An analysis of the relationship of FDI and trade
with economic growth in Central and
Eastern Europe
by
Tanja Begheijn
Advanced International Business Management & Marketing
University of Groningen 3018571
Newcastle University 160747737
Supervisors: Dr. Astarlioglu and Dr. Munro
December, 2017
Abstract
Several studies emphasize the importance of foreign direct investment (FDI) on economic growth. For example, research has found that FDI is stimulated by several factors, including trade, human capital, and domestic investment, and thereby promotes economic growth. However, prior research on Central and Eastern Europe (CEE) did not take any interaction into account. Therefore, this study analyses the relationship between FDI and economic growth in Central and Eastern Europe. Hereby, the interaction between FDI and trade is taken into account. This relationship is measured by using an OLS and a TSLS regression analysis.
The results suggest that, overall, FDI is negatively related to economic growth in Central and Eastern Europe. However, when FDI is coming from countries that are members of the Organisation for Economic Co-operation and Development (OECD), FDI is positively related to economic growth. After making a distinction between developing and developed CEE countries, more specified information has been found. This study found evidence that FDI is positively related with economic growth in developing CEE countries. On the other hand, results show that FDI is negatively related with economic growth in developed CEE countries.
Key words: Central and Eastern Europe, foreign direct investment, economic growth, trade,
Acknowledgements
This dissertation marks the end of my master studies at the University of Groningen and Newcastle University. After an intensive period of approximately five months of working on this dissertation on a daily basis, today is the day that I write these acknowledgements as the finishing touch on my dissertation. Writing this dissertation helped me developing my professional, but also my personal skills. Therefore, I would like to thank all people who contributed to this development.
First of all, I would like to thank my dissertation supervisors dr. Astarlioglu of the Economics and Business Faculty at the University of Groningen and dr. Munro of the Newcastle University Business School. Both of them were of great value during writing my dissertation. Whenever I encountered problems, they provided me with useful tools in order to continue moving in the right direction. Also, whenever I had questions regarding my dissertation, they provided me with substantiated answers.
Table of content
1. Introduction 10
2. Literature review 13
2.1 Economic growth 13
2.2 Foreign direct investment 14
2.3 Trade 17
2.4 Central and Eastern European countries 18
2.5 The relationship between FDI, trade, and economic growth in CEE countries 18
3.9.4 Outliers 35 3.9.5 Heteroscedasticity 36 3.9.6 Multicollinearity 38 3.9.7 Endogeneity 39 3.9 Robustness 40 4. Results 42 4.1 Descriptive statistics 42 4.2 Regression results 43
5. Discussion and limitations 54
6. Conclusions 60
References 61
Appendices 70
Appendix I. Test for normality 70
Appendix II. Breusch-Pagan test for heteroscedasticity 73
Appendix III. White test for heteroscedasticity 74
List of tables
Table 1. Overview variables 30
Table 2. Breusch-Pagan test for heteroscedasticity 37
Table 3. White test for heteroscedasticity 38
Table 4. Descriptive statistics 42
Table 5. OLS results H1 and H2 45
Table 6. TSLS results H1 and H2 46
Table 7. OLS results H3 - developing countries 48
Table 8. TSLS results H3 - developing countries 49
Table 9. OLS results H3 - developed countries 50
Table 10. TSLS results H3 - developed countries 51
List of figures
Figure 1. GDP growth in CEE countries 14
Figure 2. FDI inflow to CEE countries 16
List of abbreviations
CEE Central and Eastern Europe FDI Foreign Direct Investment GDP Gross Domestic Product MNE Multinational Enterprise
1. Introduction
Since the fall of Communism, there is an increasing foreign direct investment (FDI) inflow to Central and Eastern European (CEE) countries. This increase in FDI inflow, can be explained by trade liberalization policies. These policies have been essential for the privatization process during the transition period (Kornecki & Raghavan, 2008). Consequently, multinational enterprises (MNEs) from all over the world became increasingly attracted to these transition economies. The majority of CEE countries are transforming from government-owned economies to market-based economies. Moreover, the EU integration of these countries caused a strong FDI inflow due to liberalization of trade policies, the privatization of state-owned companies and the increase in open CEE markets (Koonstra, 2013). As soon as CEE countries opened up their markets for trade, FDI inflow started to increase (Kornecki & Raghavan, 2008). The reason behind this, is that international trade causes a more efficient production of goods and services because of the comparative advantages that come with outsourcing (Makki & Somwaru, 2004). This caused an increase in international trade with Central and Eastern Europe, and thereby an increase in FDI inflow.
& Somwaru, 2004; Zhang, 2001). Most of these studies state that FDI, trade, and economic growth are positively related and that trade complements FDI in this case. However, some studies found that not all countries economically benefit from FDI (Zhang, 2001; Ayanwale, 2007). This can be explained by theory of Makki and Somwaru (2017), who argue that the strength of the relationship of FDI and trade with economic growth differs across countries. Human capital, domestic investment, infrastructure, trade policies, and macroeconomic stability are all factors that influence this relationship. Because of the variation in this relationship across countries, it is an interesting observation that there are no extensive studies available that primarily focus on Central Eastern Europe. Kornecki and Raghavan (2008) studied the impact of inward FDI stock on GDP growth in this region before, but their sample consisted of only five countries and they used limited independent and control variables. Moreover, in their study they did not take the effect of trade on FDI in promoting economic growth into account. Also, their results only provide information until 2005. This lack of information regarding limits the ability to develop economic growth in Central and Eastern Europe.
Economic growth in Central and Eastern Europe is desired and therefore an interesting topic to research more in-depth. Since there is an increasing amount of FDI inflow and trade to CEE countries (Kornecki & Raghavan, 2008), it is interesting to find the relationship of FDI and trade with economic growth in these countries. Considering the lack of studies with a focus on examining the relationship of FDI and trade with economic growth in a CEE-context, this study aims to investigate this relationship. Therefore, the main research question is: “What is
2. Literature review
2.1 Economic growth
As mentioned in the introduction, MNEs from all over the world became increasingly attracted to the transition economies of Central and Eastern Europe after the fall of the Communism (Koonstra, 2013). During the Communism, CEE countries were centrally-led economies owned by the government. The fall of the Communism resulted in open markets, mass privatization of state-owned companies and liberalization of trade policies. This transformation of CEE markets, is the most important cause of the rise in economic growth (Kornecki & Raghavan, 2011). This is confirmed by Miernik (2016), who states that openness of market economies is positively related to economic growth. Together with the long history of industrialization and the relatively well-educated work force, CEE countries became an attractive investment option for firms (Stephan & Jindra, 2005). Consequently, from the 1990s, FDI inflow and trade started to increase (Kornecki & Raghavan, 2011). International trade caused a more efficient production of goods and services because of the comparative advantages for investors that come with outsourcing (Makki & Somwaru, 2004). The increase in international trade, resulted in an increase in FDI. This is discussed in the next chapter.
However, the impact of the financial crisis differed per country. Estonia, Latvia and Lithuania were the strongest affected, followed by Hungary and Romania (European Central Bank, 2010). Poland and Albania survived the crisis relatively good, and even kept showing economic growth during this period. From 2009, the economy in Central and Eastern Europe continued growing and is still growing today. The Financial Times even mentioned that nowhere in the world the growth rate expectations changed so fast and positively as in Central and Eastern Europe (Financial Times, 2017).
Figure 1. GDP growth in CEE countries
Source: World Development Indicators (2017)
2.2 Foreign direct investment
that involves significant equity stake and effective management decision power or ownership control of foreign enterprises’. FDI is also considered to ‘enclose other non-equity forms of co-operation that involve supply of tangible and intangible assets by a foreign enterprise to a domestic firm’ (De Mello, 1999, p. 135).
FDI inflow to Central and Eastern Europe has been crucial in the first stages of the transition period. As mentioned before, as soon as Central European countries started their transition from a centrally planned market economy to a market-based economy, FDI inflow started to increase. MNEs increasingly started to invest in Central and Eastern Europe in different ways and with different goals. For some investors market-seeking (horizontal FDI) was the most important goal of FDI, while others focused more on efficiency-seeking or resource-seeking (vertical FDI) (Miernik, 2016). Countries with a large domestic market (like Poland, Ukraine and Romania), or countries with a fast-growing economy (like Bulgaria or Croatia) were interesting countries for investors whose goal was market-seeking. Thereby, these countries attracted a lot of greenfield investments in the consumer goods sector (Sapienza, 2010). Investors whose goal was efficiency-seeking, started to invest in Hungary, Poland, Slovak Republic, Slovenia, and Czech Republic. These countries offered a relatively cheap, but well-educated workforce and good infrastructure (Miernik, 2016). The privatization of state-owned companies and the increase in open CEE markets resulted in a sale of state-state-owned enterprises. This attracted even more FDI from foreign investors in the form of mergers and acquisitions (Bacic, et al., 2005).
Figure 2. FDI inflow to CEE countries Source: World Development Indicators (2017)
2.3 Trade
In this study, trade is defined as the total amount of exports of goods and services and imports of goods and services. This represents the total value of all goods and other market services provided to the rest of the world, and received from the rest from the world. After the Communism, an increasing number of CEE countries opened up their markets for trade. Due to this increase in open markets and liberalization of trade policies, international trade increased (Koonstra, 2013). The reason behind this, is that international trade causes a more efficient production of goods and services because of the comparative advantages that come with outsourcing (Makki & Somwaru, 2004). Another factor that drives international trade with CEE countries, are the relatively low wages in this region. On the other hand, trade is also essential for Central and Eastern Europe itself. According to Schneider (2004), imports brings extra competitors and variation to domestic markets. This will benefit consumers. On the other hand, exports improve markets for domestic production, which will benefit businesses. Another reason for trade with Central and Eastern Europe, is the lack of certain goods. For CEE countries, the most important drivers for exports are garment, textiles, and electronics. On the other side, the most important drivers for imports are garment, chemicals and pharmaceuticals, and textiles (Seker, 2010).
hand, Russian Federation is one of the countries that is the least globally integrated in exporting and importing (Seker, 2010).
2.4 Central and Eastern European countries
This study focuses on Central and Eastern Europe. There are several reasons why it is important to investigate the relationship between FDI and trade with economic growth in this region. First, CEE countries have a high level of FDI (Koonstra, 2013). Secondly, because a lot of countries in Central and Eastern Europe still developing, there is a large desire for economic growth. Stephan and Jindra (2005) argue that particularly for countries in transition, which is the case for approximately 50 percent of all CEE countries, FDI plays an essential role. Moreover, the proximity of Central and Eastern Europe to large EU markets is an important advantage. Also, because of the relatively late start of FDI inflow to CEE countries, there is still a lot of space for economic growth and development. At last, looking at average GDP per capita, the top ten poorest countries of Europe only contain CEE countries (The World Bank, 2017). The countries on this list enclose Moldova, Ukraine, Kosovo, Albania, Bosnia and Herzegovina, Macedonia, Serbia, Montenegro, Bulgaria and Romania. This shows that there is a large need for economic growth in Central and Eastern Europe.
2.5 The relationship between FDI, trade, and economic growth in CEE countries
create competitive advantages for a firm, and thereby make the firm able to compete in markets. The amount of FDI inflow to CEE countries is rising. Recent studies show that CEE countries are a preferred choice for investors (Allen & Overy, 2006; Koonstra, 2013; Kornecki & Raghavan, 2008). Between 2000 and 2004 the share of FDI to CEE countries relative to the total FDI to Europe increased from 17% to 31%. Besides, approximately 70% of export in CEE countries is held by foreign-owned firms (Allen & Overy, 2006).
Zhang (2001) states that FDI inflow has a strong and positive effect on economic growth in China. On the other side, evidence from Nigeria shows that the impact of FDI on economic growth differs per sector (Ayanwale, 2007). Furthermore, there are studies that argue that several countries in Latin-America and South-East Asia do not economically benefit from FDI (Zhang, 2001). This can be explained by theory of Makki and Somwaru (2017), who argue that the strength of the relationship of FDI with economic grow differs across countries. This is confirmed by Zhang (2001). He states that the extent to which FDI leads to economic growth depends on country-specific characteristics. Human capital, domestic investment, infrastructure, trade policies and macroeconomic stability are all factors that influence this relationship. For example, when countries have a lack of human capital, their absorptive capacity is assumed to be smaller (Makki & Somwaru, 2004). This has a negative influence on economic growth. In order to test whether or not FDI has a positive relationship with economic growth in Central and Eastern Europe, the following hypotheses is formulated.
H1: Inward FDI has a positive relationship with economic growth in Central and Eastern Europe
confirmed by Makki and Somwaru (2004), who state that FDI and trade complement each other in promoting the economic growth. As mentioned in the introduction, factors that promote globalization are international trade and foreign direct investment. This is confirmed by Fontagne (2009), who states that FDI and trade are the main features for globalization. According to Dreher (2006), the process of globalization has effects on several purposes, among which economic growth is one of the most important ones. Moreover, the interaction between FDI and trade is considered being the core of the integration process of the world economy (Albulescu & Goyeau, 2014).
hypothesis, that focuses on whether this interaction between FDI and trade also has a positive relationship with economic growth in Central and Eastern Europe, is formulated.
H2: Inward FDI is positively moderated by trade in promoting economic growth in Central and Eastern Europe
Multiple studies state that FDI is an essential driver for economic growth in developing countries. However, the question arises whether or not there is a difference in the relationship of FDI with economic growth between developing and developed CEE countries. As mentioned before, the extent to which FDI leads to economic growth depends on country-specific characteristics (Zhang, 2001). According to Nair-Reichert and Weinhold (2001), the effect of FDI on economic growth are only positively related until a certain level. They argue that the positive effects of FDI on economic growth is not unconditional. Knowledge spillovers can ratify this discussion. Knowledge spillovers can be defined as “the process in which an inventor
gains knowledge from the research outcomes of other inventors and is able to improve her own research productivity with this knowledge without fully repaying the other inventors for the value of this learning” (Branstetter, 2006, p. 247).
& Weinhold, 2001). Because of knowledge spillovers, it is expected that inward FDI is of great benefit for developing countries (Branstetter, 2006). According to Marr (1997), FDI is particularly for developing countries very important. Unfortunately, the majority of developing countries do not receive as much FDI as they would like to, due to the uncertain prospects for investors (Koonstra, 2013). There are a lot of risks for investors when investing in developing countries; skill shortages, low technological capabilities, structural weaknesses of these economies, and inefficiencies of their small markets. However, developing countries can also be very attractive for investors because of the low cost of labour and the availability of natural resources (Marr, 1997). Because developed countries are less dependent on FDI inflow than developing countries, it is assumed that inward FDI to developing CEE countries has a greater positive effect on economic growth than FDI to developed CEE countries. Also, the possibility to gain new knowledge because of inward knowledge spillovers is larger for developing countries than for developed countries (Borensztein, et al., 1998). Secondly, the chance of outbound knowledge spillovers is significantly smaller, because of their lack of knowledge regarding technology, research and development, and manufacturing (Branstetter, 2006). Based on this information, the following hypothesis is formulated.
H3: Inward FDI to developing countries in Central and Eastern Europe has a greater relationship with economic growth than FDI inflow to developed countries in Central and Eastern Europe
particularly in less developed, transition economies in order to achieve economic growth. This is confirmed by Dabla-Norris et al. (2010), who state that FDI from emerging markets is a very meaningful source for developing countries. The main growth barrier for developing countries, is the shortage of financial resources, technology and skills. However, financial resources, technology and skills are essential factors in economic growth. Therefore, FDI inflow is needed in order to help developing economies grow. When there is a lack of FDI inflow, it would directly influence the need for external financing. Also, it would have a negative effect on investment and growth (Dabla-Norris, et al., 2010).
H4: Inward FDI from OECD countries has a positive relationship with economic growth in Central and Eastern Europe
Figure 3. Conceptual model
3. Research methods
In this chapter, the sample that is used for the empirical test of the research question and the hypotheses is introduced. Also, the dependent, independent, control, and instrumental variables are described. Thereafter, the statistical model and the estimated model are explained.
3.1 Sample
The objective of this study is to investigate the relationship between FDI, trade, and economic growth in Central and Eastern Europe. The sample consists of the following sixteen CEE countries: Albania, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Moldova, Poland, Romania, Russian Federation, Serbia, Slovak Republic, Slovenia, and Ukraine. The CEE countries Belarus, Bosnia and Herzegovina, Kosovo, Macedonia, and Montenegro are excluded from the study because of missing data regarding human capital. Consequently, this results in a sample bias.
developing CEE countries consists of Albania, Bulgaria, Croatia, Moldova, Romania, Russian Federation, Serbia, and Ukraine.
In hypothesis 4, the independent variable is FDI from OECD countries. The OECD countries are Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Latvia, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States (OECD, 2017). The total amount of FDI from these countries is considered for hypothesis 4.
3.2 Dependent variable
The dependent variable in this study is economic growth. There are several different methods in order to measure economic growth. In this study, the theory of The World Bank is used. According to The World Bank (2017), an economy’s growth is measured by the change in the volume of its output or in the real incomes of its citizens. The World Bank states that are three possibilities to measure economic growth: volume of GDP, real gross domestic income, and real gross national income (The World Bank, 2017). In this study, the per capita gross domestic product (GDP) growth rate is used, because this is the most commonly used method (Henderson, et al., 2012; Makki & Somwaru, 2004; Wang, 2009). Thereby, the per capita GDP growth rate is used as a proxy for economic growth.
The data for the dependent variable is obtained from the World Development Indicators (WDI) database from The World Bank (The World Bank, 2017). This database features all necessary data for this study regarding the dependent variable.
3.3 Independent variables
The independent variables in this study are FDI and OECD FDI. Prior studies have shown that these variables are positively related to economic growth (Marr, 1997; Allen & Overy, 2006; Makki & Somwaru, 2004; Zhang, 2001). FDI is expected to be positively related with economic growth because, as explained in the literature review, it is an important method to transfer knowledge, capital and technology.
Trade and Development (UNCTAD) database (2014). This database features FDI inflow per country between 2001 and 2012. Consequently, the hypothesis that uses OECD FDI as independent variable has a limited timeframe.
3.4 Moderator
A moderator is a variable that affects the strength or the direction of the relationship between the dependent and the independent variable (Baron, 1986). The moderator can be presented as an interaction between variables.
In this study, the variable trade is used as a moderator. As mentioned in the literature review, trade is expected to positively affect FDI and thereby economic growth. Trade is calculated as the total amount of exports and imports of goods and services. According to Makki and Somwaru (2004), the intensity of the relationship between FDI and economic growth depends on trade. By multiplying FDI with trade, interaction between these variables is taken into account. Because of the relationship between FDI and trade, it is important to take this interaction into account. With this method, biased results are reduced.
The data used for the moderator, is obtained from the WDI database of The World Bank (2017). This database features all necessary data for this study regarding this variable.
3.5 Control variables
Control variables are included in this study in order to increase the explanatory power of the model. The control variables in this study are human capital, domestic investments, tax rate, government consumption, and inflation rate. The last two control variables are used in order to check the robustness of the model. This is explained further in this chapter.
for human capital. This measure of human capital has proven to be the one most significantly related with economic growth (Barro & Lee, 1994; Wang, 2009). The data in order to measure human capital, is obtained from the Barro-Lee Educational Attainment Dataset (Barro-Lee, 2014). The control variables domestic capital investment, tax on income, profits and capital gains, and government consumption, are included in this study as proxies for institutions and infrastructure in CEE countries (Makki & Somwaru, 2004). At last, the control variable inflation rate is used as a measure of fiscal and monetary arrangements of CEE countries.
The data for all control variables, except human capital, is obtained from the WDI database of The World Bank (2017). This database features all necessary data for this study regarding the control variables.
3.6 Fixed-effect variables
Fixed-effect variables allow the analysis to control for unobservable fixed effects across countries and time (Zhuang, 2017). Fixed-effect variables are particularly used in cross-country analyses. The reason for this is that in a cross-country analysis, the sample equals the population (Green & Tukey, 1960). In this case, it is desired to control for unobservable fixed effects across time and countries. Therefore, the fixed-effect variables country and time are included in this study. The country variable consists of sixteen countries, while the time variable consists of four timeframes.
3.7 Instrumental variables
endogeneity problem. In this study, instrumental variables are used in order to address endogeneity. Instrumental variables must be uncorrelated with the error-term and correlated with the endogenous independent variable (Wooldrige, 2009). Therefore, good instruments are usually hard to obtain, because these requirements are hard to fulfil with other variables. Therefore, this study uses lagged values of FDI and trade as instrumental variables in order to address endogeneity. An extensive explanation is given further in this chapter.
Type of variable Name of variable Description
Dependent variables Economic growth Annual percentage growth rate of GDP at market prices based on constant local currency
Independent variable FDI Total FDI inflow to CEE
countries
Independent variable FDI OECD Total FDI from OECD
countries to CEE countries
Moderator Trade Interaction between FDI and
trade
Control variable Human capital Average years of male
secondary schooling Control variable Domestic investments Gross capital formation
Control variable Tax rate Tax on income, profits and
capital gains Control variable
(robustness)
Government consumption General government final consumption expenditure Control variable
(robustness)
Inflation rate The rate of price change in the economy as a whole Fixed-effect variable Country Indicates one of the sample
countries used in this study
Fixed-effect variable Time Indicates one of the time
periods used in this study Instrumental variable FDI (lagged) Total FDI inflow to CEE
countries (lagged with a one year period)
Instrumental variable Trade (lagged) Interaction between FDI and trade (lagged with a one year period)
To conclude, table 1 provides an overview of all variables used in this study. Also, a summarized description of the variables is given.
3.8 Statistical model
There are several different opinions on which variables should be used in order to measure economic growth. In this study, the model of Makki and Somwaru (2004) is used in order to test the relationship between FDI, trade and economic growth. This model is based on the endogenous growth theory of Balasubramanyam et al. (1996) and shows the idea that FDI and trade influence economic growth (Borensztein, et al., 1998). Hereby, the interaction between FDI and trade is into account. Also, control variables are included. The following equation is formed.
git = β0 + β1FDIit + β2TRDit + β3FDI*TRDit + β4Xit + γt + δi + ε
This formula is built from several independent and control variables that influence the dependent variable; g. g represents the per capita GDP growth rate, which is a proxy for economic growth. The independent variable in this model is FDI, foreign direct investment. TRD, the trade of goods and services, is the moderator. The interaction between FDI and trade is taken into account by multiplying FDI and TRD. X contains the control variables. The control variables in this model consist of HC, stock of human capital, K, domestic capital investments, and TX, tax on income, profits, and capital gains in the host country. For the robustness test, the control variables K and TX are replaced by GC, government consumption, and IRT, inflation rate. Furthermore, γt and δi are the country and time fixed-effect variables. At last, ε
3.9 Estimated model
The variables in this study are tested by an ordinary least squares (OLS) regression analysis. This is the most common regression analysis (De Vries & Huisman, 2007). According to Hutcheson (2011), an OLS regression is “a generalized linear modelling technique that may
be used to model a single response variable which has been recorded on at least an interval scale” (Moutinho & Hutcheson, 2011). OLS regression analyses assume that errors in the
dependent variable are uncorrelated with the independent variables. Because this study contains several independent variables and control variables, it is an OLS regression with multiple explanatory variables. This is called a multiple regression analysis (De Vries & Huisman, 2007). In order to be able to do a OLS regression analysis, the data for this study should comply with certain assumptions. These assumptions are: measurement, linearity, homoscedasticity, normality, outliers, multicollinearity, endogeneity and robustness (De Vries & Huisman, 2007). In the next subchapters, all assumptions are explained, tested, and solved if necessary.
In order to address endogeneity, the two-stage least squares (TSLS) regression analysis is used. This analysis is an extension of the OLS method and is used when the dependent variable’s error terms are related with the independent variables and when the model contains endogenous independent variables (Statistics Solutions, 2017). The TSLS method tests the instrumental variables and thereby solves the endogeneity problem. This problem is addressed further is this chapter.
3.9.1 Measurement
In order to do a proper regression analysis, all variables should be measured on interval- or ratio level (De Vries & Huisman, 2007).
Furthermore, the control variable human capital is measured in years, the control variables domestic capital investment and government consumption are measured as percentage of GDP, the variable inflation rate is measured as annual percentage, and the control variable tax on income, profits and capital gains is measured as percentage of revenue. The variables economic growth, FDI, trade of goods and services, domestic capital investment, inflation rate, tax on income, profits and capital gains, and government consumption are variables on interval level. The variable human capital is measured on ratio level. Therefore, it can be concluded that this assumption is met.
3.9.2 Linearity
In a multiple regression analysis, the dependent variable must be linear correlated with all independent variables together. If this assumption is met, the mean of the residuals is zero. If there is no linear correlation between the variables, the results may be biased. This may lead to incorrect conclusions regarding the hypotheses (De Vries & Huisman, 2007).
3.8.3 Normality
In a multiple regression analysis, the data must be normally distributed and have a constant standard deviation σ. If this assumption is not met, it leads to incorrect conclusions and incorrect confidence intervals (De Vries & Huisman, 2007). In order to detect whether the variables are normally distributed or not, the Q-Q plots are observed. Secondly, the Shapiro-Wilk test is used in order to analyse the exact results of the normality test. The Shapiro-Shapiro-Wilk test is the most commonly used test for normality when it regards small samples (Laerd Statistics, 2017). Therefore, this is the most appropriate measurement of normality to use in this study. If the p-value of a variable exceeds 0,05, the variable is considered normally distributed. The results showed that not all variables are normally distributed. Therefore, certain variables had to be transformed.
4 the variables became normally distributed by taking the cube root of OECD FDI, interaction between OECD FDI and trade, and the inflation rate.
After the transformations, all variables for each hypothesis are normally distributed except for two. For hypothesis 1, the control variable government consumption is not normally distributed. Also for hypothesis 4, government consumption is not normally distributed. This may be solved by removing extreme values (outliers) from the set. This is discussed in the next subchapter.
3.9.4 Outliers
Prior to the regression analysis, it is important to detect outliers in the data. An outlier is an observation that is located in an abnormal distance from the other values in the sample. By using a boxplot, outliers can be detected and eventually be deleted from the data (De Vries & Huisman, 2007). Outliers are often present in data of relatively small samples. In order to deal with outliers, the technique of winsorization is used. When an outlier is winsorized, the outlier is assigned a lower weight. This way the value will become closer to the other values in the set. Winsorizing an outlier, is preferred over removing the outlier from the set (Statistics How To, 2017).
removed because this transformed the variable from having a non-normal distribution into a normal distribution.
The normality test results after the transformations and removing outliers are specified in appendix I. After removing several outliers from the study, all variables for all hypothesis became normally distributed except for one. For hypothesis 1, the control variable government consumption is not normally distributed. This should be taken into account when interpreting the results. However, in essence only the independent variables should be normally distributed. Therefore, the assumption of normality is met.
3.9.5 Heteroscedasticity
Heteroscedasticity is also a problem that might occur in a study. Heteroscedasticity occurs when the variance of all residuals is not constant for every combination of values of the independent variables (De Vries & Huisman, 2007). This results in biased estimates of the standard errors, and thereby to incorrect outcomes and conclusions regarding the hypotheses, p-values, and confidence intervals. When heteroscedasticity occurs, more weight is given to observations with larger errors. Therefore, it is important that the data is homoscedastic. In order to do a proper OLS regression analysis, the variance of the errors should be constant (Statistics How To, 2017).
Breusch-Pagan test for heteroscedasticity are specified in table 2. The complete results are specified in appendix II. As shown in table 2, not all test rest results have a p-value of more than a significance level of 0,05. This shows that the null hypothesis that assumes that the data is heteroscedastic, should be accepted. The data used for hypothesis 1 and hypothesis 2 is considered heteroscedastic. However, the data used for the remaining hypotheses do show a p-value of more than a significance level of 0,05 and therefore, the data used for these hypotheses is considered homoscedastic.
Hypothesis F Significance
H1. Total FDI 1,587 0,011
H1. Total FDI (lagged) 3,059 0,003
H2. Interaction between FDI and trade
1,587 0,011
H3. Total FDI – developing countries
1,087 0,425
H3. Total FDI – developing countries (lagged)
1,521 0,212
H3. Total FDI – developed countries
1,648 0,160
H3. Total FDI – developed countries (lagged)
1,545 0,192
H4. OECD FDI 1,376 0,268
Table 2. Breusch-Pagan test for heteroscedasticity
Hypothesis F Significance
H1. Total FDI 0,876 0,422
H1. Total FDI (lagged) 0,594 0,556
H2. Interaction between FDI and trade
0,876 0,422
H3. Total FDI – developing countries
1,121 0,341
H3. Total FDI – developing countries (lagged)
0,789 0,465
H3. Total FDI – developed countries
0,540 0,589
H3. Total FDI – developed countries (lagged)
0,694 0,508
H4. OECD FDI 0,616 0,548
Table 3. White test for heteroscedasticity
Since the White test presents p-values of more than 0,05 for all data in each hypothesis, it can be assumed that the data is homoscedastic. Thereby, the OLS assumption regarding heteroscedasticity is satisfied.
3.9.6 Multicollinearity
makes a distinction between developed and developed CEE countries and thereby decreases both samples to eight countries, a VIF of higher than 10 is assumed to indicate multicollinearity. In appendix IV, the results of the multicollinearity tests are specified. The variance inflation factors of all variables, satisfy the need for being lower than 5. Even the VIF of the control variables do not show any values that exceed 5. This means that multicollinearity does not exist between the variables. Therefore, it can be assumed that the independent variables in the regression model are not closely related to each other. Consequently, the assumption regarding multicollinearity is satisfied.
3.9.7 Endogeneity
Endogeneity problems occur when the effect of an independent variable on the dependent variable is overestimated due to wrong correlation assumptions regarding the error-term (Makki & Somwaru, 2004). Endogeneity has three main causes, which are omitted variables, reversed causality, or measurement errors (Dranove, 2012). In this study, only the problem of reversed causality is addressed. When reversed causality occurs, X and Y are associated in another way than expected. Instead of X causing a change in Y, it is the other way around (Statistics How To, 2016).
Miernik, 2016; Tintin, 2012). According to Juma (2012), it takes some time for the effect of FDI accumulation to be felt on economic growth. She states that time lags of one year are needed in order to get reliable results when measuring the relationship between FDI and economic growth. Therefore, the endogeneity problem is addressed by using instrumental variables.
In this study, lagged values of FDI and trade are used as instruments in a two-stage least square (TSLS) method. The TSLS regression analysis is an extension of the OLS method used in this study and is used when the error of the dependent variable is correlated with the independent variables (Statistics Solutions, 2017). The instrumental variables and the TSLS regression analyses are included in the study in order to check whether estimates from models that correct for endogeneity show different results from models that do not correct for endogeneity. Therefore, the variables FDI and trade used in hypothesis 1, 2, and 3 are first tested with t=0. Thereafter, the same variables are tested with a time lag of t-1. Unfortunately, it was not possible to address endogeneity for the variables used for hypothesis 4 because of missing data regarding lagged values of OECD FDI.
3.9 Robustness
The robustness test focuses on the strength of a statistical model and the procedures of a certain statistical analysis (Taylor, 2017). Robustness tests are frequently used to determine how certain empirical coefficient estimates are. Robustness is tested by adding and removing regressors. When coefficient estimates do not change signs, and do not significantly change because of such modifications, it is assumed that estimates are robust (Koonstra, 2013).
4. Results
4.1 Descriptive statistics
In order to have a first look at the data of the variables, the descriptive statistics of all variables are observed. Table 4 provides an overview of the descriptive statistics of all variables used in this study. This table contains the raw data that is not transformed or winsorized. It is important to analyse the descriptive statistics in order to detect possible mistakes and report missing values in the study.
Variable Observations Minimum Maximum Mean Standard deviation Economic growth 64 -2,41 8,21 3,167 2,44 FDI 64 0,91 19,27 4,81 3,34 FDI (lagged) 64 0,91 23,99 4,80 3,90 OECD FDI 31 0,15 16,87 4,86 3,93 Trade 64 38,69 179,64 101,69 33,23 Trade (lagged) 64 38,50 173,55 99,58 32,84 Human capital 64 2,26 6,56 4,04 0,84 Domestic investments 64 13,24 38,18 24,87 4,88 Tax rate 61 1,82 28,18 14,58 6,29 Government consumption 64 9,81 24,04 18,47 2,78 Inflation rate 64 0,61 220,33 12,74 28,71
Table 4. Descriptive statistics
2015. In order to solve this problem, the extrapolation technique was used. Extrapolation is a technique in order to make assumptions about the near future based on known data (Statistics How To, 2013). This technique may only be used if the data is expected to be linear. Therefore, this technique was not used by the other variables with missing values.
Another problem that is solved, is the presence of outliers. As mentioned in chapter 3, outliers are detected by analysing the boxplot of all variables. Table 4 shows some outliers that were deleted from this study in order to make the data normally distributed, and thereby more convenient to analyse. The maximum of human capital was deleted and winsorized to 6,06. This value refers to Moldova in the period 1996-2000. Furthermore, the maximum of inflation rate was winsorized to a value of 64,16. This outlier was a very extreme outlier for Bulgaria in the period 1996-2000. The winsorizing technique was explained in chapter 3. It is important to take the detected problems into account when analysing the results from this study.
4.2 Regression results
In this chapter, the variables of this study are tested by the OLS regression analysis. The results show whether or not the hypotheses can be accepted or should be rejected. Results are considered significant when the value is lower than 0,10. A distinction is made between a p-value of lower than 0,10 (*), lower than 0,05 (**), and lower than 0,001 (***). This shows significance levels of 90 percent, 95 percent, and 99 percent.
with economic growth in model 2 (b = -0,127, p < 0,10). This implies that when FDI increases with 1 percent, economic growth decreases with 12,7 percent. FDI also shows a negative relationship in the other models. However, these relationships are not significant. This negative relationship is in contrary to the literature used in this study. Trade shows a negative relationship with economic growth in model 2, but a positive relationship with economic growth in model 3 and 4. These values are not significant and therefore, conclusions cannot be drawn. The interaction variable shows a negative relationship with economic growth. This would suggest that trade negatively affects FDI in promoting economic growth. However, the values are not significant and therefore, these results are not useful.
The results of the control variables are also analysed. These results show that human capital is negatively related with economic growth. However, these values are not significant. Domestic investment shows positive and significant values in model 1 (b = 0,157, p < 0,05) and in model 3 (b = 0,178, p < 0,001), which suggests that domestic investments are positively related with economic growth. At last, the variable tax rate shows different significant outcomes. In model 1, tax rate shows a significant and positive relationship (b = 0,090, p < 0,05). On the other hand, model 3 shows a negative significant relationship of trade with economic growth (b = -0,088, p < 0,10).
The R squares show that the control variables explain 8,3 percent of the model, the independent variables explain 14,1 percent, and the independent variables and the control variables together explain 20,6 percent. This leaves 79,4 percent of the model unexplained.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 3,523* 1,256 1,743 0,925 FDI -0,127* -0,157 -0,228 Trade -0,003 0,006 0,000 Interaction -0,001 -0,001 -0,001 Human capital -0,249 -0,551 -0,439 Domestic investments 0,157** 0,178*** Tax rate 0,090* -0,088* Government consumption 0,059* Inflation rate -0,019 System R2 0,083 0,141 0,206 0,197 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 5. OLS results H1 and H2
The remaining significant variables, all show the same signs. An exception is the control variable tax rate. This variable does not show a negative relationship with economic growth anymore. Model 2 (b = 0,089, p < 0,10) and model 3 (b = 0,087, p < 0,10), both show that tax rate is positively related with economic growth. Because signs of variables in the TSLS model changed in comparison to the OLS model, this alternative estimation method does not confirm the assumption that the results are robust.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 1,066 0,947 1,048 FDI (lagged) -0,158* -0,132* -0,337 Trade (lagged) 0,005 0,003 -0,003 Interaction (lagged) 0,001 0,001 0,001 Human capital -0,259 -0,482 -0,363 Domestic investments 0,158*** 0,217** Tax rate 0,089* 0,087* Government consumption 0,054 Inflation rate -0,020* System R2 0,083 0,155 0,223 0,221 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 6. TSLS results H1 and H2
With the results from the OLS and the TSLS regression analyses, hypothesis 1 has to be rejected. Results show that, overall, FDI has a negative relationship with economic growth, and not the expected positive relationship. The results also show that hypothesis 2 has to be rejected. Results regarding the interaction between trade and economic growth are not significant and therefore not useful to draw conclusions regarding the hypothesis. Consequently, it cannot be confirmed that FDI inflow is positively moderated by trade in promoting economic growth.
the influence of FDI on economic growth to developing and developed CEE countries should be measured separately. First, the regression results of the relationship between FDI and economic growth in developing CEE countries are analysed. Only model 3 and 4 show several significant results for the independent variable. Therefore, the focus will be on these models. As expected, FDI shows significant and strongly positive results in model 3 (b = 0,183, p < 0,10) and 4 (b = 0,091, p < 0,10). Model 2 shows a negative, but insignificant value. Therefore, it can be assumed that an increase in FDI to developing CEE countries, results in an increase in economic growth in Central and Eastern Europe. The variable trade shows negative values in all models, but is only significant in model 3 (b = -0,110, p < 0,10) and model 4 (b = -0,118, p < 0,10). This suggests that trade has a negative relationship with economic growth. This is in contrary to literature. However, the interaction between FDI and trade shows a positive significant relationship in model 2.2 (b = 0,027, p < 0,10) and 2.3 (b = 0,029, p < 0,10). This reveals that trade positively moderates FDI in promoting economic growth in developing CEE countries.
The results of the control variables are also analysed. The values regarding human capital are all insignificant and show positive, as well as negative values. Because the values are not significant, these values have to be neglected. Domestic investment is significant and positively related with economic growth according to model 1 (b = 0,205, p < 0,01) and model 3 (b = 0,775, p < 0,05). The control variable tax rate does not show significant values.
The model does not significantly differ from the other models regarding signs and coefficients. All variables show the same signs and similar results.
The R squares show that the control variables explain 8,5 percent of the model, the independent variables explain 16,2 percent, and the independent variables and the control variables together explain 29,9 percent. This leaves 70,1 percent of the model unexplained.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 6,078** 0,902 -10,425 -8,621 FDI -0,012 0,183* 0,091* Trade -0,015 -0,110* -0,118* Interaction 0,026 0,027* 0,029* Human capital -0,256 0,601 Domestic investments 0,205* 0,775** Tax rate -0,075 -0,018 Government consumption 0,112 Inflation rate -0,010 System R2 0,085 0,162 0,299 0,270 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 7. OLS results H3 - developing countries
model that does not address for endogeneity. This alternative estimation method also confirms that the results are robust.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 6,078** 0,454 -8,861 -10,312 FDI (lagged) -0,052 0,156* 2,348 Trade (lagged) -0,008 -0,101 -0,105 Interaction (lagged) 0,030 0,030* 0,032* Human capital -0,347 0,662 0,572 Domestic investments 0,168 0,759** Tax rate -0,012 -0,033 Government consumption 0,121 Inflation rate -0,009 System R2 0,085 0,191 0,275 0,261 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 8. TSLS results H3 - developing countries
Secondly, the regression results of the relationship between FDI and economic growth in developed CEE countries are analysed. These results are shown in table 9 and reveal important differences compared to the results for developing CEE countries. In contrary to the relationship of FDI on economic growth in developing countries, FDI shows a significant negative relationship with economic growth in developed countries. All models confirm this result. Model 2 (b = -0,201, p < 0,05), model 3 (b = -2,528, p < 0,10), and model 4 (b = -2,092, p < 0,10) all show significant results regarding the relationship between FDI and economic growth. In two out of three models, trade shows a negative relationship with economic growth. However, these values are not significant and therefore not acceptable. The interaction variable shows positive values. This suggests that FDI is positively moderated by trade. However, these values are not significant. Therefore, conclusions cannot be drawn.
show significant and positive values in model 1 (b = 0,118, p < 0,10) and 3 (b = 0,164, p < 0,10). This suggests that an increase in domestic investment, will increase economic growth in developed CEE countries. Tax rate shows a significant and positive relationship with economic growth in model 1 (b = 0,123, p < 0,10) and 3 (b = 0,118, p < 0,10).
Model 4 shows whether the results are robust. In this model, the control variables domestic investments and tax rate are replaced by the control variables government consumption and inflation rate. The control variable government consumption only shows insignificant values. Therefore, conclusions cannot be drawn. At last, inflation rate shows a significant negative value (b = -0,214, p < 0,10). This shows that an increase in inflation rate results in a decrease of economic growth. It can be assumed that the results are robust. Model 4 shows similar results for the independent variable as the other models.
The R squares show that the control variables explain 9,9 percent of the model, the independent variables explain 39,6 percent, and the independent variables and the control variables together explain 46 percent. This leaves 54 percent of the model unexplained.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
In table 10, the results of the TSLS model are shown. This model addresses endogeneity by using lagged values of FDI and trade. The TSLS models show similar results as those obtained by the OLS regression analysis. However, the TSLS model shows more significant values than the regular OLS model. An important difference in this model, that uses lagged values of FDI and trade, trade shows a significant and negative relationship with economic growth (b = -0,046, p < 0,10). This is in contrary to our expectations expressed in the literature review. However, the interaction variable is significant and positive in model 2 (b = 0,010, p < 0,10) and model 3 (b = 0,013, p < 0,05). This shows that FDI is positively moderated by trade in promoting economic growth in developed CEE countries. Another difference is that the control variable human capital now shows two significant positive values; in model 3 (b = 1,361, p < 0,10) and 4 (b = 0,1,570, p < 0,10). The remaining significant variables, all show the same signs as in the regular OLS analysis.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 2,057 3,789* 12,933* 8,362 FDI (lagged) -0,179** -2,775* -2,041 Trade (lagged) 0,002 -0,046* -0,040 Interaction (lagged) 0,010* 0,013** 0,010 Human capital 0,271 1,361* 1,570* Domestic investments 0,153** -0,029 Tax rate 0,123* 0,110 Government consumption -0,006 Inflation rate 0,110 System R2 0,099 0,420 0,508 0,488 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 10. TSLS results H3 - developed countries
model 3. However, since the independent variable is not significant, robustness cannot be confirmed. This should be taken into account when conclusions are drawn.
With the results from the OLS and the TSLS regression analyses, hypothesis 3 can be accepted. Results show that FDI has a positive relationship with economic growth in developing countries. However, in developed countries FDI has a negative relationship with economic growth. Therefore, the hypothesis that FDI has a greater positive relationship with economic growth in developing countries than in developed countries, can be accepted.
Dependent variable Economic growth
Independent variable Model 1 Model 2 Model 3 Model 4
Constant 7,899* 6,502* 4,078 3,183 OECD FDI 0,015* 0,056 -0,141 Trade 0,005 0,034 0,026 Interaction -0,001 -0,001 -0,003 Human capital -0,444 0,705 1,514 Domestic investments 0,019 -0,091 Tax rate 0,008 0,149 Government consumption -0,155 Inflation rate -0,105* System R2 0,045 0,172 0,227 0,201 * p < 0,10, ** p < 0,05, *** p < 0,001 Table 11. OLS results H4
this relationship is not significant. The interaction variable is also negative and insignificant. Therefore, conclusions regarding the relationship between these variables and economic growth cannot be drawn.
The results of the control variables are also analysed. Almost all control variables used for hypotheses 4 are insignificant. The variable human capital does not show any significant values. In two out of three models, human capital shows a positive sign. However, in model 1 human capital shows a negative sign. Because the values are not significant, these results cannot be used. The variable domestic investments also does not contain any significant values. The variable tax rate shows a consistent positive relationship with economic growth.
Model 4 shows whether the results are robust. In this model, the control variables domestic investments and tax rate are replaced by the control variables government consumption and inflation rate. Only the control variable inflation rate shows a negative and significant relationship (b = -0,105, p < 0,10) in model 4. Furthermore, model 4 suggests that the results are not robust. OECD FDI shows a different sign than it does in model 3. However, since the value is not significant, this sign can be neglected. Robustness of the model cannot be confirmed.
The R squares show that the control variables explain 4,5 percent of the model, the independent variables explain 17,2 percent, and the independent variables and the control variables together explain 22,7 percent. This leaves 77,3 percent of the model unexplained.
5. Discussion and limitations
The results showed several expected, but also surprising outcomes. In this chapter, these outcomes are discussed and linked to existing literature to support them.
Hypothesis 2 focused on whether or not FDI is positively moderated by trade in promoting economic growth in Central Eastern Europe. The interaction variable, in order to test this moderation of trade on FDI, did not show any significant values. Therefore, conclusions cannot be drawn and hypothesis 2 also has to be rejected. It cannot be confirmed that the interaction between FDI and trade results in an increase in economic growth.
Makki & Somwaru, 2004). Therefore, this outcome requires further research. The results show that hypothesis 3 can be partially confirmed. FDI has a positive relationship with economic growth in developing CEE countries. In developed CEE countries, FDI has a negative relationship with economic growth. The hypothesis stated that FDI has a greater positive relationship with economic growth in developing countries than in developed CEE countries. This suggests that both relationships are expected to be positive. However, this is not the case.
In addition to the main findings, the significant results of the control variables are discussed. Human capital shows a positive relationship with economic growth in developed CEE countries. This is in line with existing literature (Zhang, 2001; Barro & Lee, 1994) that showed that an increase in human capital results in a better absorptive capacity of knowledge and technology spillovers, and thereby contributes to economic growth.
A positive relationship between domestic investments and economic growth has been found in the results. This suggests that economic growth in Central and Eastern Europe increases when there is a larger domestic investment. This is in line with existing research of Choe (2003), who states that domestic investment is an important factor in promoting economic growth. Several other studies found evidence that domestic investments have a positive relationship with economic growth, because domestic investments improve a country’s infrastructure and institutions (Borensztein, et al., 1998; Makki & Somwaru, 2004).
The control variable inflation rate also showed a positive relationship with economic growth. This means that lowering the inflation rate, would result in a larger economic growth. This result is confirmed by Froot and Stein (1991), who state that a low inflation rate is a sign for a better climate for FDI and trade. This investment and trade, is assumed to lead to economic growth. Moreover, Fisher and Modigliani (1978), state that countries with high inflation rates are less interesting for foreign investors because of the risks and uncertainties of unstable economic.
Limitations
This study provided a detailed understanding regarding the relationship between FDI and economic growth in Central and Eastern Europe. However, the following limitations were identified.
functions of the variables. For the regression analyses, a time lag of t-1 was used in order to address endogeneity. For further research, it may be interesting to also research other time lags in order to investigate what is the long-term impact of FDI on economic growth. Moreover, the regression results showed a lot of insignificant values. Subsequently, a lot of data could not be used in order to draw conclusions. Also, this study tested covariance by using an OLS regression analysis. Therefore, it cannot be confirmed that there are causal relationships. However, this subject is addresses by testing for endogeneity. At last, the results showed some surprising outcomes. Unfortunately, not all of these outcomes could be explained by existing literature. Further research is necessary in order to investigate what exactly is the reason that FDI shows a negative influence on economic growth in (developed) CEE countries.
6. Conclusions
References
Albulescu, C. T. & Goyeau, D., 2014. The Interaction between Trade and FDI: The CEE
countries experience, Poitiers Cedex: University of Poitiers.
Allen & Overy, 2006. Foreign Direct Investment in Central & Eastern Europe, Bratislava: The Visegrad Group.
Allison, P., 2012. When Can You Safely Ignore Multicollinearity?. [Online] Available at: https://statisticalhorizons.com/multicollinearity [Accessed 30 October 2017]. Ayanwale, A. B., 2007. FDI and Economic Growth: Evidence from Nigeria.
Bacic, K., Racic, D. & Ahec-Sonje, A., 2005. The Effects of FDI on Recipient Countries in Central and Eastern Europe. Privredna Kretanja I Ekonomska Politika, 14(100), pp. 58-96.
Baguley, T., 2012. Serious Stats: A Guide to Advanced Statistics for the Behavioral Sciences. Basingstoke: Palgrave.
Balasubramanyam, V. N., Salisu, M. & Sapsford, D., 1996. Foreign Direct Investment and Growth in EP and IS Countries. The Economic Journal, pp. 92-105.
Baron, R. M., 1986. The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations. Journal of Personality
and Social Psychology, 51(6), pp. 1173-1182.
Barro-Lee, 2014. Education Attainment for Population. [Online] Available at: http://www.barrolee.com [Accessed 6 October 2017].
Barro, R. J., 2003. Determinants of Economic rowth in a Panel of Countries. Annals of
Economics and Finance, Volume 4, pp. 231-274.
Barro, R. & Lee, J. W., 1993. International Comparisons of Educational Attainment. Journal
Barro, R. & Lee, J. W., 1994. Sources of Economic Growth. Carnegie Rochester Conference
Series on Public Policy, Volume 40, pp. 1-46.
Belloumi, M., 2014. The Relationship Between Trade, FDI and Economic Growth in Tunesia: An Application of the Autoregressive Distributed Lag Model. Economic Systems, 38(2), pp. 269-287.
Bleaney, M. & Greenaway, D., 2001. The Impact of Terms of Trade and Real Exchane Rate Volatility on Investment and Growth in Sub-Saharan Africa. Journal of Development
Economics, 65(2), pp. 491-500.
Borensztein, E., De Gregorio, J. & Lee, J., 1998. How Does Foreign Direct Investment Affect Economic Growth. Journal of International Economics, 45(1), pp. 115-135.
Branstetter, L., 2006. Is Foreign Direct Investment a Channel of Knowledge Spillovers? Evidence from Japan's FDI in the United States. Journal of International Economics, 68(2), pp. 325-244.
Brecher, R. A. & Diaz-Alejandro, C. F., 1977. Tariffs, Foreign Capital, and Immiserizing Growth. Journal of International Economics, Volume 7, pp. 317-322.
Choe, J. I., 2003. Do Foreign Direct Investment and Gross Domestic Investment Promote Economic Growth?. Review of Development Economics, 7(1), pp. 44-57.
Cox, N. J., 2007. Transformations: An Introduction. [Online] Available at:
http://fmwww.bc.edu/repec/bocode/t/transint.html [Accessed 16 October 2017]. Dabla-Norris, E., Honda, J., Lahreche, A. & Verdier, G., 2010. FDI Flows to Low-Income
Countries: Global Drivers and Growth Implications.
De Mello, L. R., 1999. Foreign Direct Investment-Led Growth: Evidence from Time Series and Panel Data. Oxford Economic Papers, 51(1), pp. 133-151.
Dranove, D., 2012. Practical Regression: Introduction to Endogeneity: Omitted Variable Bias.
Kellogg School of Management, pp. 1-6.
Dreher, A., 2006. Does Globalization Affect Growth?. Applied Economics, 38(10), pp. 1091-1110.
Durham, J. B., 2004. Absorptive Capacity and the Effects of Foreign Direct Investment and Equity Foreign Portfolio Investment on Economic Grwoth. European Economic
Review, 48(2), pp. 285-306.
Eller, M., Haiss, P. & Steiner, K., 2005. Foreign Direct Investment in the Financial Sector:
The Engine of Growth for Central and Eastern Europe, Vienna: Vienna University of
Economics and Business Administration.
Economic and Social Research Council, 2015. Framework for Research Ethics. [Online] Available at: http://www.esrc.ac.uk/files/funding/guidance-for-applicants/esrc-framework-for-research-ethics-2015/ [Accessed 10 July 2017].
European Central Bank, 2010. The Impact of the Financial Crisis on the Central and Eastern European Countries. July, pp. 85-96.
European Union, 2017. About the EU. [Online] Available at: https://europa.eu/european-union/about-eu/countries_en#tab-0-1 [Accessed 3 August 2017].
Financial Times, 2017. Central and Eastern Europe Show Way on Growth. [Online] Available at: https://www.ft.com/content/ba1301c0-4c54-11e7-919a-1e14ce4af89b [Accessed 8 November 2017].
Fisher, S. & Modigliani, F., 1978. Towards an Understanding of the Real Effects and Costs of Inflation. Weltwirtschaftliches Archiv, Volume 4, pp. 33-810.
Fontagne, L., 1999. Foreign Direct Investment and International Trade: Complements or
Substitutes?, Paris: OECD Publishing.
