Sovereign wealth funds : internationalization strategies in the shadow of foreign hostility

(1)

Sovereign Wealth Funds:

Internationalization Strategies in the Shadow of

Foreign Hostility

MSc Business Administration - International Management Track

Student name: Rogier Daniël Boer

Student number: 11151714

Supervisor: Dr. V.G. Scalera

Date: January 27, 2017

(2)

2

Statement of originality

This document is written by student Rogier Daniël Boer who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Acknowledgements

The writing and completion of this thesis would not have been possible without the assistance, support and guidance of several people. First and foremost, I owe an enormous debt of gratitude to my thesis supervisor Vittoria Scalera. Vittoria’s enthusiastic explanation about sovereign wealth funds during the first meetings made it possible for me to work on a topic that was of great interest to me. Moreover, without her persistent support, critical evaluations, and expertise the completion of thesis would not have been possible. Second, I take this opportunity to record my sincere thanks to my university buddies Kevin, Roelof, Nienke and many others, who have been a constant source of joy for me.

Rogier Daniël Boer Amsterdam January 27, 2017

(4)

4

3.2 Dependent Variables ... 28 3.3 Independent Variables ... 29 3.4 Control Variables ... 31 3.5 Methodology ... 33 3.6 Descriptive Statistics ... 33 4. Results ... 37 4.1 Main Results ... 37 4.2 Robustness Check ... 39 5. Discussion ... 43 6. Conclusion ... 46 6.1 Concluding Remarks ... 46 6.2 Theoretical Contributions ... 48 6.3 Practical Implications ... 49

6.4 Limitations and Future Research... 50

(5)

5

Abstract

Sovereign wealth funds (SWFs), broadly defined as large publicly owned investment agencies, have emerged as among the most important players in global financial markets. After the financial crisis of 2008, these national investment portfolios attracted considerable public attention, due to their potential politically biased investment motives, increased size of investments, and continually evolving risk appetites. Additionally, several scholars demonstrate that cross-border SWF investments can affect the national security and global financial markets, which have led to political concerns in SWF host countries. Yet, it remains unknown which characteristics of SWF cross-border investments provoke opposition in the target country. Hence, this study examines which SWF- and deal-specific determinants affect the level of opposition faced by SWFs. In particular, this study examines more than 100 cross-border SWF acquisitions to the United States over the period 2000-2013, by adopting quantitative content analysis of American media publications.

The results demonstrate that SWFs, which engage in direct investments (as opposed to investments through investment vehicles) or majority acquisitions are more likely to face a higher level of attention and hostility in the target country. Further, fund politicization does not necessarily lead to more attention in the target country, but the expressed attention is likely to be more severe. Finally, fund opacity decreases the likelihood of facing a high level of attention and hostility in the recipient country. In this study, I control for investments in strategic sectors, bilateral trade agreements between the SWF’s and the target country, SWF size, SWF investments made during the financial crisis, and Sovereign Pension Reserve Funds versus Social Security Reserve Funds.

Key words: Sovereign wealth funds; Cross-border investments; Opposition; Investment

(6)

6

1. Introduction

Over the past decade, sovereign wealth funds (SWFs) have emerged as major investors in the global economy. Particularly after the global financial crisis of 2008, SWFs have witnessed an extensive growth of market size, which currently makes them one of the major funding sources for firms worldwide (Sovereign Wealth Fund Institute, 2016). As a result, SWFs have recently gained considerable scholarly attention (e.g. Aguilera, Capapé, and Santiso, 2015; Dewenter, Han, and Malatesta, 2010; Kotter and Lel, 2011; Truman 2009). At first blush, the investments of these governmentally owned investment funds seem to be a perfect opportunity for countries with a high variance in public revenues to ensure a durable cash flow and to invest for future generations (Bernstein, Lerner, and Schoar, 2013). However, the increasing influence of SWFs in target firms and national economies has raised questions about potential political investment objectives, which might threaten the national security of countries they invest in (Cohen, 2009; Fernandes 2014; Lee, 2010).

Indeed, the acquisition of stocks of Western financial institutions such as Barclays, Citigroup, UBS and Merill Lynch by several SWFs from non-Western countries raised concerns in recipient nations about the lack of transparency (Lee, 2010). Moreover, the establishment of the China Investment Corporation with a starting capital of $200 billion by the Chinese government in 2007 led to suspicions in Western countries about the potential consequences that politically biased investment decisions might have on national economies (Fernandes, 2014). As a result, Western countries have recently started to impose new regulations to prohibit or analyze foreign investments, after a long period of financial liberalization (Cohen, 2009). While in the year 2000 only two percent of the total regulatory changes made by the United Nations Conference on Trade and Development were restrictive rather than liberalizing, this number grew to twenty percent of the annual total in 2006 (Cohen, 2009; Golub, 2003). In addition, the International Monetary Fund established the

(7)

7

Santiago Principles to increase the level of transparency of SWFs. These are voluntary guidelines signed by thirty SWFs that represent 80% of the total assets managed by SWFs (International Forum of Sovereign Wealth Funds, 2016; Truman, 2011).

Drawing insights from the international business literature, several researchers suggest that the level of opposition faced by foreign SWFs is a crucial element that needs to be considered when analyzing SWF investments (e.g. Bortolotti, Fotak, Megginson, and Miracky, 2010; Knill, Lee and Mauck, 2012). However, little is known about how certain SWF-specific determinants and internationalization strategies can mitigate these external pressures when investing in a foreign country. An exception to this is the work of Murtinu and Scalera (2016), who examine why SWFs use intermediate investments vehicles in the form of financial or corporate companies, or SWF-controlled firms. They suggest that investment vehicles can add an extra organizational layer between the SWF and the target firm, which might reduce hostility in the recipient country. In particular, they find that SWF- and deal-specific determinants – in the form of fund opacity, fund politicization, strategic industry targets and majority ownership choices – lead to a more likely use of investment vehicles. However, the existing literature does not provide evidence based on statistical analyses whether SWF- and deal-specific determinants affect the level of hostility faced by SWFs in the recipient country.

Hence, this study aims to test whether direct SWF investments (vs. investment through vehicles), majority (vs. minority) SWF acquisitions, fund politicization, and fund opacity lead to a higher level of hostility faced by SWFs in the recipient country. Thus, the key question addressed in the present work is: how can SWFs affect the level of opposition expressed in the recipient country? To answer this question, data on SWF- and deal-specific determinants is collected through a dataset created by Murtinu and Scalera (2016), whereas data on the level of opposition is collected via the Lexis Nexis database by examining media

(8)

8

publications on SWF cross-border investments to the United States, which is proposed by Shi, Hoskisson, and Zhang (2016).

The present work contributes to the existing international business literature by examining several crucial aspects of SWFs and their investment strategies. First, a significant body of prior research in the SWF literature proposes that the hostility faced by SWFs might be due to a low transparency of SWF investment activities and the potential political investment objectives they may have (e.g. Gieve, 2008; Johnson, 2007; Kotter and Lel, 2011). However, no study has empirically examined how fund politicization and fund opacity interact with the level of opposition faced in the recipient country. Second, this work contributes to the international business literature by enlarging the understanding of internationalization strategies of SWFs. Several studies suggest that SWFs can mitigate host country hostility by showing a passive investment approach through the use of investment vehicles or minority acquisitions (e.g. Ghahramani 2013b; Murtinu and Scalera, 2016). However, statistical evidence on this is currently lacking. Hence, by quantitatively analyzing the impact of an active investment approach – in the form of direct investments rather than indirect investments through investment vehicles, and majority rather than minority ownership choices – on the level of hostility faced by SWFs in the host country, this research is sui generis in the international business literature. Finally, this study provides practical insights for target firm managers, SWF managers and policy makers into how and when SWFs investments can lead to a negative public opinion in the foreign country.

This thesis is organized as follows. Chapter 2 gives an overview of the literature on SWFs and develops the research hypotheses. Chapter 3 describes the data and methodology of this study. Chapter 4 provides the results of the statistical analyses. Chapter 5 discusses the main results and the robustness check. Chapter 6 provides the concluding remarks, including the theoretical contributions, implications, limitations, and directions for future research.

(9)

9

2. Theoretical Framework

This chapter addresses the theoretical background of this study. First, the term SWF is defined. Second, their history and current status are described. Third, the impact of SWF investments on target firms and the global economy is attended to. Fourth, the investment strategies and tactics are addressed. Fifth, the impact of government control on SWF decisions is explained. Sixth, the gap in the academic literature and the research question are addressed. Seventh, the developed hypotheses are described.

2.1 Sovereign Wealth Funds

In spite of the growing attention on SWF investments, different definitions exist on what exactly constitutes a SWF. In this report the definition of Balding (2008, p. 9) is applied, who defines SWFs as “a pool of capital controlled by a government or government related entity that invests in assets seeking returns above the risk-free rate of return”. Technically speaking, governments can exert control over this capital directly via politicians or managers appointed by politicians, or indirectly through an ad hoc appointed board (Clark, Dixon and Monk, 2013). The main purpose of SWF investments is to stabilize government and export revenues in order to offset a potential lack of specific natural resources and/or to better manage the country’s foreign reserves (Aizenman and Glick, 2008). According to Allen and Caruana (2008), SWFs can be classified into five types: “(I) stabilization funds, where the primary objective is to insulate the budget and the economy against commodity (usually oil) price swings; (II) savings funds for future generations, which aim to convert nonrenewable assets into a more diversified portfolio of assets and mitigate the effects of Dutch disease1; (III) reserve investment corporations, whose assets are often still counted as reserve assets, and are established to increase the return on reserves; (IV) development funds,

1_{The term Dutch disease refers to the economic malaise in the Netherlands as a result of gas}

(10)

10

which typically help fund socio-economic projects or promote industrial policies that might raise a country's potential output growth; and (V) contingent pension reserve funds, which provide (from sources other than individual pension contributions) for contingent unspecified pension liabilities on the government's balance sheet” (p. 5).

With regard to the latter type, Megginson, You, and Han (2013) state that SWFs are comparable to other internationally active investment arms, such as pension funds. Indeed, Blundell-Wignall, Hu and Yermo (2008) define pension funds as pools of capital controlled by the government in order to finance public pension plans. More specifically, they suggest that a distinction can be made between Sovereign Pension Reserve Funds (SPRFs), directly managed by the government, and Social Security Reserve Funds (SSRFs) that are managed by an independent firm or by a social security institution. Consequently, SPRFs are also considered in this research, as these funds are not fundamentally different compared to SWFs. This is in with the International Working Group of SWF, who define SWFs as “special purpose investment funds or arrangements, owned by the general government” (IWG, 2008, p. 27).

In terms of SWF home countries, SWFs can be broadly divided into two groups (Beck and Fidora, 2008; Megginson and Fotak, 2015). The largest group of SWFs originated in nations which are dependent on one or a few natural resources; these are also known as commodity funds. This group mainly includes SWFs sponsored by oil revenues and are established in countries such as the Arab Gulf states, Norway, Russia, Malaysia and Brunei (Megginson and Fotak, 2015). More recently, a second group of SWFs have emanated from emerging economies with large foreign currency reserves resulting from extensive balance of trade surpluses. These funds are particularly based in countries as Singapore, South Korea and China (Megginson and Fotak, 2015). In this research both commodity and non-commodity SWFs are considered.

(11)

11

2.2 History and Current Status

Notwithstanding the recently growing attention to SWFs by academic scholars, the first SWF was already established in 1953 in Kuwait, while the funds established by the United Arab Emirates and Singapore date back to the 1970s (Chhaochharia and Laeven, 2008; Kimmitt, 2008). Due to the increasing prices of natural resources and the rapid growth of foreign reserves in exporting economies, the number and size of SWFs have risen tremendously over the past decade (Beck and Fidora, 2009). Asian countries in particular have recently established SWFs in order to invest their large trade surpluses (Aizenman and Glick, 2008; Sovereign Wealth Fund Institute, 2016). Alhashel (2016) documents that the assets possessed by SWFs doubled between September 2007 and June 2014. Today, there are about 78 SWFs mainly based in Asia and the Middle East, that manage assets estimated at U.S.$ 7,200 billion (Sovereign Wealth Fund Institute, 2016). Yet, the largest group of SWFs remains oil and gas related. This group accounts for 57 percent of the total investments made by SWFs worldwide (Sovereign Wealth Fund Institute, 2016).

Over the past decade, the literature has documented a shift in the extent to which SWFs engage in creating shareholder value. Before the global financial crisis, most SWFs used to be not generally active investors (Ghahramani, 2013a; Rose, 2008). However, several authors suggest that SWFs have recently tried to create long-term value by taking an active investment approach to prevent mismanagement (Dewenter et al., 2010; Fernandes, 2014). Two examples illustrate the attempts of SWF to increase shareholder value. First, Norway’s Government Pension Fund-Global tried to increase shareholder rights in order to nominate candidates for the board of firms in the United States (Alhashel, 2016). Second, the Qatar Investment Authority played a strong role in the merger of two Swiss firms in 2013, as they did not accept the merger until a better price was offered. This active role of the Qatar Investment Authority finally resulted in more value for all shareholders (Alshashel, 2016). As

(12)

12

a result of this recent shareholder activism among SWFs, national governments raised concerns about SWFs and have begun to ponder the regulation of these investments (Bortolotti et al., 2010; Cohen, 2009; Ghahramani, 2013b).

2.3 Impact of SWF Investments

Indeed, the recent growth of SWF investments and the more active investment approach of SWFs has stirred the debate about the extent to which their investments have a positive effect on target firms. As a result, the performance of both target firms and SWFs became one of the most widely studied aspects of SWFs. Several pieces of research suggest that SWF investments have a positive short-term effect on target firms after the announcement date. First, Dewenter et al. (2010) suggest that SWF investments can yield abnormal stock returns on the short-run, as the expressed confidence in the target firm has a benign effect on its performance. Second, Kotter and Lel (2011) provide evidence that SWF investments have a positive effect on the target firm’s stock prices around the announcement date. Finally, Chhaochharia and Laeven (2008) find that the share prices of target firms respond favorably when SWFs acquire stakes, due to the fact that these investments usually take place when firms are in financial trouble. However, Bortolotti, Fotak and Megginson (2015) take a broader perspective and compare SWF investments with those of private investors. They demonstrate that target firms usually react positively to SWF investments, but do so less compared to investments by other investors, which indicates a certain “SWF discount”.

Regarding the long-term effects of SWF investments, inconsistencies exist in the literature. Kotter and Lel (2011) find no evidence that SWF investments substantially affect target firm financial performance in the long-run. However, Chhaoccharia and Laeven (2008) suggest that the long-term performance of firms acquired by SWFs tends to be poor due to both weak corporate governance and weak investment portfolio management. This is in

(13)

13

contrast with Fernandes (2008), who examines the effect of large SWF investments on target firm performance up to three years after the investment. He finds that firms have better operating performances and are valued higher after large SWF investments. This finding is in line with several other studies, which document that institutional ownership is generally associated with higher firm valuations (e.g. Ferreira and Matos, 2008; Gompers and Metrick, 2001). Thus, on a company level SWF investments are likely to yield positive short-term effects, through reducing the costs of capital for these firms and expressing confidence in their future performance. Yet, the literature is divided on the long-term effects of SWF investments on target firm performance, as these effects are likely to be influences by deal-specific determinants, such as the size of the investments.

When looking at the macro environment, Balding (2008) highlights that there is no significant difference in the impact of SWF investments on global financial markets compared to other institutional investors. According to Beck and Fidora (2008) the impact of SWF investments on financial markets is largely dependent on the investment size and motives. With regard to the investment size, especially large SWF investments are likely to affect a country’s financial market as these investments can exert a stabilizing effect on the financial market. SWFs often place long-term investments in low-performing firms and do not rely to a great extent on high leverage, unlike private equity investors (Butt, Shivdasani, Stendevad and Wyman 2007; Chhaochharia and Laeven, 2009). As regards the motives and strategies for SWF investments, several authors document a shift during the financial crisis of 2008. During this period several SWFs mainly from countries in Asia and the Middle East, provided a helping hand to Western firms through placing large investments in financially constrained companies (Gieve, 2008). Additionally, SWFs recently shifted their investment strategies from bond and index funds to assets that carry a much higher risk (Drezner, 2008). Moreover, it is documented that not all SWF investments during the financial crisis were

(14)

14

purely economically driven (Johan, Knill and Mauck, 2013). Hence, there are several reasons to believe that SWF investments may negatively affect host country economies. Indeed, SWFs under high government control might abruptly sell their assets for political reasons (Beck and Fidora, 2008). In addition, the high degree of risk for SWF investments could be a potentially destabilizing factor (Knill, Lee and Mauck, 2009). Moreover, SWFs might acquire stakes in companies in sensitive industries to support local firms (Aguilera et al., 2015; Beck and Fidora, 2008). However, as most SWFs exhibit very little information on their investments and due to the wide variety of investment motives, it remains hard to get evidence based on statistical data on the general impact of SWFs on the global economy (Bortolotti et al., 2015; Fernandes, 2014).

2.4 Investment Strategies

In contrast to the effects of SWF investments, less is known about the investment strategies of SWFs. Although most recent investment activities of SWFs were usually long-term and placed in high risk foreign assets, controversies exist in the academic literature regarding the investment strategies (Drezner, 2008; Kotter and Lel, 2011). In fact, the SWF literature can be generally divided into two schools of thought: the economic and the political. On the one hand, the political school of thought describes SWFs as purely rational market actors, who are similar to other financial investors. More specifically, Kotter and Lel (2011) quantitatively examine SWF investments and find that SWFs are similar to other institutional investors in their effect on firm performance and preferences in target firms. Additionally, Megginson et al. (2013) study the determinants of SWF investments in 78 target countries. Their results suggest that overall, SWFs act purely or principally as commercial investors. Finally, Balding (2008) examines the target countries of SWF investments and finds that SWFs act as rational, economically driven investors.

(15)

15

On the other hand, the political school of thought claims that the investment motives and strategies of SWFs differ from other institutional investors. Johan et al. (2013) provide evidence in a sample of investments in 424 target firms that SWFs have different investment practices compared to other institutional investors, such as banks, pension funds and hedge funds. Indeed, as regards the investment portfolios, Dyck and Morse (2011) suggest that portions of the investment portfolios of SWFs are driven by developmental objectives that impact the location choice. Additionally, Chhaochharia and Laeven (2008) find that SWF largely invest in industries that are not located in their home countries. Yet, they do so primarily in countries with similar cultures as their home country, which suggests that the investments of SWFs are not entirely driven by increasing shareholder value, and can also be driven by a desire to exploit informational advantages. This argument is in line with Aguilera et al. (2015), who identify that SWF investments can be driven by either a desire of home country governments to acquire foreign resources and knowhow or by a tradeoff between political legitimacy and financial returns. Moreover, Knill et al. (2012) provide evidence that political relations between countries play a role in the investment decisions of SWFs. They suggest that SWFs prefer to invest in countries with which they have weaker political relations.

2.5 Political Influence on SWF Investment Decisions

Even though most SWF investments yield positive effects for target firms, the idea that investment strategies of SWFs may be driven by non-economic goals has raised questions about the potential influence of politicians on SWF investments. The growing attention on non-economically driven SWF investments is mainly a result of large SWF investments during the financial crisis of 2008 (Fernandes, 2014). During this period, SWFs from non-Western countries purchased large amounts of stocks of firms based in Western economies (Lee, 2010). This development led to controversies in these countries because of a

(16)

16

lack of transparency of SWFs and potential politically biased investment decisions (Fernandes, 2014; Lee, 2010; Megginson et al. 2013). As a result, a few governments imposed regulations for SWFs in an attempt to make their location choices and type of purchased assets more transparent (Ghahramani, 2013b; Vasudeva, 2013). For instance, the Norwegian government established an ethical council to control for the cross-border investments of the Norwegian SWF Government Pension Fund Global. This council exerts power in two ways. First, it introduced a screening and evaluation process of target firms, based on their social, environmental governance, and ethical conduct. Second, it publicly censors and approves the cross-border investments of the Norwegian SWF (Vasudeva, 2013). In addition to the efforts of home country governments, the International Monetary Fund also called for higher transparency of SWFs by establishing the Santiago Principles in 2008. These principles are voluntary guidelines, which are signed by thirty SWFs representing 80% of the total assets managed by SWFs (International Forum of Sovereign Wealth Funds, 2016; Truman, 2011). However, despite the growing demand for a greater transparency, SWFs are, from a legal point of view, unlike financial investors not obliged to disclose information about their investment allocations (Keller, 2008). As a result, SWFs originating in the Arab Gulf countries, Russia and China are still scoring low on fund transparency (Truman, 2009).

Besides the impact of SWF investments on target firms and the low degree of transparency among several SWFs, the widely varying governance standards of SWFs and the influence of politicians on the decision-making process also lead to suspicions in recipient countries (Aizenman and Glick, 2008; Lee, 2010). More specifically, countries with a greater involvement by political leaders in SWF management are associated with different investment strategies (Bernstein et al., 2013). According to Karolyi and Liao (2010), this can be a result of agency problems, which make politicians susceptible to bribes. Moreover, although most stakeholders assume that SWFs take a passive role in the management of the

(17)

17

firm, an increasing number of critics argue that politicized SWFs sometimes do exert power over target firms behind the scenes (Fernandes, 2014). In addition, the large positions of SWFs in target firms can give them superior information, which can be transferred to firms in the SWF home country (Aguilera et al., 2015). As a result, SWFs might seek to invest in sensitive sectors such as the media, telecommunications and financial services amongst others (Balding, 2008).

Despite these potential negative consequences for host country economies, several researchers suggest that SWF investments should not be feared. For instance, there is little evidence that SWFs monitor investee firm managers, as they usually have a completely passive role in the management of target firms (Bortolotti, Fotak, Megginson, Miracky, 2010). Moreover, Avendaño and Santiso (2009) claim that the idea of SWFs, who wish to secure large stakes in Western companies is unfounded. Nevertheless, several researchers propose the introduction of more stringent transparency requirements on SWFs, as well as a greater scrutiny of foreign SWFs seeking operational control in local firms (Aizenman and Glick, 2008; Cohen, 2009). Indeed, several protective measures have been imposed by recipient countries to oppose the influence of foreign SWFs on national firms. For example, in Italy SWFs are prohibited from buying more than 5% of domestic firms, whereas in France foreign SWF investments need to be approved by the Finance minister, irrespective of their size or nationality (Lee, 2010).

Yet, it is highly questionable whether it makes sense to establish regulations on SWFs not applied to similar institutional investors, because SWFs can circumvent these rules by using certain investment vehicles, such as hedge funds and investment banks amongst others (Avendaño and Santiso 2009; Ghahramani, 2013b). Indeed, in a sample of more than 500 cross-border SWF investments, Murtinu and Scalera (2016) find that about 43% of these investments were made through an investment vehicle. More specifically, they identify three

(18)

18

types of investment vehicles that can be used by SWFs: “(I) non-SWF majority-owned financial vehicles, such as private equity funds, venture capital funds, investment banks, asset management companies, commercial banks, investment management companies, financial branches of big corporations, real estate investment trusts, and investment advisory firms; (II) non-SWF majority-owned corporate vehicles, in the form of non-financial corporations, or companies controlled by public agencies not controlled by the SWF or by the government of the country in which the SWF originates; and (III) other SWF investment vehicles, including SWF majority-owned financial and non-financial corporations (Murtinu and Scalera, 2016, p. 4).” Consequently, even though several countries imposed regulations on SWF investments, it remains unclear if these rules are truly effective.

In sum, it can be safely claimed that the lack of transparency, the potentially strategic investment motives, and the influence of politicians in the decision-making process, have led to suspicions in recipient countries, and consequently to the introduction of regulations. However, it remains unclear which SWF related aspects lead to this hostility and how SWFs can circumvent the hostility.

2.5 Research Gap

Considering the sections above, the existing literature on SWF investments is mainly focused on two aspects. First, the effects of SWF investment on target firm performance and national economies is a widely-studied topic (e.g. Bortolotti et al., 2015; Chhaochharia and Laeven, 2008; Dewenter et al., 2010). Second, many authors in the SWF literature question whether the investment decisions of SWFs differ from other institutional investors (Bernstein et al., 2013; Dyck and Morse, 2011; Johan, et al., 2013, Knill et al. 2012). Within this, many authors propose that SWF investments might provoke hostility towards efforts in the target country. However, it remains unclear in the existing SWF literature which types of SWF investment strategies and SWF-specific determinants induce this opposition. A first effort on

(19)

19

this has been made by Murtinu and Scalera (2016), who suggest that fund opacity, fund politicization, strategic industry targets, and majority ownership choices lead to a more likely use of investment vehicles. These investment vehicles may mitigate the level of hostility in the foreign country, as they can act as an extra organizational layer between the investing SWF and the target firm (Murtinu and Scalera, 2016). However, the literature does not adequately explain if firms engaging in alternative cross-border investment strategies – by means of direct investments (vs. investment through vehicles), and majority (vs. minority) acquisitions – face a lower level of hostility in the recipient country. Moreover, it remains unclear whether certain SWF-specific characteristics – such as fund politicization and fund opacity – affect the host country opposition. Given the fact that external pressures, as well as the ambition of home-country governments to gain access to assets located in foreign countries, are crucial elements that needs to be considered when analyzing SWF investment, this study examines the relationship between SWF investments and the level of opposition expressed in their recipient countries (Bortolotti et al., 2015; Chhaochharia and Laeven, 2009; Keller, 2008). Hence, the research question addressed in this work is: how do SWF- and deal-specific determinants affect the level of hostility faced by SWFs in the target country?

2.7 Hypotheses Development

Firms investing in one or more foreign countries are invariably vulnerable to political risk. Traditional scholars in this field mainly focused on assessing and conceptualizing the consequences of political risk for firms (e.g. Alon and Herbert, 2009; Kobrin, 1979). Graham and Krugman (1995) argue that there are two reasons why countries can oppose foreign direct investment. First, policies such as tax reductions or subsidies can become costlier to a nation when there is a substantial amount of foreign ownership of a firm. Second, foreign owners might try to influence the domestic political process to policies that they like. Additionally, in

(20)

20

the case of SWF investments, recipient countries might also fear strategic motivations of the SWF home country (e.g. Aguilera et al., 2015; Aizenman and Glick, 2008). In particular, SWFs can pursue political influence in key assets of the host country, or transfer knowledge or technological assets to firms in their home country, which may lead to hostility expressed in the recipient country. (Aguilera et al., 2015; Fernandes, 2014).

However, by choosing alternative internationalization strategies firms can mitigate the level of hostility faced in the recipient country. One alternative strategy that firms can adopt is the use of intermediate investment vehicles in the form of trusts, foundations, limited partnerships or corporations (Villalonga and Amit, 2009). These vehicles are commonly used for several reasons, such as tax exemptions, liability protection or strategic motivations (Tong and Li, 2011; Villalonga and Amit, 2009). Moreover, firms can use these investment vehicles to invest in firms against the wishes of local stakeholders, as the investment activities of parent acquirers then typically go unnoticed. For instance, Russian investors used a number of local investment vehicles to gain control over regional electric power companies Ukraine (Crane, Peterson, and Oliker, 2005).

According to Murtinu and Scalera (2016) SWFs can also use investment vehicles to show a passive investment approach towards the host country government. They define three forms of investment vehicles commonly used by SWFs. (I) Non-SWF majority-owned financial vehicles, which can be used for tax objectives, investment allocations and financial returns. (II) Non-SWF majority-owned corporate vehicles, which can be used for their unique assets or expertise to pursue technological, market or strategic goals. (III) SWF majority owned vehicles, which are pre-existent or set-up by the SWF and subsequently fully acquired. The latter form of investment vehicle in particular could also be used for political purposes. Balding (2008) describes this type as frequently used by Middle Eastern SWFs. For example, the Abu Dhabi Investment Authority used many foreign subsidiaries to place large

(21)

21

investments in oil-related Canadian firms, in order to prevent the local media from being able to demonstrate that the parent acquirer was a Middle Eastern SWF.

All these arguments suggest that the use of an investment vehicle by a SWF can mitigate the level of opposition expressed in the recipient country, as the investment typically goes unnoticed by both local media and host country governments. Therefore, I hypothesize that:

H1: SWFs investing directly in the target firm face a higher level of hostility in the target country than SWFs investing through an investment vehicle.

Another internationalization strategy that firms can adopt to mitigate the level of hostility is the use of alternative entry modes. The choice between non-equity entry modes and equity entry modes has major implications for resource commitment, risk, performance, and control (Pan and Tse, 2000). Several scholars in the international business literature provide insights into the different investment strategies of institutional investors. According to Shleifer and Vishny (1986), large shareholders are more likely to monitor managerial performance, since small shareholders lack incentives due to a certain free-rider problem. Moreover, institutional investors that are less connected to the focal firm are focused on monitoring and influencing, rather than trading for profit (Ferreira and Matos, 2008; Chen, Harford, and Li, 2007). In addition, Götz and Jankowska (2016) find that state-owned enterprises adapt their entry mode to the conditions of the host country, especially when the pressures for legitimacy are stronger.

With regard to SWF investments, Aizenman and Glick (2008) suggests that SWFs can use a variety of entry strategies. For example, Singapore’s SWF, Temasek, invests in long-term stakes, and seems to take a more active role in the target firm by choosing for majority ownership choices. On the other hand, the Norwegian Government Pension Fund tends to

(22)

22

purchase minority ownership stakes and hires external managers to monitor the firm. Further, Murtinu and Scalera (2016) find that SWFs that acquire majority stakes in the focal firm are more likely to use an investment vehicle. They suggest that an active role for SWFs, pursued through majority investments, may attract more hostility in the recipient country. This seems to be justified by Bortolotti et al. (2015) who emphasize that SWFs, who hold a seat on a target firm’s board negatively influence the firm’s value, especially in the case of majority acquisitions. Apart from the effects on target firm performance, large positions of SWFs in target firms can give them access to local resources and superior information, which can be transferred to firms in the SWF home country (Aguilera et al., 2015). Finally, recipient countries especially fear majority acquisitions as this can give SWFs a significant influence over target firms and national economies (Rose, 2008).

Following the above arguments, large shareholders are likely to pursue a more active role in the focal firm, which can in the case of SWF investments lead to a negative firm value, and more political influence of the SWF parent acquirer. As a result, minority acquisitions of SWFs signal a passive investment approach, and imply a very limited ability to influence the target firm’s value and managerial decisions. Therefore, I expect that SWFs acquiring majority stakes in the target firm are likely to attract more hostility in the recipient country. Hence, I hypothesize that:

H2: SWFs acquiring majority equity stakes face a higher level of opposition in the target country than SWFs acquiring minority equity stakes.

The level of hostility expressed in the foreign country may also be a result of SWF-specific determinants. Clark et al. (2013) characterize SWFs as investment funds that are owned and controlled by national governments. They ascertain that this control can be exerted through the appointment of either politicians or external managers to the SWFs

(23)

23

board. According to Shleifer and Vishny (1994), the presence of politicians in firms leads to strategies that are adapted to interest groups, which may result in activities that do not only pursue the maximization of shareholder value but also serve political objectives. This is in line with Krueger (1990), who highlights that different investment strategies of government-owned firms can be a result of a preference to hire politically connected people, rather than the best qualified people. Moreover, Karolyi and Liao (2010) emphasize that politicians present on the board of institutional investors may face agency problems, which make them susceptible to corruption and bribes.

In terms of SWF investment strategies, Bernstein et al. (2013) try to ascertain whether the presence of politicians in the board of SWFs results in different investment practices. They find that the presence of politicians in SWFs is associated with a more negative change in the price/earnings ratio in the industry in the year after the SWF investment. In addition, Bortolotti et al. (2015) provide evidence that investments by politicized SWFs are associated with a more negative effect on target firm value.

Apart from the negative association between fund politicization and target firm performance, controversies surrounding politicized SWFs are also due to other rationales. Since SWFs are currently growing at a faster pace than the global rate of new issuance of traditional reserve assets, anxiety has also been raised about the potential destabilizing effects on national economies that SWFs might have (Gieve, 2008; Summers, 2007). In particular, autocratic government owners of SWFs might seek to undermine target firm economies by acquiring strategically important assets (Barbary and Bortolotti, 2011). Moreover, SWFs can acquire large stakes in strategic industries, and potentially alter corporate strategies. For example, a state-owned enterprise of the United Arab Emirates raised controversies in the United States by purchasing a British shipping company, which allowed them to get control over several port facilities in the United States (Chhaochharia and Laeven, 2009).

(24)

24

Considering the negative effects of politicized SWFs on target firm value and the potential destabilizing effects of SWF investments on national economies, the involvement of politicians in SWFs is likely to increase the level of hostility expressed in the recipient country. Therefore, I hypothesize that:

H3. Politicized SWFs face a higher level of opposition in the target country than non-politicized SWFs.

Another SWF-specific determinant that may cause hostility in the target country is the degree of transparency. Today, the lack of transparency surrounding SWFs is one of the most frequently mentioned criticisms, as SWFs are not required to disclose information about their performance and investments strategies to shareholders (Keller, 2008). As a result, many of the largest SWFs worked with the International Monetary Fund to create a code of conduct in order to achieve a higher level of transparency of SWF. This effort resulted in the establishment of the Santiago Principles in 2008, which are voluntary guidelines signed by several large SWFs (Avendaño and Santiso, 2009; Gieve, 2008). However, today there are still only a few SWFs that disclose organizational details, which makes that many SWFs currently rank below the most secretive hedge fund in terms of disclosure on firm performance (Megginson et al., 2013; Truman, 2009; Weiss, 2008). As a result of this lack of transparency the United States will continue to be restrictive on foreign SWF investments as long as recipient countries need the investments of SWF less than SWFs need the investment capital (Rose, 2008). This is in line with Gieve (2008), who emphasizes that a higher level of transparency may help to alleviate concerns and anxiety in the recipient countries, thereby reducing protectionist pressures.

In particular, Keller (2008) gives two explanations as to why the lack of transparency is distressing to policymakers. First, it can be hard to determine whether SWFs are pursuing

(25)

25

purely commercial objectives when they do not disclose information about their investment allocations. Second, a lack of transparency makes it more difficult to identify corruption by SWF managers. Finally, with regard to firm performance, Kotter and Lel (2008) highlight that fund transparency is an important determinant of the market reaction. They find that existing shareholders of target firms reward SWF transparency by showing that investments of more transparent SWFs lead to higher cumulative abnormal returns.

Taking the above arguments into consideration, SWF opacity may lead to higher perceived risks in the recipient countries, as investments by opaque SWFs may have negative consequences for target firm performance, and national security. As a result, I claim that opaque SWFs face a higher level of hostility expressed by the host country’s government and a more negative public opinion. Therefore, I hypothesize that:

H4. Opaque SWFs face a higher level of opposition in the target country than transparent SWFs.

Thus, this research aims to examine which SWF- and deal-specific determinants affect the level of opposition in the target country. Figure 1 gives a visual representation of the dependent and independent variables considered in this research.

Figure 1. Conceptual Framework.

Direct investments

Majority acquisitions

Fund politicization Fund opacity

Level of opposition in the target country

(26)

26

3. Methods

This chapter explains the methods used for this study. First the data and sample of this research are described. Second, the dependent variables are discussed. Third, the independent variables are explained. Fourth, the control variables are given. Fifth, the regression model used in this research is explained. Sixth, the descriptive statistics and correlations between the variables considered in this research are discussed.

3.1 Data and Sample

To collect data on the level of opposition, I use the geopolitical model offered by Shi et al. (2016). They propose that scholars can rely on secondary data – such as media publications – to obtain information about the number of investments undertaken by SWFs that have been subject to government scrutiny and obstruction in the recipient country. As a result, I use secondary textual data from the Lexis Nexis academic database. This is one of the largest databases on general news publications worldwide and is frequently used in the social sciences (Deacon, 2007). From this database, I extracted newspaper articles that report the specific SWF investments, because these provide a suitable source to measure the level of opposition in recipient countries, due to their social and political orientation (Riffe, Aust and Lacy, 1993). Moreover, according to Altheide (1997), publications of the mass media on issues and problems are inevitably associated with public opinion. He suggests that the pervasiveness of fear in societies is likely to be produced through interaction with the mass media.

Further, I use a dataset created by Murtinu and Scalera (2016) to collect data on deal- and SWF-specific determinants. This dataset contains more than 1000 SWF acquisitions worldwide from 1997 to 2013. Moreover, this is the only dataset that incorporates the use of investment vehicles. From this dataset, all SWF cross-border investments to the United States are extracted in order to establish a dataset that only comprises cross-border acquisitions with

(27)

27

the United States as the target country. This selection is due to three arguments. First, to determine the level of hostility, this research is limited to English speaking newspapers. Second, several authors suggest that the level of opposition is more severe in Western countries (Cohen, 2009; Fernandes 2014; Lee, 2010). Third, this selection comprises SWF investments from all four world regions, and are placed in a large variety of industries (12).

The final dataset comprises 105 cross-border SWF acquisitions, which equals 9 percent of the dataset of Murtinu and Scalera (2016). The time-period of this dataset goes from January 1, 2000 to December 31, 2013. Table 1 shows the number of cross-border SWF investments with the United States as the target country.

Table 2. Cross-border SWF investments to the United States.

SWF Parent Acquirer SWF Country No. of investments Dubai International Capital United Arab Emirates 2 International Petroleum Investment Company United Arab Emirates 6

Istithmar United Arab Emirates 5

Mubadala Development Company United Arab Emirates 7 Caisse de Dépôt et Placement du Québec Canada 12 Canada Pension Plan Investment Board Canada 15

China Investment Corporation China 4

Korea Investment Corporation Republic of Korea 1

Kuwait Investment Authority Kuwait 2

Khazanah Nasional Malaysia 12

Stichting Pensioenfonds ABP Netherlands 5 Guardians of New Zealand Superannuation Fund New Zealand 1

Qatar Investment Authority Qatar 4

Government of Singapore Investment Corporation Singapore 8

Temasek Holdings Singapore 21

Total 105

In order to collect data on the level of opposition, I used different search terms in the Lexis Nexis database for SWFs that invested directly in the target firm and SWFs that invested through an investment vehicle. First, in the case of direct SWF investments, only the name of the SWF parent acquirer and the name of the target firm were used as a search term. Second, if the SWF did invest through an investment vehicle, I used the name of the SWF parent acquirer and the name of the target firm, as well as the name of the investment vehicle

(28)

28

and the target firm as search terms. This was done, because a preliminary study of 20 SWF investments indicated that if the SWF invested through an investment vehicle, the combination of the name of the SWF and target firm did not, in most cases, lead to newspaper articles written about the SWF investment.

Further, it must be noted that I shortened the full names of the SWF, investment vehicles, and target firms in order to get the full number of articles published about the specific investment. For example, I used Mubadala as a search term rather than Mubadala Development Company. Within the results list of the Lexis Nexis database, I extracted only articles that were published between one day before and three days after the announced date of investment, because the preliminary study indicated that all articles within this time period spoke about the specific acquisition. This resulted in an average number of 2.33 articles per investment, with a minimum of 0 and a maximum of 26.

3.2 Dependent Variables

Three variables are constructed to measure the level of hostility in the target country. For the first variable, the number of newspaper articles written about the specific acquisition are counted. This method has been previously used in several pieces of research. For example, Hopkins (2010) used it in a study to document where and when immigrants provoke opposition.

For the second variable, I adopted summative content analysis of the appearance of several key words (Hsieh and Shannon, 2005). In accordance with MacNamara (2005), this analysis involved the examination of multiple key words operationalized in a coding scheme. The preliminary study of 20 SWF investments indicated that the authors of the newspaper articles spoke about SWFs by using eight different key words. These key words can be roughly categorized into three categories. The first category comprises information about the specific SWF involved in the acquisition and contains the name of the SWF and the SWF

(29)

29

home country. It must be noted that if the SWF is based in the United Arab Emirates either Abu Dhabi or Dubai was used as the SWF home country, since the SWFs considered in this research either belong to the emirate of Abu Dhabi or Dubai. The second category contains four synonyms that were frequently used in the articles to refer to the previously mentioned SWF. These synonyms are sovereign wealth fund, pension fund, investment arm, and investment vehicle. Finally, the third category contains two verbs that were frequently used to refer to the specific SWF as well. These verbs are sponsored, and state-owned.

With regard to the third variable, I used context units to indicate whether the article spoke with a negative tone about the specific investment. As the meaning of a word depends on its syntactical role within a sentence, I counted the number of articles that contain negative sentences about the investment (Krippendorff, 2004). All three variables described above are measured by a count variable.

3.3 Independent Variables

The four independent variables examined in this research are Direct Investment, Majority, Opacity and Politicization. First, Direct Investment is measured by a dummy variable that equals one if the focal SWF made a cross-border acquisition without the use of an investment vehicle. To determine if the SWF invested through an investment vehicle in the target firm, three types of investment vehicles identified by Murtinu and Scalera (2016, p. 4) are considered, which are: “(I) non-SWF majority-owned financial vehicles, such as private equity funds, venture capital funds, investment banks, asset management companies, commercial banks, investment management companies, financial branches of big corporations, real estate investment trusts, and investment advisory firms; (II) non-SWF majority-owned corporate vehicles, in the form of non-financial corporations, or companies controlled by public agencies not controlled by the SWF or by the government of the country in which the SWF originates; and (III) other SWF investment vehicles, including SWF

(30)

30

majority-owned financial and non-financial corporations”. Within my dataset of 105 cross-border acquisitions, 56 investments (53%) were direct investments. With regard to the investment vehicles, the majority of them (52%) are based in the United Sates. For example, a Singaporean SWF (Temasek) used the American bank Morgan Stanley Venture Partners as an investment vehicle to invest in the United States’ firm SOMA Networks. Four percent of the investment vehicles were based in the home country, whereas 44% of them were based in a third country, which are mainly OECD countries. An example of the latter is an investment of the Chinese SWF China Investment Corporation. This SWF used the Canadian firm Fairfax Financial Holdings as an investment vehicle to invest in Diamond S Shipping in the United States.

Second, Majority is a dummy variable that equals one if the stake acquired by the SWF exceeds 50%. The dataset created by Murtinu and Scalera (2016) provides data on this. Within my sample, only 16 SWF investments (15%) involved majority acquisitions.

Third, for Opacity I use data provided by Bagnall and Truman (2013). They created a SWF transparency and accountability scoreboard with a scale of 0 to 100 from data provided by the Peterson Institute for International Economics. The mean transparency score for the SWFs considered in this sample is 70, which is the threshold for this study. Methodologically speaking, Opacity is a dummy variable that equals one if the SWF scores higher than this threshold. Within my dataset, 29 SWFs (28%) are regarded as opaque funds.

Fourth, as regards fund politicization, this is a dummy variable that equals one if politicians are present in the management of the SWF. Data on this is provided by a report of J.P. Morgan (Fernandez and Eschweiler, 2008). Within my dataset politicians are present in 36 SWFs (34%).

(31)

31

3.4 Control Variables

I controlled for a number of factors that may influence the level of opposition per acquisition. First, as highlighted by the theoretical framework, many concerns surrounding SWF investments are raised because of the potential influence it might have on strategic infrastructures and sensitive firms operating in key industries (Cohen, 2009). To classify if a sector is strategic, the criteria proposed by Drezner (2008) and Keller (2008) are used. As a result, I consider banks, insurance companies, construction companies, energy suppliers, primary sector companies, and transport companies as firms based in strategic sectors. Strategic sector is a dummy variable that equals one if the target firm belongs to one of these industries.

Second, I control for bilateral trade agreements between the SWF home country and target firm country, as the existence of bilateral trade agreements between countries may alleviate the risk perception of target countries, which affects the degree of opposition (Li and Vashchilko, 2010). The Non-Bilateral variable is a dummy variable that equals one if the SWFs home country and target country either have a free trade agreement or preferential trade agreement. Data on this is collected from the World Trade Organization.

Third, I control for the size of the SWF, as SWFs that are able to acquire large stakes in target firms may face a higher perceived risk in the target country (Chhaochharia and Laeven, 2009). The average size of the SWFs considered in this sample is 111 billion U.S. dollar assets under control, which is the threshold for this study. More specifically, size is a dummy variable that equals one if the focal SWF has more assets under control than this threshold.

Fourth, SWF acquisitions during the financial crisis of 2008 in particular were surrounded by suspicions in the target country (Fernandes, 2014; Lee, 2010). Thus, I control

(32)

32

for the year of investment by a dummy variable that equals one if the investment took place in the years 2008 to 2013.

Finally, I control for potential differences in strategic motivations between Sovereign Pension Reserve Funds (SPRF) and Social Security Reserve Funds (SSRF). According to Blundell-Wignall et al. (2008) these types can differ in investment strategies and transparency requirements, which might affect the hostility faced in the target country. More specifically, I use a dummy variable that equals one if the fund is defined as a Social Security Reserve Funds. Table 2 provides a description of the variables and their sources considered in this study.

Table 2. Definition of variables.

Variable Definition Source

Dependent Variables Articles SWF Appearance Negative Articles Independent Variables Vehicle Majority Politicization Opacity Control Variables Strategic sector Non-Bilateral Size Crisis SSRF

Count variable that counts the number of articles published around the date of acquisition

Count variable that counts the number of appearances SWF related key words (see section 3.2)

Count variable that counts the number of articles published around the date of acquisition that speak with a negative tone

Dummy that equals 1 if the focal SWF invests in a foreign country through a vehicle

Dummy that equals 1 if the stake owned by the SWF exceeds 50%

Dummy that equals 1 if politicians are present in the managing bodies of the SWF

Dummy that equals 1 if the focal SWF has a scoreboard lower than 70

Dummy that equals 1 if the target industry is strategic

Dummy that equals 1 if the SWF country and target country do not have a free trade agreement and/or a preference trade arrangement

Dummy variable that equals 1 if the focal SWF has assets higher than 111 billion U.S. dollars Dummy that equals 1 in if the investment took place in the years 2008–2013

Dummy that equals 1 if the fund is a Social Security Reserve Fund

My dataset My dataset My dataset

Murtinu and Scalera (2016) Murtinu and Scalera (2016) Fernandez and Eschweiler (2008) Bagnall and Truman (2013), Peterson Institute for International Economics Drezner (2008), Keller (2008) World Trade Organization Truman (2009)

Murtinu and Scalera (2016) Bagnall and Truman (2013)

(33)

33

3.5 Methodology

The statistical program SPSS is used to execute a regression analysis for each dependent variable. Since all the dependent variables are count variables that are positively skewed, both a logistic negative binomial regression and a logistic Poisson regression are appropriate regression models (Coxe, West, and Aiken, 2009). These models assume that there is no normal distribution, as they model the natural log of the dependent variable. The main difference between these models is that the Poisson regression assumes that the mean and variance are equal, whereas the negative binomial regression is less biased on this (Greene, 1994). Since there is an excess of zeros for all three dependent variables, the variance is much higher than the mean. As a result, the negative binomial regression, which has been formulated with overdispersion, is more appropriate in this study, because this makes the model less biased for heteroskedasticity (Gardner, Mulvey and Shaw, 1995). The logistic negative binomial regression can be denoted as follows:

𝑙𝑛 𝑦1, 𝑦2, 𝑦3 = 𝐵 + 𝛽1 + 𝛽2 + 𝛽3 + 𝛽4 + 𝜂𝑉

Within this equation, the y stands for the dependent variable (see section 3.2), the beta includes the independent variables (see Section 3.3), and the vector includes the control variables (see Section 3.5).

3.6 Descriptive Statistics

The descriptive statistics of the dependent, independent, and control variables of the 105 SWF cross-border investments considered in this research are shown in table 3. An average of 2.13 articles are published one day before up to and including three days after the acquisition date of the SWF investments. Additionally, an average of 7.90 key words related to the SWF were found in the articles (see table 2). From the articles that were published around the SWF acquisition date, 0.29 spoke with a negative tone about the investment.

(34)

34

As regards the independent variables, 53% of the SWF investments were made without the use of an investment vehicle, which is in line with Murtinu and Scalera (2016). Only a few (15%) investments exceeded a stake of 50% of the target firm’s equity, which is again in line with Murtinu and Scalera (2016). As expected, the minority of the investments were made by opaque SWFs (47%), whereas in line with Bernstein et al. (2013) around one third (34%) of them were politicized. As regards to the control variables, almost two third of the investments targeted a strategic sector (63%). With regard to bilateral trade agreements, the minority of the SWF investments (46%) came from SWFs in countries that do not have bilateral trade agreements with the United States, which is significantly lower than expected (Knill et al., 2012) Additionally, the majority of the SWF investments were undertaken by SWFs that have a relatively high number of assets under their control. Further, in line with Murtinu and Scalera (2016), the majority of investments (54%) were undertaken during the financial crisis. Finally, more than two-thirds (70%) of the investment were made by a Social Security Reserve Fund.

Table 3. Descriptive statistics.

Variable Mean S.D. Min Max N

Articles SWF Appearance Negative Articles Direct Investment Majority Politicization Opacity Strategic Non-Bilateral Size Crisis SSRF 2.13 7.90 0.29 0.47 0.15 0.34 0.47 0.63 0.46 0.56 0.54 0.70 4.3169 22.7107 0.8516 0.501 0.361 0.477 0.501 0.486 0.5005 0.499 0.501 0.463 0 0 0 0 0 0 0 0 0 0 0 0 26 176 5 1 1 1 1 1 1 1 1 1 105 105 105 105 105 105 105 105 105 105 105 105

To determine whether multicollinearity is present between the variables, a correlation matrix of the dependent, independent and control variables is shown in table 4. According to this table, multicollinearity seems to be present between the dependent variables. The

(35)

35

correlations between these variables are all higher than 0.45 (p-value < 0.01). As a result, I conduct a reliability check to determine if the dependent variables do not measure the same construct. The Cronbach’s Alpha among these three items is 0.41, which reassures us that there is no overlap between these three different measures. In addition, multicollinearity also seems to be present between several independent variables. However, this is likely to be a result of the fact that all these variables are all dummy variables (Farrar and Glauber, 1967). Moreover, I checked if dropping one control variable from the regression equation affects the standard errors of the other variables, which was not the case (Slinker and Glantz, 1985).

(36)

Table 4. Correlation matrix. Variable 1) 2) 3) 4) 5) 6) 7) 8) 9) 10) 11) 12) 1) Articles 2) SWF Appearance 3) Negative Articles 4) Direct Investment 5) Majority 6) Politicization 7) Opacity 8) Strategic 9) Non-Bilateral 10) Size 11) Crisis 12) SSRF 1 0.908** 0.570** -0.247* 0.258** -.060 -0.002 -0.063 -0.003 -0.021 0.020 0.111 1 0.445** -0.251** 0.204* 0.064 0.082 -0.038 -0.077 -0.088 -0.087 -0.035 1 -0.203* 0.264** -0.30 -0.045 -0.090 -0.052 0.094 0.106 0.155 1 -0.78 0.129 0.120 0.205* -0.100 -0.095 -0.061 -0.122 1 -0.27 -0.131 -0.442** 0.123 0.160 -0.143 0.295** 1 0.772** 0.348** -0.787** -0.638** 0.099 -0.478** 1 0.482** -0.674** -0.364** 0.130 -0.619** 1 -0.270** -0.592** 0.244 -0.861** 1 0.386** -0.075 0.400** 1 -0.114 0.750** 1 -0.182 1 * p-value <0.05, ** p-value <0.01

(37)

4. Results

This chapter discusses the results of the regression models tested in this research. First, the results of the negative binomial regression model are explained. Second, the results of a robustness check, in the form of a Poisson regression, are given and compared with the main results.

4.1 Main Results

As mentioned in section 3.5, a logistic negative binomial regression analysis is executed to measure if the independent variables explain some of the variance of the dependent variables. For the variable Articles, the omnibus test indicates a likelihood ratio chi-square of 57.96, statistically significant at 1%. For the variable SWF Appearance, the likelihood ratio chi-square is 124.34 (significant at 1%). As regards the variable Negative Articles, the likelihood ratio chi-square is 43.38 with a significance level of 1%. Thus, all regression models are statistically significant at a 1% confidence level.

Table 5 shows the results of the logistic negative binomial regression. All the independent variables and control variables are included in one generalized linear model for each of the three dependent variables. As the regression coefficient (B) does not have a simple interpretation, the exponential values of these coefficients are considered by calculating the exponent of B. The exponential values of B represent the odds ratios, and are shown in the columns Exp B. More specifically, the Exp B can be interpreted in terms of the change in odds (Osborne, 2008).

The results show that for SWFs who engage in direct investments, the number of Articles, SWF Appearance, and Negative Articles are all likely to increase (all three variables are significant at 1%). The odds of the number of Articles, SWF Appearance and Negative Articles are subsequently 4.01, 8.98, and 26.41 times higher for SWFs that invest directly in foreign firms compared to those that use an investment vehicle. These findings support

Sovereign wealth funds : internationalization strategies in the shadow of foreign hostility