Political risk, cultural distance and foreign direct investment

m

UNIVERSITY OF AMSTERDAM

Political Risk, Cultural Distance and Foreign Direct Investment

Paul Pater University of Amsterdam

Master Thesis

Author: University: Program: Supervisor: 2nd Supervisor: Date: Version:

Paul Dinand Pater, MSc (6051944)

University of Amsterdam, Faculty Economics and Business MSc Business Administration - International Management E. Dirksen MSc

dr. I. Haxhi 28-02-2016 Final Version

n_

Statement of originality

This document is written by Student Paul Pater 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

Table of Contents

Table of Contents 2 Table of Figures 4 Table of Tables ...•.... 4 1. Introduction 5 2. Literature 7

2.1 Foreign Direct Investment 7

2.2 Political Risk 8

2.3 Cultural Distance 11

3. Data, method and empirical models 16

3.1 Foreign Direct Investment- UNCTAD 16

3.2 Political Risk - Worldwide Governance Indicators 18

3.3 Cultural Distance - GLOBE 20

3.4 Choice of countries and other implications 22

3.5 Control variable 24 3.6 Weighted variable 25 3.7 Method 27 3.7.1 Hypothesis 1 28 3.7.2 Hypothesis 2 28 4. Results 30 4.1 Hypothesis 1 30 4.1.1 Assumptions 30 4.1.2 Regression 33 4.2 Hypothesis 2 38 4.2.1 Assumptions 38 4.2.2 Regression 39

5. Conclusions, limitations and recommendations 41

5.2 5.3 Limitations ...••. 42 i I

,.

Recommendations ::- .. --: 43

f

i 6. Bibliography 44

Appendix A Political background of chosen countries 50

A.1 Egypt ...•.. 50

A.2 Greece ...•... 51

A.3 Hong Kong 52

A.4 Morocco 53

A.5 Portugal 54

Table of Figures

Figure 1 Change in FD

I-stock per country in terms of

percentage

18

Figure 2 Culture Clusters in the GLOBE study

21

Figure 3 Changes in Level of

Governance

23

Figure 4 Worldwide FDI-stock

25

Figure 5 Hypotheses

27

Figure 6 PROCESS Model 1

29

Figure 7 ZRESID vs ZPRED Hl (SPSS)

32

Figure 8 Histogram Hl (SPSS)

33

Figure 9 ZRESID VS ZPRED H2 (SPSS)

38

Figure 10 Histogram H2 (SPSS)

39

Figure 11 Level of

Governance - Egypt

50

Figure 12 Level of

Governance - Greece

51

Figure 13 Level of

Governance - Hong Kong

52

Figure 14 Level of

Governance - Morocco

53

Figure 15 Level of

Governance - Portugal

54

Figure 16 Level of

Governance - Taiwan

55

Table of

Tables

Table 1 Cultural Distance between Culture Clusters

22

Table 2 Weighted Level of

Governance

26

Table 3 Descriptive Statistics Hl (SPSS)

34

Table 4 Correlations Hl (SPSS)

34

Table 5 Model Summary Hl (SPSS)

35

Table 6 ANOVA(a) Hl (SPSS)

36

Table 7 Coefficients (a) Hl (SPSS)

37

I I I

I

'

! Globalization has changed the world in which we live and work in in endless ways. The concept

1. Introduction

of globalization has been used in nearly all fields of academic research. Especially economic and cultural researchers have studied the concept extensively. Globalization has also changed the way we conduct business. Foreign Direct Investment (FDI) is seen as a key indicator of economic globalization, and has increased globally over the last decades: 'on average, since 1990 FDI grew by about 7 .6 percent per year ( or about $50 billion per year)' (Milner, 2014 ). According to the UNCTAD database, this growth resulted in an increase in FDI flow between

1980 and 2010 by a factor of 25 (UNCTAD, 2015).

Companies are always on the lookout for new opportunities. In this process, they tend to avoid certain risks. International businesses are nowadays more dependent on a lot of external risk factors than they were decades ago. The decisions made by managers affect the total foreign direct investments around the globe. One of the more recent factors investors take into account is political risk. 'With the trend towards globalization, political risk assessment has become an essential task for international marketing managers in making decisions with regard to whether and how to enter a foreign market' (Zou, 2015). 'The Arab Spring and the Euro crisis have shown that political factors and events can make or break markets overnight, making them of increasing importance to commercial actors when investing abroad' (Fägersten, 2015). This trend towards increased use of political risk assessment when investing abroad has been ventilated by the Economic Intelligence Unit in a very illustrating manner:

Businesspeople and investors have traditionally paid little attention in their decision- making to most forms of political risk, compared with most other important drivers of investment decisions. Macroeconomic conditions, labour availability and costs, and the overall business and policy environment in a country have been far more important issues. However, there are signs that things may be changing. Various forms of political risk have jumped towards the top of corporate agendas. (Economic Intelligence Unit, 2007).

Economic globalization, and the increase in FDI particular, linked most countries worldwide in bilateral trade relations. Whilst some anthropologists expected that cultural homogenization would occur because of this intensive worldwide economic interactions, this

was not the case (Olwig, 2013). Instead, cultural differences are still apparent. Moreover, cultural differences can be expected to play an important role in trade relations. Managers are the ones concerned with investments abroad. All managers are human beings, and in that sense they are likely to take cultural distance into account. They tend to move 'into those markets they can most easily understand' (O'Grady & Lane, 1996). The internationalization theory shows the process of companies starting investments abroad in countries that they feel familiar with, and investing further away from home gradually. In other words, they start their internationalization in countries with a low cultural distance from their home country, and progressively invest in countries with a greater cultural distance (Benito & Gripsrud, 1992). This shows us that cultural distance is of influence on business' decisions. The attitude of a company's home country towards a host country was also found to be one of the main determinants of deterring foreign direct investments (Gobinda Goswami & Haider, 2014). All these studies show that culture distance influences foreign direct investment. And as was previously mentioned, political risk is another influencer of foreign direct investment. This research is focused on the interdepence of these two factors and their effect on a host country's FDI. Since cultural distance is (more or less) constant, and political risk is more volatile, this study takes cultural distance as a moderator. This choice will be discussed in more detail further on. This brings me to the research question of this thesis.

Research question:

What is the effect of cultural distance and political risk on a country's inward foreign direct investment?

The next section contains an extensive literature review, followed by the data and method section. Consequently, the results of the analysis will be discussed. Finally, the conclusions, limitations and recommendations will be observed.

2.

Literature

This chapter will define and discuss the three concepts that are part of the research question proposed in the introduction: FDI, Political Risk and Cultural Distance.

2.1 Foreign Direct Investment

FDI is a macro-economic concept that focusses on the flow of investments between countries. Although there is no real discussion on the definition of FDI, every definition stresses another aspect of the concept. In the words of Moran (Moran, 2012), 'abstract FDI takes place when a corporation based in one country establishes a business operation in another country, through setting up a new wholly-owned affiliate, or acquiring a local company, or forming a joint venture in the host economy'. The OECD uses different words for practically the same definition, but stresses the objective of FDI: 'Direct investment is a category of cross-border investment made by a resident in one economy (the direct investor) with the objective of establishing a lasting interest in an enterprise (the direct investment enterprise) that is resident in an economy other than that of the direct investor' (OECD, 2008). A more common word for cross-border investments is bilateral FDI. There are 2 different measures involved in documenting the FDI of a particular country or area; stock and flow. FDI stock 'measure the total level of direct investment at a given point in time' (OECD, 2015), whereas FDI flows 'record the value of cross-border transactions related to direct investment during a given period of time (OECD, 2015).

Why is bilateral FDI interesting in general, and for academic research specifically? FDI is an important factor in economic relations between countries, and has multiple effects. In their much cited research, Busse and Hefeker (2007) state that 'the economic development of emerging markets and developing countries depends to a large extent on the possibility to make profitable investments and accumulate capital. According to another study by Hansen and Rand's on FDI inflows of developing countries, 'FDI causes growth' (Hansen & Rand, 2006). Furthermore, they argue that FDI has an impact on knowledge and technology spill over which increases the developing country's GDP in the long run. The company investing in the host country expands both its scale and scope. The host country has the advantage of increased

economic activity, which could lead to multiple spill-over effects. These spill-over effects can contain, but are not limited to:

•

•

•

•

•

Innovation and new technologies (Cheung & Ping, 2004) (Blomström & Kokko, 1998) (Bing, 2005)

Increased exports (Kneller & Pisu, 2007)

Increased employment (Blomström, Kokko, & Mucchielli, 2003) Increased productivity (Javorcik & Spatareanu, 2005)

Improvement of skills (Blomström & Kokko, 1998) (Cheung & Ping, 2004)

According to the OECD Benchmark Definition of Foreign Direct Investment, 'FDI is a key driver of international economic integration. With the right policy framework, FDI can provide financial stability, promote economic development and enhance the wellbeing of societies' (OECD, 2008).

Is bilateral FDI still relevant in 2015? It could be argued that bilateral FDI is losing its relevance in macroeconomic data as large multinational companies are not always bound to a single country, and are sometimes worth more than some small national economies itself. For example, Royal Dutch Shell topped the Forbes 500 list in 2013 with a revenue of 467 billion US dollars. (Forbes, 2013). In the same year, the Dutch gross domestic product (GDP) was 556 billion US dollar. This meant that in 2013 the revenue of Royal Dutch Shell related to 84% of the Dutch Gross Domestic Product (GDP) (Daniel, 2013). Other examples are Glencore's revenue comparable to 39% of Switzerland's GDP, and BP's revenue relating to 24% of the United Kingdom's GDP. From this can be derived that not all investments made by Shell are actually originated in the Netherlands. The fact that bilateral FDI deals with economic relations between two countries, could miss the essence of the investment of the multinational in the host economy. But still, bilateral FDI is the most important indicator of worldwide trade since it shows the 'latest trends in globalization of economic activities' (UNCTAD, 2015).

2.2 Political Risk

Risk is a concept that has meaning in everyday life, business environment and academic literature. In the broadest and abstract sense, risk refers to 'uncertainty about and severity of the consequences (or outcomes) of an activity with respect to something that humans value'

(A ven & Renn, 2009). What then, is the meaning of risk in business? According to Kit Sadgrove's comprehensive guide, business risk 'applies to any management decision that could have a good or bad outcome' (Sadgrove, 2015). He distinguishes two types of business risk. The first type is non-entrepreneurial risk, which is caused by fire, pollution, fraud, etc. The second type is entrepreneurial, which comes from new investments and innovations. Sadgrove emphasizes the importance of risk assessments, as 'risk management helps a company to evaluate its strengths and weaknesses [and] is a tool that makes a company grow strong' (Sadgrove, 2015).

When doing business abroad, part of the entrepreneurial type risk is the country risk. Political risk is part of a country risk analysis, which encompasses other elements of risk as well. According to Hogan, Lipton & Olson, country risk analysis is 'the evaluation that the company undertakes to determine the effect a particular foreign country or world region's policies can have on the multinational company's cash flow and thus firm value' (Hogan, Lipton, & Olson, 2012). Wagner describes country risk as follows: 'once outside its national borders, it [the company] faces a huge number of country-specific risks with which it is wholly unfamiliar. These vary in field, nature, and severity, and can range from expropriation through corruption to civil unrest' (Wagner, 2012). These two definitions of country risk both focus on a different aspect in the home country. Whereas Hogan, Lipton & Olson focus on the host country's government policies, Wagner views country risk in a broader sense and includes unintended changes not initiated by policy changes. These two views of country risk are representative for the ongoing discussion about what should be included in a country risk analysis.

What then, makes risk political? Often, it is defined negatively as non-market risk (Fägersten, 2015).

In

Kobrin's benchmark review on political risk from 1979, he defines political as something that clearly involves power or authority' (Kobrin, 1979).

In

the broadest sense, political risks are 'unwanted consequences of political activity' (Kobrin, 1979). Like I argued before, political risk is a concept that has seen increasing attention. One reason for this is the increase in FDI worldwide, which makes the risks involved in international investments relatively more important for companies (Bekaert, Harvey, Lundblad, & Siegel, 2014). Secondly, managers increasingly realize the impact of political risks on their business. Björn Fägersten recently pointed out that political risk is the newest add-on to risk management in businesses and is increasingly part of their analysis (Fägersten, 2015).

In literature, political risk is mainly focussed on intentional government action affecting the investor's business: 'Political risk refers to the risk that a government action will negatively affect the cash flows of a company conducting an international investment' (Bekaert, Harvey, Lundblad, & Siegel, 2014). This 'government action' school argues that political risk always 'stems from political behaviour and thus be actor-driven' (Fägersten, 2015). But investors are not only focused on the government's actions itself, but on the broader political context. According to Investopedia, in their own words the world's largest digital financial education platform, political risk is 'the risk that an investment's return could suffer as a result of political changes or instability in a country. Instability affecting investment returns could stem from a change in government, legislative bodies, other foreign policy makers, or military control' (Investopedia, 2015). This definition takes into account all aspects of political risk that impact investor's FDI choices. Another school in research focussed on dysfunctional institutions and systemic flaws (McKellar, 2010). This thesis will use the definition by Fägersten, which simply combines both schools of academic research: 'political risk as a potential harm to commercial activities caused by political action or arising from dysfunctional political systems' (Fägersten, 2015).

Political risk assessments are important because they are used by international investors as a tool for directing their investments. As Keiko Honda, Executive Vice President of the Multilateral Investment Guarantee Agency said in the organization's 2013 report: 'we found that investors continue to rank political risk as a key obstacle to investing in developing countries' (MIGA, 2013).

Political risk has always been a mainly subjective part ofrisk analysis. The assessment of political risk 'has traditionally been more of an art than a science' (Sachs, Tiong, & Wagner, 2008). Whilst qualitative assessment might make more sense on case studies, quantitative analysis is needed for reasonable comparisons between countries. Mainly business focussed research has focussed on quantitative political risk. According to Kaplan and Garrick, 'we are not able in life to avoid risk but only to choose between risks' (Kaplan & Garrick, 1981 ). They continue arguing that 'rational decision-making requires, therefore, a clear and quantitative way of expressing risk so that it can be properly weighed, along with all other costs and benefits, in the decision process'. For this reason, company's decision-makers prefer quantified political risk assessments to base their decisions on. The quantitative risk assessments are mostly global databases including historical data on political risk. These databases are made commercially available to businesses as a package by companies like the PRS Group and AON.

The relationship between political risk and FDI remains relatively unclear. Whilst some earlier studies reveal a negative relationship between the two (Schneider & Frey, 1985), there were other studies that find no effect (Jaspersen, Aylward, & Knox, 2000) (Hausmann & Fernandez-Arias, 2000). Furthermore, there were even studies that found different effects for different time periods_(_Loree Sf_Guisinger, 1995). More recent studies point in the direction of a negative impact of political risk on FDI. In 2012 Méon & Sekkat stated that 'foreign investment has been repeatedly found to be sensitive to political risk' (Méon & Sekkat, 2012). The testing of this relationship might thus seem redundant, but as there is still no full consensus on the topic, further research with new data is still sensible.

This brings us at the first hypothesis of my research. This hypothesis takes into account the political risk and the FDI-stock of a particular country, and expects that an increase in political risk leads to a decrease in FDI-stock. This hypothesis should be seen as collateral research leading up to the second hypothesis discussed later on.

Hypothesis 1:

Increased political risk in a particular country has negative impact on the country's inward foreign direct investment stock.

2.3 Cultural Distance

Before defining and operationalising cultural distance, a definition of culture itself is needed. Culture is used in multiple academic subjects in different ways. But the main source of research on the concept of culture itself is within the study of anthropology. The founder of cultural anthropology, Sir Edward Burnett Tylor, defined culture as 'that complex whole which includes knowledge, belief, art, morals, law, custom and any other capabilities and habits acquired by man as a member of society' (Tylor, 1920). Culture is context dependent, nurture instead of nature. Hofstede, the godfather of the concept of cultural distance (which is to be discussed later on), defined culture as 'the collective programming of the mind which distinguishes the members of one group or category of people from another' (Hofstede, Hofstede, & Minkov, 2010). An essential part of culture is the fact that people are member of a certain group of society. This implies that there are other groups or societies than the one this particular person belongs to. The difference between these groups or societies is the cultural distance.

Geert Hofstede was the first to measure cultural distance, in his case for employees of IBM worldwide. His first theory consisted of four dimensions, two other dimensions were added later after additional research. These dimensions made cultural comparisons between countries possible, and have triggered cross-cultural research in multiple academic fields. According to Michael Mink.ov, a befriended colleague researcher of Geert Hofstede, 'Hofstede argued that many national differences in work-related values, beliefs, norms, and self- descriptions, as well as many societal variables, could be largely explained in terms of their statistical and conceptual associations with four major dimensions of national culture' (Mink.ov & Hofstede, 2011 ). The six cultural dimensions of Hofstede comprise of:

•

•

•

•

•

•

Power Distance Index

Individualism versus Collectivism Masculinity versus Femininity Uncertainty Avoidance Index

Long Term Orientation versus Short Term Nonnative Orientation Indulgence versus Restraint (Hofstede, Hofstede, & Mink.ov, 2010)

Hofstede is still one of the most cited authors in social sciences, and his impact on cross-cultural studies has been praised by influential academics like Michael Bond (Bond, 2002) and Mark Peterson (Peterson, 2003).

Despite the widespread impact of Hofstede' s research, his concept of cultural distance has seen widespread criticism as well. One main critic is McSweeney, who spent an article on criticising Hofstede and concludes that 'Hofstede's claims are excessive and unbalanced' (Mcsweeney, 2002). In their guide on the essentials of and the criticism on Hofstede's model, Shaiq, Khalid, Akram & Ali summarized the criticism into the following main points (Shaiq, Khalid, Akram, & Ali, 2011):

•

•

•

•

Relevancy (validity): a survey is not a valid instrument to measure cultural distance . Cultural Homogeneity: Hofstede's survey focused on individual managers, whilst the conclusions are made on larger populations.

National Divisions: cultures are not always separated by borders; thus nations are not the right unit of analysis for culture.

Political Influences: because of the period of the research, just after the Second World War,· the political situation might have influenced dimensions like the Uncertainty Avoidance.

• One-company Approach: the research made conclusions about a nation's culture based on information from just one company.

•· Out-dated: the research is out-dated and did not stand the test of time in a the highly globalized and hybrid economy.

..

•

Too few dimensions: there are too few dimension to capture a concept as complex as culture.

Statistical integrity: statisticians doubt the statistical integrity as opposed to an analysis built upon chance.

A newer broadly accepted concept of cultural distance is the Global Leadership and Organizational Behavior Effectiveness Research (GLOBE) (House, Hanges, Javidan, Dorfman, & Gupta, 2004). The GLOBE study is partly based upon Hofstede's research and intended as a correction of his model (Minkov & Hofstede, 2011). According to the Shaiq, Khalid, Akram & Ali, 'GLOBE is a step to move further ahead from Hofstede's approach and to develop comprehensive, theoretically sound and verifiable cross cultural dimensions' (Shaiq, Khalid, Akram, & Ali, 2011). Whereas Hofstede's model used six dimensions, GLOBE puts forward nine cultural dimensions. These nine cultural dimensions are:

•

Performance Orientation

•

Uncertainty Avoidance

•

Humane Orientation

•

Institutional Collectivism

•

In-group Collectivism

•

Assertiveness,

•

Gender Egalitarianism

•

Future Orientation

•

Power Distance

They build on findings by Hofstede (1980), Schwartz (1994), Smith (1995), Inglehart (1997), and others' (Center for Creative Leadership, 2015).

Cultural differences are known to play an important role in investment choices by MNE' s. A lot of research has been done on the relationship between culture and 'entry mode choice, international diversification, and MNE performance' (Tihanyi, Griffith, & Russell, 2005). According to Gobinda Goswami and Haider, 'cultural conflicts [ .. ] are found to be

responsible for deterring FDI inflow' (Gobinda Goswami & Haider, 2014). Why would cultural distance influence the investment choices of managers? There are different factors that play a role in the decision making process. Chester Irving Barnard already differentiated between logical and non-logical mental processes in a presentation at the Princeton University in 1938 (Barnard, 1938). The obvious one is the logical or analytical factor, in which a manager weighs the objective advantages and disadvantages of certain investments. For example, cultural differences like the spoken language can create barriers for integrating two companies successfully. Moreover, the whole formal and informal institutional context can be an obstacle for investing in a particular country (Holmes, Miller, Hitt, & Salmador, 2013). Chang, Kao, Kuo & Chiu named these obstacles 'operational difficulties: 'fundamental differences in norms and values between the home country ofMNEs and the host country of their foreign operations often create operational difficulties and increase the efforts required to enter a foreign country' (Chang, Kao, Kuo, & Chiu, 2012). Managers are usually aware of these operational difficulties and consciously take them into account when making decisions.

Another factor in the decision making progress, distinguished by Barnard as well (Barnard, 1938), is less evident: the non-rational, or intuitional, part. Manager's often pride themselves with their ability for rational decision making, but intuition always has in important role. In contrast to analytical decision making, 'intuition arises through non-conscious operations' (Dane & Pratt, 2007). Cultural factors are thought to play a role in the intuitional part of decision making as well. According to Burke and Miller, part of a manager's intuition is to make decisions compatible with company culture (Burke & Miller, 1999). Culture thus plays a role in analytical as well as in intuitional managerial decision making, though the connection between the two is still unclear (Dane & Pratt, 2007).

The literature above indicates that cultural distance plays a role in managerial decision making, and thus in FDI flows. As stated in the introduction, cultural distance is a relatively stable concept. The culture of a country doesn't change within a couple of years. Thus, the cultural distance between two countries can be expected to be stable as well. My first hypothesis expects changes in political risk in a host country to influence the amount of FDI stock in that country.

For the second hypothesis, I expect cultural distance to play a moderating role in this effect. It can be expected that changes in political risk have more influence on a country further away culturally then on a neighbouring country. This brings me to my second hypothesis, in

which I expect a positive moderating effect of

cultural distance on the effect tested in the first

hypothesis:

Hypothesis 2:

There is a moderating effect of

cultural distance on the impact of

a country's political

risk on their FD

I stock.

3. Data, method and empirical models

Before explaining what this chapter consists of, the reasons for doing this research

quantitatively will be discussed. Firstly, it is the most straightforward way of testing my

hypotheses. Quantitative research methods can give reliable and objective conclusions for both.

Secondly, with statistical analysis, it is possible to generalize the results of this research.

Thirdly, as will be described elsewhere in my study in more detail, a relatively large part of

the

companies investing abroad use quantitative databases as a source for their decisions regarding

FDI.

This chapter will build upon the previously discussed literature to formulate the

quantitative part of

this thesis. First, the collected data will be discussed per variable. Second,

the direct implications of

the data on the choice of

specific countries will be discussed. Third,

the control variables will be discussed as well. Fourth, the use of a weighted variable will be

justified. Lastly, the method used to analyse the data will be introduced and described.

3.1 Foreign Direct Investment- UNCTAD

For this research, a single globally comprehensive and fairly reliable database is needed. If a

database like that is not available, multiple databases could be linked together. Most

organizations, like the OECD and APEC, only have specific bilateral FDI data for and from

specific regions. Moreover, the problem with a lot of

the FDI data available is that 'the scarcity,

unreliability and inconsistency of FDI data pose a serious challenge for policy-makers,

academics and practitioners' (UNCTAD, 2015).

In

this study, the online publicly available

database by the United Nations Conference on Trade and Development (UNCTAD) is used.

The UNCTAD is part of

the United Nations and is governed by its 194 member states.

It

is responsible for 'dealing with development issues, particularly international trade -

the

main driver of

development' (UNCTAD, 2015). The UNCTAD makes an effort to bridge the

great gap in FDI data described above. In their online publication, they use mainly national

sources for collecting their data. Where this data is not available, gaps are filled with

information from mirror countries. This means that for countries that have not published inward

FDI data, the outward FDI data of

the host countries are used as primary source to complement

the data. UNCTAD included data per country on flows as well as stock, as discussed in the

previous chapter on literature. Though not entirely complete, the UNCTAD database is the only comprehensive source available including all 206 countries which uses relatively reliable (national) sources, and is therefor the most suitable database for this particular research.

In

this thesis, FDI-stock will be used for analysis. More specifically, the change in FDI- stock will be used. As discussed before, FD I-stock measures the total value of investments at a specific moment in time. The analysis is about decisions made regarding investments, and these decisions reveal themselves in the yearly change in FDI-stock. Our hypotheses test the effect of changes in FDI, thus we need a variable that grasps changes in FDI-stock as well. Another option would be to correlate the value of political risk and FD I-stock, but these values don't tell us much. It is about managers' decisions that express themselves in mutations of the variable.

The change in FDI per country in terms of percentage of change is shown in figure 1. The countries used in this diagram are the countries used in my study to test the hypotheses. Section 2.4 will explain the choice for these countries. Without thorough analysis, but only looking at the graph there seems to be a correlation between all countries during this period. Especially the common depressions in 2005, 2008 and 2011 are noteworthy. The depression in 2008 was especially impactful on the world economy. The World Trade Organization called this the 'biggest such contraction since the Second World War' (World Trade Organization, 2009).

Figun• I C 'range tn FOT-stock per country in terms of percentage 80.00 60.00 40.00 20.00 0.00 2001 2004 2005 2006 -20.00 -40.00

-Egypt -Greece -Hong Kong -Morocco Portugal -Taiwan

Source: (World Trade Organization, 2009)

With only very limited years of

data available on only six countries, it would not make

sense to run a separate regression on the countries separately. Therefore, the change in FDI

will be generalized for all six countries.

3.2 Political Risk- Worldwide Governance Indicators

There are a lot of

different sources for data on political risk. Most academic data are qualitative

case studies on a specific country or region. Most quantitative sources are only commercially

available for companies. Fägersten's definition of political risk is used: 'political risk as a

potential harm to commercial activities caused by political action or arising from dysfunctional

political systems' (Fägersten, 2015). A database should thus contain at least some information

on the two factors covered in the definition: political action and dysfunctional political systems.

The source should include information on the level of

the political systems per country, as well

as information on chance of negative political action. There is one publicly available source

that meets this criteria: The Worldwide Governance Indicators (WGI) project (The World Bank

Group, 2015). This project measures the 'level of

governance' for all economies worldwide. It

includes the political system as well as the political action, as will be shown later on. The WGI

is a project that reports governance indicators for 215 economies over the last 20 years and is

funded by the Knowledge for Change Program of

the World Bank. The dataset is not based on

own research, but is based on information from 'a number of survey institutes, think tanks,

non-governmental organizations, international organizations, and private sector firms' (The

World Bank Group, 2015). Over 30 individual data sources make up the basis for the political

risk indicators. These sources consist of

surveys of

households and firms, commercial business

information providers, non-governmental organizations and public sector organizations. All

sources are made publicly available yearly on the WGI website. The construction of

the level

of governance score consists of three steps. First, the data is assigned to one of the six

indicators. The six dimensions are:

•

•

•

•

•

•

Voice and Accountability: Reflects perceptions of the extent to which a country's

citizens are able to participate in selecting their government, as well as freedom of

expression, freedom of

association, and a free media.

Political Stability and Absence of Violence/Terrorism: Reflects perceptions of

the

likelihood that the government will be destabilized or overthrown by unconstitutional

or violent means, including politically-motivated violence and terrorism.

Government Effectiveness: Reflects perceptions of

the quality of

public services, the

quality of

the civil service and the degree of

its independence from political pressures,

the quality of policy formulation and implementation, and the credibility of the

government's commitment to such policies.

Regulatory Quality: Reflects perceptions of

the ability of

the government to formulate

and implement sound policies and regulations that permit and promote private sector

development.

Rule of Law: Reflects perceptions of the extent to which agents have confidence in

and abide by the rules of

society, and in particular the quality of

contract enforcement,

property rights, the police, and the courts, as well as the likelihood of crime and

violence.

Control of Corruption: Reflects perceptions of

the extent to which public power is

exercised for private gain, including both petty and grand forms of

corruption, as well

as "capture" of

the state by elites and private interests. (The World Bank Group, 2015)

The second step is the rescaling of the individual source data, to make them suitable for weighting. The third step is the construction of the weighted average through an Unobserved Components Model (UCM). For a detailed description of the methodology behind the WGI index, the 2011 report by the creators Kaufman, Kraay and Mastruzzi (Kaufmann, Kraay, & Mastruzzi, 2011) should be consulted as this lies outside of the scope of my research.

Countries are rated yearly since 1999, with data going back to 1996. They are rated on each dimension on a scale from -2.5 to 2.5, in which a higher rating means higher level of governance, thus lower political risk. For rating the countries, the creators of the database assumed the same average worldwide level of governance for each year. This means that no conclusions can be made about the world's trends of increasing or increasing level of governance on a global scale. This has been done, so that conclusions can be made about the country's relative position in time and compared to other country's.

3.3 Cultural Distance - GLOBE

The GLOBE study, as introduced earlier in this thesis, will be used as measure for cultural distance. This study was done by 170 researchers from 62 countries, involving data from 17,300 middle-managers in 951 organizations from in the financial, food processing and telecommunications industries (House, Ranges, Javidan, Dorfman, & Gupta, 2004). The GLOBE study puts forward 10 culture clusters, as shown in figure 2. In the GLOBE researchers' own words, they 'used the results of previous empirical studies, geography, and religion, and, perhaps most importantly, historical accounts' for the construction for these 10 culture clusters (House, Ranges, Javidan, Dorfman, & Gupta, 2004). The clusters are named by their geographic location.

Figure 2

Culture t

luster- in the

GLOBE study

<

,,I __ ...,..,. __ ..,.,- /

--~

-.. ... '.::,..:.::_--

---

---

/'

'---~

-

--

_--

----

----

--~---

/

Source: (House, Hanges, Javidan, Dorfman, & Gupta, 2004)

Of the several reasons the authors describe for proposing the culture clusters, this thesis makes use of two: the practical and theoretical. Clustering the countries makes analysis easier and thus enables theories to be based upon them. The further clusters are apart in the circle, the bigger the cultural distance between them. For this thesis, the number of clusters that certain countries are apart in the GLOBE-circle will be the score on cultural distance. This means that cultural distance will be ranked on a scale from O to 5. For example, the cultural distance between countries in the Middle East and countries in Confucian Asia is 1, and the cultural distance between countries in the Middle East and countries in Nordic Europe is 4. Table 1 lists all the cultural distances between the culture clusters on the scale from O to 5.

Table 1

Cultural Dist.mee betw e1:n Culture Clusters A- MIDDLE EAST _{0 1 2 3 4 5 4 3 2 1} - - --- ---- B - CONFUSIAN ASIA _{1 0 1 2 3 4 5 4 3 2} -

--

--

-

--

- --

C- SOUTHERN ASIA _{2 1 0 1 2 3 4 5 4 3} - D - LATIN AMERICA _{3 2 1 0 1 2 3 4 5 4} -- --- ---- - - E - NORDIC EUROPE _{4 3 2 1 0 1 2 3 4 5}

--

-- - F -ANGLO _{5 4 3 2 1 0 1 2 3 4} - --- --- - - - G - GERMANIC EUROPE _{4 5 4 3 2 1 0 1 2 3} - H - LATIN EUROPE _{3 4 5 4 3 2 1 0 1 2} I - SUB-SAHARA AFRICA _{2 3 4 5 4 3 2 1 0 1} J - EASTERN EUROPE _{1 2 3 4 5 4 3 2 1 0} -- --- - - - --- -

This research is restricted by the GLOBE study, as it only uses 61 countries. Therefor,

the choice of

host countries is limited as discussed in the next paragraph. Moreover, the amount

of

home countries available for analysis is limited by the choice for the GLOBE study as well.

3.4 Choice of countries and other implications

The host countries that will be used for analysis in this thesis are: Egypt, Morocco, Greece,

Portugal, Hong Kong and Taiwan. The choice of

these countries was reliant on the choice of

the data sources and the availability of

data for the three variables, the independent as well as

the dependent. First, the countries should be part of one of

the culture clusters of

the GLOBE

study. This left 61 countries to be chosen from. Second, the country should have had a

considerate change in political risk according to the WGL This left eight out of

the 61 countries.

Third, there should be sufficient FDI-stock information available through UNCTAD for

significant analysis. This made the total number of countries suited for analysis six. These six

countries are spread over four different culture clusters, and consists of four states that have

seen an increase in political risk (Egypt, Morocco, Greece and Portugal) and two states that

have experienced significant decline in political risk (Hong Kong and Taiwan). Also, four out

of

six have relatively high level of

governance compared to two with a score below zero. Figure

3 shows the rise and decline of

the level of governance for these countries between 2002 and

2012.

2.00 1~0' •Q,~------, ~--- - __ --- 1.00 QJ u ,:: ta ,::

...

QJ > 0 _0.50 l!)

-

0 äi > QJ ...J 0.00 2002 2004 2005 2006 2007 2008 2009 2010 2011 2012 -0.50 -1.00

Egypt Greece -Hong Kong -Morocco -Portugal -Taiwan

Source: (The World Bank Group, 2015)

The used data and the choice of host countries has implications for the processing of

data on the home countries. Whilst the GLOBE study puts forwards cultural divides within

countries, the UNCTAD data on bilateral FDI is naturally based on the formal borders. This

meant that the information on four home countries had to be processed differently and will be

discussed hereafter.

The GLOBE study differentiates Germany East and Germany West. Nonetheless, both

are located in the Germanic Europe cluster, which means the difference is insignificant for this

thesis. Consequently, Germany East and Germany West are simply combined to Germany.

The GLOBE study distinguishes between the French (FR) and German (DE) cultures

within Switzerland, which are located in respectively the Latin Europe and Germanic Europe

clusters. This entails that a choice should be to which culture cluster Switzerland's FDI should

be subject to. According to the Swiss Federal Statistical Office (Federal Statistical Office,

2013), 63.5 percent of the population indicates German or Swiss German as their main

language. This opposed to French, which only 22.5 percent of

the population state as their main

language. Because German is the main language in the country, the FDI will be allocated to Switzerland-DE in the Germanic Europe cluster.

Another cultural divide that the GLOBE study makes is between Northern and Southern Korea, whilst the UNCTAD does not make that distinction. Whereas for Switzerland the cultures are relatively spread out across the country, the North and South of Korea are separated by a border. According to the Bank of Korea (South Korea's national banlc) in their report on North Korea's economic performance in 2014, they conclude that the North Korean GDP is only 2.3 percent of South Korea's GDP (The Bank Of Korea, 2015). The impact ofNorth Korea on the total Korean economy is thus marginal. Therefor, South Korean position in the Confucian Asia culture will be used in combination with Korean FDI statistics. The fact that there is a significant difference in political risk between North and South Korea is no issue here, as only the data on FDI is used in the analysis.

The last and most troublesome country to be discussed in this part is South Africa. The GLOBE clusters divide the country in South Africa White and South Africa Black. According to Census 2001, the black Africans make up the majority of the population with 79.2 percent. Coloured and white people only account for 8.9 percent each (SouthAfrica.info, 2015). The racial discussion is an important part of the country's history. The current president of South Africa, Jacob Zuma, discussed the racial impact on their economy in a speech in 2012: 'The ownership of the economy is still primarily in the hands of white males as it has always been' (Laing, 2012). This quote was based on an influential study by Trevor Chandler called Black Economic Interest on the Exchange. In

his study, he researched what part of

the Johannesburg

Stock Exchange (JSE) was owned by native black South Africans (Alternative Prosperity,

2015). He concludes that less than 10% of

the stock exchange is owned directly by the black

majority, which is according to himself

a 'good proxy for the market'. Most of

the market is

thus white-owned, and thus probably seen as a predominantly white culture. Therefor, for this

thesis we use South Africa White in the Anglo culture cluster as source of

South African FDI.

3.5 Control variable

Political risk is not the only factor affecting FDI, thus a thorough analysis should include one

or more control variables. A control variable is an independent variable that is thought to effect

the dependent variable, but is held constant in order to explore the relationship between other

variables.

In

the same article discussed before on political risk and FDI, the authors Meon and

Sek.kat argue that the 'literature emphasizing the role of local political risk does not seem to have yet taken into account the role of global factors (Méon & Sek.kat, 2012). They call the fact that global factors are not included in the analysis 'striking'. As could be seen in the first visual exploration of the FDI data of the six countries, there seems to be some correlation between them. In this case, one control variable is inserted into the data: world FDI stock. This . is the only control variable used in this thesis because other factors of influence on FDI are mainly on the national level of the host country, and can thus not be generalized for all countries in this research. This control variable is used to correct the data for the worldwide growth in trade due to globalization. For reasons already described, data from the UNCTAD database is used. The data shows that world FDI stock has risen sharply between 2001 and 2012 (Figure 4).

Figure 4 Worldwide FDJ-stock

$25,000 $20,000 .... m 0 $15,000 0 Vl ::::i C 0 $10,000 co $5,000 $0 2000 2002 2004 2006 2008 2010 2012 2014 Source: (UNCTAD, 2015) 3.6 Weighted variable

As stated before, the WGI database comprises six dimensions. These six variables construct a total score on the level of governance. These variables have distinct impacts on a manager's investment choices. WGI is a database on general level of governance in a country. As argued before, level of governance is a good indicator for political risk. But not all included dimensions

are of equal weight for investors. Off course, different types and fields of investment call for the consideration of different weights, but for this thesis that is not the point. This thesis focusses on the political risk investors, in general, include in their investment considerations. In general, all six dimension can be assumed to be part of this consideration as they are part of political risk, though some more heavily weighted than others. To include this weighting of investors into this study, all six dimensions will be made a weighted variable. The percentage each dimension is weighted (as shown in table 2) will be argued in the next paragraphs.

Table 2 Weighted Level of Governance

Voice and Accountability

Political Stability and Absence of Violence/Terrorism Government Effectiveness Regulatory Quality Rule of Law Control of Corruption 0.10 0.30 0.15 0.15 0.15 0.15

What is the reason the Voice and Accountability plus the Political Stability and Absence of Violence/ Terrorism are weighed differently from the other dimensions? The definition I used of political risk includes political actions and dysfunctional political systems. Whilst all dimensions can be expected to play a role within the boundaries of this definition, two of these dimensions will get weighted differently than the others: 'Political Stability and Absence of Violence/Terrorism' and 'Voice & Accountability'.

Businesses are, more than with the other dimensions, concerned with political stability in a country. As seen in historical accounts, 'business managers name Africa's political instability as a key obstacle to economic development' (Frynas, 1998). This has been proven to be true in a research conducted by Alesina, Özler, Roubini and Swagelon. They concluded that the economic development of 113 countries in the period between 1950 and 1982 was dependent on the height of political unrest. The results of this study showed that 'in countries and time periods with a high degree of political instability, [economic] growth is significantly lower than otherwise' (Alesina, Özler, Roubini, & Swagel, 1996). This means that 'political instability and absence of violence/ terrorism' should be weighed higher than average.

The 'government effectiveness', 'regulatory quality', 'rule of law' and 'control of corruption' are weighted as the mediocre dimensions. The effect of all these dimensions make

sense, and have been discussed extensively in literature (Busse & Hefeker, 2007) (Hausmann & Fernandez-Arias, 2000).

'Voice and accountability' is the dimension that has the lowest score in my weighting. There is no business research available that constitutes the effect of 'voice and accountability' on business decisions and performance. However, there is an indirect effect that is important for companies. It has been proven that increased voice and accountability can improve 'poverty outcomes for poor people' (Thirkell, Trepanier, & Earle, 2009), which can in tum increase business opportunities and positive outlooks. Thus, whilst voice and accountability is the least influential dimension, it would not make sense to completely ignore its effect.

3.7 Method

To test the hypotheses, previously discussed data will be combined in and consequently assessed with SPSS. Both previously stated hypotheses are portrayed schematically in Figure 5. The full line represents hypothesis 1, the dotted part represents hypothesis 2.

Figure 5 Hypotheses Cultural Distance I I I I I I I I

[ p:!~al ]---·---{

__

s_!o_

~_i

__

]

For political risk, the change in the previous three years are taken into account, because we assume effect of these years on the change in FDI in that specific year. This is done because a certain delay can be expected. When investing, the company's decision is influenced not only by the current political risk, but also by its trend. Moreover, a lot of companies have investment plans that are spread over a couple of years, which creates a delay in their reaction to political risk changes. Therefor, taking into account the previous three year seems reasonable. However, this choice remains highly subjective and, is in need of further research in the future, but this lies out of the scope of this research.

3. 7 1 Hypothesis 1

Hl:

Increased political risk in a particular country has negative impact on the country's inward foreign direct investment stock.

In this hypothesis we expect change in FDI stock to decrease when political risk increases. The change in the past three years in the country's political risk is expected to effect the change in the FDI stock of that same country. For this, the change in total FD I-stock is used as dependent variable. We control this effect for the change in worldwide FDI. Keep in mind that political risk is measured through the level of governance. A high score of change in the level of governance means a low score on the change in political risk.

This gives us the following equation, for which we expect b to be positive:

Equation 1:

t.TFDI

=a+

b (t.WFDI)

+ c

(t.PR012)

in which, for a specific year, t.TFDI is the change in a host country's total FDI-stock, t.WFDI is the change in the worldwide FD I-stock, and t.PR012 is the change in the level of governance in the host country in the current and past two years.

For analysing this hypothesis, a multiple linear regression was executed by means of SPSS.

3.7.2 Hypothesis 2

H2:

There is a moderating effect of cultural distance on the impact of a country's political risk on their FDI stock.

This hypothesis adds the cultural distance to the equation, used in testing Hl. This time, we expect the effect of political risk on the FDI to change more when the cultural distance between the two countries is bigger. Opposed to the country's total FDI-stock used in the first hypothesis, we use bilateral FDI, as discussed previously. Like with the first hypothesis, we

control for the effect of worldwide FDI. Again, keep in mind that a low score for PR indicates a high political risk.

This gives us the following equation, for which we expect b

to be positive:

Equation 2:

f).TFDI

=

a+

b (f).WFDI)

+

c

(f).PR0l2)

+

d ( f).WFDI * f).p

R012)

in which, for a specific year,

f).T FD!

is the change in a host country's total FD I-stock, f).WFDI

is the change in the worldwide FDI-stock, and

f).PR0l2

is the change in the level of

governance

in the host country in the current and past two years.

For analysing this hypothesis, again SPSS will be used. More specifically, the

PROCESS macro designed by Andrew F. Hayes will be used. According to the Hayes' website,

this is an 'add-on for SPSS [ .. ] for statistical mediation, moderation, and conditional process

analysis' (Hayes, 2015). This macro makes analyses including moderation in SPSS less

demanding. See figure

6

for a schematic view of

the moderation model in PROCESS.

Figure 6

PROCESS Model

I

Conceptual Model

Statistical Model

X M XM ey I , 1 ,, y

Source: (Hayes, 2015)

4. Results

In this part of the thesis, the assumptions and results of the analyses in SPSS will discussed

separately for hypothesis 1 and 2. In SPSS, the following labels represent the three variables

of

the equations:

t:.TFDI

=

CTFDIPCT t:.WFDI

=

CWFDIPCT

t:.PR012

= CPRt012

4.1 Hypothesis 1

In this chapter the results of hypothesis 1 will be discussed. First, the assumptions will be

tested. Consequently, the multiple regression will be analysed step-by-step.

4.1.1 Assumptions

The data used for the multiple regression should be checked for assumptions, to make sure the

model can be generalized beyond the sample used. The following five assumptions will be

tested using SPSS:

•

Multicollinearity

•

Serial correlation

•

Homogeneity of

variance

•

Linearity

•

Homoscedasticity

By checking the variance inflation factor (VIF) values, the assumption of multicollinearity is

tested. The VIF shows if there is a strong linear relationship between different independent

variables of

the model (Field, 2009). For the first hypothesis, this means the VIF value checks

if

there is a strong linear relationship between the change

in

FDI worldwide and the change in

FDI in a specific country. Scores above 10 are reason to be concerned with the assumption of

multicollinearity. SPSS gives us a VIF value of 1.005, which is far below the threshold of 10.

It can thus be concluded that for hypothesis

I

there is no problem with the assumption of m ultico llineari ty.

-By doing a Durbin-Watson test, the assumption of serial correlation is tested. The Durbin-Watson test tests the residuals of the estimated equation for serial correlation. Using · Field's (2009) rule of thumb, scores lower than one or higher than three are problematic. The Durbin-Watson test for the first hypothesis gives a score of 1.558. This score lies far within the boundaries stated to be problematic, which means there is no problem with the assumption of serial correlation.

By plotting the graph of ZRESID against ZPRED, the following assumptions are tested (Field, 2009):

•

•

•

Homogeneity of variance: the variances should be the same throughout the dataset . Linearity: the mean value of the outcome variable lies along a linear line .

Homoscedasticity: residuals at each level of the predictors should have the same variance.

ZRESID stands for the standardized residuals, ZPRED stands for for standardized predicted values. SPSS produces the following graph (figure 7):

Figure 7 7RESfD \-_ ?PRED Hl (SPSS)

Scatterplot

Dependent Variable: t:,. TFDIPCT

QI ~ 1- > "t:l QI .., u il

l:

o- -c QI N ~ ru 'tl C ru -1- .,; C 0 'ijj "' QI

...

áJ'I -2 a: 0 0 0 00 0 oO O 0 0 0 0 _{0 0} 0 ₀ 0 0 00 0 0 _{0 00} 0 0 00 o@ 0 0 0 ₀ 0 0 0 0 0 0 0 0 -2 -1 ' 0 ' ' l 2 ' I 3

Regression Standardized Residual

The graph shows a fairly random distribution of the plot, which indicates that there is no problem with the above stated assumptions. The plot does show a small curve, which could indicate a problem with linearity. This minor flaw does not lead to a significant problem with the assumption.

By looking at the histogram of the frequency of the dependent variable (Figure 8), the assumption of normal distribution is tested. The histogram shows a normal distribution, which means there is no problem with the assumption of normal distribution.

Figure 8 It 1s•t ,gram fl 1 (SPSS)

Histogram

Dependent Variable:

t:,.

TFDIPCT

15 10 Mean= -1.25E 15 Std. Dev.= 0.977 N = 4S -2 -1 0 l 2 3

Regression Standardized Residual

It can be concluded that there are no problems with the assumptions for hypothesis 1, as all assumptions are met.

4.1.2 Regression

To execute the multiple regression, two different blocks will be used. The use of blocks creates a hierarchy in the regression. This means that variables are entered in a certain order to check their influence. In the first block, only the worldwide FDI will be inserted as independent variable. In the second block, political risk will be added as another independent variable. This is done to check the added value of political risk in the regression. In both cases, FDI stock is the dependent variable. Cases are excluded listwise, which means that cases with at least one

missing value are completely left out of the analysis. The p-value gives the significance, which means that it measures the meaningfulness of the variable. For this research, a level of significance of 95% is used. This corresponds to a p-value that is smaller then 0.05.

First, SPSS produces a Descriptive output and a Correlation matrix. See the tables below (Table 3, 4):

Table 3 Descriptiv c Statistics Hl l SPSS)

Mean Std. Deviation N

--

---

CTFDIPCT

11,4129 19,72419 45

CFWDIPCT

10,6018 11,74644 45

CPRt012

-,0611 ,15939 45 Table 4 Correlations Hl (SPSS)

CTFDIPCT

CWFDIPCT

CPRt012

Pearson Correlation

CTFDIPCT

1.000 .690 .184

CWFDIPCT

.690 1.000 -.048

CPRt012

.184 -.048 1.000

Sig. (1-tailed)

CTFDIPCT

.000 .114

CWFDIPCT

.000 .377

CPRt012

.114 .377

N

CTFDIPCT

45 45 45

CWFDIPCT

45 45 45

CPRt012

45 45 45

The correlation matrix is useful for getting a rough feeling for the data. For the multiple regression testing Hl, the N is 45. The matrix also shows that the correlation between worldwide FDI and a country's FDI is .690 and is significant (p

=

.000), which is a significantly high correlation. The correlation between political risk and FDI is .184 and is not significant (p

=

.114). The correlation between political risk and worldwide FDI is low (-.048) and is not

significant (p= .3 77), thus the variables measure different things, as could logically be expected.

A step further than correlation is the multiple regression,

in

which an outcome variable is predicted by changes in several predictor variables. A regression gives us information about the causality. As discussed before, model 1 includes only the worldwide FDI as independent variable. Model 2 adds political risks as the second independent variable. See the model summary below (Table 5):

Table 5 \rfodel Summary fl I (SPSS)

Change Statistics Adjusted Std R

R R Error Square F Sig. F Model R Square Square Estimate Change Change Dfl Df2 Change

1 .690(a) .476 .463 14,44899 .476 38.993 43 .000

2 . 723(b) .523 .500 13,94770 .047 4.146 42 .048

a. Predictors: (Constant), CWFDIPCT

b. Predictors: (Constant), CWFDIPCT, CPRt012

The R-square in the model summary tells us that model 1 accounts for 47.6% of the variability in the dependent variable. When the second independent variable is added, 52.3% of the variability can be accounted for by the model. This means only a small increase in explanatory power when political risk is added to the model. Nonetheless, the added effect from model 2 is significant (p=.048). The Adjusted R-square for both models has a value that is fairly close to the R-square, which means that the model can be generalized.

The next part of our analysis involves the ANOV A, which tests whether the model is significantly better in predicting the outcome of the dependent variable than the 'best guess' method using the mean. See the ANOVA (analysis of variance) table from the SPSS output below:

Tah1e

6 "\ ".JU\ "\(al Hl (:-,PSS) Model 1 2 Sum of Squares df

Regression

8140.667

1

Residual

8944.258

43

Total

17117.926

44

Regression 8945.310

2

Residual

8170.616

42

Total

17117.926

44

Mean Square F

Sig.

8140.667

208.773

4472.655

194.538

38.993

22.996

.OOO(b)

.OOO(c)

a. Dependent Variable: CTFDIPCT b. Predictors: (Constant), CWFDIPCT

c. Perdictors: (Constant), CWFDIPCT, CPRt012

For model 1, the F-score is above one and is highly significant (p=.000). This means that the

improvement due to this model is greater than the inaccuracy within that same model. For

model 2 similar results are shown. The model is highly significant as well (p=.000) and has an

F-score above

1.

Above tables have shown that both models are able to improve our ability to predict the

outcome. The next table shows the specific coefficients with which that effect happens.

Table 7 Coefficients (a) If 1 1SPSS)

Unstandardized Standardized

Coefficients Coefficients

Model B Std. Error Beta t Sig.

1

(Constant)

-.864

2.916

-.296

0.769

CWFDIPCT

1.158

.185

.690

6.244

.000

2

(Constant)

.594

2.905

.204

.839

CWFDIPCT

1.176

.179

.700

6.559

.000

CPRt012

26.894

13.207

.217

2.036

.048

a. Dependent Variable: CTFDIPCT

The unstandardized coefficients give us the following equation for predicting the change in

FD I-stock:

b.TFDI

=

0.594

+

1.176

(b.WFDI)

+

26.894 (b.PR012)

These values indicate certain relationships between the variables. Both coefficients are

positive, which indicates that an increase in the independent variable leads to an increase in the

dependent variable. Thus, when worldwide FDI-stock and/or level of governance increases,

then the FDI-stock in a particular country increases. For this case, an increase of 1 percent in

worldwide FDI led to an increase of 1.176 percent in the country's FDI. This would not make

sense if

we would have analysed all countries in the world, but in this case it means that the six

countries analysed are influenced more heavily by the worldwide FDI than average (which

would be 1). For the political risk indicator level of governance, the results show that an

increase in level of

governance of 1 (which is a very large difference on a scale from -2.5 to

2.5) increases a country's FDI-stock with almost 27%.

4.2 Hypothesis 2

In this part, the results of

the second hypothesis will be discussed. First, the assumptions will

be tested. Consequently, the multiple regression will be analysed step-by-step.

4.2.1 Assumptions

The data used for the multiple regression should be checked for assumptions, to make sure the

model can be generalized beyond the sample used. These are the same test as carried out for

the first hypothesis, and will thus be not discussed in detail again.

The VIF value of

the data for the second hypothesis is is 1.000 and 1.007, which means

the assumption of

multicollinearity is met. The score on the Durbin-Watson test is 2.57, thus

the assumption of

serial correlation is met. SPSS gives figure

9

when plotting ZRESID against

ZPRED. From this graph, it can be concluded that the assumptions of

homogeneity of

variance,

linearity and homoscedasticity can be met. SPSS produces the histogram in figure

10

from the

data used. By looking at this histogram, the assumption of

normal distribution is met.

Figure 9 ZRl.:.SlD VS ZPRED H2 (SPSS)

Scatterplot Dependent Variable: b. FDI

(1,1 ::, ~ > 2 "C (1,1

-

u ~ l 15:: "C (1,1 N ~ 0

_...

111 '"C C 111 ,;:ï -l C 0

,:;;

"'

~ -2 01 ~ ·3 0 0 co 0 0 0 0 0 0 -20 _ 10 0 10 20 30