The effect of industrial relations determinants in the host country
on FDI inflow
Panel-data analysis of 27 OECD member countries for the period 2000-2013.
Ans De Maeyer
Supervisor: Dr. D. H. M. Akkermans
iii Abstract
This research investigates the effect of industrial relations determinants on foreign direct investment (FDI) inflow into host countries via a panel data analysis of 27 OECD member countries. More specifically, the effect of labor market protection, collective bargaining coverage and union power on FDI inflow is examined. In addition, it is tested whether the type of FDI the country is likely to attract has an influence on the effect of union power. To do this, secondary data is used from various databases. The results show that a higher level of union power has a significant positive effect the FDI inflow into host countries, regardless of the type of FDI the host country is likely to attract. A non-significant, positive effect can be found for labor protection. A higher level of collective bargaining coverage has a negative, non-significant effect.
v Contents
Abstract ... iii
List of figures ... vii
List of tables ... vii
List of abbreviations ... viii
1. Introduction ... 1
2. Literature review ... 2
2.1Foreign direct investments ... 2
2.1.1 Types of FDI ... 3
A) Vertical and horizontal FDI ... 3
B) FDI inflow and FDI outflow ... 3
C) Greenfield investment and merger and acquisition ... 4
2.1.2 Effects of FDI ... 4
A) Possible positive effects ... 4
B) Possible negative effects ... 5
2.1.3 Trends in FDI inflow ... 6
2.1.4 Determinants of FDI inflow ... 6
2.2 Industrial relations ... 7
2.2.1 Industrial relations determinants ... 7
A) Labor market protection ... 7
B) Union power ... 9
C) Collective bargaining coverage ... 10
D) Control variables ... 11
2.2.2 Conceptual model ... 14
3. Methodology and data ... 15
3.1 Sample selection ... 15
vi
3.2.1 Dependent variable ... 16
3.2.2 Independent variable ... 16
3.2.3 Interaction effect ... 17
3.2.4 Control variable ... 18
3.3 Statistical model and estimation method ... 19
3.4 Econometric Principles ... 20
3.4.1 Stationarity ... 20
3.4.2 Homoscedasticity ... 21
3.4.3 Residual autocorrelation ... 21
3.4.4 Normal distribution of error term ... 21
4 Results ... 23
4.1 Descriptive statistics ... 23
4.2 Correlation matrix and multicollinearity ... 24
4.3 Regressions ... 27
5. Discussion and conclusion ... 30
6. References ... 33
7. Appendices ... 38
Appendix 1: Overview of countries in sample ... 38
Appendix 2: Definition and source of variables ... 39
Appendix 3: Levin-Lin Chi test for unit root. ... 40
Appendix 4: Likelihood Ratio test (LR) for heteroscedasticity ... 41
Appendix 5: Woodridge test for autocorrelation ... 41
Appendix 6: Normal distribution error terms ... 42
Appendix 7: Fixed vs Random effect model ... 43
vii List of figures
Figure 1. Conceptual model 14
Figure 2. FDI inflow (% GDP) over time 45
List of tables
Table 1. Descriptive statistics 23
Table 2 Correlation coefficients 26
Table 3. Regression results 29
Table 4.OECD countries included in the sample 38
Table 5. Definition of variables 39
Table 6. Levin-Lin-Chu test for unit root 40
Table 7. Likelihood Ratio test (LR) for heteroscedasticity 41
Table 8. Woodridge test for autocorrelation 41
Table 9. Shapiro-Wilk W test for normal data 42
Table 10. Hausman test 43
viii List of abbreviations
CPI corruption perceptions index
FDI foreign direct investment GDP gross domestic product
ICTWSS database on institutional characteristics of trade unions, wage setting, state intervention and social pacts
ILO international labor organization IMF international monetary fund MNE multinational enterprise
OECD organization for economic co-operation and development OLS ordinary least squares
1 1. Introduction
The world is your backyard! The last decades, the world has become more and more globalized. Borders have opened for trade, people travel on a more regular base and distance has become a relative concept. Many firms aim at entering foreign markets, more money is spend on investments abroad and multinational enterprises (MNE) become bigger every year (UNCTAD, 2015).
This trend is reflected in the figures of foreign direct investment (FDI) flows. Apart from some fallbacks during economic difficult periods, an overall tendency of increase in FDI flows can be observed over the last decades. Driven by the possible positive spillovers of FDI inflow and encouraged by international organizations that stimulate economic globalization, local governments try to attract as much FDI inflow by creating a favorable investment policy (Charlton, 2003).
Creating an attractive investment climate and the right incentives for foreign investors is a delicate issue. Wrong use of incentive programs can cause the benefits of FDI inflows to disappear, or even create negative spillovers. Questions that arise in this context: Which
factors determine the attractiveness of a host country for FDI? and To which extent do country-specific factors affect the attractiveness of a host country for FDI inflows?
(Kleinknecht 1998; Charton, 2003).
Scholars have found evidence that one of the factors that can influence the attractiveness of the host country is the local industrial relations policy. As part of the institutional approach, it might be interesting to take a closer look at the relationship between these two concepts (Li & Ren, 2010; Radulescu & Robson, 2008; Ham & Kleiner, 2007).
The aim of this research is to examine the link between differences in industrial relations determinants of countries all over the world and their FDI inflow. Besides that, it will also be investigated whether the type of FDI a host country attracts, influences the way the determinants are perceived by the investing firms. By doing so, one can look whether there are certain industrial relations determinants of a host country which trigger or negatively affect FDI inflow into the host country.
To find an answer on what is stated above, the following research question can be formulated:
2
inflow to this country?. Ham and Kleiner (2007) executed similar research in the past for the
period 1985-2000. However, research on more recent data is missing. To fill this gap, this research will perform a panel data analysis using a sample of the 27 Organization for Economic Co-operation and Development (OECD) member countries for the period 2000-2013. This research adds an extra dimension by investigating the differences in effect of determinants based on the type of FDI a host country is likely to attract (Ham & Kleiner, 2007).
The research is structured as following: The next section provides a literature review in which foreign direct investment, industrial relations determinants and the link between these two concepts are discussed in detail. Based on the literature, three hypotheses are formulated and a conceptual model is provided. The methodology and data used in this research is discussed in section three. The analysis of the data and the results can be found in section four. Section five draws a conclusions and limitations of the research. Besides that, it also gives some recommendations for future research.
2. Literature review
To examine the impact of industrial relations determinants on FDI inflow it is essential to have a good understanding of both concepts. For this reason, the existing literature for industrial relations and foreign direct investments is reviewed in this section. Based on this literature review, the hypotheses of this research will be formulated.
2.1Foreign direct investments
Firms that engage in productive activities outside the country in which they are incorporated are called multinational enterprises (Dunning , 1977). A possible way for this type of firm to enter a market in a foreign country is via a foreign direct investment. This country is called the host country. In literature, the abbreviation FDI is often used. The OECD (2009) defines FDI as: “a cross-border investment by a resident entity in one economy with the objective of
3 2.1.1 Types of FDI
Foreign direct investment can be classified into different types based on the motive of the investing firm, the investment flow and the investment mode.
A) Vertical and horizontal FDI
A first way to make a distinction is by looking at the motive of the investing firm. According to Dunning and Lundan (2008), MNE’s can have four different motives for making a foreign investment e.g. resource seekers, market seekers, efficiency seekers and strategic asset seekers. Depending on the purpose of the investing firm, it will choose a country with different characteristics. Large differences exist in particular between a foreign direct investment executed by a resource seeking firm and market seeking firms (Beugelsdijk, Smeets , & Zwinkels, 2008; Dunning & Lundan, 2008).
One the one hand, foreign direct investments can be done by resource seeking firms. This type of FDI is called vertical FDI. Firms often engage in this type of FDI because of a lack of resources in the home country. These firms are in need of specific resources, which are of better quality and cheaper to acquire in a foreign market. Generally, MNE’s search for host countries where they can get access to three types of resources: cheap suppliers and labor forces, physical resources and thirdly technological capability (Dunning & Lundan, 2008; UNCTAD, 2015).
On the other hand, companies can do a horizontal foreign direct investment. This type of FDI is done by market seekers. It is an investment that aims at conquering or maintaining a foreign market. Very often, a company is already active in that particular market via e.g. export. Changes in host country policies might oblige this company to change its strategy from export to FDI. Market seeking behavior of companies can be explained by five factors: the possibility of market growth, anticipating on host government’s actions, becoming more familiar with local preferences, decrease in production or transaction costs and physical presence as part of the strategy of increased globalization (Dunning & Lundan, 2008; UNCTAD, 2015; Helpman, Melitz, & Yeaple, 2003).
B) FDI inflow and FDI outflow
4 The term FDI outflow, on the contrary, captures all direct investments made by an inhabitant of a specific country to a foreign economy. In this research, the focus will be on FDI inflows rather than outflows. The reason for this is that inward flows can be seen as an indication for the attractiveness of a specific economy for foreign investors (OECD, 2009).
C) Greenfield investment and merger and acquisition
When focusing on FDI inflows, scholars have distinguished two major types of FDI inflow modes: greenfield investments and mergers and acquisitions. Each of these two types has its own characteristics (Wang & Wong, 2009).
The main difference is that a merger and acquisition involves an acquisition of the existing assets of a firm in the host country. A greenfield investments, on the other hand, start from scratch and requires building new facilities. Firm specific factors of the investing firm, e.g. efficiency and incentives, as well as the host country’s market characteristics like competition in the local market will influence the entry mode decision of the investing company (Wang & Wong, 2009).
The type of FDI, greenfield or merger and acquisition, will influence the extent to which the investment will create economic growth in the host country. Wang and Wong (2009) found that a greenfield investment in general has the tendency to be more beneficial for the host country than a merger and acquisitions. This can be explained by the fact that a greenfield investment starts from scratch and brings new assets to the region, while this is not always the case for a merger and acquisition (Wang & Wong, 2009).
In this research, the effect of industrial relations determinants on FDI inflow into host countries in general will be examined. No distinction will be made between greenfield investments and mergers and acquisitions.
2.1.2 Effects of FDI
The previous part already indicated that FDI inflow will influence the host country’s economy. In what follows, the possible positive and negative effects of FDI inflow will be discussed in detail.
A) Possible positive effects
5 host country. Additionally, it is a way to bring new technological knowledge into the country. It can also create jobs, competition and lead to optimal asset allocation (UNCTAD, 2015). These possible positive spillover effects for the host country make local policymakers eager to attract as much FDI inflow into their country. However, one should be aware that FDI is not the magical trick to create growth. First of all, the positive spillover effects can occur but are not presumed. Spillovers will only take place if local firms in the host country are able and motivated to absorb the technology and skills provided by the foreign firms. Besides that, they might also negatively affect the local economy. These possible negative spillover effects will be discussed in the next paragraph (Blomström, Kokko, & Mucchielli, 2003; UNCTAD, 2015).
B) Possible negative effects
FDI can generate positive effects for a host country. Still, it is important to remain skeptical and be aware of the possible negative spillovers of FDI inflow for a country. Examples of possible negative effects are e.g. income inequality, lower local wages, increased interest rates as MNE finances investment in host country and aggressive competition (Aitken & Harrison, 1999; Lipsey & Sjöholm, 2005; Sylwester, 2005)
Local government policies might unintentionally facilitate these negative effects of FDI. Large competition exists between different countries and regions to attract FDI. To do so, each country or region tries to create the most attractive environment for foreign investors by providing massive incentive packages for investing firms. Examples of incentive instruments that are often used by governments are tax benefits, cash grants, loans and infrastructure subsidies (Charlton, 2003; Blomström, Kokko, & Mucchielli, 2003; UNCTAD, 2015).
When providing these incentive packages, local policymakers need to assure that these interventions do not cause the benefits of FDI to move from the host country to the investing MNE. The automobile industry in Brazil is a good example of how competition between regions caused local policymakers to go crazy with incentive packages (Blomström, Kokko, & Mucchielli, 2003; UNCTAD, 2015; Charlton, 2003).
6 2.1.3 Trends in FDI inflow
FDI inflows have been increasing steadily over the past twenty years. Over time, some fallbacks are noticed during the aftermath of the dotcom bubble at the beginning of the twenty first century and the financial crisis of 2008. Since the beginning of the latest century developed countries, which were initially the major receivers of FDI inflow, were overtaken by the developing countries (UNCTAD, 2015).
Figures for most recent years indicate that this trend might have been interrupted. While developing countries and transition economies are facing decreasing FDI inflows, developed countries are able to remain to have a stable inflow. This leads to developed countries receiving on average 55% of the total FDI inflow (UNCTAD, 2015).
The increasing FDI flows all over the world over time can be explained by possible positive effects for both the investing firm as the host country. Also international organizations e.g. International Monetary Fund (IMF) and the World Bank encourage local governments to pursue a policy of economic liberalization and international (Boockmann & Dreher, 2003). 2.1.4 Determinants of FDI inflow
In the previous section it was already mentioned that local governments try to increase the attractiveness of their region by creating a favorable investment climate. In order to do this, it is necessary to know which factors MNE’s take into consideration when choosing a location for their investment.
According to Ham and Kleiner (2007), the following factors will determine the investment location of the MNE. They provide the following equation 2.1 (Ham & Kleiner, 2007): FDI = f(cost of capital, labor costs(wage, industrial relations system),other factors). (2.1) The way FDI determinants influence the attractiveness of a host country might depend on the motive of the investing firm. Certain factors will be seen as more or less important, depending on whether the MNE engages in vertical or horizontal FDI (Ham & Kleiner, 2007).
7 2.2 Industrial relations
Industrial relations is a complex concept. Over time, scholars have come up with a large variety of definitions for industrial relations. Flanders (1965) defines industrial relations as a study of the institutions of job regulations. In this research, the definition of Salamon (2000) is used. He defines industrial relations as phenomena inside and outside the workplace which are concerned with determining and regulating the employment relationship (Flanders, 1965; Salamon, 2000).
Across the globe, different systems of industrial relations exist. Each of these systems has a long historical background. Some scholars believe convergence between the different industrial relations systems will take place over time due to technological determinism. Others, who don’t believe in this trend, claim that efficient institutions can be achieved via multiple, distinct ways or systems (Traxler, Blaschke, & Kittel, 1996; Whitley & Kristensen, 1996).
The definition of industrial relations, stated earlier, indicates that industrial relations is a concept that captures multiple aspects. Due to this, multiple variables need to be taken into consideration when investigating industrial relations systems. (Ham & Kleiner, 2007).
This approach of using multiple variables to investigate industrial relations has been used by many scholars in the past. Some reoccurring determinants are compensation cost for workers, work council, layoff restrictions, unionization, bargaining level, labor law index, labor, employment protection, involvement in policymaking…Only a small selection of these determinants will be used in this research due to limited availability of data (Cooke & Noble, 1998; Radulescu & Robson, 2008; Ham & Kleiner, 2007; Brechter, Brandl, & Meardi, 2012). 2.2.1 Industrial relations determinants
In this research the effect of the following industrial relations determinants on FDI inflow will be examined: labor market protection, union power and collective bargaining coverage. Based on the current literature dealing with these industrial relations determinants, three hypotheses will be formulated.
A) Labor market protection
8 Previous research indicates that the flexibility of the labor market in a host country does influence the investment decision of an MNE. Javorcik and Spatareanu (2005) investigated the relation between FDI inflow and differences in labor market rigidities of host and home countries for European countries for the period 1998-2001. They state that a more flexible labor market in the host country will positively affect the probability and volume of FDI inflows (Javorick & Spatareanu, 2005).
A plausible reason is the idea that the cost of doing business in a certain area decreases when rules and legislations to protect an employee in the region become more flexible. A higher level of flexibility will make this region more attractive for FDI inflow. This idea of positive relations between labor market flexibility and FDI inflows in host country is in line with the term race to the bottom, introduced by Cary in 1974 (Cary, 1974; Olney, 2013).
Cary (1974) describes this term as the trend to lower corporate standards by governments and local firms in order to attract new firms to a region. Across the world, this trend of loosening employment protection can be observed. Especially in the period where the financial crisis hit the world the hardest, many countries decreased the strictness of their regulations (OECD, 2013; Cary, 1974).
Even though less rigid labor protection is expected to positively influence the FDI inflow into the host country, it is also associated with the negative effects like lacking innovation. Kleinknecht (1998) argues that less rigid labor markets discourage innovation. Due to a high amount of low paid temporary workers, which has a higher chance to exist in markets with low levels of market protection, the effort of companies to search for ways to optimize productivity of employees via innovation might diminish (Kleinknecht, 1998).
As a consequence, it is expected that less rigid markets will be beneficial in the short run. In the long run, however, too flexible labor market might decrease innovation and growth and eventually influence employment and profit growth in a negative way (Kleinknecht, 1998).
9 Hypothesis 1: A higher level of labor market protection in a host country will decrease the attractiveness of this country for FDI inflow.
Another important determinant of industrial relations in a specific country is the way the social dialogue takes place. Social dialogue can be defined as: “All types of negotiation,
consultation or simply the exchange of information between representatives of governments, employers and workers on issues of common interest. It covers tripartite processes and institutions of social dialogue, such as social and economic councils; institutions, such as trade unions and employers’ organizations; and processes such as collective bargaining.”
(Hayter & Stoevska , 2011, p. 1)
Determinants which are often used to represent the social dialogue are trade union density and collective bargaining coverage. In this research, these two determinants will also be used and their relationship with FDI inflow into a host country will be examined (Hayter & Stoevska , 2011).
B) Union power
The first indicator for the social dialogue in a specific country is trade union density. This can be seen as a proxy for the union power in a specific country. Union power can affect the investment decision in two distinct ways, depending on the motive of the investing firm. On the one hand, it is expected that high union power will negatively affect the FDI inflows into countries in case of vertical FDI, when firms search for cheap resources. Radulescu and Robson (2008) state the following concerning this based on research of nineteen OECD countries: “The conventional wisdom on this issue is that a strong trade union presence will
tend to make a country a less attractive location for foreign direct investment, due to concern on the part of multinational enterprises that the rent-extraction activities of trade unions will tend to limit the profitability of their investments.” (Radulescu & Robson, 2008, p. 662).
10 On the other hand, high union power makes it possible for unions to put more pressure on firms and to have a higher influence on the way of working. Consequently, high union power will lead to wage rises, reduction of the wage inequality and increases the probability of industrial peace in a country. Overall, this can increase stability and purchasing power of the local market. This is important when firms do a horizontal FDI, which aims at conquering or maintaining a foreign market (Dunning & Lundan, 2008; Haskel & Slaughter, 2001).
These two contradicting points of view urge us to create a conditional hypothesis. This hypothesis will test the interaction of union power and vertical and horizontal FDI attracting countries. It will investigate whether the type of FDI a country is likely to attract influences the effect of union power on FDI inflow into a country (Brambor, Clark , & Golder, 2006; Ham & Kleiner, 2007).
As a result of this, the following conditional hypothesis can be formulated:
Hypothesis 2: A higher level of union power in a host country is associated with an increase in attractiveness for FDI inflow when the host country is a horizontal FDI attracting country (market seeking), but will lead to a decrease when the host country is a vertical FDI attracting country (resource seeking).
It is important to mention that this determinant should be analyzed with caution as trade union density does not fully reflect the influence of unions in a specific country. Countries like France have rather low levels of trade union density while collective bargaining is very important in this country. Eastern European countries, on the other hand, have really high trade union density. To present a realistic picture of the social dialogue in a specific country, the collective bargaining coverage needs to be considered (Hayter & Stoevska , 2011).
C) Collective bargaining coverage
Collective bargaining coverage is an indicator which provides information about the extent to which collective negations influence the terms of workers’ employment and whether the collective agreements rule the terms and conditions of all employees (Traxler, 1994).
Windmuller (1987) defines collective bargaining as: ”a process of decision making between
11 High collective bargaining coverage will decrease the freedom for MNE to implement the home country’s human resource management in a host country. This might make foreign investors reluctant to invest in countries with high collective bargaining coverage (Freeman & Medoff, 1984).
On the contrary, collective agreement can create industrial peace. It makes it possible for all parties involved to get a better understanding of the industry and the working conditions. This decreases the probability of strikes, misunderstanding and dissatisfaction among different parties (Wellington & Winter, 1969).
The question is whether these findings from Freeman and Medoff (1984) are still valid today in a world of increased globalization. Due to this, the third hypothesis is:
Hypothesis 3: Host countries with higher collective bargaining coverage will be less attractive as a host country for FDI.
High levels of trade union density do not necessary imply a high collective bargaining coverage. Often, high collective bargaining coverage rates are a result of the extension mechanism which exists in the specific country. The extension mechanism is the extension of collective bargaining condition to employees who are not union members and employers who are not affiliated (OECD, 1994).
D) Control variables
Equation 2.1 formulated by Ham and Kleiner (2007) indicates that the FDI inflow to a specific host country is not solely influenced by industrial relations. Other factors also play a role in the location decision of investing firms. In order to provide a realistic view of how industrial relations determinants influence FDI inflow, some control variables need to be included in the model (Ham & Kleiner, 2007).
12 - Openness to trade
The level of openness to trade in a country is expected to influence the investment decision of foreign investors. It can be seen as a measurement for trade restrictions. Generally, higher levels of trade openness are associated with lower trade protection and transaction costs in a host country. Due to this, it is expected that a higher level of openness to trade will positively affect the attractiveness of host countries for FDI inflow (Asiedu, 2002).
- Inflation
Inflation is the second control variable included in the model and can be seen as a proxy for macroeconomic stability in the host country. Schneider and Frey (1985) state that a high inflation rate can be seen as an indication of economic tension in the host country. Accordingly, foreign investors will be less willing to invest in host countries with high inflation levels as it creates uncertainty (Schneider & Frey, 1985; Alfaro, Chanda, Kalemli-Ozcan, & Sayek, 2004).
- Human capital
In the past, multiple scholars have stated that the quality of human capital in a country does affect the inflow of FDI. Human capital, the third control variable, covers all knowledge, attributes and skill competences that an individual possesses and that gives the possibility to create personal, social and economic well-being (Rodriguez & Loomis, 2007).
Educational level is often used as a proxy for human capital. A higher educational level in a country can be associated with a higher level of human capital. A well-educated population surges FDI inflow in a specific country and helps achieving economies of scale and scope. These economies of scale and scope will lower the marginal costs and increase the efficiency, and attractiveness of a host country for FDI inflow (Bhavan, Xu, & Zhong, 2011; Kinoshita & Campos, 2003).
On the contrary, having a highly educated laborer comes with higher wage. This will increase the labor cost of an investing firm. Due to this, the relationship between human capital and FDI inflow can be both negative and positive, depending on the motive of the firm (Noorbakhsh, Paloni, & Youssef, 2001).
13 - Market potential
High market potential is expected to have a positive influence on the attractiveness of a country for FDI inflows. Market potential, measured by the level of population growth, can be an indication of possible increase in the size of the future market where the products can be sold. Due to this, MNE might be interested in investing in these kinds of countries (Aziz & Makkawi, 2012).
Other scholars confirm this statement from Aziz & Makkawi (2012). Culem (1988) finds a positive relationship between market potential and investments for data of six industrialized countries for the period 1969-1986. A similar result is found for the investment behavior of French multinationals (Aziz & Makkawi, 2012; Chen & Moore , 2010; Culem, 1988; Al-Sadig, 2009).
- Corruption
The fifth control variable is corruption. It can be defined as :”Arrangement that involves a
private exchange between two parties (the 'demander' and the 'supplier'), which (1) has an influence on the allocation of resources either immediately or in the future, and (2) involves the use or abuse of public or collective responsibility for private ends.” (Macrae, 1982, p.
678).
Corruption can be further categorized. First of all, a distinction can be made between corruption at country-level (macro) and industry-level (micro). Most scholars focus on corruption at country-level. In this research, the focus will also be on country-level (Asiedu & Freeman, 2009; Kaufman & Vincente, 2011).
Corruption can also be categorized differently: legal and illegal corruption. Illegal corruption consists of all types of corruption that are in conflict with the laws the country’s legislation. Legal corruption, on the other hand, does not conflict the laws in a particular country and is often less obvious (Kaufman & Vincente, 2011).
14 hand, corruption might also increase the speed of decision making and reduce bureaucracy (Mauro, 1998; Al-Sadig, 2009).
Overall, it is expected that the negative effects of corruption, higher costs and increased uncertainty, will be higher than the possible positive effect of corruption.
- Time dummy financial crisis
Finally, a time dummy variable will be included to capture the effect of the financial crisis which started in 2008. This crisis has had a large impact on the global economy and investments. To see to which extend this event has influenced the FDI inflow into host countries, a time dummy has been included. A negative effect of this control variable is expected (Moran, 2012).
2.2.2 Conceptual model
Based on the hypotheses stated before, the following conceptual framework can be constructed:
15 3. Methodology and data
In the previous section, an overview of the literature has been provided. Based on this, hypotheses have been formulated and presented in a conceptual model. To test these hypotheses, an empirical research will be conducted. In this section, the methodology and data used in this research will be discussed in detail. First, the selection of the sample will be discussed. Next, the different variables will be operationalized. Furthermore, the statistical model will be discussed and robustness tests will be executed.
3.1 Sample selection
To test the different hypotheses, a sample of 27 OECD member countries is used for the period 2000-2013. The OECD consists of 34 mainly high income countries, which are spread all over the world. Out of this group of 34, 27 countries are included in the sample. Appendix one provides an overview of all countries included in the sample. The majority are European countries. Beside that, a few Asian and North-American countries are also part of the sample. In order to be included in the sample, two criteria must to be met. First of all, the country needs to engage in FDI and receive FDI flows. This is the case for the majority of countries around the world. The second and most important criterion is the availability of data concerning industrial relations determinants. Databases like ICTWSS database and OECDstat provide this information. Since the effect of industrial relations determinants is the core of this research, only countries with sufficient data concerning industrial relations can be included in the sample (Visser, 2015).
Unfortunately, these databases provide mainly very recent data for only a small selection of countries. After analyzing the availability of the required data, it becomes clear that most information can be found for OECD member countries. The focus on OECD member countries, which are mainly developed countries, can be seen as a limitation of this research since the literature part indicates that also developing countries have an import role in the FDI flows across the world (UNCTAD, 2015) .
16 industrial relations determinants and FDI inflow might have evolved. This creates the need for more research on this topic based on more recent data. Due to this, the period 2000-2013 is chosen for this research.
3.2 Variables
Secondary data is used in this research. Data is obtained from e.g. The World Bank, Barro-Lee dataset, Transparency International, OECDstat, ILOstat and UNCTADstat database. Below, the different variables used in this research are presented. A detailed overview with the definitions and data sources of the different variables can be found in appendix two. 3.2.1 Dependent variable
To measure the dependent variable, FDI inflow into the host country (FDI), the annual FDI inflow as a percentage of GDP is used. The same variable was used in the research of Ham and Kleiner (2007). Data is retrieved from UNCTAD. Figure two in appendix eight gives a graphical overview of the FDI inflows (% GDP) to the different countries in the sample over time (Ham & Kleiner, 2007; UNCTADstat, 2016).
3.2.2 Independent variable
To test the effect of the first industrial relations determinant, labor market protection (Labprot), the variable strictness of employment protection is used. In previous research, the level of employment protection has often been used as a proxy for labor market protection. Data for the period 2000-2013 is retrieved from OECDstat (OECDstat, 2016a; Javorick & Spatareanu, 2005).
To construct the variable strictness of employment protection, three aspects are taken into consideration. The first aspect is the procedural inconveniences that employees face when starting a dismissal process. Besides that, the notice period and severance pay is also taken into consideration. The third aspect is the difficulty of dismissal, determined by the circumstances in which it is possible to dismiss workers. The data has some missing values for Estonia, Luxembourg and Slovenia. The last value carried forward method is used to solve this issue. This method can be used since data for labor protection is relatively stable over the entire sample period for the different countries (ILO, 2015).
17 employees work as an independent contractor. This type of worker is often not covered by laws concerning labor protection. Taking this into consideration, one should be aware that the level of strictness of employment protection is not always fully reflects the reality (Van Peijpe, 1999).
The second independent variable, union power (Trunion) is measured via the trade union density in the host country. This is the ratio of wage and salary earners that are trade union members, divided by the total number of wage and salary earners. Data is extracted from OECDstat (OECDstat, 2016c).
The third main industrial relations determinant in this research is collective bargaining coverage (Colbar). This variable provides information concerning the total number of employees whose pay and/or conditions of employment are determined by one or more collective agreements. Data is extracted from ILOstat (ILOstat, 2016).
3.2.3 Interaction effect
Hypothesis two is a conditional hypothesis that looks at the interaction between union power and the type of FDI a host country attracts. To test this interaction effect, a distinction must be made between vertical and horizontal FDI attracting countries. This can be done by looking at the labor cost (Labcost) in the host country. To measure the labor cost of the different countries over time, the average annual wages per full-time and full-year equivalent employee in the total economy is used. Data for Slovenia has some missing values. In order to deal with this, the last value carried on method is used (OECDstat, 2016b).
Based on the labor cost a host country, the sample is divided in two groups by creating a dummy variable (DLabcost). Dunning and Lundan (2008) believe that MNE that engage in vertical FDI search for cheap resources. Low levels average annual wages in a country can indicate cheap labor forces. As a consequence of this, countries with low levels of annual wages will be categorized as vertical FDI attracting countries. These countries are given value zero (Dunning & Lundan, 2008).
18 The value 26 000 is the threshold point (mean of labor cost – one standard deviation). Countries with an average annual wage higher than 26 000 USD are given the value one and are considered to be horizontal FDI attracting countries. Countries with values below this threshold point are given value zero and are vertical FDI attracting countries. Based on this criterion, the following two groups of countries are created. Vertical FDI attracting countries are: Czech Republic, Estonia, Hungary, Portugal and Slovakia. All other countries in the sample are classified as horizontal FDI attracting countries. One can argue that this threshold point is rather arbitrary. For this reason, the interaction between trade union density and labor cost will also be tested, without using a dummy.
3.2.4 Control variable
As mentioned in the literature part, some control variables need to be included. Below, the data used for these variables are discussed in detail. The control variables in this research are openness to trade (Openness), market potential (Marpot), inflation (Infl), human capital (Human), corruption (Corr) and the time dummy (DTime).
A widely used measure for the openness to trade of a country (Openness) is the sum of exports and imports of goods and services as a share of GDP. This research also uses this measure. A second control variable, the market potential of the host country (Marpot), is measured via the annual population growth. To find the level of inflation (Infl), inflation as measured by the annual growth rate of GDP implicit deflator is used. Data for all three of these control variables is retrieved from The World Bank (Asiedu, 2002; The World Bank, 2016a; The World Bank, 2016c; The World Bank, 2016b).
Educational attainment can be used as a proxy for human capital (Human), the fourth control variable. The Barro-Lee database provides data for educational attainment. Unfortunately, this data has a five-year interval. In order to create a balanced dataset with data for each year and each country, the five-yearly data is interpolated (Barro & Lee, 2016).
To capture the corruption level in the different countries over time (Corr), data from the corruption perception index (CPI) is used. Higher levels indicate the perception of a lower level of corruption in the public sector in a specific country (Transparancy International, 2016).
19 2008. Years up till 2008 have the value of zero, while the period from 2008 onwards has the value of one.
3.3Statistical model and estimation method
This research is executed via a panel data analysis. In the past, many researchers have been using a panel data set to examine determinants of FDI flows. A panel data set can be defined as a data set that follows a given sample of units over time, and thus provides multiple observations on each individual in the sample (Vijayakumar, Sridharan, & Rao, 2010; Ham & Kleiner, 2007; Hsiao, 2014).
A major advantage of this type of analysis is that it increases the efficiency of the econometric estimates. Panel data analysis consists of a large number of data points. Due to this, there will be less collinearity between variables and a higher degree of freedom. Besides that, this type of data makes it possible to provide an answer to hypotheses which are impossible to test with other types of data sets. Especially dynamic effects can be captured using panel data (Hsiao, 2014).
Panel data regression can be expressed as followed:
Yit = α+βXit + ϵit (3.1) Y represents the dependent variable, FDI inflow (% GDP). The independent and control variables are represented by X. I and t indicate respectively the indexes of units and the indexes of a period. The constant term in this model is α. β represents the slope of the curve and ϵ captures the error term of the model (Verbeek, 2008).
For testing the proposed hypotheses, the following statistical models will be used:
Model 1: FDIit = α + β1 Opennessit + β2 Inflit + β3 Human it + β4 Marpotit + β5 Corrit + β6
DTimeit + ϵit (3.2)
Model 2: FDIit = α + β1 Opennessit + β2 Inflit + β3 Humanit + β4 Marpotit + β5 Corrit + β6 DTimeit + β7 Labprotit + β8 Colbarit + β9 Trunionit + ϵit (3.3) Model 3: FDIit = α + β1Opennessit + β2 Inflit + β3 Humanit + β4 Marpotit + β5 Corrit + β6 DTimeit + β7 Labprotit + β8 Colbarit + β9 Trunionit + β10 DLabcostit + β11 Trunionit x
20 Model 4: FDIit = α + β1Opennessit + β2 Inflit + β3 Humanit + β4 Marpotit + β5 Corrit + β6 DTimeit + β7 Labprotit + β8 Colbarit + β9 Trunionit + β10 Labcostit + β11 Trunionit x
Labcostit+ ϵit (3.5)
The initial model, model one, looks at the effect of the different control variables on FDI inflow. In the second model, the industrial relations determinants are added in order see how these determinants relate to FDI inflow. The effect of union power might vary depending on whether the country is more like to attract vertical FDI or horizontal FDI. In order to test this interaction effect, model three and four are constructed. Model three looks at the interaction between the variables trade union density and the dummy of labor cost. Model four uses the original values of labor cost, instead of the dummy variable, to investigate the interaction. This last model can be seen as robustness check of the results found in model three.
Different techniques can be used to analyze panel data. In order to know which type of regression one should use, the Hausman test and the Breusch-Pagan Lagrange multiplier test can be executed. These tests help to decide whether a random effect, fixed effect or normal OLS regression is most suitable for the specific model. The results of these tests indicate that a random effect should be run for model two and three. A fixed effect model is the best option for model one and four. Results for these tests can be found in appendix seven (Torres-Reyna, 2007).
3.4 Econometric Principles
Before the regression analysis can be carried out with the statistical program STATA, some tests need to be run. In order to provide reliable results, the assumptions regarding stationarity, homoscedasticity, autocorrelation and normality need to be met. To see whether this is the case, several tests are executed and discussed in this section. More detailed information about these tests can be found in the appendices.
3.4.1 Stationarity
For every variable, data is available for multiple years. For this type of data, one should check whether the data is stationary e.g. properties do not change over time. If this is not the case, variables are non-stationary. Non-stationary variables tend to generate unreliable regression coefficients and can create a spurious regression (Koop, 2006).
21 is the case for this dataset. The results of the Levin Li & Chin test can be found in appendix three.
The null hypothesis can be rejected for almost all variables, which indicates that these variables are stationary. Only in the case of labor protection unit root is present. To deal with this problem, the first difference of this variable can be used in the regression. After rerunning the Levin Len & Chi test for the first difference of this variable, the null hypothesis can be rejected and the variable is no longer non-stationary. Taking the first difference of this variable affects the true meaning of this variable. Due to this, the unadjusted version of the variable will be used instead of the first difference of Labor protection (d.Labprot).
3.4.2 Homoscedasticity
A second assumption is that the variance of the error terms is constant, homoscedastic. If this is not the case, heteroscedasticity occurs. This has implications for the regression as it can create invalid T-statistic values. To test this, STATA proposes to run a likelihood-ratio test for heteroscedasticity for all four models used in this research. The results in appendix four show that in all four models heteroscedasticity is present. To make sure that this does not bias the results, the robustness option is used when running the models in STATA (Koop, 2006). 3.4.3 Residual autocorrelation
Another possible pitfall is residual autocorrelation. A data set is auto-correlated if the error terms of different cross-sectional observation are correlated over time. Autocorrelation leads to less efficient results, and should be avoided. To test whether autocorrelation is present, the Wooldridge test for residual autocorrelation can be used. Results of this are provided in appendix five. In all four models, autocorrelation is present. Similar to the issue of heteroscedasticity, the robustness option will be used (Koop, 2006).
3.4.4 Normal distribution of error term
23 4 Results
In this section, the descriptive statistics, correlation matrix of the variables and the results of the different models are discussed and presented.
4.1 Descriptive statistics
The initial step is to generate the descriptive statistics. This information is presented in table one. It provides information about mean, median, maximum, minimum, standard deviation and amount of observations. The descriptive statistics show a balanced dataset with no missing values. For each variable 378 observations (N) can be found.
Table 1. Descriptive statistics
N Mean
Standard
deviation Minimun Maximum FDI (% GPD) 378 4.515 11.060 -60.314 141.485 Openness to trade 378 96.930 58.360 20.260 357.475 Inflation 378 2.320 2.260 -5.200 15.430 Human captial 378 38.400 12.300 9.520 69.750 Market potential 378 0.540 0.600 -1.700 2.890 Corruption 378 71.510 16.850 34.000 100.000 Union power 378 29.740 18.790 5.650 79.080
Coll. bargaining coverage 378 60.330 27.920 10.700 100.000
Labor protection 378 2.180 0.770 0.260 4.580
Year dummy 378 0.43 0.50 0.00 1.00
24 The mean of the dependent variable FDI (% GDP) is 4.515. This indicates that the average annual FDI inflow in the sample of countries is 4.51% of a country’s GDP. The standard deviation, minimum and maximum of this variable show that large differences exist in annual FDI inflow (% GDP) among the different countries in the sample. The highest FDI inflow (% GDP) can be found in Luxembourg. This country is a very popular destination for FDI, especially the banking and insurance sector, because of its favorable tax policy. The negative minimum indicates reverse investments or disinvestments.
The mean value of 29.74 of trade union density illustrates that, on average, almost 30% of the employees are trade union members. The maximum of 79.08% can be found in Sweden. Other Scandinavian countries e.g. Denmark, Norway and Finland have also high levels of trade union density. The lowest levels can be found in Estonia, France, Republic of Korea, United States and United Kingdom.
The collective bargaining coverage gives an indication of the percentage of employees that is covered by a collective agreement. This is on average 60%. The highest levels of collective bargaining coverage can be found in Austria, Belgium France, Slovenia and Sweden. The Republic of Korea, United States and United Kingdom have low levels of collective bargaining coverage.
The strictness of labor market protection is measured via an index. The highest level of strictness of labor market protection can be found in Portugal, Czech Republic, Estonia, Greece, Sweden and the Netherlands. Angelo-Saxon countries like United Kingdom, United States and Canada tend to have low levels of labor market protection.
Overall, high standard deviations and large differences between minimum and maximum can be found for the different variables in this research. This can partly be explained by the diversity of countries in the sample. Besides that, the data includes information about a period of thirteen years. During this time, the global economy has been affected by the financial crisis. A time dummy variable (DTime) for the economic crisis which started in 2008 is included to capture possible effects of this event.
4.2 Correlation matrix and multicollinearity
25 independent and dependent variable. To test this, one can look at the correlation matrix of all variables. To make sure that the estimations and predictive power of the models are not affected, correlation coefficients should not be higher than 0.7 (Dormann, et al., 2013; Koop, 2006).
The correlation between the different variables is shown in table two. First, correlation between the different variables is investigated. When focusing on the variable FDI, it becomes clear that only market potential and corruption are significantly positively correlated with FDI inflows. The positive correlation between the variable corruption and the dependent variable might seem surprising but can be explained by the fact that higher levels of CPI indicate less corruption (Transparancy International, 2016).
26 Table 2. Correlation coefficients
FDI Openness Infl Human Marpot Corr Trunion Colbar Labprot DTime DLabcost
Trunoin x DLabcost Labcost Trunion x Labcost FDI 1 Openness -.4659* 1 Infl 0.1626* 0.1234* 1 Human -0.0791 0.1063** 0.0617 1 Marpot 0.191* 0.159* 0.0298 -0.275* 1 Corr 0.05 -0.001 -0.087*** -0.2387* 0.3945* 1 Trunion -0.0667 0.0947*** 0.0292 -0.1524* 0.1087** 0.4921* 1 Colbar -0.0087 -0.0023 0.016 -0.1655* 0.0424 0.218* 0.5197* 1 Labprot -0.0007 0.1043** 0.0828 0.0312 -0.3372* -0.3229* 0.0558 0.5004* 1 DTime -0.0545 0.1096** -0.2773* 0.0764 0.0175 -0.029 -0.092*** -0.0545 -0.0627 1 Dlabcost -0.0388 -0.224* -0.2373* -0.3302* 0.463* 0.533* 0.3207* 0.2828* -0.3904* 0 1 Trunion x DLabcost 0.0434 0.0289 0.0475 -0.223* 0.2268* 0.6028* 0.9485* 0.5177* -0.094*** -0.0556 0.5843* 1 Labcost 0.1242** 0.0794 -0.2251* -0.3202* 0.6175* 0.7019* 0.2411* 0.1256* -0.4920* 0.1354*** 0.7752* 0.7752* 1 Trunion x Labcost 0.1223* 0.1825* -0.0283 -0.2275 0.2970* 0.6336* 0.9460* 0.4761* -0.0854*** -0.0117 0.4684* 0.9562* 0.5027* 1
27 4.3 Regressions
The results of all four models are presented in table three. The first model looks at the effect of the different control variables. Openness to trade has, as expected, a positive significant effect on FDI inflow into a host country. The results also show that a higher level of human capital has a negative and significant impact on FDI inflow. This can be explained by the fact that a high level of human capital, a skilled labor force, is associated with higher wages. Increasing levels of human capital will be seen as something negative when the foreign investor is in need for low-skilled laborers (Noorbakhsh, Paloni, & Youssef, 2001).
A rather surprising result is the negative significant effect of the variable corruption. As discussed before, higher levels of this variable indicate less corruption. This outcome might indicate that the possible positive effects of corruption, the ability to speed up the decision making process, might be more important than expected. All other control variables do not significantly influence FDI inflow into a host country. The overall R² (0.1768) indicates that the control variables have limited explanatory power.
The effect of the industrial relations determinants on FDI inflow is tested in model two, three and four. The initial step is to look at the effect of the variables collective bargaining coverage and labor market protection in order to see whether hypotheses one and three can be supported. This can be tested via model two. This model shows that labor market protection has a positive effect on FDI inflow. It indicates that higher levels of market protection lead to more FDI inflow into the host country. This does not meet the expectations, but is also not significant. Due to this, hypothesis one cannot be supported.
In line with what is stated in previous literature, a negative effect of higher collective bargaining coverage on FDI inflow can be observed. However, this result is without significance. Due to this, hypothesis three cannot be fully supported. The overall R² of model two is noticeably higher than the overall R² of model one. This indicates that, even though the hypothesis cannot be supported, adding industrial relations determinants to the model increases the explanatory power.
28 depending on the type of FDI the country is likely to attract. To test this conditional hypothesis, an interaction term is added in model three and four.
Results for these models indicate that, whenever dummy of labor cost is zero, trade union density will have a significant positive effect on the FDI inflow into a host country. This positive effect can also be observed in case dummy of labor cost is one, namely when host country is a horizontal FDI attracting country. The results show that an increase in union power in a host country has a significant positive effect on the FDI inflow into this country, regardless of the type of FDI the host country is likely to attract. The expected relationships stated in hypothesis two cannot be found. Due to this, hypothesis two cannot be supported.
29 Table 3. Regression results
Variable Model 1 Model 2 Model 3 Model 4
Constant 22.8576** -3.338 -9.5900 -7.7125 (0.022) (0.443) (0.172) (0.7) Openness 0.1519** 0.0835* 0.0904* 0.1327 (0.027) (0.000) (0.000) (0.018) Infl 85.5433 61.4063 61.9923 76.1934 (0.135) (0.165) (0.175) (0.174) Human -28.8740** -9.3000* -9.6640* -23.4545*** (0.039) (0.001) (0.002) (0.084) Marpot 0.9576 0.9986 0.8531 1.002 (0.728) (0.604) (0.635) (0.695) Corr -0.3391** -0.0303 0.0115 -0.3289*** (0.05) (0.334) (0.708) (0.08) DTime -0.5501 -0.3204 0.2489 -0.7822 (0.713) (0.875) (0.908) (0.479) Labprot 0.3959 -0.1018 2.8594 (0.638) (0.917) (0.289) Colbar -2.3784 -2.0203 -1.87255 (0.349) (0.481) (0.784) Trunion 1.5694 38.8181** 87.8721** (0.594) (0.035) (0.028) Labcost 0.009 (0.163) DLabcost 8.0926*** (0.075) Trunion X DLabcost 1 -34.472** (0.041) Trunion x labcost -0.0031 (0.219) No. of observations 378 378 378 378 No. of groups 27 27 27 27 Wald Chi2 5.83 656.60 746.76 9.48 Prob > Chi2 0.006 0.000 0.000 0.000 R-square within 0.0702 0.0538 0.0619 0.0960 between 0.5584 0.866 08621 0.1406 overall 0.1768 0.2586 0.2612 0.0617
30 5. Discussion and conclusion
FDI inflows have been increasing in the past decades and the possible positive effects for host countries are enormous. To attract FDI inflows and to benefit from possible spillover effects, local governments find it important to create a good investment climate. This should be done with caution to make sure that the incentive programs do not take away the possible benefits for the host country. Due to this, local governments want to know which factors do influence the inflow of FDI to their country. In this research, the aim was to look how industrial relations determinants influence the FDI inflow into host countries and to find an answer on the question: What is the effect of certain industrial relations determinants in the host country
on the FDI inflow to this country?
Providing an answer to this question can help policymakers and managers to know on which industrial relations factors one should focus. Previous studies have examined this topic, but there is a lack of research based on more recent data, especially for developed countries. This research adds an extra dimension by including the interacting between the type of FDI a country is likely to attract and the industrial relations determinant union power. To examine all of this, three hypotheses are tested for a sample of 27 OECD countries for the period 2000-2013.
Four models were run in order to find an answer to the three hypotheses. Model one, which includes all control variables, has low explanatory power. Higher levels of trade openness and lower levels of human capital do lead to a significant increase of FDI inflow into a host country. The significant negative effect of the variable corruption is unexpected. This indicates that positive effects of corruption, the ability to speed up the decision making process, might be more important than initially expected. All other control variables do not have a significant effect on FDI inflow into a host country.
31 for labor market protection in model two can be an indication that companies are aware of this possible negative effect in the long run (Kleinknecht, 1998).
Model two also provides information about the effect of collective bargaining coverage on FDI inflow into a host country. Collective bargaining coverage can be seen as one of the determinants of social dialogue in a country. It is expected that, whenever collective agreements play a more important role in the terms and conditions of employees, this would negatively affect the FDI inflow into a host country. Model three confirms this idea, but not significantly (Freeman & Medoff, 1984; Hayter & Stoevska , 2011).
Dunning and Lundan (2008) state that MNE’s can have varying motives to invest abroad. They distinguish four motives for FDI which can lead to a different type of FDI. Model three and four investigate whether the type the FDI the host country is likely to attract does influence the effect of the industrial relations determinant union power (Dunning & Lundan, 2008).
The results show that there is a difference in effect between vertical and horizontal FDI attracting countries. Though, it is not the effect which was initially expected. In the literature, it is stated that lower levels of trade union density are preferred in vertical FDI attracting countries as this decreases the rent-extracting activities of trade unions. Also, a lower level of union power is also likely to lead to lower wages. This is important for MNE’s that seek cheap resources.
The results in model three and four show a different effect. In vertical FDI attracting countries, an increase of the variable trade union density leads to an increase in FDI inflow. A possible explanation for this is the fact that high levels of trade unions, one of the determinants of social dialogue, can increase the probability of industrial peace. It might indicate that a stable local market is not only important for market seeking firms, but also for resource seeking firms.
32 The analysis shows that industrial relations determinants do affect FDI inflow into countries, even though the relationship can not always be significantly proven. Besides that, it becomes clear that the type of FDI the host country is likely to attract has only limited influence on the effect of the industrial relations determinant. A positive relation exists between trade union density and FDI in both types of countries. Only the strength of this effect is different.
Overall, this research shows that the idea that MNE’s are likely to have a preference for a less rigid industrial relations policy in the host country might be a too simplistic way of thinking. It seems that MNE’s are aware of the possible negative effects on the long term of loosened rules e.g. less innovation and less industrial peace. Because of this, MNE’s favor stricter rules some cases.
For local governments, this might indicate that loosening rules and regulations can be beneficial for some determinants, but should not be seen as the magical recipe to increase FDI inflow into a host country. As already mentioned in the literature part, this loosening of local industrial relations policies needs to be done with caution as it diminish the positive effects of FDI inflows for host countries. The findings of this research show that less rigid regulations are preferred, as long as it does not affect the industrial peace in the host country.
33 6. References
Aitken, B. J., & Harrison, A. E. (1999). Do domestic firms benefit from direct foreign investment? Evidence from Venezuela. American economic review, 605-618. Alfaro, L., Chanda, A., Kalemli-Ozcan, S., & Sayek, S. (2004). FDI and economic growth:
the role of local financial markets. Journal of international economics, 64(1), 89-112. Al-Sadig, A. (2009). Effects of Corruption on FDI inflows. The. Cato J, 29, 267.
Asiedu, E. (2002). On the determinants of foreign direct investment to developing countries: Is Africa different? World Development, 30(1), 107-199.
Asiedu, E., & Freeman, J. (2009). The Effect of Corruption on Investment Growth: Evidence from Firms in Latin America, Sub-Saharan Africa and Transition Countries. Review of
Development Economics, 13(2), 200-214.
Aziz, A., & Makkawi, B. (2012). Relationship between Foreign Direct Investment and Country Population. International Journal of Business and Management, 7(8), 63. Barro, & Lee. (2016). Data Sets on Long-Run Enrollment Ratios and Educational Attainment.
Retrieved May 3, 2016, from Barro-Lee: http://barrolee.com/
Beugelsdijk, S., Smeets , R., & Zwinkels, R. (2008). The impact of horizontal and vertical FDI on host's country economic growth. International Business Review, 17(4), 452-472.
Bhavan, T., Xu, C., & Zhong, C. (2011). Determinants and growth effect of FDI in South Asian economies: Evidence from a panel data analysis. International Business
Research, 4(1), 43.
Blomström, M., Kokko, A., & Mucchielli, J. L. (2003). The economics of foreign direct investment incentives. Springer Berlin Heidelberg, 37-60.
Boockmann , B., & Dreher, A. (2003). The contribution of the IMF and the World Bank to economic freedom. European Journal of Political Economy, 19(3), 633-649. Brambor, T., Clark , W. R., & Golder, M. (2006). Understanding interaction models:
Improving empirical analyses. Political analysis, 14(1), 63-82.
Brechter, B., Brandl, B., & Meardi, G. (2012). Sectors or countries? Typologies and levels of analysis in comparative industrial relations. European Journal of Industrial Relations. Cary, W. L. (1974). Federalism and corporate law: reflections upon Delaware. The Yale Law
Journal, 83(4), 663-705.
34 Chen, M. X., & Moore , M. 0. (2010). Location decision of heterogeneous multinational
firms. Journal of International Economics, 80(2), 188-199.
Chen, X., Ender, P. B., Mitchell, M., & Wells, C. (2016). Chapter 2 - Regression Diagnostics. Retrieved June 9, 2016, from Stata web books: regression with Stata:
http://www.ats.ucla.edu/stat/stata/webbooks/reg/
Cooke, P. (2001). Regional innovation systems, clusters, and the knowledge economy. Industrial and corporate change. 945-974.
Cooke, W. N., & Noble, D. S. (1998). Industrial relations systmens and US foreign direct investment abroad. British Journal of Industrial Relations, 36(4), 581-609.
Culem, C. G. (1988). The locational determinants of direct investments among industrialized countries. European economic review, 32(4), 885-904.
Dormann, C. F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., et al. (2013).
Collinearity: a review of methods to deal with it and simulation study evaluating their performance. Ecography, 36(1), 27-46.
Dunning , J. H. (1977). Trade, location of economic activity and the MNE: a search for an
eclectic approach. Palgrave Macmillan UK.
Dunning, J. H., & Lundan, S. M. (2008). Multinational enterprises and the global economy. Edward Publishing.
Flanders, A. D. (1965). Industrial relations: What is wrong with the system? An essay on its theory and future. Faber & Faber.
Freeman, R. B., & Medoff, J. L. (1984). What do unions do. Indus. & Lab. Rel. Rev, 244. Ham, H., & Kleiner, M. M. (2007). Do industrial relations institutions influence foreign direct
investment? Evidence from OECD nations. Industrial relations: A Journal of
Economy and Society, 46(2), 305-328.
Haskel, J., & Slaughter, M. J. (2001). Trade, technology and UK wage inequality. The
Economic Journal, 111(468), 163-187.
Hayter, S., & Stoevska , V. (2011). Social Dialogue Indicators International Statistical Inquiry 2008-09 Technical Brief. pg 1.
Helpman, E., Melitz, M. J., & Yeaple, S. R. (2003). Export versus FDI. National Bureau of
Economic Research.
Hsiao, C. (2014). Analysis of panel data. Cambridge university press.
ILO. (2015). Employment protection legislation: summary indicators in the area of
terminating regular contracts (individual dismissals). Geneva: International Labor
