Investigating political-institutional characteristics and Chinese foreign direct investment in Europe

(1)

INVESTIGATING POLITICAL-INSTITUTIONAL

CHARACTERISTICS AND CHINESE FOREIGN DIRECT

INVESTMENT IN EUROPE

A THESIS ON THE EXPLANATORY FACTORS OF CHINESE

FDI

Presented to the faculty of Governance and Global Affairs of

Leiden University

in partial fulfilment of

the requirements for the degree

MASTER OF SCIENCE PUBLIC ADMNISTRATION

by

B. KEUZENKAMP BSc

Stud. Nr. 2093308 Track

Economics & Governance Supervisor

N. van der Zwan 2nd reader W. Bolhuis Den Haag, Netherlands

(2)

1

I. List of figures and tables

Figure 1. Outward FDI flows of China, 2002 - 2016 ... 4

Table 1. Descriptive statistics of the outcome and main determinant variables ... 21

Table 2. Descriptive statistics of the control variables ... 25

Table 3. Summary of the used variables and their operationalisation. ... 28

Table 4. OLS-regression results ... 33

Appendix A.. Pairwise correlations ... 46

Appendix B.. RvF-plot: Linear relation between independent and outcome variables ... 47

Appendix C. Descriptive statistics of model 4 (Low credit rating). ... 48

(4)

3

II. List of abbreviations

CCP = Communist Party of China

FDI = Foreign Direct Investment

IEF = Index of Economic Freedom

IMF = International Monetary Fund

IPRI = International Property Rights Index

OECD = Organisation for Economic Co-operation and Development

OLS = Ordinary-Least-Squares

MNE = Multinational Enterprise

MOFCOM = Ministry of Commerce (of the People’s Republic of China)

(5)

4

1. Introduction

Policymakers around the world try to make their countries’ investment climate as attractive as possible for foreign multinational enterprises (MNEs) to invest in. They do this because foreign direct investments (FDI) can generate all kinds of positive externalities such as increased employment, diffusion of technology, and productivity improvements (Javorcik, 2002). As China is currently emerging as a major source of FDI (see Figure 1), and Chinese MNEs are increasingly investing in Europe (Hanemann & Huotari, 2018), it is in the interest of both policymakers and society as a whole to understand the driving factors of Chinese FDI in Europe.

The classic theory on the determinants of FDI is that macro-economic factors are the main determinant factors that MNEs take into consideration when they have to decide in which location they will invest (Caves, 1974; Cheng & Kwan, 2000; Dunning, 1980; Grosse & Trevino, 1996; North, 1981; North, 1990). However, more recent studies find that, aside from macro-economic factors, political-institutional factors also affect the location decision of FDI. (Baek & Qian, 2011; Barassi & Zhou, 2012; Holmes, Miller, Hitt, & Salmador, 2013; Pajunen, 2008; Pao & Tsai, 2011; Siegel, Licht, & Schwartz, 2013). However, Chinese FDI does not seem to behave in line with the conventional theories that are presented by these authors. As Shan, Lin, Li, and Zeng (2018) describe, Chinese MNEs do not perceive weak institutions as barriers to investment and are comfortable with taking higher risks than their Western counterparts. Therefore, new theories are necessary to help understand how FDI made by Chinese MNEs differs from FDI made by Western MNEs. Furthermore, as most studies have been conducted on FDI originating in Western countries into developing countries, new theories are also necessary to understand how FDI from developing countries into developed countries behaves. This study helps to close these gaps in the literature by providing insight in how macro-economic and political-institutional factors influence the location decision of FDI by Chinese MNEs in Europe.

Preliminary research on Chinese FDI has put forward multiple theories that help understanding how Chinese FDI differs from traditional FDI, and what the possible causes of these differences are. One of the theories, following from research by Buckley et al. (2007), Kolstad and Wiig (2012), and Shan et al. (2018), is that Chinese FDI is attracted to, instead of repelled by, poor political-institutional qualities of a potential investment location. This theory is quite

(6)

5 fascinating as it could mean that Chinese FDI behaves opposite Western FDI and current conventional trade theories.

Figure 1. Outward FDI flows of China, 2002 - 2016. Adapted from "Statistical Bulletin of

China's Outward Foreign Direct Investment 2017" by Ministry of Commerce of People's Republic of China (2018).

There are three distinct differences between Chinese MNEs and traditional Western MNEs that can explain why Chinese MNEs make a different evaluation of potential investment locations than their Western counterparts. The first difference is that Chinese MNEs are being steered by the Communist Party of China (CCP) and are thus serving other (geopolitical) goals than maximalisation of return on investment (Kolstad & Wiig, 2012; Morck, Yeung, & Zhao, 2008). The second difference is that Chinese MNEs face less backlash in their home-country if they violate ethical standards abroad, allowing them to do business in locations where Western MNEs can not or will not (Kolstad & Wiig, 2012). The third and final difference is that Chinese MNEs have more experience than Western MNEs in navigating their business in countries with poor political-institutional qualities, as China itself is a country with poor political-institutional quality (Morck et al., 2008; Yeung & Liu, 2008). The theory chapter of this study elaborates more on the abovementioned differences between Chinese and Western MNEs.

Many (South-Eastern) European countries are searching for alternatives to EU investments, as EU investments usually come with (the perception of) a too large political package of requirements for loans which these countries can not or do not want to fulfil (Okano-Heijmans, 2019). Therefore, the question is posed what these countries can do to attract alternative, Chinese, investments. Contributing to an answer to that question, this study tests whether the

2,7 2,85 5,512,26 21,1626,51 55,91 56,53 68,8174,65 87,8 107,84 123,12 145,67 196,15 0 50 100 150 200 250 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Outward FDI flows of China, 2002 - 2016

(7)

6 theory that Chinese FDI is attracted to locations with poor political-institutional quality, holds for Europe. Ensuing, the main research question of this study is:

Are political-institutional factors of European countries correlating with the level of foreign direct investment that they receive from Chinese multinational enterprises?

To research if the theory holds, an ordinary-least-squares (OLS) regression analysis on country level was conducted on the relation between six political-institutional determinants of FDI and Chinese FDI in Europe. The six studied determinants are: corruption, protection of property rights, environmental protection policies, corporate tax rate, political stability, and democracy. These six are chosen as together they provide a good representation of the overall political-institutional quality of a country and are proven to influence traditional FDI flows (Tocar, 2018). Each of the six determinants has their own sub-question: how does [determinant 1-6]

effect Chinese FDI in Europe? Aggregating the answers to these sub-questions will enable the

main research question to be answered.

The organisation of the remainder of this study is as follows; the theory chapter presents the used concepts, their relations, and related hypotheses, and embeds them in the available academic literature in the field; the subsequent methodology chapter presents the research design, case selection, the operationalisation of the used concepts, and the considerations made in the selection of the data; the following results chapter presents the results of the OLS-regressions; next is the analysis chapter in which the results are interpreted and the hypotheses are tested; the final chapter is the conclusion of the study in which the main research question is answered, the most important results are presented, possible limitations of the research discussed, and various suggestions for further research are provided.

(8)

7

2. Theory

2.1 Characteristics of foreign direct investment

2.1.1 Definition

Foreign direct investment is defined by the International Monetary Fund (IMF) and the Organisation for Economic Co-operation and Development (OECD) as “international investment by an entity resident in one economy in an enterprise resident in another economy that is made with the objective of obtaining a lasting interest” (IMF & OECD, 2003, p. 23). This implies that there should be some form of a long-term relationship between investor and investee, in which the investor has a considerable amount of control over the invested enterprise (IMF & OECD, 2003). The OECD argues that an investor from one country must acquire at least a 10% voting power share of an enterprise in another country for the transaction to be regarded as FDI (OECD, 2008). Apart from the initial financial transaction, FDI usually constitutes a broader co-operation between investor and investee. This co-operation can generate positive outcomes for both investor and investee. These outcomes include the transfer of knowledge through training and skill acquisition, the transfer of technologies and organisational expertise, the introduction of new production procedures, the investment in backward and forward linkages, access to foreign markets, and access to cheap labour and natural resources (Lamsiraroj & Ulubasoğlu, 2015).

2.1.2 Necessary conditions

Dunning (1980) argues that there are three conditions that must be met for FDI to take place. First, the investing firm must have an ownership advantage. Ownership advantages are assets that give the investing firm a competitive edge over the host country’s domestic firms, which is necessary to outweigh the transaction costs related to operating in a foreign country. Second, the investing firm must have an internalisation advantage. When an MNE benefits more from establishing production internally overseas than it would from licensing or franchising its production to foreign firms, the MNE has an internalisation advantage. Finally, the host country market must present the investing firm with a locational advantage. The locational advantage includes all relevant factors that make a location attractive for foreign firms to invest in. These advantages include but are not limited to factor prices, trade policies, capital flows, political-institutional factors, and access to markets.

(9)

8

2.1.3 Different types of foreign direct investment

Dunning’s eclectic paradigm on FDI discerns three main motives for MNEs to expand their business activities abroad (Dunning, 1993). Some MNEs are seeking to take advantage of labour cost and factor price differences between countries around the world. This is profitable for MNEs as they increase the efficiency of their production process by fragmenting it vertically across countries (Riedl, 2010). This type of vertical integration of MNEs is called

efficiency-seeking FDI or manufacturing FDI (Dunning, 1993; Yeaple, 2003). Other MNEs

seek to gain access to foreign markets and avoid international trade costs. MNEs that deliver services or products that are non-tradeable, usually have a need to be located near their consumers. By expanding their business activities abroad, they can gain access to new markets (Riedl, 2010). Furthermore, when foreign markets are blocked by high import tariffs or other trade and/or transport costs, MNEs can gain access by transferring their production process abroad, circumventing these costs (Chakrabarti, 2001). Most MNEs active in this form of FDI keep their headquarters in their home country and establish identical production plants in each new host country, each serving its own market (Chakrabarti, 2001). This type of horizontal integration is called market-seeking FDI (Dunning, 1993; Yeaple, 2003). Finally, some MNEs seek to gain access to natural resources by expanding their business abroad. This type of FDI is called resource-seeking FDI (Dunning, 1993).

All the above-mentioned types of FDI can take on different forms. These forms are; mergers and acquisitions, the construction of new production plants abroad, the reinvestment abroad of profits made by host country firms, and intra-company loans. The establishment of a new firm abroad is regarded as a greenfield investment. This title refers to an investment that creates a new production facility in the host country. The acquisition of an already existing foreign firm is regarded as a brownfield investment (Çolak & Alakbarov, 2017).

2.2 Benefits and drawbacks of FDI

Foreign direct investment (FDI) is regarded as one of the engines of economic growth. Many studies have found that FDI positively contributes to economic growth (e.g., Almfraji & Almsafir, 2014). According to Jensen (2003), FDI leads to economic growth through three avenues; employment, technological diffusion, and productivity improvements. Each of these will be discussed separately below. Hoping to gain these advantages, governments across the globe vehemently try to attract FDI (Javorcik, 2002). Moreover, for many countries, policies

(10)

9 and incentives to attract FDI have become an elemental part of their economic development strategies (Jensen, 2003).

2.2.1 Employment

The relation between flows of FDI and employment rates is unclear. The studies that have been conducted on the topic obtained varying results. Many find no aggregate effect of FDI on employment rates, such as Braunstein and Epstein (2002) and Brincikova and Darmo (2004). Other studies find a positive aggregate effect of FDI on employment rates, such as Vacaflores (2001) and Geishecker and Hunya (2005). Contrarily, Liu (2012) even finds a short-term negative effect of FDI on employment, especially in the primary sector. Thus, it can be concluded that FDI can have either positive or negative effects in terms of employment. Most studies agree that the sign of the effect of FDI on employment depends on what form of FDI is concerned. It is argued that a distinction must be made between greenfield and brownfield investments when assessing the impact of FDI on employment. Greenfield investments create new production capacity and new jobs and thus have a positive effect on employment rates. On the contrary, brownfield investments can have negative effects on employment, as more efficient management systems and equipment, and technological innovations, may reduce the number of workers needed for the host country’s firm’s production process. Additionally, both forms of FDI may reduce employment through forces of crowding-out, as entering the market increases the competition, which forces domestic firms to improve their competitiveness by reducing employment (Brincikova & Darmo, 2014).

2.2.2 Technological diffusion

According to Borensztein, De Gregorio, and Lee (1998), the mechanism through which FDI causes economic growth is one of technological diffusion. Borensztein et al. (1998) build on Romer’s (1994) theory of endogenous growth. In this theory it is argued that economic growth is the result of investments in education, innovation and knowledge. Borensztein et al. (1998) argue that the growth rate of a developing country can partially be explained by examining their technologic level relative to that of the rest of the world. Implementing modern technologies can lead to a greater economic efficiency, which in turn leads to economic growth. Therefore, a developing country which has a relatively low technological standard can achieve economic growth by implementing modern technologies which have already been developed in technologically advanced countries. Thus, successful development of a developing countries’ economy requires a system in which the ideas and knowledge from the advanced

(11)

10 countries are transferred to the developing country (Borensztein et al, 1998). However, as this study is focussed on Chinese (developing country) investments in Europe (developed countries), the transfer of knowledge might work the other way around. In this case, Chinese investors would be providing the funds and the European investees the necessary knowledge, simultaneously transferring the knowledge to the developing country.

Romer (1993) argues that there are different ways in which the government from a developing country can promote this diffusion of technology. For example, by trying to attract foreign individuals with specific (tacit) knowledge, or by promoting students to obtain advanced education abroad, or by simply importing technologically advanced products. However, for the most impoverished countries these methods would be too expensive.

Another avenue, which is available for all developing countries, is creating an environment in which foreign MNEs are incentivised to share their knowledge with their domestic partners. Markusen (1998, p. 753) argues that “MNEs are closely associated with knowledge-based assets rather than physical capital. Knowledge capital includes but is not limited to the human capital of employees, patents, blueprints, formulae, managerial and work procedures, marketing knowledge, reputations, and trademarks”. By sharing this knowledge with a domestic partner in a developing country, MNEs can capitalise on the increased efficiency that is gained by using the MNEs’ technological knowledge. As FDI involves a significant component of technological transfers and spill-over, FDI is one of the main ways in which MNEs can capitalise on their technological edge over developing countries (Romer, 1993). Simultaneously, this technological diffusion lifts the technological standard of the developing country and thus increases its economic efficiency which in turn leads to greater economic growth (Barrel & Pain, 1997).

2.2.3 Productivity improvements

It is conventionally agreed upon by both scholars and policymakers that, on average, foreign-owned enterprises (FOEs) perform better than domestic firms (e.g., Bevan & Estrin, 2004; Caves, 1996; Doms & Jensen, 1998; Dunning, 2013; Girma, Thompson, and Wright, 2002). Bevan and Estrin (2004) and Borensztein (1998) argue that it can be assumed a priori that FOEs are more productive than domestic firms. They suggest that FOEs suffer additional transaction costs (i.e. language barriers, customs) compared to domestic firms. Thus, in order to be able to compete on a domestic host market, they would have to be more productive. This theory is supported by findings of Girma et al. (2002) who observe that FOEs are between eight and

(12)

11 fifteen percent more productive than their domestic counterparts. They argue that this is caused by MNEs bringing in certain assets, such as brand name capital and organisational efficiency, that lead to productivity advantages. Dunning (2013) argues that economies of scale and intra-group trading play an important role. MNEs can provide services or products at lower cost if they produce at higher volumes. Furthermore, intra-group trading allows FOEs to acquire products, goods, services, or assets at lower costs than they would on the open market, further increasing the productivity of capital of FOEs. Additionally, Girma et al. (2002) find that these improvements in productivity result in higher wages for employees of FOEs of four to five percent, compared to their domestic counterparts.

2.3 Political-institutional and macro-economic driving factors

As this study seeks to test whether there is a negative relation between political-institutional quality and Chinese FDI in Europe, the study focusses on the political-institutional determinant factors of Chinese FDI. However, as scholars have identified macro-economic factors such as labour costs, agglomeration factors, and market size as key driving factors of FDI in general (Chakrabarti, 2012; Tocar, 2018), macro-economic factors are also included in this study as control variables.

2.4 Institutional-political driving factors of FDI

It is generally accepted that political factors (corruption, political regime, stability, risk etc.) influence the location choice of FDI. Likewise, the institutional framework, which includes protection of investments, (intellectual) property rights, and enforcement of regulations, also affects location choice (Tocar, 2018). North (1991) argues that institutions provide the necessary framework for social, political, and economic interaction. This suggests that when the institutional framework is of better quality, transaction and production costs are lower, which in turn attracts FDI. This is confirmed by most of the academic literature on the subject which finds that countries with better institutional quality attract more FDI (Globerman & Shapiro, 2002). Contrarily, Shan et al. (2018), Kolstad & Wiig (2012), and Buckley et al. (2007), report opposite findings for Chinese FDI. It is of interest to find out whether this reported negative relation between Chinese FDI and political-institutional quality is also true in Europe, and if so, which specific political-institutional factors drive this counter-intuitive

relation. As there aren’t many studies on the topic,there is not yet much conclusive evidence

on which political-institutional factors influence FDI. Nonetheless, Tocar (2018) discerns six political-institutional factors that are highly likely to have a significant relation with FDI, which

(13)

12 are therefore also used in this study. These factors are: (a) corruption, (b) property rights, (c) environmental policies, (d) corporate tax rate, (e) political stability, and (f) democracy.

2.4.1 Corruption

Many studies (e.g., Barrasi & Zhou, 2012; Javorcik & Wei, 2009) find that the level of corruption is negatively related to FDI. Shleifer and Vishny (1993) argue that corruption can be viewed as an added tax on investment and/or as a factor that increases the insecurity about investment costs, which both deter FDI. However, when Chinese FDI is concerned, the conducted studies find either a positive relation with corruption (Cheung, De Haan, Qian, & Yu, 2012), or indifference to corruption (Gu, 2009). Multiple authors (e.g., Cheung et al., 2012; Kolstad & Wiig, 2012; Morck et al., 2008; Yeung & Liu, 2008) explain this counter-intuitive relation by arguing that MNEs who have experience with operating within a corrupt environment in their home country, are not deterred by corruption barriers, but instead prefer to invest in corrupt countries. Furthermore, as the institutional standard in China is much lower (i.e. 39/100 score on Transparency International’s 2018 Corruption Perception Index) than in the Western source countries of FDI, Chinese MNEs are less vulnerable to reputational and/or financial costs associated with engaging in ethically questionable behaviour abroad. Therefore, Chinese MNEs perceive risks differently than MNEs from developed countries which allows them to invest in places where MNEs from developed countries can not (Kolstad & Wiig, 2012). This leads to the first hypothesis tested in the empirical analysis:

Hypothesis 1. European countries that are more corrupt receive more Chinese FDI than European countries that are less corrupt.

2.4.2 Property rights

There is strong evidence of a positive relation between strong property rights and FDI (Tocar, 2018). Increased counterfeiting and piracy in developing countries such as China presents ‘Western’ MNEs with a significant threat to their overseas business interests. Thus, as firms want to reduce the risk of their products being imitated, the presence of strong property rights attracts investments from ‘Western’ MNEs (Du, Lu, & Tao, 2012). On the contrary, Chen, Li, and Shapiro (2015) explain that MNEs originating from a country with weak property rights are confronted with two types of difficulty when expanding their business into countries with strong property rights. First, the MNEs will have to get acquainted with the regulations and procedures for registering and maintaining their property rights such as copyrights, trademarks, and patents. Second, they can’t make use of counterfeit goods to make a profit. As China itself

(14)

13 has weak property rights protection, Chinese MNEs will have to overcome these difficulties if they want to do business in a country with strong property rights. Thus, strong property rights can act as an additional barrier to Chinese FDI. This leads to the second hypothesis tested in the empirical analysis:

Hypothesis 2. European countries that have strong property rights protection receive less Chinese FDI than European countries that have weak property rights protection.

2.4.3 Environmental policies

Most scholars agree that environmental policies have a significant negative effect on FDI (Tocar, 2018). The effect can be explained with the ´pollution haven theory´. Given that the purpose of FDI is to maximise profit, highly polluting MNEs will seek to relocate to areas where there are relatively low environmental standards (pollution havens) and thus relatively low related costs (Dinda, 2004; Pao & Tsai, 2011). Siegel et al. (2013) find that the greater the distance between environmental regulatory regimes of the home and host country, the lower the level of FDI. They suggest that MNEs’ technologies and organisational structures are tailored to fit certain environmental regulatory standards, and thus MNEs choose to invest in countries with similar standards. As the environmental standards in China are relatively low, and because of the reduced risk for Chinese MNEs to engage in ethically questionable behaviour (see par. 2.4.1), such as environmental pollution, it is expected that Chinese MNEs will perceive environmental regulation as an impediment to investment. Thus, the third hypothesis tested in the empirical analysis is:

Hypothesis 3. European countries that have strong environmental regulation receive less Chinese FDI than European countries that have weak environmental regulation.

2.4.4 Corporate income tax rate

Corporate income tax rates are one of the key drivers of FDI (Arbatli, 2011; Morck et al., 2008; Tocar, 2018). A host country’s corporate income tax rate directly affects the profit an MNE makes from its FDI. Thus, it seems logical to assume that if MNEs seek to maximise the return on their investment, they perceive high corporate income tax rates as a deterrent for FDI (Tocar, 2018). Following this logic, one would expect Chinese outward FDI to be negatively related to corporate income tax rates. However, most Chinese MNEs that invest abroad are state-owned enterprises (SOEs). For example, SOEs were responsible for 71 percent of Chinese FDI in the EU in 2017 (Hanemann, Huotari, & Kratz, 2019). Generally, the top executives of Chinese SOEs are directly appointed by a party secretary, who is often handing down directives from

(15)

14 the CCP’s Organisational Department (Morck et al., 2008). This means that Chinese SOEs’ FDI location decisions are not only incentivised by profit maximalisation but are also affected by the CCP’s political objectives, such as supporting Chinese foreign policy or regime survival (Kolstad & Wiig, 2012). Therefore, it could be that the corporate income tax rate has no significant effect on Chinese FDI. Thus, it will be interesting to see whether the fourth hypothesis holds:

Hypothesis 4. European countries that have a high corporate income tax rate receive less Chinese FDI than European countries that have a low corporate income tax rate.

2.4.5 Political stability

In the academic literature, it is generally accepted that political stability has a positive effect on FDI (Chakrabarti, 2001; Pajunen, 2008; Schneider & Frey, 1985). The term political stability encompasses a large collection of concepts, ranging from changes in macroeconomic management or the regulatory environment to civil wars (Chakrabarti, 2001). When a country is politically unstable, MNEs will resort to exporting or licensing instead of investing in own local production, as having large sunken investments in a politically unstable country is a potential risk (Buckley et al., 2007).

Nonetheless, it appears that Chinese FDI is mostly taking place in countries with high levels of political instability (Buckley et al., 2007; Shan et al., 2018). Multiple explanations are provided for this unconventional behaviour of Chinese MNEs. First, it is argued that countries with high levels of political instability receive little FDI from developed countries. This means that Chinese MNEs experience limited competition in these unstable countries and are thus able to broker better deals with the host country’s government, which in turn makes the country more interesting for Chinese MNEs to invest in (Buckley et al., 2007). Second, inexperience and naivety in the assessment of political risk could be a reason why Chinese FDI is not deterred by political instability (Ma & Andrews-speed, 2006; Wong & Chan, 2003). Third, many of the politically instable countries where Chinese MNEs invest in are either ideologically similar or (former) communist countries. Therefore, it may be not political instability that drives the effect on FDI, but ideological and political affiliations (Baek & Qian, 2011; Buckley et al., 2007; Ma & Andrews-speed, 2006). As there is no definitive conclusion on the relation between political stability and Chinese FDI, it will be interesting to find out whether the fifth hypothesis will hold:

(16)

15

Hypothesis 5. European countries that are politically unstable receive more Chinese FDI than European countries that are politically stable.

2.4.6 Democracy

The academic literature presents no conclusive explanation of the relation between a country’s democratic level and FDI. The risk of policy change is lower in democracies as there are more veto players involved in policymaking, acting as checks and balances to the executive power (North & Weingast, 1989; Tsebelis, 1995). Therefore, democracies are more predictable and thus more attractive to invest in (Jensen, 2003). However, the electoral process can also cause instability and unpredictability due to the frequent turnover of elected officials (Holmes et al., 2013). Furthermore, authoritarian regimes can provide better entry deals to MNEs and protection (repression) from labour unions, because they are not accountable to electorates (Jensen, 2003; Asiedu & Lien, 2011). But, increasingly, pressure from non-governmental organisations makes MNEs subject to reputation loss if they cooperate with repressive regimes (Busse, 2003). Spar (1999) explains that a distinction must be made between resource-seeking FDI and other forms of FDI. He argues that resource-seeking MNEs are restricted in their location choice by the presence of natural resources, and thus often have no other option than to cooperate with authoritarian regimes due to the ‘resource curse’. Asiedu and Lien (2011) confirm this theory and find that an authoritarian regime only has a positive effect on FDI in countries that rely heavily on exporting natural resources, and a negative effect in all other cases.

The small number of studies on Chinese FDI specifically, also present opposing results (Cheung et al., 2012; Kolstad & Wiig, 2012). But, as Chinese MNEs are not seeking cheap labour, nor natural resources when investing in Europe (Haneman & Huotari, 2018), it is expected that the democratic level has a positive relation with Chinese FDI in Europe. Therefore, the sixth hypothesis that will be tested in this study is:

Hypothesis 6. European countries that have a high level of democracy receive more Chinese FDI than European countries that have a low level of democracy.

2.5 Macro-economic explanatory factors of FDI

As mentioned in par. 2.3, the focus of this study is on the political-institutional determinant factors of FDI, on which there is not much conclusive evidence. On the contrary, there is much conclusive evidence on the relations between macro-economic explanatory factors and FDI. This makes macro-economic factors widely accepted as significant explanatory factors of FDI

(17)

16 in general (Chakrabarti, 2001; Tocar, 2018). Therefore, in order to test whether the estimated effects of the political-institutional factors do not capture those of the macro-economic determinants, the four most significant macro-economic factors are included in this study as control variables. These are: a) market size, b) labour cost, c) agglomeration factor, and d) infrastructure.

2.5.1 Market Size

The macro-economic determinant factor that is mentioned in nearly every study on determinants of FDI is market size (Chakrabarti, 2001; Tocar, 2018). This determinant becomes specifically important in the case of market-seeking FDI, as larger markets allow MNEs to take advantage of economies of scale. This is confirmed by Velde and Bezemer (2006), who find that when regional integration of markets takes place, for example through trade agreements, the larger integrated markets are more likely to attract FDI. Chinese FDI in the EU seems to be concentrated more and more in the largest economies (UK, Germany, and France). However, as many of the European countries are part of the European Economic Community (EEC), they share the same market. Thus, it will be interesting to see whether market size has a significant effect on Chinese FDI in Europe.

As Chinese FDI in the EU is for a large part market-seeking, it is expected that the relation between market size and Chinese FDI in Europe shows a positive sign.

2.5.2 Labour cost

Low labour costs can decrease the production costs and thus increase the efficiency of an MNE (Kersan- Škabić, 2013; Riedl, 2010). Therefore, labour costs are generally regarded to be negatively related to FDI (Chakrabarti, 2001; Tocar, 2018). However, workers with higher wages are usually also higher skilled workers, which in turn may be an attractive factor for FDI (Filippaios & Papanastassiou, 2008; Du et al., 2012).

As most of the Chinese MNEs that are active in Europe are not seeking cheap labour but are investing in high-tech and advanced manufacturing industries (Hanemann & Huotari, 2018), it is expected that there will be a positive relation between labour costs and Chinese FDI in Europe.

2.5.3 Agglomeration factor

The third macro-economic factor which is used in this study is the presence of agglomerations or the ‘agglomeration factor’. Agglomerations are clusters of industry specific firms which are

(18)

17 located in a specific geographic area. The increasing contingent nature of the international business environment has forced firms to become more flexible. This is necessary as they need to be able to quickly adapt to changes in their environment (Buckley & Casson, 2000). This process has led firms to focus on their core-business and outsource their ancillary activities. As a result, vertically integrated firms become less common and networks of independent firms have come in its place (Harrison, 1994). These networks (clusters) of firms generate positive externalities such as technological spill-overs, access to industry specific suppliers (forward linkages) and customers (backward linkages), and access to a shared pooled market for skilled labour (Marshal, 2009). Furthermore, agglomerations strengthen the bargaining power of individual MNEs versus local governments, which is likely to result in more favourable conditions for MNEs to operate in (Du, Lu, & Tao, 2008). All these positive externalities make an industrial agglomeration an attractive location for foreign MNEs to establish a firm (Boudier-Bensebaa, 2005).

Therefore, it is expected that there will be a positive relation between the presence of industrial agglomerations and Chinese FDI in Europe.

2.5.4 Infrastructure endowments

The presence of good infrastructure is commonly agreed upon as a determinant factor of FDI (Du et al., 2012; Tocar, 2018). Foreign MNEs are likely to invest in countries where their cost of doing business is low, as this ensures higher profitability. The costs of doing business are dependent on the presence of road and port facilities, telephone and internet networks, and supplies of electricity and energy (Jones, 2000). Therefore, countries that have a high quality and availability of transportation-, communications-, and energy infrastructure are likely to attract FDI (Findlay, 1996; Mateev, 2009; Mottaleb & Kalirajan, 2010; Nicoletti, Golub, Hajkova, Mirza, & Yoo, 2003). Additionally, around 40% of Chinese FDI in the EU is spent on brownfield investments in transport, utilities, and infrastructure (Hanemann et al., 2019). Therefore, it can be argued that countries that have large infrastructure endowments in place present more opportunities for Chinese MNEs to invest in.

(19)

18

3. Methodology

3.1 Research design

The main research question of this study warrants the use of a quantitative research design, as the study tries to establish whether there is a relation between political-institutional quality and Chinese FDI in Europe. To uncover the rationale or the “why” of the relation, further, more qualitative, research is necessary. However, by disaggregating the main research question into six sub-questions, more insight is already given into what factors drive the relation between political-institutional quality and Chinese FDI in Europe. This study uses an OLS-regression analysis to provide an answer to the main research question. This type of analysis is used as it reports the significance, estimated strength and sign of a relationship. Another option would have been to use a fixed-effects analysis. However, as time and data constraints limited the usable data to three years, outcomes would not be reliable.

Five different models are used to estimate the relation between political-institutional quality and Chinese FDI in Europe. The first model uses the main explanatory variables of interest and is used to test hypothesis 1 to 6. The second model adds control variables to check the robustness of the relations discovered with the first model. In the third model, a time-factor is added as a robustness check. The fourth and fifth model are the same as model two, but split up in separate regressions for European countries with a low credit rating (model 4) and European countries with a high credit rating (model 5).

All the variables used are transformed to its natural logarithm because using a log-log model provides smaller residuals (Blonigen & Davies, 2004), and results are easier to interpret, i.e. a 1% increase in an explanatory variable results in an [coefficients value] % change in outcome variable.

(20)

19 The five models are specified as follows:

Model 1: Ln(𝐶𝑂𝐹𝐷𝐼_𝑖𝑡=0) = (𝑏₀+ 𝑏₁Ln(M𝐶𝑜𝑟_𝑖𝑡−1) + 𝑏₂𝐿𝑛(𝑀𝑃𝑟𝑜𝑝_𝑖𝑡−1) + 𝑏₃𝐿𝑛(𝑀𝐸𝑛𝑣_𝑖𝑡−1) + 𝑏₄𝐿𝑛(𝑀𝐶𝑡𝑎𝑥_𝑖𝑡−1) + 𝑏₅𝐿𝑛(𝑀𝑃𝑜𝑙𝑠_𝑖𝑡−1) + 𝑏₆𝐿𝑛(𝑀𝐷𝑒𝑚_𝑖𝑡−1)) + 𝜀_𝑖 Model 2, 4 and 5: Ln(𝐶𝑂𝐹𝐷𝐼𝑖𝑡=0) = (𝑏0+ 𝑏1Ln(M𝐶𝑜𝑟𝑖𝑡−1) + 𝑏2𝐿𝑛(𝑀𝑃𝑟𝑜𝑝𝑖𝑡−1) + 𝑏3𝐿𝑛(𝑀𝐸𝑛𝑣𝑖𝑡−1) + 𝑏₄𝐿𝑛(𝑀𝐶𝑡𝑎𝑥_𝑖𝑡−1) + 𝑏₅𝐿𝑛(𝑀𝑃𝑜𝑙𝑠_𝑖𝑡−1) + 𝑏₆𝐿𝑛(𝑀𝐷𝑒𝑚_𝑖𝑡−1) + 𝑏₇𝐿𝑛(𝐶𝑀𝑎𝑟𝑠𝑖𝑧𝑒_𝑖𝑡−1) + 𝑏₈𝐿𝑛(𝐶𝐿𝑎𝑏𝑐𝑜𝑠𝑡_𝑖𝑡−1) + 𝑏₉𝐿𝑛(𝐶𝐴𝑔𝑔𝑙𝑜_𝑖𝑡−1) + 𝑏₁₀𝐿𝑛(𝐶𝐼𝑛𝑓𝑟𝑎_𝑖𝑡−1)) + 𝜀_𝑖 Model 3: Ln(𝐶𝑂𝐹𝐷𝐼𝑖𝑡=0) = (𝑏0+ 𝑏1Ln(M𝐶𝑜𝑟𝑖𝑡−1) + 𝑏2𝐿𝑛(𝑀𝑃𝑟𝑜𝑝𝑖𝑡−1) + 𝑏3𝐿𝑛(𝑀𝐸𝑛𝑣𝑖𝑡−1) + 𝑏₄𝐿𝑛(𝑀𝐶𝑡𝑎𝑥_𝑖𝑡−1) + 𝑏₅𝐿𝑛(𝑀𝑃𝑜𝑙𝑠_𝑖𝑡−1) + 𝑏₆𝐿𝑛(𝑀𝐷𝑒𝑚_𝑖𝑡−1) + 𝑏₇𝐿𝑛(𝐶𝑀𝑎𝑟𝑠𝑖𝑧𝑒_𝑖𝑡−1) + 𝑏₈𝐿𝑛(𝐶𝐿𝑎𝑏𝑐𝑜𝑠𝑡_𝑖𝑡−1) + 𝑏₉𝐿𝑛(𝐶𝐴𝑔𝑔𝑙𝑜_𝑖𝑡−1) + 𝑏₁₀𝐿𝑛(𝐶𝐼𝑛𝑓𝑟𝑎_𝑖𝑡−1) + 𝑏₁₁2015. year + 𝑏₁₂2016. year) + 𝜀_𝑖 2015. 𝑦𝑒𝑎𝑟 = {1, 𝑖𝑓 𝑦𝑒𝑎𝑟 = 2015 0, 𝑖𝑓 𝑦𝑒𝑎𝑟 = 2014 2016. 𝑦𝑒𝑎𝑟 = {1, 𝑖𝑓 𝑦𝑒𝑎𝑟 = 2016 0, 𝑖𝑓 𝑦𝑒𝑎𝑟 = 2014

3.2 Case selection

This study is based on data collected from 40 European countries from 2014 to 2017. These countries are the 28 EU member states plus Albania, Belarus, Bosnia and Hercegovina, Iceland, Kosovo, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, and Ukraine. Europe was chosen as the geographical area of interest as it is one of the main developed regions of the world where China is investing in, and aggregate data on country level is widely available.

For model 4 and 5 the included countries are divided into two groups based on their sovereign debt credit rating. The ratings used are an average of the individual ratings presented by Moody’s, Fitch, and Standard & Poor’s, which are the largest credit rating agencies in the

(21)

20 world. The included countries are divided into countries with a high credit rating (better than A-/A3) and countries with a lower credit rating (A-/A3 or worse). The sovereign debt credit rating is a representation of the risk that a country will default on its debts, and is often used to gauge the investment climate of a country (Elfakhani & Mackie, 2015). The country ratings are important as they affect the ratings of borrowers of the same nationality. For example, private companies or local governments are rarely given higher ratings than that of their home country (Cantor & Packer, 1996). If countries have a bad credit rating, they pay more interest on their loans as to compensate for the higher risk of default (Elfakhani & Mackie, 2015). As mentioned before, Chinese MNEs may have political goals in their investment strategy. Therefore, it will be interesting to see whether their FDI location decision depends on other factors in countries that have a hard time to attract capital than in countries where capital is more accessible. For example, it could be that Chinese MNEs are willing to invest in investment opportunities that are at serious risk of default, as long as the weak political-institutional framework of that country allows them to utilise the investment for their political goals, so called ‘debt trap politics’. Examples of this strategy being employed are the take-over of the Greek port of Pireaus (Reguly, 2019), and the Sri Lankan port of Hambantota (Abi-habib, 2018).

For each observation, the dependent outcome variable (Chinese FDI in Europe) is taken for the years 2015 – 2017 (t=0). The related determinant variables are taken a year earlier (2014 – 2016, t-1) than the corresponding dependent variable. This one-year lag between determinant and outcome is used because it is likely that a year passes between the moment that MNEs gather information on location factors and the moment of the actual investment i.e. the investment is based on determinant data of the year before the investment.

3.3 Operationalisation of main variables

The main variables of interest in this study are the outcome variable and the main determinant variables. These are: a) Chinese FDI (outcome variable), b) corruption, c) property rights, d) environmental regulation, e) corporate income tax rate, f) political stability, and g) democracy. The descriptive statistics of the main variables are reported in table 1. Notable are the high kurtosis values reported for political stability and democracy. Further elaboration on how this study deals with the high values is given in paragraph 3.6.

(22)

21

Table 1. Descriptive statistics of the outcome and main determinant variables.

Variable Obs. min max mean st. dev. skewness kurtosis

lnCFDI 113 -1.139 10.42 5.191 2.743 -0.142 2.315 lnmcor 117 1.445 4.376 3.280 0.765 -0.621 2.230 lnmprop 117 2.996 4.500 4.070 0.383 -1.009 3.510 lnmenv 117 3.824 4.507 4.333 0.143 -1.456 4.944 lnmctax 117 1.082 3.456 2.179 0.494 -0.047 2.695 lnmpols 117 3.434 4.486 4.243 0.191 -1.984 8.990 lnmdem 117 3.158 4.602 4.401 0.244 -3.491 17.16

3.3.1 Chinese outward foreign direct investment

The main outcome variable is Chinese FDI in European countries. Reliable and consistent data on Chinese FDI in Europe is difficult to gather (Meunier, 2012). There are three reliable resources that could be used. The first resource is the Rhodium Group’s China Investment Monitor, but this dataset is not publicly available and thus poses problems in terms of accessibility and replicability. The second resource is The Heritage Foundation’s China Global Investment Tracker. This is a dataset that contains information on individual investment transactions made by Chinese parent companies. However, it only includes transactions over 100 million U.S. dollar and does not provide aggregate country data. This is problematic as it potentially ignores a large amount of Chinese FDI, and it does not provide enough observations to make a significant regression analysis. The third resource is the Ministry of Commerce (MOFCOM) of the People’s Republic of China’s 2017 Statistical Bulletin of China’s Outward Foreign Direct Investment. This report provides worldwide data on Chinese FDI on aggregate country level and is consistently measured over countries and time. As this data is provided by the Chinese government, the validity of the data can be questioned. However, as data provided by the Rhodium Group and MOFCOM have high similarity (Meunier, 2012), it is deemed valid enough to use in this study. Thus, the values for Chinese FDI are taken from MOFCOM’s 2017 Statistical Bulletin of China’s Outward Foreign Direct Investment and is measured in million U.S. dollar.

(23)

22

3.3.2 Corruption

The level of corruption is measured by taking the average of Transparency International’s Corruption Perception Index and the World Bank’s World Governance Indicator. As corruption is a hard to measure concept, both used indexes rely on survey answers. This means the data used can possibly be tainted by inter-rater reliability issues. However, as both indexes are widely acknowledged by scholars in the field, the overall reliability and validity of the measures as a representation of corruption is acceptable to use in this study (Barassi & Zhou, 2012). The Corruption Perception Index and World Governance Indicator range from 0 to 100, with a low score representing a more corrupt host government. For this study, the score is inversed, so that a higher score corresponds with a more corrupt government. For example, an average score of 20 becomes a score of 80.

3.3.3 Property rights

The level of property rights protection is measured by taking the average of the 2016 Heritage Foundation’s Index of Economic Freedom (IEF), and the 2016 Property Rights Alliance’s International Property Rights Index (IPRI). The IEF measures the degree to which a country’s laws protect private property rights and the degree to which its government enforces those laws. Additionally, the independence of the judiciary, and the ability of individuals or companies to enforce contracts is included in the measure. As the IEF score is an aggregate index of data provided by the World Bank, World Economic Forum, and the Credendo Group, the index can be assumed as free of political bias and thus valid for use in this study. The IEF ranges from 0 – 100, with a score of 100 representing maximum protection of property rights.

The IPRI is composed by the Property Rights Alliance, in co-operation with 113 think tanks around the world. The index is a measure of a country’s legal and political environment, and the degree to which it enforces both physical property rights and intellectual property rights. Because there are so many different partners on which the IPRI relies for their data, inter-rater reliability could be an issue. However, as there are no large discrepancies in the data provided by the two indexes, the average of both indexes used should provide a reliable and valid representation of the concept of property rights. The IPRI ranges from 0 – 10, with a score of 10 representing maximum protection of property rights. This IPRI score is multiplied by 10 for this research in order to obtain a 0 – 100 score.

(24)

23

3.3.4 Environmental regulation

The measure for stringency of environmental regulation used in this study is taken from the 2016 Environmental Performance Index. This index is provided by Yale University and Colombia University in co-operation with the World Economic Forum. The index provides a score ranging from 0 to 100 resulting from aggregate outputs of 24 indicators within 10 issues; air quality, water & sanitation, heavy metals, biodiversity & habitat, forests; fisheries; climate & energy; air pollution; water resources; and agriculture. Environmental outputs are not necessarily a good measure of environmental regulation stringency as it provides problems in terms of validity. However, it is the most commonly agreed upon proxy used in the academic literature (Dinda, 2004; Pao & Tsai, 2011; Naughton, 2014; Siegel et al., 2013), and is therefore also used in this study. Even though it is not the most valid measure of environmental stringency, it does provide a comprehensive score that is comparable between countries. Because the index is reported bi-annually, the missing data for 2015 is substituted with averages of the 2014 and 2016 scores.

3.3.5 Corporate tax rate

The measure for corporate tax rate is taken from the Government Revenue Dataset. The Government Revenue Dataset is compiled by the International Centre for Tax and Development and the United Nations University World Institute for Development Economics Research (UNU-Wider). This dataset includes a variable that measures the taxes on income, profit, and capital gains of corporations and enterprises, including taxes on resource firms. Because this provides a comprehensive and consistent measure of corporate tax, it is deemed valid to use in this study. Furthermore, the Government Revenue Dataset is specifically created to provide transparent data of high quality to researchers; sources for every datapoint are given; data choices are documented; and problematic observations are explicitly flagged. As there are no flagged observations in the data used in this study, it can be assumed that the data is reliable.

3.3.6 Political stability

There are multiple sources of cross-country data that seek to report the differences between countries in terms of political risk and stability. However, as discussed in the previous chapter, political stability encompasses many different aspects, which makes it a complex variable that is hard to quantify. One of the publicly available indexes that reports political stability is the WGI provided by the World Bank Group. The political stability score measures perceptions of the likelihood of political instability and/or politically motivated violence, making it a valid measure

(25)

24

of the concept political stability as described in paragraph 2.4.5. Even though the ratings are subjective, they are calculated consistently across countries, which makes it a reliable index to use in this study (Arbatli, 2011). The data is transformed in such a way that the scores range from 0 to 100, with a higher score corresponding to a politically more stable country.

3.3.7 Democracy

In this study, three different indices are used to measure the level of democracy in European countries. These indices are the The Economist Intelligence Unit’s Democracy Index, the Freedom House’s Freedom in the World Report, and the Center for Systemic Peace’s Polity IV Index. Each index is transformed to a scale from 0 to 100, with a high score representing a more democratic country. The average of the three transformed indices is taken as the score for democracy used in this research. Each of the indices uses a different method for conceptualisation, measurement, and aggregation of their data. The Democracy Index is based on ratings for the electoral process and pluralism, civil liberties, the functioning of government, political participation, and political culture. The Freedom in the World Report reviews a country’s electoral process, political pluralism and participation, functioning of government, freedom of expression and belief, associational and organisational rights, rule of law, and personal autonomy and individual rights. Finally, the Polity IV Index examines political rights, civil liberties, and constraints on executive power. Nonetheless, the indices show a high level of correlation, indicating that they represent the same underlying phenomena. This makes them complementary to each other and increases the reliability and validity of the aggregate outcome as a measure of democratic level.

3.4 Operationalisation of control variables

To measure the effects of the political-institutional factors on Chinese FDI in Europe, a selection of control variables is added to the study. As mentioned in the theoretic chapter, the most commonly acknowledged explanatory factors of FDI are macro-economic factors. Therefore, the study controls for: a) market size, b) labour cost, c) agglomeration factor, and d) infrastructure endowments. The descriptive statistics of the control variables are reported in table 2. Notable is the lower amount of reported observations for labour cost, which could bias the reported results in model 2 and 3.

(26)

25

Table 2. Descriptive statistics of the control variables

Variable Obs. min max mean st. dev. skewness kurtosis

lncmarsize 117 8.307 15.18 11.74 1.761 0.0811 2.132

lnclabcost 101 0.548 4.270 2.868 0.915 -0.668 2.673

lncagglo 113 3.724 4.408 4.127 0.180 -0.368 2.039

lncinfra 117 3.103 4.239 3.823 0.246 -0.616 3.612

3.4.1 Market size

Total GDP per country is the measure most used for market size when studying the relation between market size and FDI (Chakrabarti, 2001; Tocar, 2018). Arbatli (2011) shows that measurement problems arise when GDP per capita is used as a measure for market size, when studying its effect on FDI. Therefore, for this study total GDP per country is considered a valid measure of market size. Data for this variable is taken from the World Bank’s World Development Indicators. As data is provided by the different national governments, each country uses different methods, definitions, and reporting standards. Furthermore, one of the main concerns of the compilers is that the informal economy, which is especially large in developing countries (with a small market size), is not incorporated in the indicator. However, as Chinese MNEs are unlikely to invest in the informal economy, and national accounts data is reviewed by the compilers and if necessary adjusted to be consistent with international guidelines, the indicator provided by the WDI is considered reliable enough to use in this study.

3.4.2 Labour cost

There are various ways in which labour cost can be measured. The most preferable measure would be the differential in unit costs between the source and host country. However, data constraints make it impossible to use this measure. Therefore, this study uses the average hourly labour cost per employee. The used data is taken from ILOSTAT, provided by the International Labour Organisation. ILOSTAT’s labour cost indicator comprises remuneration for work performed, bonuses, employers’ social security expenditures, costs to the employer for vocational training, and other miscellaneous costs regarded as labour related costs to the employer, making it a comprehensible and valid measure of labour cost.

To make the data comparable, local currencies are converted to U.S. dollars using market exchange rates, making it a reliable measure of labour cost over different countries. market

(27)

26 exchange rates are the preferred tool for conversion over the generally favoured purchasing power parities, as market exchange rates are the basis on which MNEs build their foreign labour costs calculations, as opposed to purchasing power parities which are more often used in research related to welfare issues (OECD, 2006). As is shown in Table 2, there are less observations for labour cost than the other variables. The countries that have missing values are Albania, Bosnia & Hercegovina, Macedonia, Montenegro, and Switzerland. As most of these countries are Eastern European, non-EU countries, the reported results could be biased.

3.4.3 Agglomeration factor

Agglomerations can either be clusters of industry specific firms, or areas in which there is a significant presence of foreign business activity, both of which provide positive externalities for MNEs.

To measure the presence of clusters of industry specific firms, using the ratio between industry specific GDP and total sector GDP would be the preferred measure (Riedl, 2010). However, as cross-country industry specific data is not publicly available, it is impossible to use this measure in this study. Therefore, in this study the presence of agglomerations is measured with data from the World Economic Forum’s Executive Opinion Survey. The survey is taken by over 14700 business executives in 141 countries, providing a cross-country comparable score. In this study, the average scores of question 6.11: “In your country, how prevalent is foreign ownership of companies?”; and question 11.3: “In your country, how widespread are well-developed and deep clusters (geographic concentrations of firms, suppliers, producers of related products and services, and specialised institutions in a particular field)?” are taken, and transformed to a scale ranging from 0 to 100. As it is not a direct measure of agglomerations but a measure of perception, validity concerns must be taken into account when drawing conclusions based on results of this variable. Additionally, as the same questions are asked over time and cross-country to multiple executives per country, the measure can be considered reliable.

3.4.4 Infrastructure

As mentioned in the theoretic framework, infrastructure endowments can be divided into transport facilities, communication networks, and energy supplies. Du et al., (2012) use highway density as a proxy for transport facilities. However, as there is no reliable cross-country aggregate data on highway length publicly available for all European countries for 2014 - 2016, this study uses railway density instead. This variable is measured as the length of

(28)

27 railway tracks per square kilometre in a country. Following the operationalisation of Mottaleb and Kalirajan (2010), communication infrastructure is measured by taking the number of internet and telephone (both fixed-line and mobile) subscriptions per 100 people. Finally, the energy supply of a country is measured by taking the approach of Khachoo and Khan (2012), who use the electric power consumption per capita (kWh per capita) as a proxy for infrastructure. All the data used, except electric power consumption, is taken from the World Development Indicators provided by the World Bank Group. Data on electric power consumption is provided by the International Energy Agency’s World Energy Balances. Data for infrastructural endowments is sometimes not internationally comparable as the collection of infrastructure data has not been standardised. For example, the data used for railway length is based on reporting of the individual railway companies. Furthermore, data on communication facilities might not be strictly comparable as some countries end their fiscal (measured) year on a different moment in time. Therefore, the strict validity and reliability of the data can be questioned, but in will in general be valid and reliable enough to make cross-country comparisons. A weighted average is taken from the five indicators resulting in one aggregate score for development of infrastructural endowments ranging from 0 to 100.

3.5 Summary

A summary of all the used variables, their abbreviations, and their operationalisation is presented in table 3.

(29)

28

Table 3. Summary of the used variables and their operationalisation.

Variable Proxy Expected sign Main or control variable

Dataset Original Data Source Chinese FDI (CFDI) Capital flows in $ million Outcome Statistical Bulletin of China’s Outward Foreign Direct Investment People’s Republic of China’s Ministry of Commerce Corruption (MCor) Perception of corruption + Main Corruption Perception Index, Governance Indicator Transparency International, The World Bank Group Property rights (MProp) Property rights protection - Main Index of Economic Freedom, International Property Rights Index Heritage Foundation, Property Rights Alliance Environmental policy stringency (MEnv) Aggregate environmental policy outcomes - Main Environmental Performance Index Yale University, Columbia University, World Economic Forum Corporate income tax rate (MCTax)

Taxes on income, profit, and capital gains of

corporations

- Main Government

Revenue Dataset

International Centre for Tax and Development, and UNU-Wider Political stability (MPols) Political risk score - Main World Governance Indicators

The World Bank Group Democracy (MDem) Civil liberties, political rights, and constraints on executive power

+ Main Freedom in the

World, Democracy Index, Polity IV Freedom House, Economist Intelligence Unit, Center for Systemic Peace Market size (CMarsize) Gross domestic product in $ Bn + Control World Development Indicators

The World Bank Group

Labour cost (CLabcost)

Average hourly labour costs in $

+ Control ILOSTAT International

Labour Organisation Agglomeration (CAgglo) Average survey score + Control Executive Opinion Survey World Economic Forum Infrastructure (CInfra) Railway density, internet and telephone subscriptions, electric power consumption + Control World Development Indicators, World Energy Balances

The World Bank Group,

International Energy Agency

(30)

29

3.6 Data analysis

The data is tested for linearity, independence, normalcy, and equality of variance. The relation between the determinant variables and the outcome variable is a linear one. This is established by examining the scatterplot between fitted values and residuals presented in appendix A. Independence of the variables can be discussed as the pairwise correlation table, presented in appendix B, shows that there are some variables that have a high correlation. To test whether multicollinearity is a problem, a test for variance inflation factors is performed on the data which results in a high (>10) value for lnclabcost and lnmprop. However, this is not problematic because the estimated coefficients for lnclabcost in model 2 and 3 are insignificant. Thus, multicollinearity between lnclabcost and lnmprop has no negative effects on the results. The normalcy is of the data is tested by examining the skewness and kurtosis. The descriptive statistics in tables 2 and 3 show several variables with a high kurtosis. The high kurtosis values for lncinfra and lnmdem are caused by Kosovo which has missing data on those variables and is thus excluded in the regression analysis. The high kurtosis value for lnmpols is caused by Ukraine’s outlier values. This problem is addressed by substituting Ukraine’s values with the

value of the bottom 10th_{percentile. To test variance, a Breusch-Pagan / Cook-Weisberg test for}

heteroskedasticity was performed on the data which resulted in a value of 0.0020. As this value is smaller than 0.05 the null hypothesis of the data having a normal variance is rejected. This means the data has a heteroskedastic variance. To address this issue, robust standard errors are used.

(31)

30

4. Results

The main results from the OLS regression of China’s FDI in Europe on the five specified models are presented in table 4. The coefficient columns show the estimated coefficients for the equation specified in the models. Because the models used in this study use a large number of variables, the R-squared values may not give a correct representation of the goodness of fit of the estimated coefficients and the actual observations. Furthermore, the R-squared value may report a too high value due to limited sample sizes, overfitting the model as a result. Therefore, to compensate for these two limitations, the adjusted R-squared values are also reported.

Main model (1).

The results of the main model as presented in table 4, show a significant negative effect of democracy on Chinese FDI, suggesting that Chinese MNEs prefer to invest in European countries where political freedom and civil liberties are limited. Furthermore, the results show a significant positive effect of environmental regulation and corporate tax rate on Chinese FDI in Europe for the period 2014-2016. This would suggest that Chinese MNEs are motived by strong environmental regulation and high taxes. Because this would be highly unlikely, the estimated effects could be interpreted as indicators of a faulty model. When controlling for macro-economic determinants in model 2, it seems that this is indeed the case.

Control model (2)

The second model adds four (macro-economic) control variables to the regression equation. Compared to the main model, the R-squared value has almost doubled by adding the control variables, greatly increasing the model’s explanatory power. The results show a small significant negative effect of corruption on Chinese FDI, suggesting that Chinese MNEs prefer to invest in European countries where corruption is under control. Additionally, a large significant negative effect on Chinese FDI is shown for property rights. This suggests that Chinese MNEs favour to invest in European countries where there is little property rights protection. Additionally, the results show significant positive effects of market size, agglomeration, and infrastructure on Chinese FDI. Suggesting that Chinese MNEs preferably invest in European countries with a larger market size, more commercial agglomerations, and more infrastructural endowments. Finally, the results show that the significant effects for environmental regulation and democracy estimated in the main model, are no longer significant when the control variables are added. This suggests that the estimated effects in the main model

Investigating political-institutional characteristics and Chinese foreign direct investment in Europe