Corporate Taxes and Headquarters Relocation:
A comparative study of relocation decisions by multinational enterprises within the European Union
Master Thesis
MSc International Business & Management
Tobias Schlüter
t.schluter.2@student.rug.nl Student number: S3558770
University of Groningen, The Netherlands Faculty of Economics and Business
Supervisor: Prof.dr. D.J. Bezemer Co-assessor: Dr. K. van Veen
ABSTRACT
The focus of this study is on the influence of corporate income taxes on the likelihood of headquarters relocation. Therefore, two different taxes are taken into consideration, the statutory corporate income tax rate and the effective average tax rate. Although there exists broad academic literature on the influence of taxation on firm’s location decisions, the role of different corporate income taxes on headquarters relocation within the European Union remains widely unclear. This study aims at filling this gap by examining 404 intra-EU headquarters relocations in the 2009-2017 period. Since most MNE headquarters relocations are taking place a consequence of mergers and acquisitions this study makes use of cross-border M&A deals to investigate the influence of taxation on relocations. Therefore, this study controls for country-specific variables and examines the influence of firm-country-specific moderators. The results support the findings of former research as a decrease in corporate income tax payable has a significantly positive impact on the possibility of HQ relocation. Furthermore, this study comes up with a unique feature, due to the research design results allow to compare the impact of two different tax rates on HQ relocation. Contrary to the common notion of literature this study presents evidence that MNEs are more responsive on statutory corporate income taxes rather than on effective average tax rates.
TABLE OF CONTENTS
LIST OF TABLES AND FIGURES ... I LIST OF ABBREVIATIONS ... II 1. INTRODUCTION ... 1 2. LITERATURE REVIEW ... 4 2.1. HEADQUARTERS RELOCATION ... 4 2.1.1. Internationalization ... 4 2.1.2. Headquarters location ... 4
2.1.3. Prior research on headquarters relocation ... 5
2.1.4. Prior research on the influence of taxation on headquarters relocation ... 5
2.1.5. M&As and relocation ... 6
2.2. CORPORATE TAXATION AND TAX AVOIDANCE ... 7
2.2.1. Neo-classical location theory ... 8
2.2.2. Statutory corporate tax ... 8
2.2.3. Corporate effective average tax rate ... 9
2.2.4. Double taxation and the repatriation tax ... 11
2.2.5. Tax competition ... 11 2.2.6. Tax avoidance ... 13 2.2.6.1. Profit shifting ... 13 2.2.6.2. Transfer pricing ... 14 2.2.6.3. Debt shifting ... 16 2.3. CONCEPTUAL MODEL ... 17
3. DATA & METHODOLOGY ... 18
3.1. SAMPLE ... 18
3.2. DATA ... 19
3.2.1. Independent Variables ... 19
3.2.1.1. Statutory Corporate Income Tax Rate ... 19
3.2.1.2. Corporate Effective Tax Rate ... 19
3.2.2. Dependent Variable ... 23 3.2.3. Moderator Variables ... 25 3.2.4. Control Variables ... 25 3.3. EMPIRICAL MODEL ... 26 3.4. RESULTS ... 26 4. CONCLUSION ... 30 4.1. DISCUSSION ... 30 4.2. LIMITATIONS ... 32
4.3. FUTURE RESEARCH DIRECTIONS ... 33
4.4. CONCLUSION ... 33
REFERENCES ... 35
APPENDICES ... 41
APPENDIX A:DESCRIPTION OF VARIABLES (INCL. DATA SOURCES) ... 41
APPENDIX B:EXAMPLE-CALCULATION OF THE EATR ... 42
APPENDIX C:STATUTORY CORPORATE INCOME TAX RATES ... 44
APPENDIX D:EFFECTIVE AVERAGE TAX RATES ... 45
APPENDIX E:UNEMPLOYMENT RATES IN THE EU-28 ... 46
APPENDIX F:GPDS WITHIN THE EU-28 ... 47
APPENDIX G:GDP GROWTH RATES ... 47
LIST OF TABLES AND FIGURES
Figure 1. Conceptual Model ... 17
Figure 2. Initial Dataset (source: Orbis and Zephyr databases) ... 18
Figure 3. Intra-European Headquarters Relocations - An Overview ... 24
Figure 4. Test of Normality ... 27
Figure 5. Summary of descriptive Statistics and Correlations ... 28
Figure 6. Binary Logistic Regression ... 29
Figure 7. Description of Variables (incl. data sources) ... 41
Figure 8. Statutory Corporate Income Tax Rates (source: European Commission) ... 44
Figure 9. Effective Average Tax Rates (source: European Commission) ... 45
Figure 10. Unemployment Rates of EU member states (source: Eurostat) ... 46
Figure 11. GDPs of EU member states (source: Eurostat) ... 47
Figure 12. GDP growth rates (source: Eurostat) ... 47
LIST OF ABBREVIATIONS
BvD - Bureau van Dijk
CEO - Chief Executive Officer CFC - Controlled Foreign Company CIT - Corporate Income Tax EATR - Effective Average Tax Rate EMTR - Effective Marginal Tax Rate EMEA - Europe, the Middle East and Africa ETR - Effective Tax Rate
EU - European Union
FDI - Foreign Direct Investment FIFO - First In – First Out
GDP - Gross Domestic Product
HQ - Headquarters
LPI - Logistics Performance Index M&A - Mergers & Acquisitions MNC - Multinational Company MNE - Multinational Enterprise
OECD - Organization of Economic Co-Operation and Development
1. INTRODUCTION
„Yes, of course Google minimizes its tax bill by operating in Bermuda and Ireland”, this statement made by Google’s President of EMEA Business & Operations Matt Brittin underlines the importance of taxes on the location of multinational’s headquarters. In addition, he justifies Google’s approach to avoid taxes by the duty to minimize its shareholder’s costs (Knight, 2012). Other popular examples of tax avoiding multinational enterprises (MNEs) in Europe are the US coffee company Starbucks, the Swedish furniture empire IKEA, and the very recent case of the British-Dutch consumer goods giant Unilever (Knight, 2012; Pieters, 2018; Ewing, 2018). The strategies of tax avoiding behavior vary between the different companies, some MNEs try to reduce their tax bill by placing their headquarters to low-tax jurisdictions, others minimize their shareholder’s expenses by moving their headquarters to countries which do not levy any dividend taxes. Exemplary for the latter approach is the case of Shell, the British-Dutch oil and gas company established a special tax ruling with the Dutch Tax Authority in 2005, which ensured that its British shareholders will not have to pay dividend taxes in the Netherlands, before the MNE moved its headquarters to The Hague (Pieters, 2018). After the cross-border relocation of IKEA’s headquarters to Leiden, officials in Brussels accused the Netherlands of violating European Union’s rules by helping the Swedish flagship to minimize its tax payments (Ewing, 2018). Nevertheless, the Netherlands are not the only EU member country which tries to attract the world economy’s big players in order to create jobs and boost the domestic economy.
The aim of this study is to measure the tax-related cross-border mobility of multinationals’ headquarters within the European Union during the time span from 2009 to 2017. In accordance with the broad literature on this topic this study presents the effective average tax rate as an additional tax rate which is considered to be a very good proxy for whole tax systems, as it takes different economic and domestic tax parameters into account. Both the statutory and the effective average tax rate are well established in the theoretical literature and are considered to be the relevant tax variables for multinational companies (Schaffer & Turley, 2000; Devereux & Griffith, 2003; Overesch & Rincke, 2011).
Although there exists extensive research on the relationship of taxation on a company’s location decisions, the role of a country’s tax policy remains unclear. While Clausing (2010) can’t find significant evidence for a relationship between headquarters’ locations and taxes a vast majority of research presents a contrary view on that. There is a lot of literature about this topic focusing on various geographic regions (e.g. German municipalities, US states or a global perspective) and different taxes (withholding taxes, repatriation taxes and controlled foreign corporation legislations). This work contributes to the existing literature by focusing on the impact of two different corporate tax rates on multinationals’ headquarters intra-Europe, cross-border relocations. Therefore, the role of the statutory corporate income tax and the effective average tax rate will be evaluated and compared. The results are suggested to give a meaningful response to the following research question:
How do statutory and effective taxes on corporate income affect the probability of headquarters relocation within the European Union?
This study answers the research question using a dataset with a sample of 6300 multinational firms based in the actual 28 EU member countries, taking into account headquarters relocations between 2009 and 2017. In accordance with previous large-scale, firm-level data approaches, this study makes use of mergers and acquisitions (M&As) to identify cross-border headquarters relocations within the EU-28. While controlling for country-level effects, this study is considering firm-specific tax-avoiding strategies (transfer pricing, profit- and debt-shifting) within its conceptual framework. This paper will henceforth not only give a response to the research question, it will also offer a new perspective on the reasonableness of tax competition between the European countries.
2. LITERATURE REVIEW
2.1. Headquarters relocation
This section will initially introduce the different degrees of internationalization. In a further step, headquarters location will be discussed and the most appropriate definition for this study will be presented. The subsequent sections present an overview about the current research in this field, especially discussing the role of taxation and M&As in greater detail.
2.1.1. Internationalization
When firms want to maintain their international competitive advantage, they have to add to their current advantages or offset disadvantages by internationalization into foreign markets (Porter, 1990). Forsgren, Holm and Johanson (1995) distinguish three degrees of internationalization. The first degree covers all the initial steps of internationalization such as exporting, outsourcing of production, sales and R&D (Benito, Larimo, Narula & Pedersen, 2002; Barner-Rasmussen, Piekkari & Björkman, 2007). The second degree refers to more advanced stages of internationalization in which subsidiaries may develop into strategic centres with more responsibilities than in the first degree. Building on the initial work of Forsgren et al. (1995), international headquarters relocation may be recognized as the third degree of internationalization (Barner-Rasmussen et al., 2007; Laamanen, Simula & Torstila, 2012).
2.1.2. Headquarters location
depending on the operation?. This work defines headquarters as the residence (country or jurisdiction) of a firm where the majority of shares are owned. It can be assumed that the place of the majority ownership and the tax residence coincide in that country (Voget, 2011).
2.1.3. Prior research on headquarters relocation
Prior research on international HQ relocations has focused mainly on the effects of various firm-specific characteristics such as the relationship between corporate and divisional headquarters, type of ownership and the extent of international involvement (Laamanen et al., 2012). Other studies focused on the main determinants of headquarter locations in a national and an international context (Klier & Testa, 2002; Forsgren, Holm & Johanson, 1995). Brouwer, Mariotti and Ommeren (2004) examined the mobility of firms in an international relocation context. Gregory, Lombard and Seifert (2005) used a sample of 167 corporate headquarters relocations to investigate a de- or increase in operating performance after relocation. Another study of Birkinshaw, Braunerthjelm, Holm and Terjesen (2006) documented different factors contributing to relocation of corporate headquarters, while an extensive study of Strauss-Kahn and Vives (2009) focused on specific location features as well as drivers that enhance relocation (firm- and location-specific).
2.1.4. Prior research on the influence of taxation on headquarters relocation
cross-state (within the United States) headquarters relocation ranging from 9.3% to 27% (Chow, Huang, Klassen & Ng, 2018).
Becker, Egger, and Merlo (2012) made use of a dataset of more than 11,000 German municipalities in order to investigate the sensitivity of MNEs’ locations. Therefore, they examined the relationship of taxes which are levied by municipalities on the number of relocations. Their results indicate that a decline in corporate tax rates leads to increase in MNEs’ headquarters within a certain municipality. There are three papers that employ large-scale firm-level data (Clausing, 2010). First Barrios, Huizinga, Laeven and Nicodème (2012) were looking at the influence corporate taxation (including double taxation) on firms’ location decisions. Their outcomes present evidence that taxation in the home country is influencing MNEs’ location decisions. A similar approach of Voget (2011), showed that repatriation taxes are influencing multinationals’ location decisions significantly. A sample of more than 2,000 MNEs and 140 relocations (in the 1997-2007 period) were taken into account. The findings of this study provide proof that a ten percent increase in repatriation taxes leads to a 2.2 percent growth of cross-border relocations. Also controlled foreign corporation (CFC) legislation is found to positively influence the probability of headquarters relocations (Voget, 2011). Another study within this field provides evidence that firms engage in tax avoiding behavior after cross-border mergers and acquisitions. Huizinga and Voget (2009) therefore investigated the parent-subsidiary structure of MNEs. Their research involved the US, Japan as well as some European jurisdictions between 1985 and 2004.
2.1.5. M&As and relocation
One possible way for a multinational firm to engage in tax avoidance is to install a subsidiary in a low-tax in order to shift its profits to that jurisdiction. There are two common types to set up a subsidiary in a foreign tax system – foreign direct investment (FDI). First, greenfield investment, which means to build up a company and its operation from ground up in a foreign country. Second, cross-border mergers and acquisitions (M&As), which refers to the fusion with- or the bargain of an existing company abroad. Since this study makes use of cross-border M&A deals as a measure of headquarters relocation, the following part will focus on that specific type of FDI and underline the growing importance of that entry mode (von Hagen & Prettl, 2017; UNCTAD, 2017).
(UNCTAD), he also predicts that the value of M&A deals will exceed 80 percent of the global FDI flows within the next years. Furthermore, Ali-Yrkkö & Ylä-Antilla (2004) underline the importance of M&As by stating that in almost all cases, the relocation of Finnish headquarters has taken place as a consequence of a M&A transaction. Similar to the Finnish case, Braunerhjelm’s study (2004) shows that a great majority of Swedish headquarters relocations are the result of M&As.
The basic assumption of M&As is that these deals are settled due to ‘ownership advantages’. Such advantages occur when the ownership change is expected to add further value to the company. Such value creations can appear as increase or as decrease in the target’s future cashflows or a decrease in the risk a company is exposed to (Belz, Robinson, Ruf & Steffens, 2013). A typical example for an action that enhances the firm’s value is the replacement of a poor management team (Belz et al., 2013; Auerbach & Reishus, 1987).
Ownership advantages can occur in various different manifestations. Still, most of the existing literature concerning FDI focuses mostly on greenfield investment or the change in a company’s operating performance after a M&A deal took place. As a result, the importance of taxation and its management remains almost entirely unexplored, although lowering a firm’s tax liabilities would provide a great opportunity to bring forth ownership advantages (Belz et al., 2013; Becker & Fuest, 2011).
Nonetheless, M&As are the FDI type which gained more and more importance over last years. In 2011, the value of global cross-border M&A deals rose by over 50 percent to $526 billion, greenfield investment projects accounted for a value of $904 billion (UNCTAD, 2012). After chirpy activities in developed markets over the past decade, the value of global FDI flows total up to $1.746 billion, in 2016. The deal value covered by M&A deals accounts for $869 billion (18 per cent increase compared to 2015) and exceeds the total value of greenfield investment projects which is at $828 billion (7 per cent increase from 2015) (UNCTAD, 2017). To sum up, the entry mode of mergers and acquisitions becomes increasingly important and offers several opportunities for MNEs and their shareholders to avoid taxes. Therefore, corporate tax rates are expected to influence investment decisions (Auerbach & Reishus, 1987; Belz et al., 2013; Blouin, Fich, Rice & Tran, 2018).
2.2. Corporate taxation and tax avoidance
why different governments are engaged in a global arena of tax competition. Afterwards this section presents how firms deal with given corporate taxes and how they try to minimize their tax payments in order to maximize their after-tax profits.
2.2.1. Neo-classical location theory
This study makes use of the neo-classical location theory that focuses on the premise of the rationale firm that maximizes profit in choosing the optimal location (Bouwer, Mariotti & Ommeren, 2004). According to this theory, managers strive to maximize after-tax value. On the other hand, governments attempt to strengthen the economy while raising tax revenues. During this process, firms often try to organize their operations to reduce exposure to higher-tax jurisdiction (Chow, Huang, Klassen & Ng, 2018).
2.2.2. Statutory corporate tax
The statutory tax rate is a tax imposed by a jurisdiction on the corporate income of a company (Slemrod, 2004). The literature distinguishes between two commonly accepted reasons for levying corporate taxes. First, corporate taxes serve as a kind of withholding tax that function as backstop for personal tax. The second reason is that corporate taxes can basically seize the payments earned by owners of fixed factors without distorting the investment decision (Becker & Fuest, 2007).
There is extensive empirical research on statutory corporate taxation and its formation, such literature is mostly based on the interdependence of different tax jurisdictions and various economic variables (e.g. country size, a country’s budgetary needs, and the openness of a country to foreign direct investment and MNEs) (Gérard & Ruiz, 2009). Other research within this field is focused on the corporate tax rate of neighbouring countries and suggests it as a main determinant of a country’s taxation (Devereux, Lockwood & Redoano, 2008). A large part of the literature is considered to be motivated by the research on tax competition, which is sharply increasing since the 1980’s (Gérard & Ruiz, 2009).
28 per cent in 2008) and Italy (from 33 to 27,5 per cent in 2009) (Loretz, 2008). A similar development has been found for the whole European Union, the average statutory tax rate shrank from 35.5% in 1995 to 25.3% in 2006. What’s interesting is that the falling tax rates haven’t been found replicated in changes of the revenues from corporate income taxes. While the importance of them increased substantially, the share of corporate income taxes to GDP increased from 2.7% to 3.3% within the same time period (Piotrowska & Vanborren, 2008). Policy makers often describe these actions as responses to tax cuts in their neighbouring countries (Loretz, 2008). Section 2.2.5. will discuss the competition on corporate income taxes in greater detail.
These findings suggest that differentials in corporation income tax rates influence foreign direct investment considerably, as well as the choice of company’s locations (Bénassy-Quéré, Fontagné & Lahréche-Révil, 2005; Keuschnigg, 2009). This makes it very important for policy makers to deeply understand the determinants of corporate income tax revenues (Piotrowska & Vanborren, 2008). This is reflected in Hypothesis 1.
Hypothesis 1: Differences in the statutory corporate income tax between home and host country are positively linked to the possibility of headquarters relocation.
Although, the next section (2.2.3.) describes the corporate effective average tax rate as a more appropriate measure to determine headquarters relocations, the statutory tax still has its justification. Some reasons therefore are the fact that the EATR converges to the statutory rate as profits increase. Sometimes the statutory tax rate may be the best information available for multinationals. In addition, if affiliates are just used as profit shifting vehicles, the statutory tax rate matters, because no real investment activity is involved. And finally, some central elements of the effective average tax rate aren’t important for the MNE’s affiliates (e.g. depreciation allowances for non-manufacturing subsidiaries) (Overesch & Wamser, 2009).
2.2.3. Corporate effective average tax rate
the amount to tax a business ultimately pays on its income”. Effective tax rates allow to take account of such factors, that determine a tax system and are therefore considered to better measure actual tax payments. Furthermore, they help to make meaningful comparison of different tax jurisdictions (Hansson, Porter & Williams, 2012; Katsikas & Lewis, 2016).
Similar to the developments of the statutory corporate tax rate, also the corporate effective tax rate applied on a company’s income declined over the last years. A possible explanation for these trends is provided by globalization and tax competition (Slemrod, 2004; Hansson, et al., 2012; Dias & Reis, 2018). These findings are supported by the findings of Dyreng, Hanlon, Maydew & Thornock (2017), they found evidence that effective tax rates have decreased by 0.4 percentage points per year over the past 25 years. Another study found that there exist substantial differences in the level of statutory and effective tax burden, especially among a group of older and newer EU Member States (Andrejovská, Mikóková & Martinková, 2017). Despite those differences, both the statutory as well as the effective corporate income tax have been found to influence a company’s foreign direct investment decision significantly (Bénassy-Quéré, Fontagné & Lahrèche-Révil, 2000).
A lot of attention came to multinational firms and their cross-border activities, due to their ability to shift taxable income from high- to low-tax countries (Dharmapala & Riedel, 2013; Dyreng et al., (2017). Therefore, multinationals are suggested to examine a smaller effective corporate income tax burden (Rego, 2003).
Contrary to the assertion of Dias & Reis (p. 4-5, 2018), that Dyreng et al. (2017) found “the ETR decreased in a stronger way in multinational companies than in domestic firms”, Dyreng et al. (2017) didn’t find any support of a concentrated decrease in effective tax rates in multinational firms. They found “essentially the same decrease in effective tax rates over time among purely domestic firms as among multinationals” (p. 461).
Devereux and Griffith (1998; 2003) argue that the location choice of MNEs strongly depends on the earnings after the deduction of taxes, and therefore describe the effective tax as the right indicator to measure this relationship.
To sum up, such as the statutory corporate tax rate, the corporate effective tax rate is considered to have a distinct impact on a firm’s choice of headquarters location. Due to their ability to shift profits across borders, MNEs are especially vulnerable to this tax rate. This reasoning is reflected in Hypothesis 2.
Hypothesis 2: Differences in the effective average tax rate between the home and the host country are positively linked to the possibility of headquarters relocation.
2.2.4. Double taxation and the repatriation tax
“Corporate firms are subject to the corporate income tax and the income tax that applies to either profit distributions or realized capital gains, while taking into account the double tax relief if appropriate” (de Mooij & Nicodème, p. 480, 2008). When speaking about the double tax relief it is very important to mention the OECD model tax convention, which explains the recommended procedure, and therefore offers tax jurisdictions the choice between an exemption and a foreign tax credit as the only two possibilities to relieve for double taxation (Huizinga & Voget, 2009). While most countries (e.g. EU countries) make use of an exemption system that relieves foreign income of their resident firms, others (e.g. the United States, Russia and India) apply a tax credit system, taxing foreign income when it is repatriated to the home country and offering tax credit for taxes paid to foreign governments. Many countries use hybrid systems that exempt some types of income, but also tax other types of foreign income (de Mooij & Nicodème, 2008). International double taxation occurs when the same tax and taxable materials are collected for the same period of time, by the public authorities from different countries (Radu, 2012).
Since this study focuses solely on member countries of the European Union, double tax reliefs will not be taken into account, because most of these states exempt for profits made abroad. Nevertheless, for the sake of completeness it is important to mention double taxation within this paper.
2.2.5. Tax competition
(Knight, 2012). The literature surprisingly paid very little attention on defining tax competition, a very wide definition describes every form of non-cooperative tax setting by independent governments as tax competition. Such a broad definition covers vertical as well as horizontal tax competition (Wilson & Wildasin, 2004). “Vertical tax competition takes place among different levels of government such as federal government and state governments within the US” (Wilson, p. 289, 1999). Since this study aims to compare the influence of taxation on headquarters relocation from different and independent EU member countries, a narrower definition will be used. Contrary to the vertical form, horizontal tax competition focuses on the arena where governments at same level are competing (Wilson, 1999; Wilson & Wildasin, 2004). Fox, Hill & Murray (2015) describe the corporate income tax as one of the primary taxes at the same level. They also found rich anecdotal evidence of horizontal tax competition across different countries.
The literature describes three common types of tax competition. The first type occurs when individuals shift their portfolio capital to another jurisdiction in order to avoid paying tax on their capital gains. Usually such capital is hidden behind the covering of secrecy imposed by, for instance, banks in tax havens such as Switzerland. The second type refers to multinationals that shift their paper profits from high-tax jurisdictions to low-tax jurisdictions (e.g. Ireland and Luxemburg) by using different accounting techniques (e.g. intra-firm transfer pricing) in order to hide the profits from the government. The third type of tax competition refers to independent governments that adjust their policies (e.g. lowering corporate tax rates, imposing subsidization) in order to entice foreign direct investment (FDI) from multinational enterprises (McLure, 1970; Desai, Foley & Hines, 2006; Dischinger, Knoll & Riedel, 2014; Leung, 2017).
A strong tendency toward less than efficient level of output of local services will possibly be the result of tax competition. In that sense that political decision makers may hold spending under the levels for which marginal benefits equal marginal costs, in order to attract FDI to their countries (Oates, 1972). According to Wilson (1999), “a central message of the tax competition literature is that independent governments engage in wasteful competition for scarce capital through reductions in tax rates and public expenditure levels” (p. 269).
(Winner, 2005). Therefore, these factors are suggested to influence the relationship of taxes on the possibility of headquarters relocation. Consequently, these factors will be taken into account – as control variables – within the empirical framework, which will be presented later within this work.
2.2.6. Tax avoidance
2.2.6.1. Profit shifting
As mentioned above, profit shifting is one of the major types of tax competition. The increasingly accelerating globalization is the main reason why tax havens (e.g. Switzerland for secrecy reasons and Ireland as well as Luxemburg for their low-tax jurisdiction) have become so important over the past decades. There are increased opportunities for MNEs to shift their profits towards countries with a low corporate tax. These developments have changed the strategic tax game for international profits and pose new challenges to policy makers (Krautheim & Schmidt-Eisenlohr, 2011).
Common profit shifting strategies contain the tax-favored distortion of intra-firm transfer prices and the debt-equity structure or opening an affiliate in a tax haven and shifting profits as well as highly profitable assets (e.g. patents and trademarks) abroad (Krautheim & Schmidt-Eisenlohr, 2011; Dischinger et al., 2014). Such highly profitable intangible assets are more and more seen as the key to competitive success as well as the main driver of corporate profit. In addition, patents and trademarks offer a lot of profit shifting opportunities for MNEs, because of a highly non-transparent and firm-intern process of profit shifting (Dischinger & Riedel, 2011).
Especially multinational firms are attracting a lot of attention within the literature on profit shifting, due to the fact that those firms usually have an extended ability to shift profits and assets cross borders in order to avoid paying taxes in a higher-tax jurisdiction (Dyreng, Hanlon, Maydew & Thornock, 2017).
Therefore, it makes perfect sense that this study focuses on the headquarters of tax-avoiding MNEs. Referring to one of the common profit-shifting strategies – opening an affiliate in a low-tax jurisdiction – many former studies (Brouwer et al., 2004; Richardson & Lanis, 2007; Strauss-Kahn & Vives, 2009; Fernández-Rodriguez & Martinez-Arias, 2011; Wang, Campbell & Johnson, 2014) stated that firm-specific factors, like the firm size, are influencing the (re-) location decision significantly. Consequently, this paper suggests firm size to play an important role when it comes to the relocation of a corporate headquarter. This is reflected in the Hypothesis 1a) and Hypothesis 2a).
H1a: A target firm’s size positively affects the relationship of statutory corporate tax rate differentials between home and host country.
H2a: A target firm’s size positively affects the relationship of effective average tax rate differentials between the home and the host country.
2.2.6.2. Transfer pricing
There is an increasing amount of new multinational enterprises resulting from the globalization of the world economy. Therefore, the number of cross-border transactions between their affiliates is growing extensively (Yao, 2013). MNEs usually have great reticence about their structures and the way how they engage in internal trading, which makes it easy for them to allocate parts of their income from high-tax countries to affiliates from low-tax jurisdictions in order to increase their after-tax profits (OECD, 2012; Grubert & Mutti, 1991). A common vehicle to shift profits across different countries is the pricing of goods exchanged between related parties – known as transfer pricing. Some multinationals manipulate these intra-firm prices by under- or over-invoicing in order to exploit cross-border tax differences. Under-invoicing occurs when – for instance – the exports from a high-tax to a low-tax country are sold extensively cheaper than either the production or market prices. Over-invoicing is related to shifting deductible expenses towards high-tax countries in order to reduce the overall corporate tax payments (Christea & Nguyen, 2016; Eden, 2001). A survey of MNE’s tax directors, states that transfer pricing strategies and practices vary substantially across multinationals (Klassen, Lisowsky & Mescall, 2017).
Erosion and Profit shifting of the OECD (2013), a tax haven is not just defined as a low-tax jurisdiction, furthermore tax havens seek to facilitate tax avoidance by “artificially segregating taxable income from the activities that generate it” (p.13) (Davies, Martin, Parenti & Toubal, 2018).
The fact that the countries of the European single market levy different corporate income taxes and separate taxation of profits of each foreign subsidiary, gives MNEs the opportunity to avoid taxes (Dischinger, 2007). An example from Denmark underlines these findings. After the acquisition of an affiliate in a low-tax county, Danish multinationals reduce the prices on their exports to that country on average by 5.7 to 9.1 percent. Such a reduction in price translates to $141 million in under-reported export revenues, which then equals about 3 percent of Danish MNC’s tax returns (Cristea & Nguyen, 2016). Findings like this are the reason why it is of utmost importance for governments, particularly high-tax jurisdictions, to protect their decreasing national corporate tax revenue against any kind of transfer price manipulating behavior (Dischinger, 2007).
Therefore, the OECD (2012) imposed transfer pricing rules that “require MNEs to price, for tax purposes, their internal or intra-group transactions and calculate profits as if the transaction had taken place between independent businesses – the arm’s length principle” (p.14). The goal of this principle is to ensure a consistent basis for profit allocation beneath all affiliates of an MNE and all transactions are treated with fairness and legality (OECD, 2012; Yao, 2013).
Although there are initiatives such as the OECD’s transfer pricing roles, profit shifting strategies are still common in the global tax arena and some MNEs are more likely to engage in such tax avoiding behaviour and consequently relocating their assets to low-tax jurisdictions. Firms from the knowledge-intensive sector are suggested to exhibit a different degree of mobility as well as lower taxes due to the prominent role of intangible assets (Voget, 2011). It is reasonable that intangible assets such as patents are easier to shift across borders. Taking the knowledge-intensiveness into account, these findings are reflected in the following hypotheses.
H2b: The degree to which a firm’s core business is knowledge-intensive positively affects the relationship of differences in the effective average tax rate and the likelihood of headquarters relocation.
2.2.6.3. Debt shifting
An additional type of profit shifting is the so called, debt shifting or thin-capitalization. Basically, debt shifting is the procedure of multinational companies earning interest income in low-tax jurisdictions and deducting those payments in higher-tax jurisdictions with the aim that tax savings arising from the deductions in higher-tax area exceed the corresponding tax liabilities in a low-tax jurisdiction (Schindler & Schjelderup, 2013; Huizinga, Laeven & Nicodeme, 2008).
Thin-capitalization can be divided into two sub-types: external and internal debt shifting. The external form is characterized by the extensively loading of those affiliates generating high tax savings with external debt, while reducing the use of external debt in affiliates with low tax savings. The internal form of debt shifting refers to a strategy of borrowing and lending among linked affiliates within the MNE’s group. The mode of operation is to deduct interest in high-tax countries and to gain interest in low-tax countries in such a way that the tax savings in high-tax countries exceed the increased tax liability in low-tax countries (Ruf & Schindler, 2015; Schindler & Schjelderup, 2013). However, in the corporate finance literature each of the two manifestations of debt shifting, a multinational’s capital structure is always for central matter of concern. The most obvious advantage of debt compared to equity financing relates to the general tax-deductibility of interest expenses at the corporate level (Merlo & Wamser, 2014; Blouin, Huizinga, Laeven & Nicodème, 2014).
Debt shifting and thin-capitalization of firms have attracted increased attention of literature in corporate finance as well as policy makers. The ongoing integration of national financial markets and the fact that there is an extensive growth in the number of multinational enterprises globally are just some reasons that there is increasing concern that governments lose significant parts of their corporate tax revenues (Schindler & Schjelderup, 2013; Ruf & Schindler, 2015). Huizinga et al. (2008), suggest that ignoring the international shifting activities which are arising from differences in national corporate taxes downplay the impact of national taxation taxes on debt policies by about 25 percent.
unrelated parties (OECD 2012; Schindler & Schjelderup, 2013). According to Desai, Foley and Hines (2004), such “rules typically are vaguely worded and seldom, though arbitrarily, imposed, making their effects difficult to analyze quantitively; any impact they have is likely to reduce the estimated significance of factors influencing total indebtedness”. Blouin et al. (2014), complement these findings by stating that the effectiveness of thin-capitalization rules depends on the extent to which they are enforced by responsible tax administrations.
The central measure, when it comes to debt shifting, is the capital structure and is therefore considered to influence the relocation decision of MNEs’ headquarters significantly. Especially firms which engage in debt shifting strategies seem to be sensitive to the leverage. According to the literature, it seems to be beneficial for a company if it examines a high grade of indebtedness in a high tax country. Consequently, this work suggests that a high leverage negatively affects the relationship of taxation on headquarters relocation.
H1c: A high level of leverage negatively affects the influence of differences in the statutory corporate tax on the possibility of headquarters relocation.
H2c: A high level of leverage negatively affects the influence of differences in the effective average tax on the possibility of headquarters relocation.
2.3. Conceptual Model
This section presents the conceptual model behind this study and therefore presents all hypotheses made in the previous sections and the respective relations.
3. DATA & METHODOLOGY
3.1. Sample
“United in diversity” is the motto of the European Union, and in fact, there are big differences, not only in cultures, languages and people, but also huge differences in the member’s economies. This diversity is reflected in the dataset which will be used to give a proper response
to the research question. The European Union counts 28 member-countries (EU-28) that are forming an internal single market. Although, there are big differences in the taxes that are levied from the 28 different jurisdictions which is also represented in the dataset.
The initial sample covers 32.665 multinational enterprises and 607 cross-border M&As during the period 2009 to 2017. The data was collected from multiple sources. Therefore, firm-level data was retrieved from the Orbis database, which contains extensive financial and ownership information about companies worldwide. And all data concerning the cross-border
Figure 2. Initial Dataset (source: Orbis and Zephyr databases)
M&A-deals was compiled from the Zephyr database. Both databases belong to Bureau van Dijk, a major publisher of business information, which specialises in private company data. In 2017 Bureau van Dijk was acquired by Moody’s. Tax rates and country-specific information were gathered from other reputable sources such as the OECD and Eurostat databases.
3.2. Data
The following section provides information about the approach of data collection, calculation and selection. Appendix A presents a comprehensive table in which each variable is defined and described.
3.2.1. Independent Variables
Due to the reason that this paper is examining if multinational companies with headquarters within the European Union are more sensitive to statutory corporate income taxes or rather to corporate effective tax rates, this study comes up with two independent variables. The two independent variables reflect difference between the tax rates (CIT and EATR) in home and host country. A positive Delta means a reduction in taxes after a relocation, for companies that stayed in their home country the value of the variables takes zero.
3.2.1.1. Statutory Corporate Income Tax Rate
Since this study only takes European Union member states into account, the European Commission (2018) presents all the relevant statutory corporate income taxes on their database. The data covers a timespan from 2009 to 2017 and therefore includes all the information needed for the analysis. The complete list of the statutory CIT rates for all EU-28 member states is presented in Appendix C.
3.2.1.2. Corporate Effective Tax Rate
Commission (Spengel, Schmidt, Heckemeyer & Nicolay, 2016). All approaches of calculating the corporate EATR used in this paper are based on the work of Devereux and Griffith (1998, 2003). While the major part of research has concentrated on marginal choices, the central point of their research is a multinational’s location choice if there exist a limited number of mutually exclusive locations for its investment, and the impact of varying levels in taxation between locations. Taking the neo-classical location theory into account, this approach is mainly based on value-maximizing firms (Devereux & Griffith, 2003).
The work of Devereux and Griffith (1998, 2003), is based on a standard approach to measuring the effective marginal tax rate (EMTR), which resulted from the research of, for instance, Auerbach (1979), King & Fullerton (1984) and Keen (1991). But, the EMTR primarily influences the size of an investment. “But the choice of location depends on the level of post-tax net present value (NPV); for a given pre-post-tax NPV. This can be measured by an “effective average tax rate” (EATR) which will be described in the following section (Devereux & Griffith, 2003).
In a first step, the nominal interest (𝑖) will be calculated. Therefore, the real interest rate will be supplemented by inflation rate. The resulting expression 𝑖 can be seen as a shareholder’s discount rate in the absence of personal taxes:
𝑖 = (1 + 𝑟)(1 + 𝜋) − 1.
In a second step, the statutory tax levied on the realized capital gains (𝑧∗)will be
discounted by ,1 − 𝑚./𝑖. Therefore, the numerator multiplies the personal tax rate on capital
gains with the fraction of shares that are sold within the year of observation, because 𝑧∗ is only
payable upon the sale of shares. The expression in the denominator stands for the fraction of net interest income multiplied with the nominal interest rate. The result (𝑧) can be defined as the accruals-equivalent capital gains tax rate 𝑧.
𝑧 =34,35612∗ 7/.81 +34,356351 7/.+ 934,356351 7/.:;+ 934,356351 7/.:<+ ⋯ > = 14,35612∗ 7/.
• 𝑚.: personal tax rate on interest income;
• 𝑧∗: personal tax rate on capital gains realizations;
• 𝑟: real interest rate; • 𝜋: inflation rate.
Afterwards, it is possible to calculate the shareholder’s discount rate (𝜌). For this purpose in a first step the net interest income (%) will be divided by the net capital gains (%) of accruals-equivalents. The result stands for the net income from interest and capital gains of accruals-equivalents. In a second step the value will be multiplied by the nominal interest rate (incl. real interest and inflation rates):
𝜌 = A3563527B 𝑖,
For further calculations it is necessary to express a term which measures the tax discrimination between new equity and distribution (𝛾) as well as the net present value of capital allowances (𝐴). The calculation of 𝛾 reflects the net income of dividends by talking into account that imputation tax systems allow for a reduction in the income tax payable on distributions. Therefore, the net accruals-equivalent capital gains tax rate and the imputation credit are positively related to tax discrimination (a high 𝑧 or 𝑐 leads to a higher 𝛾).
𝛾 =(352)(35G)(356F) .
• 𝑚H: personal tax on dividend income;
• 𝑧: accruals-equivalent capital gains tax rate; • 𝑐: imputation credit1.
In a next step the net present value of capital allowances (A) will be computed. Therefore, the reduction (in the amount of the allowance) of the corporate tax rate is expressed in the numerator. In order to obtain the NPV of allowances per unit of investment, the value of the numerator have to be divided by the sum of the allowance rate and the shareholder’s discount rate. The result of this approach are capital allowances on a declining basis, in other words, a constant growth of the allowance rate, results in a shrinking growth of the NPV of the capital allowances.
1 Dividend imputation tax systems provide shareholders with a credit for corporate taxes paid that can be used to
𝐴 =34KIJ L1 + A(34K)35JB + A(34K35JB;+ A(34K35JB<+ ⋯ M = J4KIJ .
• 𝜏: corporate tax rate; • 𝜑: allowance rate;
• 𝜌: shareholder’s discount rate.
F describes the additional costs for the usage of external financing. It can be seen as the net present value of the cashflows that emerge when financing stems from new equity or debt. If an investment is financed by retained earnings:
𝐹QR = 0
If an investment is alternatively financed by new equity, the NPV of the resulting cashflows is described by the numerator which contains the product of the complementary rate of the tax discrimination (between new equity and distribution), the effective real estate rate, and the shareholder’s discount rate. The latter two variables are positively linked to the cashflow resulting from the additional costs of new equity. Contrary, the complementary rate of the tax discrimination negatively influences that cashflow. Similar to the calculation of the NPV of 𝐴, discounting the cashflow of new equity financing is calculated on a declining basis.
𝐹TR = −K(35U)(34V)
34K .
If an investment is otherwise financed by debt, the NPV of the resulting cashflows is described by the numerator which expresses the positive influence of the effective real estate rate and the tax discrimination measure. The nominal interest rate 𝑖 is reduced by the corporate tax rate (this reflects the tax advantage of debt). By dividing the product of these variables by (1 + 𝜌) the cashflow which results from debt financing is discounted on a declining basis as well.
𝐹W = U(34V)(K5.(35I))
(34K) .
In order to calculate for the effective average tax rate (EATR) it is necessary to define the economic rent in the presence (𝑅) and in the absence (𝑅∗) of taxes:
𝑅 = 𝛾
1 + 𝜌{(𝜌 + 𝛿)(1 + 𝜋)(1 − 𝜏) − 𝑣𝜏𝜋 − [𝜌 + 𝛿(1 + 𝜋) − 𝜋](1 − 𝐴) − (1 + 𝜌)𝑒} + 𝐹TR+ 𝐹H
𝑅∗ = `5a 34a.
• 𝛿: economic depreciation rate;
• 𝑣𝜏𝜋: this term reflects the taxation of inventories & financial assets valued on a FIFO basis (𝑣 takes the value one in case of FIFO, otherwise it takes the value zero); • 𝑝: pre-tax real rate of return.
In a final step the effective average tax rate (EATR) can be calculated:
𝐸𝐴𝑇𝑅 =Q∗`/(34a)5(352)Q.
The methodology described above is considered to be well-fitted for the simulation of tax reforms on investment incentives. Devereux & Griffith (2003) suggest the potential co-ordination of tax regimes within the European Union as a very interesting area. The complete list of corporate EATRs it presented in Appendix C.
3.2.2. Dependent Variable
The dependent variable of this paper’s conceptual model is a binary variable and therefore takes the value one if a multinational headquarters has relocated to another EU member state. Otherwise it takes the value zero (Voget, 2011; Laamanen et al., 2012).
numbers) to the Zephyr database in order to find all the mergers and acquisition deals (4) within the European Union (acquirer- as well as target-firms are members of the EU-28) from 2009 to 2017.
The first column (Figure 2) shows all the firms registered in the Orbis database in 2018, only firms with at least 2 million Euros of total assets were taken into account. The headquarters of column 3 are defined as multinational if they own at least one foreign subsidiary. Since the Zephyr database covers a broad variety of different M&A deals, this study just takes transactions into account, through which the control of the firm (majority of shares) changed from one country to another.
In the course of the data collection the dataset was narrowed significantly, companies with incomplete or missing information were deleted from the sample. As a result, 6300 companies remained in the sample. 404 of these firms relocated their headquarters within the period from 1st January 2009 to 31st December 2017, this corresponds to a relocation ratio
(within the EU-28) of 15.59 percent. Within the analysis the 5896 that did not relocate serve as a control group.
Figure 3. Intra-European Headquarters Relocations - An Overview
Relocating from Code Relocating to
3.2.3. Moderator Variables
Three firm-specific moderating variables are considered to affect the relationship between corporate income taxes and the possibility of headquarters relocation. First, the firm size, which describes a firm’s sales level or total turnover. Second, the core business, which is a binary variable and takes the value one if a firm’s NACE core code is categorized as knowledge-intensive according to Eurostat classifications, alternatively it takes the value zero. Finally, a firm’s leverage has to be taken into account. The leverage is the ratio of a firm’s non-equity liabilities (the sum of current and non-current liabilities) to its total assets (Voget, 2011; Strauss-Kahn, 2009). All the data which is necessary to define the firm-specific moderators has been derived from the Orbis database.
3.2.4. Control Variables
3.3. Empirical Model
As already mentioned in the previous section, the dependent variable of the model is headquarters relocation and since there are only two options for MNEs (relocate or not relocate), this variable is binary. Consequently, this study makes use of two binary logistic regressions analysis, due to the fact that two different taxes (statutory corporate income tax and the effective average tax rate) are examined. This kind of analysis is considered to be the most appropriate statistical approach if the dependent variable is of dichotomous and of non-normally distributed nature (Hair, Anderson, Tatham & Black, 2005). Moreover, the binary regression analysis has commonly been used within the academic field of headquarters relocation and the research on foreign direct investments (e.g. mergers and acquisitions) (Voget, 2009; Strauss-Kahn & Vives, 2009; Laamanen et al, 2012).
The main objective when conducting a binary regression analysis is to predict the probability to which a certain event (e.g. headquarters relocation) occurs. The results of such statistical approaches allow to describe to what extent the independent variables influence the dependent variable. Additionally, interaction effects can be tested within the analysis by means of strengthening or weakening the influence of the independent (moderator) variables on the dependent variable.
Applied on this study, a positive regression coefficient (e.g. the B-value) of an independent variable would indicate a positive effect towards relocation (coded as one). Contrary, a negative B-value describes a tendency towards no relocation, which is coded as zero.
3.4. Results
model differ from a normal distribution. These findings are strengthening the assumption that the binary logistic regression is the right choice for analyzing the data.
In a next step, all variables were tested for multicollinearity, a high correlation between two independent variables negatively affects the outcomes of the regression analysis. Therefore, a correlation matrix was conducted and analyzed for each variable. According to Pallant (2013), correlations below the threshold of 0.7 are no concern for posing any problems during the statistical analysis. As a result, the two control variables Market Size (correlation of .971 with Log (GDP)) and Log (Assets) (correlation of .832 with Log (Firm Size)) have been excluded from the sample. It is clearly no surprise that there are higher correlation values and significance between the country-specific variables (Log (GDP), GDP Growth Rate, Market Size, LPI and Unemployment) as well as the firm-specific predictors (Log (Firm Size), Log (Assets), KIS and Leverage). Also, the significant correlations between the two tax rates and relocation are not very astonishing, since this is in accordance to the hypotheses (1&2). In contrast to that, it is very unexpected that the correlations between the CIT (-.182) and the EATR (-.115) are negatively related to the dependent variable, because these results indicate that a high home country taxation positively affects the possibility of rather staying at the home country than relocating across borders.
Figure 4. Test of Normality
Statistic df Sig. Statistic df Sig.
Relocation 0.539 6246 0.000 Relocation 0.539 6246 0.000
Log (GDP) 0.292 6246 0.000 Log (GDP) 0.292 6246 0.000
GDP Growth Rate 0.213 6246 0.000 GDP Growth Rate 0.213 6246 0.000
Market Size 0.299 6246 0.000 Market Size 0.299 6246 0.000
LPI 0.212 6246 0.000 LPI 0.212 6246 0.000
Unemployment 0.190 6246 0.000 Unemployment 0.190 6246 0.000
CIT (%) 0.315 6246 0.000 EATR (%) 0.150 6246 0.000
Δ CIT (%) 0.478 6246 0.000 Δ EATR (%) 0.475 6246 0.000
LOG (Firm Size) 0.042 6246 0.000 LOG (Firm Size) 0.042 6246 0.000
Log (Assets) 0.058 6246 0.000 Log (Assets) 0.058 6246 0.000
KIS 0.473 6246 0.000 KIS 0.473 6246 0.000
Leverage 0.038 6246 0.000 Leverage 0.038 6246 0.000
In a last step, the remaining variables were examined by conducting a binary logistic regression analysis via IBM’s SPSS-Statistics. Therefore, this study makes use of 11 models in order to test the hypotheses made in the previous sections of this paper. Model 12 includes all control
variables, each one shows significant influence on headquarters relocation. As a result, the whole model significantly explains the dependent variable with a 𝑋; of 181.901 (p<.01) and a
correct classification ratio of 93.8. Only the Nagelkerke 𝑅; of .075 shows a weakness, because
the model only explains 7.5 percent of the variance of the dependent variable. This weakness is declining when other variables are added within the following models. In Model 2 the difference (between home and host country) in CIT was added to the control variables and it is found to be significantly positvely related to the dependent variable. Adding the delta-CIT variable improved the ability of the model to predict a headquarters relocation, 𝑋;=211.644
(p<.01), Nagelkerke 𝑅; of .088 and 93.8 percent of correct classified observations. The Models
3,4 and 5 contain each of the moderator variables and the corresponding interaction effects on the relationship between the home country CIT and the moderator variable. Each of the moderator variables show significant positive influence (at the .01 level) on the relationship of CIT differences and headquarters relocation. Testing the moderating variables improves the ability of the model to explain the variance of the dependent variable. Also, the ratio of correct classifications shows a slight increase within Model 3 and 4 (94.0 & 93.9), while there was no
2 Additional models with other control variables were tested: firm age, number of employees (log), GDP per
capita, fixed assets (log), tax growth rate in the year before the relocation; but there weren’t any improvements of the models.
Variables Mean SD Min Max N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 Relocation .064 .245 0 1 6300 1 . 2 Log (GDP) 13.64 1114 9.557 15.003 6299 -.025 1 (.050) . 3 GDP Growth .023 .031 -.144 .192 6299 .046** -.072** 1 (.000) (.000) . 4 Market Size .082 .058 .001 .165 6297 -.067** .971** -.095** 1 (.000) (.000) (.000) . 5 LPI 3.85 .225 2.728 4.226 6299 -.014 .362** .154** .279** 1 (.252) (.000) (.000) (.000) . 6 Unemployment .085 .026 .029 .275 6300 .018 .338** -.395** .346** -.199** 1 (.150) (.000) (.000) (.000) (.000) . 7 CIT (%) .316 .059 .100 .387 6300 -.182** .612** -.237** .619** .048** .661** 1 (.000) (.000) (.000) (.000) (.000) (.000) . 8 Δ CIT (%) .0003 .020 -.202 .236 6292 -.036** .046** -.018 .065** .031* .012 .106** 1 (.004) (.000) (.145) (.000) (.015) (.340) (.000) . 9 EATR (%) .289 .065 .088 .384 6300 -.115** .768** -.265** .765** .067** .626** .887** .075** 1 (.000) (.000) (.000) (.000) (.000) (.000) (.000) (.000) . 10 Δ EATR (%) .0004 .020 -.192 .243 6292 .051** .078** -.006 .093** .011 .052** .070** .641** .102** 1 (.000) (.000) (.624) (.000) (.383) (.000) (.000) (.000) (.000) .
11 Log (Firm Size) 10.966 2.137 .693 19.109 6257 -.020 .302** .038** .274** .257** -.075** .054** 0.000 .119** -.012 1
(.122) (.000) (.003) (.000) (.000) (.000) (.000) (.996) (.000) (.337) . 12 Log (Assets) 11.114 1.961 7.636 19.551 6300 0.000 .305** .014 .270** .262** -.018 .089** .004 .148** .002 .832** 1 (.974) (.000) (.275) (.000) (.000) (.162) (.000) (.724) (.000) (.850) (.000) . 13 KIS .240 .427 0 1 6300 .041** .108** -.041** .099** .024 .084** .104** -.014 .114** .004 -.112** .017 1 (.001) (.000) (.001) (.000) (.053) (.000) (.000) (.269) (.000) (.770) (.000) (.185) . 14 Leverage .584 .234 0 1 6300 -.012 .085** -.022 .080** .031* .066** .106** .011 .093** .001 .173** .056** .013 1 (.342) (.000) (.079) (.000) (.015) (.000) (.000) (.388) (.000) (.940) (.000) (.000) (.292) . For complete variable names see the variable list (Appendix A). **Correlation is significant at the .01 level, *Correlation is significant at the .05 level (2-tailed).
change within Model 5. Model 6 exhibits the highest Nagelkerke 𝑅; (.222), which indicates a
good ability in explaining the variance of the dependent variable. The only independent variable that remains significantly linked to the dependent variable is the firm size (.484; p<.01). Similar to the Model 2, adding delta-EATR to the control variables in Model 7 improves its ability, in terms of predicting headquarters relocations, significantly. Also, all the control variables show significance, but this delta-variable exhibits a smaller regression coefficient than delta-CIT in Model 2 (7.640 < 10.969). Similar to the delta of the corporate income tax, the delta-EATR also shows a significant positive influence on the dependent variable. Model 8, 9 and 10 examine the influence of the moderators on the relationship of delta-EATR on the probability of relocation. Each of the moderators shows a significant (p<.01) positive influence on the likelihood of headquarters relocation. This changes when all moderators are taken together in Model 11. Within this model, the only moderator that remains significantly influencing the dependent variable is the firm size (.627; p<.01). KIS and Leverage show positive B-values, but they are not significant. The last model exhibits a 𝑋; of 507.805, a Nagelkerke 𝑅; of .207
and a correct classification ratio of 94.0 percent, which underlines the quality of the model.
Figure 6. Binary Logistic Regression
Dependent Variable: Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Headquarters Relocation Control Variables Unemployment Rate .217** .227 .246** .231** .246** .246** .208** .300** .217** .276** .301** Log GDP -.267** -.283** .429** -.197** .258** .414** -.283** .861** -.167** .520** .850** LPI 1.274** 1.325.** 2.764** 1.568** 2.647** 2.715** 1.261** 2.135** 1.272** 2.068** 2.057** GDP Growth Rate 11.202** 12.138** 5.980** 10.628** 7.541** 5.771** 11.040** 7.592** 10.644** 8.942** 7.485** Independent Variables Δ CIT (%) 10.969** 14.280** 12.150** 13.407** 14.363** Δ EATR (%) 7.640** 9.767** 8.231** 8.619** 9.861**
Moderators & Interactions
H1a: Log (Firm Size) .578** .484** Log (Firm Size) * CIT (%) -2.132** -1.779**
H1b: Log (Firm Size) .686** .627**
Log (Firm Size) * EATR (%) -2.624** -2.357**
H2a: KIS 4.669** 1.110 KIS * CIT (%) -13.786** -2.313 H2b: KIS 3.557** 1.009 KIS * EATR (%) -11.030** -1.940 H1c: Leverage 9.315** 2.024 Leverage * CIT (%) -31.042** -5.991 H2c: Leverage 9.305** 1.405 Leverage * EATR (%) -33.820** -4.564 Observations (N) 6299 6292 6249 6292 6292 6249 6292 6249 6292 6292 6249 Chi-square 181.901** 211.644** 529.518** 282.928** 467.719** 545.91** 194.953** 490.923** 243.778** 400.458** 507.805** -2 log Likelihood 2813.611 2745.264 2421.780 2673.980 2489.189 2405.383 2761.955 2460.375 2731.131 2556.450 2443.493 Nagelkerke R² .075 .088 .216 .117 .191 .222 .081 .201 .101 .164 .207 % classified 93.8 93.8 94.0 93.9 93.8 93.9 94.1 93.9 94.1 94.1 94.0
4. CONCLUSION
This section concludes the thesis. First, the findings of the empirical analysis and theory-based hypotheses are compared and discussed in the section 4.1. Afterwards the limitations section discusses possible weaknesses of the work. Based on the limitations, section 4.3. presents recommendations for Future Research. In the final section all findings of this work are taken together in order to give a meaningful response to the research question.
4.1. Discussion
This work examined the influence of statutory corporate income taxation and effective average tax rates on the possibility of multinational’s headquarters relocation. Additionally, different factors that are considered to influence MNE’s location decisions were investigated by means of three hypotheses (for each of the two tax rates). The research setting was designed to capture tax-motivated relocation behaviour. Therefore this study makes use of cross-border M&As within the European Union (2009 – 2017). The design of this research makes it possible to compare the influence of two different types of tax rates on the likelihood of headquarters relocation.
are supported by the findings of this study, although their impact on the dependent variable was different than expected.
This study did also find empirical support for hypotheses H1a and H2a in the way that therefore the results of the regression analysis found evidence that a firm’s size (by means of operating turnover) positively influences the likelihood of MNE’s headquarters relocation. According to many studies before, this work considers these findings to be supportive for the assumption that MNEs shift their profits across borders (to lower tax jurisdictions), in order to minimize their tax bill. These results are also consistent with the theory that MNEs make use of their international nature when pursuing their objective of profit maximization, which clearly reflects the underlying neo-classical location theory (Bouwer et al., 2004).
Also, hypotheses H1b and H2b can be confirmed by the results of the binary logistic regression analysis. The findings show a significantly positive influence of degree to which a firm’s core business is knowledge-intensive on the possibility of headquarters relocation. These outcomes are in line with the findings of Voget (2011), who states that firms from the knowledge-intensive sector exhibit a higher degree of mobility than others. An explanation therefore is that intangible assets (e.g. patents) are easier to shift across borders, and also the pricing of such goods is more complicated and less transparent compared to others.
The study did not find empirical support for hypotheses H1c and H2c, which suggested that a higher leverage is negatively correlated to the likelihood of headquarters relocation. Contrary to the assumption made in the hypotheses, leverage seems to be positively influencing the dependent variable. These results are also contradicting the common notion of literature. A possible explanation for these interesting findings may be the fact that a lot of companies from the sample are moving from high-tax to lower tax jurisdictions and therefore a firm’s indebtedness doesn’t play a big role or even a reverse role to the hypotheses H1c and H2c.
Even though a firm’s knowledge-intensiveness and leverage show significant influence within the models in which they were tested separately, both turn out to have an insignificant impact on headquarters relocation when they are implied to the comprehensive models (6 & 11). Consequently, a firm’s size remains the only moderator variable with a significant influence on the dependent variable with slightly higher impact in model 11, which takes the difference in effective average tax rates into account.
high unemployment rate reflects a weak economic situation in the home country. Also, the other controlling variables show irrational influence on the dependent variable, since GDP, its growth rate and LPI are considered to reflect a high-quality environment for MNEs. Therefore, these findings suggest that a higher quality business environment in the home country is positively related to the possibility of headquarters relocation. Possible explanations for these controversial outcomes are discussed within the next section.
4.2. Limitations
As with every research, this study has some limitations that arise from the research design and the nature of the data used in the analysis. As already mentioned in the previous section some controversial findings may stem from the dataset.
Since this study draws statements from an EU perspective the focus of this section is on the nature of the data. The European Union has very different member countries, which are varying in terms of size, economic power, taxation, quality of economic environment and also in their geographic location. While there are very big and well-developed economies (e.g. Germany, France, Great Britain and Italy accounting for 53.9% of the EU population in 2017), there are also very small and underdeveloped countries (e.g. Bulgaria, Cyprus, Estonia, Latvia, Lithuania, Slovakia accounting for 3.3% of the EU population in 2017) (Eurostat, 2019). Nearly half (48.3%) of the 404 companies that moved their headquarters across borders relocated to one the four biggest EU economies. 63 cross-border M&As took place within these four countries, that corresponds to 15.6 percent of all deals (Figure 3 presents a comprehensive table of all M&A deals of the sample). One common feature of the four biggest EU economies is that each one levies comparatively high corporate income taxes. As result of these findings higher home country income taxes (CIT & EATR) are negatively linked to the dependent variable, which can be considered as an irrational behaviour (these outcomes are reflected in the interaction effects within the regression analysis). Consequently, drawing assumptions from the results should be done very carefully.
