1
Sovereign Wealth Funds:
The influence of political systems on performance
Master Thesis
MSc. in Business Administration, International Management University of Amsterdam
Student: Robin Klaver (11398124) Thesis supervisor: Dr. V.G. Scalera Date: January 26, 2018
2
Statement of originality
This document is written by Student Robin Klaver 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
Table of contents
1 Introduction... 5
2 Literature review ... 9
Sovereign Wealth Funds ... 9
Sovereign Wealth Funds investments ... 11
Investment motives ... 12
2.3.1 Political motives ... 13
Political systems characteristics ... 15
Investment performance ... 18 Research gap ... 20 3 Hypothesis development ... 22 4 Methodology ... 27 Data collection ... 27 Sample ... 28 Variables ... 30 4.3.1 Dependent variables ... 30 4.3.2 Independent variables ... 31 4.3.3 Moderator variables ... 31 4.3.4 Control variables ... 32 4.3.5 Analytical models ... 34 5 Results ... 36
Model assumptions and descriptive statistics ... 36
Correlation analysis... 39 Regression analysis ... 42 6 Discussion ... 47 Academic relevance ... 49 Managerial implications ... 50 Policy implications ... 50
Limitations and suggestions for future research ... 51
7 Conclusion ... 52
Acknowledgements ... 55
References ... 56
4
Abstract
Sovereign Wealth Funds (SWFs) control capital on behalf of governments to achieve a rate of return. They have gained increasing attention because of their investments in national economies and critical industries. This attention is also caused by the presence of politicians in SWFs, leading to negative reactions in target countries because of unclear SWF motives. Partly because of this, SWF investments with politicians involved are associated with higher drops of target firm value. To date, it was unclear whether political systems with different incentive structures and political motives have distinctive performance effects. Therefore, this study examines the relationship between investment performance and political systems in both the SWF home country and target country. In addition, the effects of political stability, the presence of politicians in SWFs and the use of intermediate investment vehicles to indicate a passive stance towards target countries are examined.
A sample of 189 investments of 24 different SWFs is used to perform a cross-sectional regression analysis. The results confirm that investment performance is positively influenced by democracy in the home country. No evidence was found for the moderating effect of democracy in the target country, presence of politicians in a SWF or the use of an investment vehicle. Contrary to the expectations, political stability is negatively associated with investment performance. These findings supplement the literature on host-country-specific characteristics that influence acquisition performance by combining the literature on SWF investment performance and the influence of political systems.
Keywords: Sovereign Wealth Funds; acquisition; investment; performance; political systems; democracy
5
1 Introduction
Sovereign Wealth Funds (SWFs) have gained increasing attention because of their investments in large Western firms and size with assets estimated to be worth $7.32 trillion (Alhashel, 2015; Sovereign Wealth Fund Institute, 2017). They became key players by supporting banks during the 2008 financial crisis, developing their own national economies and investing in critical industries like transportation and utilities (Aguilera, Capapé, & Santiso, 2016). This has not gone unnoticed in the newspapers, with topics including increasing investments in emerging markets by SWFs and growing protectionism in the United States (Malay Mail Online, 2017), socialists intervening with the investment strategy of a potential Swiss SWF (Worstall, 2017) and fear of political risks regarding usage of funds for electoral benefits when setting up a SWF in the United Kingdom (Kerr, 2017). From these articles it turns out that the government-owned aspect of SWFs adds an extra dimension of governments influence, which raises interesting questions (Aguilera et al., 2016).
Generally, SWFs control capital on behalf of the government to achieve a rate of return (Balding, 2008). They commonly diversify assets that originate from natural resource revenues (Aguilera et al., 2016; Aizenman & Glick, 2009; Johnson, 2007). A SWF portfolio normally consists out of public and private equities and fixed income securities (Preqin, 2015). This research specifically focuses on acquisitions, because of the long-term orientation of these investments and the possibility for a SWF to influence acquisition targets (Bernstein, Lerner, & Schoar, 2013). The objectives of SWF include diversifying assets to reduce dependency on underlying commodities, increasing return on reserves to provide for future generation, promoting industrialisation by acquiring expertise or promoting political intentions (Aguilera et al., 2016; Blundell-Wignall, Hu, & Yermo, 2008). The intention of SWFs to maximize
long-6
term financial returns might for example be influenced by investing in domestic underperforming industries due to political influences (Bernstein et al., 2013). Fund politicization also leads to negative reactions in recipient countries because of unclear motives (Cohen, 2009; Knill, Lee, & Mauck, 2010). Distrust mainly occurs as a result of different norms between the SWF home country and the recipient target country (Monk, 2009). It is expected that political motives and responses to SWF investments differ by political system because of different incentive structures (Wu, 2012). This affects the fund performance, SWFs operating in political opaque surroundings can, for example, make investments with more agility than transparent counterparts (Drezner, 2008). On the other hand, corruption is more commonplace due to a lack of checks and balances (Kurer, 1993; Sandholtz & Koetzle, 2000; Wu, 2012). To date, the distinctive influence of political systems on the performance of their SWFs remains unclear.
SWFs generally gain a positive investment return on the short term, followed by negative effects on the long term (Bortolotti, Fotak, & Megginson, 2015; Chhaochharia & Laeven, 2008; Dewenter, Han, & Malatesta, 2010; Kotter & Lel, 2011). However, politicization of SWFs results in higher drops in firm value due to the previously mentioned reasons among others (Bernstein et al., 2013; Bortolotti et al., 2015; Murtinu & Scalera, 2015). This explanatory research examines the influence of different political systems on the performance of SWFs acquisitions. To do this, political systems are placed on a scale from autocracies to democracies with anocracies as middle category (Goldstone et al., 2010; Hague, Harrop, & McCormick, 2016). Main differences between the systems are the length of governance and thus a stable investment policy, the amount of repressive elements in a society and participation opportunities for citizens (Hague et al., 2016; Siaroff, 2013). These elements influence SWF performance, because other shareholders might for example fear possible tunnelling and
7
expropriation behaviour if participation in a political system is limited (Dewenter et al., 2010; Fotak, Bortolotti, Megginson, & Miracky, 2008). By combining data on political systems and SWF acquisition performance, this research answers the following question: How do political systems in the SWF home country and target country influence the performance of acquisitions? This research question will be answered using an Ordinary Least Squares regression analysis on 189 cases of 24 different SWFs (Field, 2013). The data contains cross-sectional SWF deal data, investment performance data and country-specific information regarding the political system in both the SWF home and target country. This approach supplements the existing literature on the influence of country characteristics on SWFs. Existing research reveals that SWFs prefer to invest in countries with relatively weaker bilateral political relations and in relatively developed countries (Knill, Lee, & Mauck, 2012; Kotter & Lel, 2011). However, little is known about the influence of home and host country-specific characteristics on acquisition performance (Murtinu & Scalera, 2015). In addition, there is an ongoing debate on the motives of SWFs. Some authors describe that SWFs are used for social or political goals of politicians and may be used as an arm of foreign policy (Cohen, 2009; Fernandes, 2014). Others disagree and describe SWFs as purely market investors (Kotter & Lel, 2011; Megginson, You, & Han, 2013). This research adds quantitative insights to this debate by researching whether there is a performance difference in their investment patterns to demonstrate if certain political systems systemically outperform others. Existing research has found a drop in firm value with SWF politicization (Murtinu & Scalera, 2015). It is however unknown whether this drop differs by political system, with other associated corporate governance and transparency standards. Furthermore, current literature on comparative politics focuses on regime characteristics. This research looks at whether these characteristics also apply to their SWFs to see how they are affected by local institutions. Lastly, the outcomes are relevant in practice for
8
shareholders by showing how the performance of firms is influenced by SWF acquisitions, allowing them to adjust their investment strategy. After all, managers and shareholders of target firms are directly impacted by SWF investments.
This thesis starts with a literature review that leads to the research gap and hypotheses. The fourth chapter discusses the methodology, followed by the results section and the discussion section. Lastly, chapter seven contains the conclusion.
9
2 Literature review
This chapter provides an evaluation of existing literature on SWFs and political influences. First, SWFs are introduced (2.1) followed by a description of their investments (2.2). The third section provides an overview of their investment motives (2.3). The characteristics of political systems are then described (2.4). Lastly, the investment performance of SWFs is discussed (2.5) and a research gap regarding performance implications by different political systems is identified (2.6).
Sovereign Wealth Funds
The first non-federal SWF established in 1854 to support educational institutions in Texas. Almost a century later in 1953 Kuwait was the first sovereign state to establish a SWF, with the intention to manage its oil surpluses (Alhashel, 2015). Today, more than 70 SWFs exist of which Norway’s Government Pension Fund and Abu Dhabi Investment Authority are the largest with assets estimated to be respectively 922 and 828 billion US dollars (Sovereign Wealth Fund Institute, 2017). These government-owned investment funds are distinct from State-Owned Enterprises (SOEs) because of their large pool of assets and international long-term orientation (Aguilera et al., 2016). Yet, the difference with SOEs is not always unquestionable. Both SOEs and SWFs can engage in the same type of businesses. Therefore, differentiation is possible based on owner’s funding source and objective of the entity rather than the businesses involved. SWFs typical business is investing, while SOEs typically have other everyday activities apart from investing (Backer, 2010).
Numerous definitions of SWFs exist, most referring to the source and overall goal of the funds. The OECD defines SWFs in a working paper as “a fund set up to diversify and improve the return on foreign exchange reserves or commodity (typically oil) revenue, and sometimes
10
to shield the domestic economy from (cycle inducing) fluctuations in commodity prices. As such most invest in foreign assets” (Blundell-Wignall et al., 2008, p. 4). A broader definition describes 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” (Balding, 2008, p. 9). Funding for SWFs include foreign exchange reserves, sales of (non-)commodity resources or tax revenues (Blundell-Wignall et al., 2008).
The SWF definition also covers some Public Pension Reserve Funds (PPRFs). Sovereign Pension Reserve Funds (SPRFs) are considered a SWF and the classification of Social Security Reserve Funds (SSRFs) differs by case, depending on the degree of legal independence and integration of balances in government accounts (Blundell-Wignall et al., 2008).
The SWFs are increasingly located in emerging and developing countries, because export-led economies and economies with natural resources benefit from global integration and higher commodity prices (Aguilera et al., 2016). This leads to current account surpluses that are not directly needed. These reserves are diversified using SWFs to attain more profitable assets (Aizenman & Glick, 2009; Johnson, 2007).
Existing studies on SWFs include the asset allocation of SWFs, with allocation characteristics like location and stake of the SWFs being examined (Bortolotti, Fotak, Megginson, & Miracky, 2009; Chhaochharia & Laeven, 2008; Kotter & Lel, 2011). SWF investments also spurred the debate on possibly imposing additional regulations, with proponents pointing to potential strategic behaviour by foreign SWFs in the future and opponents emphasizing existing regulation and SWFs acting as model investors (Bahgat, 2008; Epstein & Rose, 2012; Gilson & Milhaupt, 2007). This research builds on the asset allocation literature and focusses on the performance of investments, specifically acquisitions made by
11
SWFs. This research examines how political systems, with associated political agendas, influence the performance of acquisitions. These acquisitions will be discussed in the next section.
Sovereign Wealth Funds investments
SWFs manage investment portfolios on behalf of sovereign states, often taking large stakes in companies (Dewenter et al., 2010). The portfolio consists for a large part out of relatively stable public and private equities and fixed income securities. However, SWFs also increasingly diversify in real assets, such as real estate and infrastructure to meet their objectives. The New Mexico State Investment Council (SIC) for example funded the real estate fund Blackstone with 1.5 billion US dollars in March 2015. SWFs are active in hedge funds and private debts are limited, with respectively 33% and 23% of the SWFs investing in these asset classes (Preqin, 2015). This research focuses on acquisitions being “purchases of private equity, and structured equity positions in public firms” (Bernstein et al., 2013, p. 224), because more SWFs invest in these assets than the previously mentioned alternative assets. Furthermore, SWFs can exert more influence in long-term investments, which makes the effect SWF influence measurable (Bernstein et al., 2013).
Acquisition targets are regularly located in the United States and other Western countries. Those markets finance current account deficits with SWF investments and are liquid enough to receive large inflows (Drezner, 2008). Correspondingly, SWFs most likely invest in economically developed countries with protection of property rights (Ciarlone & Miceli, 2016). SWFs also target domestic companies and prefer to invest in the financial sector (Drezner, 2008; Fotak et al., 2008). In addition, SWFs increase acquisitions in countries in economic crisis to prevent further spread of economic turmoil (Bortolotti et al., 2015; Ciarlone & Miceli,
12
2016). Lastly, foreign acquisitions in countries with large SWFs are uncommon because they are the most protectionist concerning inward investments (Drezner, 2008).
When looking more closely to the acquisition methods used by SWFs, the sample of Fotak et al. (2008) shows that they prefer to acquire stakes using privately negotiated stock purchases versus the open market. Also, SWFs can either acquire a company directly or use an investment vehicle, such as a financial or corporate intermediate company. The intermediate can also be controlled by the SWF in the form of a subsidiary. The motives of a SWF are not always clear for host countries. Using an investment vehicle might indicate a passive stance and thereby reduce the hostility in the host country (Murtinu & Scalera, 2016). The acquisition motives will be further discussed in the next section.
Investment motives
SWF objectives include diversifying assets, increasing return on reserves, providing for future generations when natural resources are exhausted, promoting industrialisation or promoting political intentions (Aguilera et al., 2016; Blundell-Wignall et al., 2008). The first objective, diversifying assets, is especially applicable to funds with main incomes from one underlying commodity. Those SWFs reduce their dependence on fluctuations of the commodity price by diversifying their assets. Correspondingly, most SWFs are located in countries with commodities like oil, diamonds, copper and other raw materials (Fotak et al., 2008). Those funds are also known as stabilization funds (Allen & Caruana, 2008).
Increasing return on reserves is necessary in some societies to supply for aging societies and meet pension labilities (Fotak et al., 2008). This motive is similar to providing for future generations, where SWFs convert natural resources into a diversified portfolio and mitigate the effects of the Dutch Disease (Murtinu & Scalera, 2016). The Dutch Disease is named after the
13
economic development in the Netherlands during the first petroleum boom. An increased price for natural gas with associated payment surpluses lead to an appreciation of the local currency. A high currency valuation is detrimental to the export level of the manufacturing sector. A country and the investments by a SWF can prevent the effects of the Dutch Disease by diversifying assets and achieving growth in all sectors, rather than growing in a single sector (Bature, 2013).
Lastly, promoting industrialisation is achieved by acquiring expertise or technology to facilitate development in the home country and complement home country specific advantages (CSAs) (Drezner, 2008). Drezner (2008) mention the example of Arab SWFs acquiring stakes in the energy sector to complement their industries involving natural resources. Singapore’s SWF is another example that is more likely to acquire port facilities, which complements their trade economy as 14th largest exporter and 16th largest importer in the world (Drezner, 2008;
The World Bank, 2015). The promotion of political objectives is further discussed in the next section.
2.3.1 Political motives
Long-term maximization of financial returns might be influenced by social or political goals of politicians (Fernandes, 2014). SWFs with involvement of political leaders invest more in domestic firms and in underperforming local industries that need support (Bernstein et al., 2013). Political relations also influence investment location decisions because SWFs prefer, contrary to the Foreign Direct Investment (FDI) literature, to invest in nations with weaker political relations (Knill et al., 2012). Furthermore, SWFs tend to invest in countries with the similar cultures (Chhaochharia & Laeven, 2008). This may indicate that SWF exploit their informational specific advantages in similar cultures. However, the cultural bias is more apparent in comparison to other global investors. This deviation from rational investments
14
partially suggests that non-financial motives impact investment decisions (Chhaochharia & Laeven, 2008; Knill et al., 2012). Chhaochharia & Laeven (2008) conclude that government decisions affect the capital allocation of SWFs and “that such culture-induced variation in preferences may serve political rather than social objectives” (p. 33). Vasudeva (2013) argues that SWF motivations may include exerting normative pressures to influence firms’ investment behaviour. SWFs also obtain technology for domestic strategic interests (Drezner, 2008; Monk, 2009). Finally, SWFs can serve as a means through which political systems can safeguard them against movements in the free market and thereby avoiding toppling power structures. This motive is especially applicable to political systems with a strong hierarchical power structure (Drezner, 2008).
Target countries occasionally question investments because of unclear motives that potentially destabilize markets, create implications for national security and pursue non-market strategies (Knill et al., 2010). Distrust by target countries also arises as a result of different norms between the SWF home country and recipient target country (Monk, 2009). Opaque SWF structures and activities further contribute to these concerns and lead to higher perceived risk in the target country, accompanied by more political pressures there (Kotter & Lel, 2011; Murtinu & Scalera, 2016). These political pressure could influence investment decisions because some recipient countries pre-examine selected investments and may impose protectionist restrictions (Cohen, 2009; Monk, 2009).
Although SWFs may be used as an arm of foreign policy (Cohen, 2009) and are “blurring the lines between finance and politics” (Aguilera et al., 2016, p. 3), other research describes SWFs as purely market investors. Kotter and Lel (2011) find that SWFs behave like passive institutional investors. A study of country-level determinants have also found that SWFs act principally as commercial investors (Megginson et al., 2013). Lastly, investment allocation
15
decisions do not depend on the political system in the home and host country when comparing with mutual funds (Avendaño & Santiso, 2009). Nevertheless, the performance of SWFs acquisitions can still be affected by (the appearance of) political involvement. The characteristics of political systems and the performance implications for SWF investments are further discussed in the next sections.
Political systems characteristics
The political motives and responses to SWF investments differ by political system. This is because each system has different incentive structures, which influences actors and ultimately the economic performance (Wu, 2012). A SWF originating in a non-democratic country will, for example, face difficulties convincing others of their commercial goals in democratic countries. This distrust or lack of legitimacy places a discount on these SWF investments (Monk, 2009). On the other hand, SWFs operating in politically opaque surroundings can make investments with more agility than transparent SWFs. The interaction between the home country and it’s SWF is also clear from the correlation between civil liberties and the transparency of its SWF. SWFs can make unpopular investments more easily in political systems wherein dissidents are suppressed. Unpopular investments often generate higher profits in the long run, allowing autocratic states to have an advantage (Drezner, 2008). Making unpopular investments is difficult in democratic societies where politicians have to respond to the demands of the majority. Political leaders also cannot exploit their power because of higher contestation levels and horizontal checks (Wu, 2012).
Political systems have been researched in many respects. Wu (2012) found that democracies only outperform the economic performance of autocracies where structural factors exist, such as external circumstances unfavourable to growth. Other researched subjects include
16
the influence of political systems on FDI inflow levels (Jensen, 2003), the durability of political systems (Kailitz, 2013) and social requisites for democracies (Lipset, 1959).
To compare political systems, typologies are used to divide states into groups. Although there is no universally agreed division, the most common ones will be used here (Hague et al., 2016). Political systems can be broadly placed on a scale from autocracies to democracies with anocracies as an ambiguous middle category (Goldstone et al., 2010; Hague et al., 2016).
Full autocracies “combine an absence of effective contestation for chief executive with repressed or suppressed political participation” (Goldstone et al., 2010, p. 195). Features of this system may include non-existent or limited free elections, power in the hands of a limited number of people, weak institutions, limited political participation with possible constrained opposition, limited protection of individual rights and a limited media establishment that shares the state’s opinion. Changing leaders is usually difficult in autocratic regimes, resulting in long governance periods (Hague et al., 2016). Corruption is also higher in autocratic systems that exert greater control over the economy and in nations with weaker democratic norms and practices, resulting in misallocation of resources (Kurer, 1993; Sandholtz & Koetzle, 2000). This misallocation may be described as patronage, using state resources to reward supporters of the regime (Hague et al., 2016). Autocracies exist in at least three variants. Totalitarianism is characterised by one centre of power, lack of economic pluralism, an elaborate ideology as the basis for policies and citizen participation in the single political party. Sultanistic regimes are built around a glorified leader that lacks legitimacy and spreads fear and paranoia. This system now only occurs in Equatorial Guinea and North Korea. Authoritarian regimes are the residual category and lack the underlying ideology and mobilization characteristics like in totalitarian regimes (Siaroff, 2013).Those authoritarian regimes can be divided in at least five forms, of which an absolute monarchy with one ruler and members of the royal family in
17
politics and the military is mostly seen in the Middle East. Several former Soviet republics, Angola and Zimbabwe have ruling presidents that control the government and media reports. Ruling parties with a strong president are seen in many Communist and African states after their independence. Military rule with a government by the military hardly exists today. Lastly, a theocracy is seen in Iran, where religious leaders govern (Hague et al., 2016).
Anocracies are associated with weak central governments combining democratic with authoritarian repressive aspects (Vreeland, 2008). Democratic institutions are used to exert political authority in these hybrid states, but rules are violated to such an extent that minimum requirements for a democracy are not met. Characteristics include weak political participation, unfair elections, corruption, influenced media and weak regulations. Anocracies are most seen in poor states with a high level of ethnic, religious or economic heterogeneity and states that face external threats (Hague et al., 2016).
Democratic political systems are based on a fair and open mandate from citizens and are characterized by a representative government, free elections, freedom of speech, multiple parties with participation platforms, independent media, legislation that protects individual rights and political procedures with checks and balances. A country is a direct democracy when all citizens participate in the decision-making processes. Since this is not practical, self-government has evolved to an elected self-government in what is called a representative democracy. Citizens hold representatives accountable for their interests and can vote in elections. However, most contemporary democracies are classified as a liberal democracy, in which the characteristics of a representative democracy prevail with the addition that governments are limited in power to protect citizens' rights. Power is dispersed in these societies and procedures include checks and balances to ensure institutions work together (Hague et al., 2016; Siaroff, 2013). Political parties in this political system are commonly classified on a scale from left to
18
right. The left parties are more commonly associated with social security, education, trade unions and equality. Right parties favour the free market, free trade, law and order, freedom and rights. However, one should note that the distinction between left and right has blurred and the labels are ill-defined (Hague et al., 2016).
In sum, investments by SWFs that operate on state mandate are influenced by institutional variations in political systems. The next section will continue with the mixed investment performance by SWFs.
Investment performance
A wide body of research has examined SWF performance and the performance of their acquisition targets. SWFs are distinct from private investors because they are more often large shareholders. Theory suggests that large shareholders improve monitoring of managerial performance, because they have strong incentives to monitor their acquisitions. This in contrast to small shareholders, which can free-ride on large shareholders. It is thus expected that acquisition by a SWF will increase the performance of acquisition targets (Fotak et al., 2008). This is also partly because SWFs can act as lobbyists on behalf of a target. However, the drawbacks include possible tunnelling and expropriation costs (Dewenter et al., 2010). Large shareholders may influence firms against the interests of other shareholders, causing agency costs on the firm. These agency costs specifically arise from opaque SWFs because their behaviour is unknown to other shareholders (Fotak et al., 2008).
In general, target firm price reactions show a positive return after announcements of SWF investments on the short-term followed by falling shares or no substantial effect on the long-term (Bortolotti et al., 2015; Chhaochharia & Laeven, 2008; Dewenter et al., 2010; Kotter & Lel, 2011). The average initial positive price response is caused because SWFs often buy firms
19
in financial distress. The long-term results can be attributed to corporate governance problems between incumbent management and SWFs. Conflicting interests between SWFs and other shareholders will negatively influence the firm performance (Chhaochharia & Laeven, 2008). Another explanation for the long-run effects is limited shareholder activism by SWFs (Kotter & Lel, 2011). This explanation is however not supported by Kotter & Lel (2011), who find that target firms are subject to monitoring and influencing by SWFs.
The performance implications are also dependent on target and SWF characteristics. Transparent SWFs have a greater impact on target performance than opaque SWFs because of lower agency costs (Fotak et al., 2008; Kotter & Lel, 2011). Correspondingly, Bortolotti et al. (2015) find that SWFs under strict government control are associated with a large decrease in target firm value and performance because of possible political influences. Likewise, Murtinu & Scalera (2015) and Bernstein et al. (2013) find that higher politicization of funds leads to higher drops of firm value. Investments in strategic industries also drop more in stock price than non-strategic industries. This might be caused by controversy in the target country and a correlation between strategic industries and political objectives. Furthermore, target firm performance may be hindered by high levels of regulation in strategic industries (Murtinu & Scalera, 2015). Finally, cross-border investments increase more in stock price than domestic ones (Murtinu & Scalera, 2015). According to Bernstein et al. (2013) this is because political pressures influence SWFs to support domestic underperforming industries.
In sum, many factors including political involvement influence SWF investment performance. However, how the influence of political system characteristics upon investment performance remains unclear.
20
Research gap
Although the performance of SWF investments seems clear with short-term positive returns and negative effects on the long term (Bortolotti et al., 2015; Chhaochharia & Laeven, 2008; Dewenter et al., 2010; Kotter & Lel, 2011), fund politicization moderates this relation resulting in higher drops in firm value (Bernstein et al., 2013; Bortolotti et al., 2015; Murtinu & Scalera, 2015). It remains however unclear whether this applies for all political system or relates to (a combination of) specific political systems.
SWFs from democratic societies have to show more accountability and validity than autocratic societies to achieve legitimacy with their citizens, which might lead to longer procedures for SWFs and making timely decisions for investment opportunities more difficult (Clark & Monk, 2010). On the other hand, SWFs originating from more opaque non-democratic countries have to deal with lower acquisition performance because of their perceived risk and illegitimacy in some target countries (Knill et al., 2010; Kotter & Lel, 2011; Monk, 2009). Other shareholders might fear those SWFs because of possible tunnelling and expropriation behaviour, resulting in a drop of acquisition performance (Dewenter et al., 2010; Fotak et al., 2008). Furthermore, political corruption in these systems is also associated with misallocation of resources and might influence decisions by SWFs (Kurer, 1993; Sandholtz & Koetzle, 2000). The same SWFs might however make unpopular investments more easily, because of lower participation by citizens in these political systems (Drezner, 2008).
In sum, theory regarding the influence of political systems on the performance of their SWFs is mixed. Hence, the following research question will be examined: How do political systems in the SWF home country and target country influence the performance of acquisitions?
21
This research answers the call for more research on the effect home and host country characteristics on performance (Murtinu & Scalera, 2016). It adds an extra, yet undiscovered, dimension relating to performance of SWF investments to the literature. The political regime is also a proxy for the (non-)strategic objectives of SWFs. Bernstein (2013) argues that some political systems support local underperforming industries. Drezner (2008) adds that political systems with strong power structures use SWFs to safeguard them against movements of the free market. This research adds quantitative insights to this ongoing debate about the motives of SWFs and their regimes by researching whether there is a performance difference in their investment patterns. Furthermore, the theory on comparative politics mainly focuses on characteristics of political systems. This research adds insights on the extent to which these characteristics also apply to their SWFs. Lastly, the research outcomes are relevant in practice for shareholders by showing how the performance of firms is influenced by SWF acquisitions originating from certain political systems.
22
3 Hypothesis development
This study examines the influence of political systems on the performance of SWFs acquisitions. Acquisitions are examined because SWFs can exert more influence in long-term investments than in short-term trade (Bernstein et al., 2013). The performance of SWF investments is expected to be influenced by political systems with different structures and incentive systems (Wu, 2012). As described, political systems can be placed on a scale from autocracies to democracies (Goldstone et al., 2010; Hague et al., 2016). It is expected that SWFs with their home country in non-democratic countries face difficulties because they lack legitimacy in other regimes (Monk, 2009). Other investors fear tunnelling and expropriation by SWFs because their behaviour is unknown and might be conflicting with other shareholders (Dewenter et al., 2010; Fotak et al., 2008). Furthermore, horizontal checks are less common in autocracies resulting in corruption and misallocation of resources (Kurer, 1993; Sandholtz & Koetzle, 2000; Wu, 2012). The benefit of SWFs from non-democratic countries is, however, that they make unpopular investments more easily because power is centralized and citizens participate less in controlling those regimes (Drezner, 2008). Based upon the aforementioned literature and research question, the following hypotheses are made.
H1: There is a positive relation between the level of democracy in the home country and their SWF acquisition performance.
The performance is not only influenced by home country, but also by the norms in the target country. Autocracies have different standards and norms in comparison to democratic countries when it comes to ensuring property right protection. Furthermore, differences exist in the negotiation power of leaders and labour unions between democracies and autocracies. Lastly, the level of democracy affects the checks and balances as well as the accountability
23
mechanisms in place. These mechanisms decrease political risks and potential expropriation costs (Jensen, 2003). The level of checks and balances and constraints on political leaders is on average higher in democratic systems. This results in democratic systems expropriating foreign investments less frequently (Li, 2009). Furthermore, in a similar vein as H1, horizontal checks are less common in autocracies and result in corruption and misallocation of resources and will also effect investment in those political systems (Kurer, 1993; Sandholtz & Koetzle, 2000; Wu, 2012). Therefore, I hypothesize as follows.
H2: There is a positive relation between the level of democracy in the target country and their SWF acquisition performance.
It is expected that host country political stability leads to policy continuation for investments (Feng, 2003; Wu, 2012). SWFs are less surprised by significant changes and stability is more hospitable to investments. Several studies have found that political stable countries are associated with higher economic growth than political instable environments (Aisen & Veiga, 2013; Alberto Alesina, Özler, Roubini, & Swagel, 1996). Near elections, democracies are for example more prone to immediate consumption and changing demands. In general, authoritarian rules better resist demands and are in office for much longer time than elected governments (Feng, 2003; Wu, 2012). Firms also prefer to do business in politically stable countries, as FDI inflows are higher for political stable countries (Asiedu, 2006). Political instability may lead to new legislation that directly causes a threat a company’s goals and their assets. A new political system can also lead to renewed government policies that affect businesses, including fiscal and monetary reforms, expropriation or trade barriers. Business managers may even be uncertain about the government’s commitment to maintain existing statutes, making it difficult to take long-term decisions (Frynas, 1998; Miller, 1992). If companies cannot make optimal decisions in political unstable environments because of
24
uncertainty, this could have an even greater impact upon SWFs when politicians are involved. This is generalized to the following hypothesis with political stability defined as the number of years since the last big transition (Feng, 2003; Wu, 2012).
H3: There is a positive relation between political stability of the target country and SWF acquisition performance.
The hypotheses are expected to be moderated by the actual involvement of politicians in SWFs. When politicians are present in a fund, the characteristics of the political system will also be more prevalent. The presence of democratic standards, or lack therefore, are better transferred to the SWF if politicians themselves are present in the fund. It is thus expected that H1 will be more significant if the SWF is politicized. Furthermore, host countries occasionally question investments because of unclear motives and the pursuit of non-market strategies (Knill et al., 2010). It is for example more difficult for a SWF originating from an autocratic political system to convince a democratic host country, with more checks and balances in place, of their commercial motives (Monk, 2009). Other investors will fear tunnelling and expropriation by SWFs with politicians involved from opaque political systems because their behaviour is unknown (Dewenter et al., 2010; Fotak et al., 2008). This influences investors’ expectations about the performance of the target firm. Likewise, Murtinu & Scalera (2015) and Bernstein et al. (2013) find that higher politicization of funds leads to higher drops of firm value. The presence of politicians makes interaction between the SFW and political objectives more accessible and increases opportunities for preferential treatments, while possibly reducing the overall investment skills relative to professionals (Bernstein et al., 2013). In sum, I hypothesize as follows.
25
H4: The relations in H1 and H2 are more significant when the acquiring SWF is politicized than when no politicians are present inside the SWF.
Lastly, SWFs can use intermediate investment vehicles, such as a financial, cooperate or SWF majority-owned subsidiary intermediate. SWF investments are often faced with hostility because they are possibly influenced by politicians. This leads to a higher risk perception in the host-country as the true investment motives are opaque. To minimize the risk perception in the target country, SWFs want to indicate a passive stance towards to host countries and thereby reduce hostility. The use of an investment vehicle adds an organizational layer between the SWF (with associated politicians) and the target firm. Thereby, direct influence on the target firm is also limited (Murtinu & Scalera, 2016). The use of an investment vehicle enables SWFs to originate their investment from more democratic home countries, which lead to a lower significance of H1. Lastly, SWFs can originate their investments from countries with better relations that align with the norms of the target country and thereby decrease their influence. In sum, the additional layer between politicians and SWFs will influence will affect the previous hypotheses containing characteristics of the SWF home country in the following way.
H5: The relations in H1 and H2 are less significant when the acquiring SWF uses an investment vehicle compared to the SWF acquiring a firm directly.
26
27
4 Methodology
The hypothesis will be answered using a quantitative research design. A deductive approach is taken to test the theories using data. This enables the analysis of a great number of cases based on standardized data. This study uses secondary data from online data sources and previous studies (Saunders, Lewis, & Thornhill, 2016).
First, the data and sample are described followed by the dependent, independent, moderating and control variables. The chapter ends with analytical models and descriptive statistics.
Data collection
The main source of this research is the cross-section dataset of SWF transactions compiled by Murtinu & Scalera (2016). Their study used deal data obtained by Bureau van Dijk’s Zephyr. This data, from 1998 to 2013, includes the acquirer name, target name, target industry, country code and region, deal type, stake, deal status, deal value, use of an investment vehicle, acquisition completion date, the involvement of politicians in SWFs and transaction performance. Bureau van Dijk’s ORBIS is used to obtain target firm performance in the year following and preceding a transaction (Bureau van Dijk, 2018). Furthermore, they complemented the dataset with the total assets of each SWF in 2007 (Truman, 2009).
The data from Zephyr, Truman (2009) and ORBIS was merged using unique ID’s and checked manually to ensure validity. The deal data was then combined with political system characteristics of the Polity IV database (Marshall, Gurr, & Jaggers, 2016). Multiple researchers in the Polity project conducted a periodic coding review to ensure reliability and consistency within the dataset. They have collected data on the variables using a checklist of country attributes. This database contains variables about the level of institutionalized democracy and institutionalized autocracy. Institutionalized democracy is composed of data involving the
28
presence of mechanisms that allow citizens to effectively exercise preferences for other policies and leaders and the existence of constraints on the power by executives. Institutionalized autocracy is composed of data involving the restriction of political participation, selection of chief executives in the political elite and the exercise of power with limited institutional constraints. These variables are operationalized using the combined Polity Score, that places countries on a scale from autocracies (score -10) to democracies (score 10) (Marshall et al., 2016). Other authors use Polity IV database for similar purposes (Bortolotti et al., 2015; Sauvant, Sachs, & Jongbloed, 2012).
Lastly, country specific variables are gathered from the World Bank to control for local economic development of target countries. The variables FDI inflows, GDP growth and GDP per capita have been exported from the World Bank for all countries between 1998 and 2013. The deal characteristics are merged with World Bank data using unique ID’s, that include the target country code and year of the investment. Some investments contained country codes for which no information was available. Therefore, the data associated with China was used in transactions involving Hong Kong. Data associated with the United Kingdom was used for the Cayman Islands and Bermuda, as they are both British Overseas Territories. Lastly, data associated with the United States was used for the United States Virgin Islands.
Sample
The dataset of Murtinu & Scalera (2016) contains 1172 transactions over the years 1998 to 2013. After filtering for transactions involving a minority stake or acquisition, the availability of performance and control data and removing outliers (see section 5), the sample has been limited to 189 cases of 24 different SWFs. The SWF’s home country, SWF name and their number of investments in the sample are specified in Table 1.
29
Table 1: Number of investments by SWFs sorted by country
SWF country SWF parent acquirer No. of investments
Canada Caisse de dépôt et placement du Québec
28 Canada Canada Pension Plan Investment
Board
4
China China Investment Corporation 7
China China National Social Security
Fund 2
Ireland National Pensions Reserve Fund, The
2
Korea Korea Investment Corporation 2
Kuwait Kuwait Investment Authority 3
Libya Lybian Investment Authority 1
Malaysia Khazanah Nasional Bhd 10
Netherlands Stichting Pensioenfonds ABP 22 New Zealand Guardians of New Zealand
Superannuation Fund
1
Norway Government Pension Fund -
Global
4
Qatar Qatar Investment Authority 5
Saudi Arabia Saudi Arabian Monetary Agency 1
Singapore Goverment of Singapore
Investment Corporation Pte Ltd, The
41
Singapore Temasek Holdings Pte Ltd 35
United Arab Emirates Abu Dhabi Investment Authority 1 United Arab Emirates DIFC Investments LLC 1 United Arab Emirates Dubai International Capital LLC 5 United Arab Emirates International Petroleum
Investment Company
6 United Arab Emirates Investment Corporation of Dubai 2
United Arab Emirates Istithmar PJSC 3
United Arab Emirates Mubadala Development
Company Pjsc 2
United States California Public Employees Retirement System
1
Total 189
The investments in the sample are mostly targeted at firms in the Asia – Pacific region (39%), followed by Europe (30%), North America (25%), South Asia (10%) and MENA (1%).
30
The regions Latin America and Sub-Saharan Africa both received one SWF investment in the sample. The number of SWFs investments by target country is specified in Table 2.
Table 2: Number of SWFs investments in target countries
Target country Investments Target country Investments
Australia 6 New Zealand 1
Austria 1 Norway 1
Belgium 1 Papua New Guinea 1
Brazil 1 Philippines 1
Canada 31 Portugal 2
China 23 Russia 1
France 9 Singapore 12
Germany 3 South Africa 1
Greece 2 Spain 3
India 9 Sri Lanka 1
Indonesia 4 Switzerland 1
Ireland 3 Thailand 2
Italy 9 Turkey 1
Japan 2 United Kingdom 17
Korea 7 United States 10
Malaysia 13 Total 189
Netherlands 10
Variables
The operationalization of the hypotheses dependent, independent, moderator and control variables will be discussed. All variables are summarized in Table 3: Definition of variables. 4.3.1 Dependent variables
The Dependent Variable (DV) for this research is acquisition performance. It has been defined in a similar manner as Bortolotti, Fotak, & Megginson, 2015. The Return on Assets (ROA) at the end of the year preceding the investment is subtracted from the ROA at the end of the year after the investment to determine the profitability. The ROA is used because it gives an indication of the profit from the deployed assets, regardless of the SWF size. A time span of
31
multiple years is used because SWFs generally have long investment horizons (Bortolotti et al., 2015).
4.3.2 Independent variables
The three independent variables are democracy home country, democracy target country and political stability. Institutionalized democratic and autocratic scores per country are set-up by the Polity IV project using characteristics regarding competitiveness of executive recruitment, openness of executive recruitment, constraints on chief executives, regulation of political participation and competitiveness of political participation (Marshall et al., 2016). These variables are weighted for each country by the Polity IV project using a checklist for country attributes. For the democracy home and target country variables, the unified Polity score is used which subtracts the institutionalized autocratic score from the institutionalized democratic score. Hereby, it is possible to relatively compare all countries on democratic and autocratic elements present in the government. Lastly, the political stability variable indicates the number of years since the last big regime change in the target country and is retrieved from the Polity IV database. They flag a regime change as “a three-point change in the Polity score over a period of three years or less” (Marshall et al., 2016, p. 17) or the end of an unstable period without political institutions. This variable is used as a proxy for policy continuation, as government interruption leads to economic uncertainty in the short-term which affects investment decisions (Feng, 2003). Other authors use the Polity IV database for similar purposes (Bortolotti et al., 2015; Sauvant et al., 2012).
4.3.3 Moderator variables
The moderating factors in this study are the political involvement in SWFs and the use of investment vehicles by SWFs (Bernstein et al., 2013; Bortolotti et al., 2015; Murtinu & Scalera, 2016). Political involvement is a dummy variable that equals one if politicians are present in
32
the SWF management. The database of Murtinu & Scalera (2016) is used, who gathered the data from a report provided by J.P. Morgan (Fernandez & Eschweiler, 2008). Fund politicization is used as an indicator of whether the norms of the government are more directly present in the SWF.
The second variable originates from the same database and equals one if the SWF invested using an intermediate investment vehicle. The three considered types are financial vehicles like banks and financial branches, non-financial corporate vehicles that are not owned by the SWF and majority-owned SWF subsidiaries (Murtinu & Scalera, 2016). This data has been collected from Bureau van Dijk’s (2018) Zephyr database. The use of an investment vehicle indicates a passive stance towards the target country, or the use of an intermediate with more equal standards to the target country (Murtinu & Scalera, 2016).
4.3.4 Control variables
Five control variables are identified to control for factors influencing investment performance. First, SWFs are accused of investing in domestic underperforming industries due to political influences (Bernstein et al., 2013). SWFs also face negative reactions in recipient countries because of unclear motives (Cohen, 2009; Knill et al., 2010). The dummy variable cross border, that equals one for investments abroad, controls for these theories to check whether there is a performance difference.
Second, a dummy is added that equals one if the investment is made in the years 2008-2013 to account for the economic recession that effects acquisition performance.
The three other control variables are host country-specific and collected from the World bank. Corresponding with the approach of Bortolotti et al. (2015), this research control for GDP per capita and GDP growth to account for local economic circumstances. The GDP per capita
33
is defined as the target country gross domestic product divided by the midyear population in constant 2010 U.S. dollars. The GDP growth is the annual percentage growth of GDP at market prices on local constant currency (The World Bank, 2015). This research adds FDI inflows as a control variable for the investment climate, which is defined as the foreign direct investment net inflows to acquire a management interest divided by the GDP (Sekkat & Veganzones-Varoudakis, 2007; The World Bank, 2015). Table 3 provides a description of all variables included in this study and their respective source.
Table 3: Definition of variables
Variable Definition Source
Dependent variables
Investment performance The return on assets at the end of the year following the investment, subtracted by the ROA at the end of the year preceding the investment
Murtinu & Scalera (2016), Bureau van Dijk (2018)
Independent variables
Democracy home country Scale from +10 (strongly democratic) to -10 (strongly autocratic) in the SWF home country
Project Polity IV (Marshall et al., 2016)
Democracy target country Scale from +10 (strongly democratic) to -10 (strongly autocratic) in the target country
Project Polity IV (Marshall et al., 2016)
Political stability Number of years since the last big political system transition in the target country.
Project Polity IV (Marshall et al., 2016)
Moderating variables
Politicization SWF Dummy that equals 1 if politicians are present in the SWF management
Murtinu & Scalera (2016), Fernandez & Eschweiler, 2008
Investment vehicle Dummy that equals 1 if the SWF invests using an intermediate investment vehicle
Murtinu & Scalera (2016), Bureau van Dijk (2018)
34
Cross border investment Dummy that equals 1 if the
investment is made abroad. Murtinu & Scalera (2016)
Crisis Dummy that equals 1 if the
investment is made in the years 2008-2013
Own dataset
FDI inflow Net inflows of investment to acquire a lasting management interesting from foreign investors divided by the GDP
The World Bank (2015)
GDP growth Annual percentage growth
rate of GDP at market prices based on constant local currency
The World Bank (2015)
GDP per capita Gross domestic product divided by midyear population in constant 2010 U.S. dollars
The World Bank (2015)
4.3.5 Analytical models
The sample will be analysed with a multiple Ordinary Least Squares (OLS) regression analysis in SPSS for the dependent variable (Saunders et al., 2016). Several other related studies are also conducted with a regression analysis (Bortolotti et al., 2009; Chhaochharia & Laeven, 2008; Fernandes, 2014). Before doing this, the data is checked on correlation and multicollinearity (Field, 2013). As the dependent variable is not normally distributed due to extreme kurtosis, it is transformed first using the following formula to ensure a normal distribution.
(X-Sample mean)*((X-Sample mean)2)a, with a as a constant.
The following model will be used to test the hypothesis.
(Y-Sample mean)*((Y-Sample mean) )2)a = !0 + !1 X1 + !1*!2 X2 + ε1, where a is -0,22.
The regressions will be performed using a stepwise hierarchical linear regression (Field, 2013). The control variables and moderators are added first, followed by the
35
independent variables and the interaction between the independent variables and the moderator.
36
5 Results
This section starts with the descriptive statistics of the sample (5.1), followed by the correlation analysis (5.2) and a description of the main results including implications for the hypotheses (5.3).
Model assumptions and descriptive statistics
Before starting with the descriptive statistics, the data was checked on the necessary assumptions to apply a linear regression model. The skewness and Kurtosis should be close to zero, with values between -2 and +2 considered acceptable for a normal distribution (Field, 2013). The dependent variable shows a high kurtosis of 5,26. The control variable target FDI Inflow also shows a high skewness of 10,06 and a Kurtosis of 118,49. Although the regression analysis does not assume a normal distribution for all variables, it does include the assumption of independent errors, normally distributed errors, homoscedasticity and a linear relationship between the IVs and the outcome variable (Field, 2013). With the current data, the residuals are heteroscedastic, meaning that the variances are unequal at each level of the IV. Furthermore, initial scatterplots indicate non-normal distributed residuals. Therefore, an outlier analysis was conducted because regression models are heavily biased by unusual cases (Field, 2013).
Outliers outside three standard deviations were checked for the ROA and the FDI Inflow. To check the outliers, the data was supplemented with two variables containing the ROA for a period of four and six years. It revealed that the outliers in the sample were generally cases that lack ROA data for a longer period than the two years in the analysis. These cases with limited ROA data thus lack reliable information outside the analysis period, but do have a high impact on the outcome as an outlier. Hence, these cases have been excluded from the analysis. Furthermore, three cases with abnormal z-scores outside three standard deviations in at least two of the three years after an acquisition have been excluded. Lastly, one case with only two
37
years of performance data and a FDI inflow z-score of 12,18 has been excluded. This highly influential case with ID 621 is targeted at the host country Luxembourg, with a FDI Inflow of 44,8% of the GDP (The World Bank, 2015). All excluded cases are listed in Table 4.
Table 4: Excluded cases based on outlier analysis
ID Acquiror name Target name year Z-Score
ROA for two years Z-Score ROA for four years Z-score ROA for six years 435. Temasek Holdings Pte
Ltd Amyris Inc.
2012 -3,35
500. Benchmark Capital Marin Software Inc.
2012 5,08 611. Stichting
Pensioenfonds ABP Valad Group Property
2008 -4,09 -0,44 621. Canada Pension Plan
Investment Board
Invista European Real Estate Trust SICAF
2012 -2,75
630. Caisse de dépôt et
placement du Québec Kangaroo Capital Inc.
2005 3,53
652. Caisse de dépôt et
placement du Québec Atna Ltd Resources 2005 3,13 -0,2 6,17 671. Caisse de dépôt et placement du Québec Chariot Resources Ltd 2005 6,72 1,27 6,30 824. Government of Singapore Investment Corporation Pte Ltd, The Brixton plc 2007 -3,18 1201. China Investment
Corporation SouthGobi Energy Resources Ltd
2009 5,62 12,16 8,91
The exclusions did not lead to a sufficient solution for the violated assumptions. Therefore, the dependent variable was transformed using the transformation in section 4.3.5. The constant a was determined by executing a Kolmogorov-Smirnov tests on the range of possible values. Then the least significant value was used as a to ensure a normal distribution.
38
When using the transformed variable, the residuals are normally distributed and are homoscedastic in P-P plots. Homoscedasticity is determined by plotting the standardized residuals against standardized predicted values and checking whether the points are randomly distributed (Field, 2013).
The descriptive statistics of the dependent, independent, moderator and control variables of the cases are shown in Table 5. The sample shows that the performance increases on average with the years relative to the preceding years. Furthermore, the dummies show that 20% of the funds in the sample are politicized, 16% of transaction were made using an investment vehicle, 68% of the investments is made abroad and 44% of the investment were made during the years 2008-2013. The percentage of politicized SWFs in the sample is relatively low compared to Murtinu & Scalera (2016), this is partly because performance data is not available for some politicized funds. When not filtering on the availability of performance and control data within the dataset, the percentage of politicized funds in the adapted sample would increase to 32%.
Table 5: Descriptive statistics
Variable N Min Max Mean S.D.
Investment performance 189 -29,12 44,82 -0,82 8,70
Investment performance (transformed)
189 -6,59 8,43 -0,19 2,68
Democracy home country 189 -10 10 1,14 6,96
Democracy target country 189 -7 10 6,23 6,03
Political stability 189 0 203 74,08 49
Politicization SWF 189 0 1 0,20 0,40
Investment vehicle 189 0 1 0,16 0,37
Cross border investment 189 0 1 0,68 0,47
Crisis 189 0 1 0,44 ,050
FDI Inflow 189 -2,74 87,44 5,94 9,28
GDP growth 189 -3,77 14,23 4,14 3,67
39
Correlation analysis
Table 6 presents the correlation matrix between the variables. Spearman’s correlation coefficient is used as a non-parametric statistic based on ranked data, because it does not assume normality like Pearson’s correlation coefficient (Field, 2013). The IV democracy home country shows a significant positive relation to the investment performance (r = 0,173, p = 0,05). The IV political stability (r = -0,144, p = 0,05) shows a negative relation to the performance. The dummy variable cross border investment (r = -0,14, p = 0,1) and crisis have a negative relation to investment performance (r = -0,248, p = 0,01). The IV democracy home country shows a significant positive relation to democracy in the target country (r = 0,288, p = 0,01) and a positive relation to political stability (r = 0,237, p = 0,01). Furthermore, this IV is negatively related to cross border investments (r = -0,513, p = 0,01), GDP growth (r = -0,14, p = 0,1) and positively to GDP per capita (r = 0,287, p = 0,01). Democracy in the target country is positively related to political stability (r = 0,567, p = 0,01) and GDP per capita (r = 0,701, p = 0,01). This IV is significantly negatively related with GDP growth (r = -0,722, p = 0,01). Political stability is negatively related to GDP growth (r = -0,297, p = 0,01) and positively related to the controls FDI inflow (r = 0,189, p = 0,01), and GDP per capita (r = 0,572, p = 0,01). The moderator politicization of a SWF is negatively related to democracy in the home country (r = -0,503, p = 0,01) and political stability (r = -0,236, p = 0,01). This moderator is also positively related to the use of an investment vehicle (r = 0,179, p = 0,05). The use of an investment vehicle is also negatively related to political stability (r = -0,203, p = 0,01) and GDP per capita (r = -0,161, p = 0,05). Lastly, several control variables show significant correlations. First, cross border investments are negatively related to FDI inflow (r = -0,249, p = 0,01) and GDP per capita (r = -0,324, p = 0,01). Second, the crisis variable is negatively related to FDI inflow (r = -0,228, p
40
= 0,01). Third, FDI inflow is positively related GDP per capita (r = 0,253, p = 0,01). Lastly, GDP growth is negatively related to GDP per capita (r = -0,621, p = 0,01).
Multicollinearity exist if there is a strong correlation between multiple predictors, which makes it difficult to untangle estimates of the regression coefficients. The Variance Inflation Factor (VIF) will be used to indicate the presence of strong linear relationships between predictors for each regression. Values below the 10 and an average VIF of 1 indicate no presence of multicollinearity (Field, 2013). All VIF values are specified in appendix I.
41
Table 6: Spearman's correlation matrix
Variables 1 2 3 4 5 6 7 8 9 10 11
1) Investment performance (transformed)
2) Democracy home country ,173*
3) Democracy target country 0,01 ,288**
4) Political stability -,144* ,237** ,567**
5) Politicization SWF -0,04 -,503** -0,04 -,236**
6) Investment vehicle 0,00 -0,08 -0,08 -,203** ,179*
7) Cross border investment -0,14+ -,513** -0,08 -0,11 0,06 0,05
8) Crisis -,248** -0,08 -0,09 -0,02 0,06 0,08 0,04
9) FDI Inflow -0,02 0,08 -0,09 ,189** -0,05 -0,12 -,249** -0,10
10) GDP growth 0,01 -0,14+ -,722** -,297** -0,03 0,04 0,06 -,228** 0,14
11) GDP per capita -0,03 ,287** ,701** ,572** -0,11 -,161* -,324** 0,09 ,253** -,621**
+. Correlation is significant at the 0.1 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
42
Regression analysis
A hierarchical multiple regression was performed to examine how democracy in the SWF home country, democracy in the target country and political stability predict the ROA of investments, after controlling for international investments, investments in the crisis period, target country FDI inflow, GDP growth and GDP per capita (Table 7). Furthermore, the interaction effect of fund politicization with both democracy in the home country and target country was tested. Lastly, the interaction effect of using an intermediate investment vehicle with democracy in the home country and target country was also tested.
In the first step of the hierarchical regression, the variables cross border investment, crisis, FDI inflow, GDP growth and GDP per capita, politicization SWF and investment vehicle were entered as predictors. Respectively, the independent variables democracy home country, democracy target country and political stability were each added in the second, third and fourth step. The fifth step contains all IVs, excluding the interaction terms with the moderators. The sixth step adds the interaction between politicization and democracy of the home and target country to the model. In the seventh and last step, model five is extended with the interaction between investment vehicle and both the democracy in the home and target country.
The first model with control and moderator variables is statistically significant (F = 2,53; p < 0,05) and explained 8,9% of the variance in the ROA. The control variables cross border investment (β = -0,78; p < .10), crisis (β = -1,16; p < .01) and FDI Inflow (β = 0,04; p < .10) are statistically significant, while the variables GDP growth (β = -0,10; p > .10), GDP per capita (β = 0,00; p > .10), politicization SWF (β = -0,32; p > .10) and investment vehicle (β = -0,34; p < .10) are not significant at the 0,10 level. From this result, the difference between investments during the crisis and in other periods outside the crisis, holding other predictor variables constant, is -1,47 for the original untransformed ROA value. The difference between domestic
