively in t1 , the solution to this first public finance issue is a Nash equilibrium.

Proposition 1 Under an SA system, higher profit mobility triggers a race to the bottom in tax rates: i) when profit is immobile, the equilibrium tax rate only depends on the production 10

technology; ii) the equilibrium tax rates are increasing with the cost parameter of profit shifting, b. Proof. See Appendix 1. When b → +∞, the cost of profit shifting is so high that all incentives to shift profit disappear and the tax rate is simply chosen to maximise the tax revenue obtained from the ”productive” tax base (i.e. the tax base that would be determined only by the allocation of capital, in the absence of profit shifting). Therefore, (??) is simplified to: ∂ [τi F (Ki∗ ; Hi )] /∂τi = 0. As a result, the tax rate chosen by country i will be such that: τi∗ = τimax = 1 − βi . Beyond this threshold tax rate, the corporate tax burden on the firm in this country would be so high that any marginal increase in the tax rate would reduce tax revenue (this is a traditional Laffer curve effect of over-taxation). Therefore τimax is the tax rate that maximises tax revenue in country i when profit is immobile. Interestingly, it only depends on the share of human capital involved in the production process. This is due to the fact that Hi is provided by the government through the public input gi . If the production process required no human capital (hence no gi ), there would be no corporate taxation. Conversely, the more intensively the public input is used in the production process, the more corporate tax the firm is willing to pay and the higher τimax . When the cost of shifting profit is prohibitive, national corporate tax rates are independently decided from one another, without any strategic interaction. Governments can therefore set the rates at τimax , the highest possible tax rate before reaching the downward slope of the Laffer curve (dashed lines on Figure 4 ). Let us now consider the impact of profit shifting on the equilibrium tax rates. The second part of Proposition 1 tells us that the equilibrium tax rates will decrease as soon as profit mobility is more affordable to the firm. When profit is mobile, tax rates cannot be chosen so as to maximise the tax revenue from the productive tax base, and strategic interaction occurs. The tax rate equilibrium will therefore be at the intersection of the two reaction functions, τA∗ (τB ) and τB∗ (τA ). [Figure 4] The explanation is as follows: when the cost of profit shifting decreases, the firm has an incentive to evade taxation. This creates competition among governments and to preserve their tax base, they set a statutory tax rate below τimax . There is thus a traditional race-to-the-bottom mechanism whereby higher tax base mobility fosters tax competition. Simulations reported in Appendix 2 illustrate how changes in parameters affect the tax rate equilibria. We start with a benchmark case where countries are perfectly symmetric in terms of stocks of human capital and production technologies. First, we calculate the different equilibrium tax rates for alternative values of b (cases A.1 to A.3 ). Our simulations are consistent with the analysis described above, which shows that tax rates decrease in a more competitive environment 11

where profit shifting is more affordable to the private sector. In the extreme case where b → 0, profits would be entirely shifted to the country with the lowest tax rate at no cost and competition would force both countries to simply abandon corporate taxation. Second, we assume that, for exogenous reasons, the stock of human capital in both countries increases. Because of its positive effect on the marginal productivity of capital, the firm would accept a larger corporate tax burden, driving tax rates to a higher level in both countries (A.4 ). Similarly, a higher amount of human capital would mitigate the impact of a more competitive tax environment (A.5 ). Lastly, a change in production technology also affects the tax rate equilibrium: when production is more human capital intensive, the impact of tax competition on tax rates is less strong (A.6 ).

3.2.2

Composition of public spending

Let us now solve the second public finance issue. The ratio gi /Gi describes how public resources are allocated to the provision of the two types of public goods. Equation (??) tells us that society’s relative preference for gi is αi , but when deciding on the composition of public spending, the decision maker must also take into account the budget constraint (??). Therefore the composition choice may differ from αi and introduce a distortion compared to society’s relative preferences.

Proposition 2 Provided that hi 6= 0, the composition of public expenditure will not only depend on society’s relative preferences for the two public goods. There will be a systematic bias in favour of the public input. Proof. Maximising (??) subject to (??) with respect to gi and Gi gives the following relation: gi∗ = G∗i 1−

αi ∂Ri (τi∗ ;Ki∗ ;S ∗ ) ∂gi

(9)

Distortion compared to society’s relative preferences is defined as gi∗ /G∗i 6= αi and occurs if and only if: ∂Ri (τi∗ ; Ki∗ ; S ∗ ) = hi τi ∂gi



βi Fk (Ki∗ ; Hi )

βi /(1−βi ) >0

(10)

No distortion occurs if at least one of these two conditions is fulfilled: hi = 0 or τi = {0; 1}. The public decision introduces a distortion compared to society’s relative preferences because the relative share of public spending allocated to gi and Gi does not exactly reflect αi .

More precisely, as long as additional public input can increase corporate tax revenue

(∂Ri (τi∗ ; ki∗ ; S ∗ )/∂gi > 0), the ratio gi∗ /G∗i will always be greater than αi . This distortion results from the productive nature of the public input. Raising gi can attract a larger volume of capital 12

invested in country i (i.e. a larger corporate tax base), and therefore increase the tax revenue for a given tax rate. From (??) it appears that the degree of distortion increases with hi , the transformation rate of public input into human capital. This means that the distortion increases with the ability of the public input to attract corporate tax revenue. Yet this solution to the allocation problem is consistent with maximising society’s utility: allocating a larger fraction of public resources to the public input enables the government to attract a larger tax base and therefore raise the total amount of public resources dedicated to the production of both gi and Gi .

Proposition 3 When countries are perfectly symmetric, an increase in tax competition entails a reallocation of public spending in favour of the general public good.

Proof. Equation (??) tells us that the evolution of gi /Gi relative to τi follows a hump-shaped curve. In addition,  arg max  1 − τi

 

αi ∂Ri (τi∗ ;Ki∗ ;S ∗ ) ∂gi

= 1 − βi = τimax

(11)



When countries are symmetric, τAmax = τBmax corresponds to the highest possible tax rate under SA (see Proposition 1 and Appendix 1 ). Therefore, under SA, gi /Gi increases with τi and b. Since providing more public input can be used to attract the corporate tax base (provided that hi 6= 0), the degree of corporate tax competition has a significant effect on the composition of public spending: the two public finance issues addressed in this model are closely intertwined. Proposition 3 describes this interdependence. The lower graph on Figure 5 shows the pattern of distortion in the allocation of expenditure with respect to the corporate tax rate for country A. As explained above, solving the allocation issue will almost never yield gA /GA = αA , except when τA = {0; 1} or hA = 0. If there is indeed no corporate taxation, there is no need to attract a larger corporate tax base. Conversely, if all taxable profit is captured, the firm will entirely shift its profit to the other country. In both cases, the fiscal authorities do not get any revenue out ¯ A ). of corporate taxation and can only rely on other revenues to finance public expenditure (R No distortion occurs because the public input has no ability to increase tax revenue. Between these two extreme cases, and provided that hi 6= 0, the pattern of distortion with respect to the corporate tax rate follows a hump-shaped curve. Interestingly, the threshold tax rate, τAmax , is also the highest possible corporate tax rate when profit is immobile. Therefore tighter tax competition (i.e. higher profit mobility), represented here by a shift from equilibrium E0 to E1 , reduces distortions in the composition of public expenditure. [Figure 5] 13

Proposition 3 might seem counterintuitive. As tax competition becomes tighter, we would expect the decision maker to attract the more mobile tax base not only by cutting taxes, but also by allocating a greater share of public expenditure to public input provision. In their model, KM also display a systematic bias in the composition of public spending. Nevertheless, they show that tighter tax competition leads countries to spend even more on the productive public good. Starting from a similar initial bias, the mechanism displayed in our model yields the opposite outcome. The key for understanding our result is the double nature of tax base mobility in this paper: the specificity of corporate tax base mobility is that it depends on both capital and profit mobility. As profit mobility increases (i.e. b decreases), it matters gradually more than capital mobility in determining the tax base. The relative ability of the public input to raise corporate tax revenue therefore declines, whereas profit can be easily attracted through tax cuts. In the extreme case where b = 0 (i.e. under perfect profit mobility), the final location of the tax base depends entirely on profit mobility, and the public input has no impact on it. In other words, when profit is perfectly mobile, distortions in public expenditure do not pay and society’s relative preference is fully respected.

3.3

Asymmetric countries

Let us now assume that A is endowed with a higher initial amount of human capital than B. We find that A can set a higher corporate tax rate than B. As indicated in equation (??), this is because a higher amount of human capital in a given country increases the marginal productivity of capital in that country. Simulations A.1 and A.7 in Appendix 2 show that, all other things being equal, a higher stock of human capital in country A significantly increases the tax rate in that country. This allows country B to increase its own tax rate in turn. Starting from a case where countries are perfectly symmetric, this shock is illustrated in the upper left graph of Figure 6 by the new reaction function (dashed) of country A. As in Proposition 3, it entails in turn a shift in favour of public input provision in both countries (lower graphs). In other words, competition for the mobile tax base changes and focuses more on public spending. It is noteworthy that the tax rate increases relatively more in A, as Figure 6 shows. Furthermore, this country shifts a larger share of its budget to public input provision than country B. By contrast, B relies relatively more on its tax rate rather than public input provision to attract the mobile tax base. When facing competition, asymmetric countries will therefore adopt different types of strategies: the country with a higher initial amount of human capital focuses more on the expenditure tool, whereas the other country has a low-tax strategy. [Figure 6] There are other sources of asymmetry in our model. Assume for instance that production in country A is more human capital intensive, then this country will be able to set a higher tax

14

rate than B (see Appendix 2, A.8 ). Compared to the benchmark case, this will also attenuate the impact of tax competition on the equilibrium tax rates. As mentioned above, the two public finance issues are intertwined in our model, therefore asymmetries are likely to affect the composition of public spending. Based on (??) and (??), Figure 7 illustrates the pattern of public spending in the two countries for different sources of asymmetry. The left-hand side illustrates differences in terms of production technologies. When βA < βB , the high-tech country A can afford not only a higher tax rate but also a larger bias in favour of productive public spending. The right-hand side of Figure 7 describes a situation where the public input in country A is more productive (hA > hB ): we also observe a higher degree of distortion in A. [Figure 7] Clearly, these cases of asymmetry tell us that countries compete by adopting different strategies based on their comparative advantages. Countries where public input provision is likely to attract significant corporate tax revenues are less willing to compete primarily through tax cuts. Conversely, ”low tech” countries offer lower tax rates that reduce the net cost of capital to the private sector.

4

Corporate tax coordination

We now assume that a favourable political and economic environment allows tax coordination among the two countries. We investigate the case of full harmonisation, whereby a supranational central planner sets a unique tax rate τ h in both countries.5 We impose two additional assumptions. First, although the tax rate is decided at a centralised level, corporate tax revenues are still collected at the national level. Therefore, harmonising tax rates does not involve any tax base consolidation between the two countries, and the SA system still prevails in this section. Second, only the resolution of the first public finance issue (taxation) is delegated to the supra-national level; the composition of public expenditure remains the responsibility of the two countries.

4.1

Coordination among symmetric countries

5 Another

option for coordination could be that a supranational central planner decides on country-specific corporate tax rates. For symmetric countries, the rates would be equal to the unique, harmonised tax rate. Differences could only appear in the case of asymmetric countries. However, it is unlikely that computing countryspecific tax rates could be feasible in practice, and we therefore focus on full harmonisation.

15

Proposition 4 Coordinating corporate tax rates affects the two public finance issues: i) on the revenue side, tax rates no longer depend on profit mobility; ii) on the expenditure side, the bias in the composition of public spending in favour of the public input increases.

Proof. i) Maximising the central planner’s utility function: U C = UA (gA ; GA ) + UB (gB ; GB ) subject to (??) yields:   ∗ ∗ ∂ RA (τ h ; KA ) + RB (τ h ; KB ) =0 ∂τ h

(12)

This first order condition tell us that the central planner maximises τi F (Ki∗ ; Hi ) in country i, which is the condition that prevails under the decentralised solution when profit is perfectly immobile (see Appendix 1 ). As βA = βB = β, we find that τ h = τ max = 1 − β. ii) Given the harmonised tax rate τ h , each national decision maker solves its second public finance issue: gi∗ = G∗i 1−

αi ∂Ri (τ h ;Ki∗ ;S ∗ ) ∂gi

(13)

For any value of b, τ h = τ max is larger than τi∗ under tax competition (see Propositions 1 and 4i). Given (??), gi∗ /G∗i is therefore larger when tax rates are harmonised than in the decentralised case. In the previous section, we saw that profit mobility was the major source of tax competition. When we impose centralisation, (??) tells us that the central planner takes into account the incidence of the domestic tax rate on tax revenue in the other country. In other terms, centralisation internalises the strategic interaction that is responsible for the race to the bottom. With a unique tax rate in both countries, the rationale for profit shifting disappears. Therefore, b does not have any impact on the centralised tax rate equilibrium: the central planner sets the same corporate tax rate as if profit were immobile (b → +∞) in the decentralised case (see simulations B.1 to B.6 in Appendix 2 ). Nevertheless, the mechanism exposed in the previous section tells us that harmonisation does not leave the structure of public spending unchanged. Figure 8 shows that it would impose a shift in the composition of public spending in favour of the public input. More precisely, centralisation would impose the highest degree of distortion compared to society’s relative preference. [Figure 8] The explanation is as follows. Under centralisation, incentives to shift profit disappear, and the location of the corporate tax base only depends on where the firm allocates capital. Under this setting, the ability of the public input to raise corporate tax revenue reaches its peak. Public 16

spending is therefore the main instrument that countries use to attract the mobile tax base. As a result, coordination may be efficient in ruling out a race to the bottom in tax rates, but it does not eliminate international competition over the mobile tax base: it simply changes its instrument.

4.2

Coordination among asymmetric countries

We now analyse the case where production in country A (thereafter referred to as the ”hightech” country) requires more human capital that in country B, so that: βA < βB . The other assumptions on symmetry remain unchanged and, in particular, the two countries are assumed to have the same economic – or political – weight.

Proposition 5 When production technologies differ between the two countries, centralisation imposes over-taxation in the low-tech country.

Proof. The decision maker sets a harmonised tax rate τ h such that: τBmax = 1 − βB ≤ τ h ≤ 1 − βA = τAmax

(14)

B . The harmonised tax rate is an average of the highest possible tax with: τ h = 1 − βA +β 2 rates in the two countries.

In the high-tech country A, the harmonised tax rate is higher than under decentralisation, however this country can afford a higher tax rate because it remains below τAmax . On the other hand, full harmonisation can be particularly problematic for the low-tech country on which a corporate tax rate is imposed beyond τBmax , on the downward slope of its Laffer curve. The simulation B.8 reported in Appendix 2 shows that the harmonised tax rate lies between τAmax (see case B.6 ) and τBmax (see case B.1 ). More moderate arrangements than full harmonisation could soften this result, but they would in any case imply some over-taxation in country B. Figure 9 illustrates the consequences of tax coordination on the second public finance issue. In line with Proposition 3, the high-tech country allocates a larger share of expenditure to the public input than under decentralisation. The consequences for the low-tech country depend on the extent of over-taxation. If over-taxation remains limited, country B will also spend relatively more on the public input. If the harmonised rate is much higher than τBmax , B will spend relatively less on the public input. The intuition behind this outcome is as follows. As harmonisation disables competition by means of tax rates, countries compete using public input provision. However, providing more public input does not pay off in all cases. One can show that the derivative of Fk with respect to g is increasing with β for low values of β, but decreasing with

17

β for higher values of β on [0, 1]. In other terms, beyond a certain threshold, the less production is human-capital intensive, the less additional public input increases the marginal productivity of private capital, and the less distortions in the composition of public expenditure can attract private capital. If βB is too high compared to βA , country B is better off coming closer to society’s preference than competing with country A to attract more capital. [Figure 9] The theoretical findings in this section contribute to explaining why the coordination mechanisms that have been discussed in the EU so far have failed to be implemented. What we have exposed here is an extreme case, where the central planner has full power to control strategic interactions and impose tax rates on countries. In reality, political economy factors would reduce the central planner’s leeway. In particular, one can hardly expect that coordination would drive tax rates up to, not to mention beyond, their maximum levels in the sense of the Laffer curve. The actual proposals are usually less restrictive than full harmonisation and take, for instance, the form of a minimum tax rate required among a group of countries, but they still imply an increase in tax rates in the low-tax countries – which these countries are not willing to accept. When tax rates are harmonised rather than set at the national level, the countries that had low tax rates lose their comparative advantage in terms of tax rates, and cannot easily compete in terms of public input provision. This highlights the difficulties raised by tax rate coordination among a heterogeneous group of countries.

5

Common Corporate Tax Base

Given the difficulties in coordinating corporate tax rates in the EU, the European Commission has changed its approach and proposed adopting a common and consolidated corporate tax base (see EC, 2001[?]). This proposal was followed by the creation of a Common Consolidated Corporate Tax Base Working Group (CCCTB WG) in 2004. Questions on the practical implementation of such a scheme are currently being debated within the CCCTB WG and do not constitute the focus of this section. Our objective here is to investigate how shifting to a CCTB system could affect the two public finance issues presented in our analysis. Under a CCTB system, the profit of the multinational firm is consolidated. The common tax base, denoted BU , is then apportioned according to a given formula. In the large federal states where this system is currently implemented (such as the US and Canada), the apportionment formula depends on one or several reliable indicators that describe the geographical distribution of the firm’s activity (usually sales, payroll and the amount of capital in each jurisdiction). In our model, η describes how the common tax base is apportioned among the two countries: η is allocated to country A and 1 − η to country B. The only assumption that we impose on the formula is that it depends on the amount of capital in each country. η is therefore not a simple 18

parameter but a function of capital: η(KA ; KB ), such that ∂η/∂KA ≥ 0 and ∂η/∂KB ≤ 0. Note that the formula could include other factors in addition to capital, but we restrict the analysis to capital, all other things being equal. Moreover, since adopting a CCTB system does not require any coordination of tax rates, we assume that each country decides independently on the tax rate which applies to its fraction of the firm’s consolidated profit. Figure 10 summarises how the common tax base is modelled in this section. [Figure 10] The rationale for having a CCTB system is that the tax base is pooled. As a result, there cannot be any profit shifting. Under the assumptions the common tax base only depends on capital allocation: BU (KA ; KB ) = F (KA ; HA ) + F (KB ; HB ), the firm’s total net profit is as follows: Πcom = (1 − τA ) ηBU (KA ; KB ) + (1 − τB ) (1 − η) BU (KA ; KB ) ¯ −r(KA + KB ) − W

(15)

The timing of events is the same as in the previous sections: in t2 , the firm decides where to allocate its taxable profit, given the tax rates that national decision makers set simultaneously and non-cooperatively in t1 . In addition, the firm and the national decision makers are assumed to play strategically as regards apportionment, in the sense that they internalise the impact of their decisions on the apportionment formula η(KA ; KB ). This is of crucial importance for the understanding of the analysis below. Precisely because the apportionment formula depends on capital allocation, the firm and governments have incentives to use capital allocation to influence the definition of the apportionment formula. For this reason, adopting a CCTB system is likely to affect public decisions in t1 as well as the firm’s behaviour in t2 .

5.1

The firm’s decision under a CCTB system

Proposition 6 Under a CCTB system, the firm anticipates the impact of capital allocation on the apportionment formula. Proof. The firm maximises (??). The first order condition below gives us the optimal allocation of capital in country A: ∂η ∗∗ ∗∗ ∗∗ BU (KA ; KB ) (1 − τˆ) Fk (KA ; HA ) = r + (τA − τB ) ∂K A

(16) with: τˆ = ητA + (1 − η) τB Under a CCTB system, the firm decides on the allocation of capital based on an average tax rate, τˆ, which depends on the national tax rates and the apportionment formula. This

19

is as expected, since the two accounting systems are merged. However, this is not sufficient to describe the firm’s behaviour. Indeed, provided that the apportionment formula reacts to capital flows (i.e. ∂η/∂Ki 6= 0), the allocation of capital also responds to the tax rate differential. This clearly reflects the strategic behaviour mentioned above. Assume for instance that τB < τA . By shifting capital from country A to country B, the firm can modify η so that a larger fraction of the common tax base will be allocated to the low-tax country B. This will reduce the average tax rate τˆ applied on its consolidated profit and therefore the total corporate tax burden on the firm. Under a traditional SA system, the firm can influence its overall tax burden by means of profit shifting operations. A CCTB system rules out this possibility and therefore eliminates the side effects of profit mobility. Nevertheless it does not necessarily reduce the tax base volatility, as capital can be shifted as well. While profits cannot be shifted, firms may arbitrage ex ante on the location of their activity. The more the apportionment formula reacts to capital flows, the more the firm has an incentive to act strategically and make capital allocation respond to the tax rate differential.

5.2

Public decision under a CCTB system

As adopting a CCTB system changes the firm’s behaviour in t2 , it also affects public policy choices in t1 . Proposition 7 Under a common tax base system, changes in the tax rate of one country affect capital allocation in both countries. Proof. To illustrate the analysis, let us introduce two assumptions on how the apportionment formula reacts to capital: (

∂η ∂KA ∂η ∂KB

= =

γFk (KA ;HA ) BU (KA ;KB ) k (KB ;HB ) − γF BU (KA ;KB )

(17)

Without loss of generality for the results presented in this paper, we assume that the variability of the formula relative to capital is proportional to the marginal productivity of capital, i.e. it declines as the level of capital gets higher. γ ≥ 0 is the parameter that measures the degree of sensitivity of the formula to capital, such that when γ = 0, capital does not enter the definition of the formula. We can re-write (??) as follows: (

∗∗ [1 − τˆ + (τB − τA ) γ] Fk (KA ; HA ) = r ∗∗ [1 − τˆ + (τA − τB ) γ] Fk (KB ; HB ) = r

20

(18)

We can now describe how the firm adjusts the allocation of capital when country A changes its corporate tax rate:   

∗∗ ∂KA ∂τA

A HA = − (η+γ)β 1−βA r



βA ∗∗ ;H ) Fk (KA A

 

∗∗ ∂KB ∂τA

B = − (η−γ)β 1−βB



βB ∗∗ ;H ) Fk (KB B

HB r

βA /(1−βA ) βB /(1−βB )

(19)

For comparison we make the same calculation under an SA system, using the first order condition (??):   

∗ ∂KA ∂τA ∗ ∂KB ∂τA

βA HA = − 1−β A r



βA ∗ ;H ) Fk (KA A

βA /(1−βA ) (20)

=0

Comparing (??) and (??), we see that moving from an SA to a CCTB system affects the scope of strategic interactions: under an SA system, the allocation of capital in a given country depends only on the local tax rate, whereas under a CCTB system the domestic tax rate also has an impact on the amount of capital in the partner country. As expected, a cut in the corporate tax rate in one country leads the firm to increase the amount of capital in this country under both accounting systems. Under SA, tax cuts have no direct impact on capital allocation in the partner country because strategic interactions take place exclusively through profit shifting. This is different under a CCTB system where strategic interactions operate directly through capital allocation. The implications depend on the sensitivity of the apportionment formula to capital. When the apportionment factor does not react much to capital flows (γ < η), the firm’s decision depends above all on the average tax rate τˆ. In that case, as a tax cut in country A reduces τˆ, it raises the overall after-tax marginal productivity of capital and the firm can therefore afford to increase the amount of capital in both countries. For higher values of γ, however, the firm reacts more strategically to the tax rate differential. As a result, a tax cut in A leads the firm to transfer capital from B to A. Let us now investigate how strategic interactions under a CCTB system affect the two public decisions. Beginning with the issue of tax rates, country A sets a corporate tax rate subject to the following first order condition (the symmetric condition applies for country B ): η

∗∗ ∗∗ ∂[τA BU (KA ;KB )] ∂τA

+

∗∗ ∂η ∂KA ∗∗ ∗∗ ∂KA ∂τA τA BU (KA ; KB )

=0

(21)

The first term on the left-hand side of (??) is similar to (??) and (??), in the sense that tax rates are set in a way that maximises tax revenues. As in Section 4, decision makers maximise total tax revenues – i.e. in this case, the revenues that are collected from the whole common ∗∗ ∗∗ tax base, BU (KA ; KB ). In other words, adopting a common tax base leads countries to behave

like a central planner. Nevertheless, this first order condition also contains a second component.

21

The second term on the left-hand side of (??) indicates that countries need to maximise not only total tax revenues, but also the share of the tax base that will be allocated to them. In this regard, they need to internalise the impact of their decisions on the apportionment of the tax base by anticipating the strategic response of the firm to tax rate differentials. This leads them to adjust tax rates downwards. In the previous section, we showed that harmonising tax rates results, as expected, in higher tax rates than in the decentralised case. Adopting a CCTB brings about some form of coordination, however countries are led to compete on tax rates to attract capital inflows. Therefore, tax rates under a CCTB system differ from the harmonised tax rate calculated in Section 4. We can expect that the more the apportionment formula is sensitive to capital, the more the firm reacts to discrepancies in tax rates, and the more countries are likely to reduce their tax rates. In other words, the higher γ, the more likely a race to the bottom in tax rates is to occur. As a result, adopting a CCTB may prove inefficient in reducing tax competition even though it eliminates profit shifting. Competition by means of lower tax rates can still occur and might even be fostered if the apportionment formula is defined in a way that allows the firm to manipulate it significantly.

6

Concluding remarks

We have presented a model of corporate tax competition where decisions on tax rates are intertwined with decisions on the composition of public spending, as we believe that this joint analysis more accurately captures the strategy used by countries to attract a mobile corporate tax base. Another important assumption in this paper is that the mobility of the tax base does not only derive from capital mobility, but also from profit shifting. In this framework, countries have two policy instruments at their disposal to compete for the tax base: providing more public input and cutting tax rates. Our first conclusion is related to the impact of corporate tax competition on the composition of public spending. Contrary to KM, we have shown that tighter tax competition would not necessarily entail a shift to more productive public spending. We explain this result by the fact that the more mobile profit is, the more the final location of the tax base depends on profit shifting and the less the initial allocation of capital matters. This is a fresh finding compared with the traditional tax competition literature, which usually considers that more tax base mobility forces countries to restructure public expenditure and reallocate it in favour of productive purposes. Our model rejects this conclusion. Looking at alternative measures of coordination, we have shown first that the harmonisation of corporate tax rates among homogeneous countries would result in higher tax rates and a reallocation of public spending in favour of the productive public good. Indeed, even though harmonisation would prevent a race to the bottom in tax rates, it would not eliminate competition. 22

Countries would still seek to attract the mobile tax base, but attractivity would depend more on public input provision. In this regard, further investigation would be required to assess the potential welfare implications. A shift towards relatively more productive public spending would not necessarily deteriorate welfare, as the public input is also used by households. The outcome of the welfare analysis would depend on two elements: the households’ relative preferences for the productive and general public goods, and the consequences of tax rate harmonisation on the level of total public expenditure. This analysis could be conducted in a general equilibrium framework that would capture the overall effect of tax rate harmonisation. We have also established that setting a unique tax rate among a group of asymmetric countries may have harmful consequences, because it may prevent some countries (the formerly low-tax countries) from choosing a corporate tax strategy consistent with their structural characteristics. This problem is well reflected in two conflicting views on the current discrepancies in corporate tax rates in Europe. The first view, put forward by high-tax countries, argues that tax competition is harmful and advocates more coordination. The second view uses comparative advantage arguments to justify the strategy adopted by low-tax countries. More fundamentally, this debate leads us back to the underlying question: from what degree of heterogeneity does coordination become harmful? Lastly, in view of the difficulties in coordinating tax rates in practice, the countries willing to co-operate could consider a common tax base as an alternative option. However, we have shown that, when the apportionment formula gives firms a chance to manipulate the allocation of the tax base (for instance through capital flows), a new form of strategic interactions may appear. This means that a race to the bottom can still occur and a common tax base would not succeed in protecting the high-tax countries from competition on tax rates. The key issue is the definition of the formula itself. In line with Gerard (2006), our model advocates a formula that focuses on factors that cannot be easily manipulated by the firm. For instance, sales depend mainly on demand and could therefore be considered as a reasonably exogenous factor. We have presented an analysis of some of the main questions raised by corporate tax competition and coordination, but other issues are still to be tackled in future work. Among them, political economy factors may be paramount. The current absence of any corporate tax coordination in the European Union results undoubtedly from the decision-making process at the European level, which requires unanimity for decisions on taxation. The discussions on whether to adopt a common tax base and, in particular, what apportionment factors to choose, may once again prove sensitive. In addition, a common tax base would imply the end of discrepancies in the legal definitions of tax bases. Maintaining discrepancies among countries and, hence, preventing international comparisons of statutory tax rates in a transparent manner, may also well be deliberately part of government strategies.

23

References [1] Baldwin R.E. & Krugman P., 2004, ”Agglomeration, integration and tax harmonisation”, European Economic Review, 48(1), 1-23. [2] Alesina A. & Glaeser E.L., 2005, Fighting poverty in the US and Europe, Oxford University Press, Oxford. [3] Bartelsman E.J. & Beetsma R.M.W.J., 2003, ”Why pay more? Corporate tax avoidance through transfer pricing in OECD countries”, Journal of Public Economics, 87(9-10), 222552. [4] B´enassy-Qu´er´e A., Gobalraja N. & Trannoy A., 2005, ”Tax competition and public input”, Working Papers 2005-08, CEPII research center. [5] B´enassy-Qu´er´e A., Lahr`eche-R´evil A. & Fontagn´e L., 2005, ”How does FDI react to corporate taxation?”, International Tax and Public Finance, 12(5), 583-603. [6] De Mooij R. & Ederveen S., 2003, ”Taxation and foreign direct investment: a synthesis of empirical research”, International Tax and Public Finance, 10(6), 673-93. [7] Devereux M.P., Lockwood B. & Redoano M., 2008, ”Do countries compete over corporate tax rates?”, Journal of Public Economics, 92(5-6), 1210-35. [8] Dhillon A., Wooders M.H. & Zissimos B., 2007, ”Tax competition reconsidered”, Journal of Public Economic Theory, 9(3), 391-423. [9] Commission of the European Communities, 2001, ”Towards an Internal Market without tax obstacles. A strategy for providing companies with a consolidated corporate tax base for their EU-wide activities”, COM(2001) 582 final. [10] European Communities, 1998, Conclusions of the ECOFIN Council meeting on 1 December 1997 concerning taxation policy (including code of conduct for business taxation). Official Journal of the European Communities 98/C 2/01, Brussels. [11] Eggert W. & Haufler. A., 2006, ”Company tax coordination cum tax rate competition in the European Union”, FinanzArchiv, 62(4), 579-601. [12] Fuest C., 1995, ”Interjurisdictional competition and public expenditure: is tax co-ordination counterproductive?”, FinanzArchiv, 52 (4), 478-96. [13] Gordon R.H. & Wilson J.D, 1986, ”An examination of multijurisdictional corporate income taxation under formula apportionment ”, Econometrica, 54 (6), p. 1357-73. [14] Hauptmeier S., Mittermaier F. & Rincke J., 2009, ”Fiscal competition over taxes and public inputs. Theory and evidence ”, ECB Working Paper 1033. 24

[15] Keen M. & Marchand M., 1997, ”Fiscal Competition and the Pattern of Public Spending”, Journal of Public Economics, 66(1), 33-53. [16] Kolmar M. & Wagener A., 2006, ”The role of the tax base in tax competition with formula apportionment”, University of Vienna, mimeo. [17] McLure C.E., 1981, ”The elusive incidence of the corporate income tax: the state case”, Public Finance Quarterly, 9(4), 395-413. [18] Neumark-Report, 1963, The EEC Reports on Tax Harmonization. The Report of the Fiscal and Financial Committee and the Reports of the Sub-Groups A, B and C, IBFD, Amsterdam. [19] Oates W.E., 1972, Fiscal Federalism, Harcourt, Brace & Jovanovich, New York. [20] OECD, 1996, Model tax convention on income and on capital, Paris. [21] Sørensen P.B., 2004, ”Company tax reform in the European Union”, International Tax and Public Finance, 11(1), 91-115. [22] Wilson J.D, 1999, ”Theories of tax competition”, National Tax Journal, 52(2), 269-304. [23] Wilson J.D, 1986, ”A theory of interregional tax competition”, Journal of Urban Economics, 19(3), 296-315. [24] Zodrow G.R & Mieszkowski P., 1986, ”Pigou, Tiebout, property taxation, and the underprovision of local public goods”, Journal of Urban Economics, 19(3), 356-70.

25

Appendix 1: Tax rate equilibrium among symmetric countries under SA (proof ) Country i maximises its corporate tax revenue: Ri (τi ; Ki∗ ; S ∗ ) = τi [F (Ki∗ ; Hi )−δS ∗ ] with respect to τi and given τj . As the countries are symmetric, the Nash equilibrium yields: τA∗ (τB∗ ) = τB∗ (τA∗ ). (i) When b → +∞ (no profit mobility), there is no profit shifting (S ∗ = 0) and the issue is purely domestic. Each decision maker simply needs to maximise the corporate tax receipts levied on the ”productive” tax base, i.e. the tax base that results from the allocation of capital   1 1−τ ∗ 1−β before profit shifting takes place: τi F (Ki∗ ; Hi ). With Ki∗ = Hi β r i given by the firm’s decision in t2 , we find that the reaction functions are simplified to: τi∗ = τj∗ = τ max = 1 − β

(A1)

(ii) In the general case where 0 < b < +∞, and with Ki∗ and S ∗ resulting from the firm’s decision in t2 , developing (??) gives country i’s best response to country j’s corporate tax rate, τi∗ (τj ). The Nash equilibrium is described by(??), with the symmetric condition for country j : " Hi

1 − τi∗ β r

β  1−β

β 2 τi∗ − 1−β r



1 − τi∗ β r

#  2β−1 1−β

 τj∗ τi∗ − =0 + 2b b 

(A2)

We can re-write (??) as follows:   ∗ τj ∂ [τi F (Ki∗ ; Hi )] τ∗ − i =0 + ∂τi 2b b

(A3)

(??) shows that the reaction of country i has two components. The left-hand part describes the maximisation of corporate tax receipts levied on the ”productive” tax base whereas the righthand part describes the maximisation of corporate tax revenue resulting from profit shifting. On the left-hand side, τi F (Ki∗ ; Hi ) displays a traditional hump-shaped Laffer curve that is increasing with τi on [0; τ max ] and decreasing with τi on [τ max ; 1]. As τi∗ = τj∗ , the right hand side is always negative. Under an SA system, country i will therefore set its tax rate in the interval [0; τ max [ where

∂[τi F (Ki∗ ;Hi )] ∂τi

> 0. As a result, the tax rate in country i is increasing with b:

τi∗ =

  τj∗ ∂ [τi F (Ki∗ ; Hi )] +b 2 ∂τi

26

(A4)

Appendix 2: Tax rate equilibria under SA (simulations)

τA

τB

A Decentralised equilibrium

B Tax rate harmonisation

Symmetric countries 0.27

1

Benchmark case*

2

Higher cost of profit-shifting

b = 1 .5

3

Lower cost of profit-shifting

b = 0.05

4

Higher stock of human capital

HA = HB = 5

5

(3) + (4)

6

Change in technology

0.27

0.5 0.5

0.45 0.45

0.5 0.5

0.11 0.11

0.5 0.5

0.35 0.35

0.5 0.5

0.18 0.18

0.5 0.5

0.43

βA = βB = 1 3

0.43

0.66 0.66

Asymmetric countries 0.29

7

Asymmetry in human capital

HA =5

0.34

8

Asymmetry in technology

βA =1 3

0.35

0.5 0.5

0.29

0.59 0.59

* Simulations in the benchmark case are based on the following values: H i = 3; b = 0.2; r = 1; β = 0.5.

40

%

35

Ruding proposal 30 25

EU-27 average

20 15 10

Cyprus

Ireland

Bulgaria

Latvia

Lithuania

Romania

Hungary

Poland

Slovakia

Estonia

Czech Republic

Portugal

Slovenia

Greece

Denmark

Austria

Netherlands

Finland

United Kingdom

Sweden

Germany

Spain

Luxembourg

Italy

France

Malta

0

Belgium

5

Fig.1. Top statutory corporate income tax rates in EU countries, April 2008 In black: euro area Member States as of April 2008 (Slovakia joined in January 2009) Source: KPMG's Corporate and indirect tax rate survey, 2008.

12 10 8

EA-16 average

EU-25 average

6 4 2 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Fig. 2. Corporate income tax revenue as percentage of total taxation Source: European Commission, Taxation Trends in the European Union, 2009.

~ Hi > Hi r 1 −τ i ~ Fk ( K i ; H i ) Fk ( K i ; H i ) Ki Fig. 3. Capital allocation

τB Highest possible tax rates

1

τ *A ( τ B ) τ Bmax τ *B ( τ A ) τ *B

0

τ *A

τ Amax

Fig. 4. Tax rate equilibrium

1

τA

τB

τ *A ( τ B )

1

τ *B ( τ A ) EO

E1 0

1

τA

1

τA

gA GA EO

E1

αA

0

τ

max A

Fig. 5. The two intertwined public finance issues

τB

τB

τ *A ( τ B )

1

1

τ *B ( τ A ) E1 EO

H A1 > H A0 = H B0 1

0

gA GA

τA

0

1

τB

1

τB

gB GB

E1

E1

EO

EO

α

α

0

1

τA

0

Fig. 6. Asymmetric countries

hA > hB

β A < βB gi Gi

gi Gi

gA GA

gB GB

gB GB

α

0

α

1

τ Bmax

τ Amax

gA GA

τi

0

1

τ Amax = τ Bmax

Fig. 7. Two examples of asymmetry

gi Gi

Eh

EO

α

0

1

τ

* i

τ =τ h

τi

max

Fig. 8. The impact of harmonisation among symmetric countries

τi

τB

τ *A ( τ B )

1

1

τ Amax Eh

τh

τh τ τA ) * B(

τ *B

EO 0

τ

gA GA

τ

* A

τ

h

τ Bmax

1τA max A

1

0

gB GB

Eh EO

Eh

EO

αB

αA

1

0

τ *A

τ

h

τ

max A

τA

0

1

τ τ * B

max B

τ

h

Fig. 9. Harmonisation among asymmetric countries

τA

η

τB

BU

1-η

Multinational activity

Fig. 10. The CCTB system

τB