1
University of Groningen Faculty of Economics and Business
Thesis in International Economics and Business (IE&B)
Topic: FDI INFLOWS AND INSTITUTIONAL QUALITY.
Systematic review and econometric application to the CEE and the Balkan region
Ilias Zaptsis (s2021609) E-mail: i.zaptsis@student.rug.nl
Supervisor: Dr. Ger Lanjouw (g.j.lanjouw@rug.nl) Co-supervisor: A.J.Mulder (arnold.mulder@rug.nl) Co-assessor: G.De Jong (g.de.jong@rug.nl)
2 Abstract
This paper analyzes whether the -relatively low- attractiveness of Foreign Direct Investment (FDI) flows to Balkan countries in comparison to CEE countries is a result of their weak quality of institutions. Based on a panel data analysis of 14 selected economies we advocate that nations which exhibit high performance, measured by different institutional indicators, have higher presence of foreign investors since we find evidence that stable institutions cultivate a healthier business environment in a country. Additionally, we implement a more detailed econometric analysis (i.e factor analysis) in order to investigate the effect of the institutional quality on FDI inflows. The results suggest that governments should pay more attention to institutions such as rule of law and control of corruption since they are strongly connected to FDI inflows.
Key-words: Foreign Direct Investment Inflows, Institutional Quality, Factor Analysis, Panel
Data
Acknowledgements
3 Table of contents Abstract 2 Acknowledgement 2 Table of contents 3 Introduction 4 Added value 5 Literature review 6 Part 1 : Theoretical background of FDI 6
Part 2 : Relationship between FDI and growth, FDI determinants 8 Part 3 : Institutional quality 10 Description of variables 15 Dependent variable 16 Independent variables 16 Factor analysis procedure 20 Hypotheses 21 The model 22 Diagnostic tests 25 Normality 25 Multicollinearity 25 Heteroskedasticity 25 Autocorrelation 26 Results 27 Conclusion and Implication 30
Limitations and further research 32 References 33 Data sources 37
4 Introduction
In the last 20 years, economic development and economic growth have been the target of the majority of governments as the means to increase the overall welfare of their countries. Therefore, methods of promoting growth have been examined by many researchers (Lyroudi et
al., 2004; Alfaro et al., 2006). In that respect, Foreign Direct Investment (FDI) is an area with
great interest since FDI is regarded as a growth-enhancing machine that causes transfer of knowledge, entrance of foreign capital and increase in the employment rate (Alfaro et al. 2006; Golub, 2009)
The global distribution of FDI shows that during the last twenty five years the FDIs has increased dramatically (Navaretti and Venables, 2004). Regarding to the allocation of FDI inflows at the Europe, countries which are located at the Central Eastern and South Eastern Europe (CEE and SEE) have received a small share of the world’s FDI inflows. Furthermore, this share is unequally distributed at this region since the Balkan countries have received less FDI per capita comparing to the other countries of this area. (Pournarakis and Varsakelis 2004; Fabry and Zeghni 2010).
But what could elucidate this relatively low attractiveness of FDI flows to Balkan countries? Observation of the literature regarding the reasons which explain the level of FDI inflows in a country reveals that the following factors may clarify their distribution at the Balkan region.
5
The institutions, as conceptualized by D. North (1991), are “… the humanly devised constraints which structure political, social and economic interaction. They consist of both informal constraints (such as sanctions, customs, taboos, traditions and codes of conduct) and formal rules (such as constitutions, property rights and laws). Throughout history, institutions have been devised by human beings to create order and reduce uncertainty in exchange”(North, 1991). Therefore it could be argued that effective or well-developed institutions cultivate the ideal circumstances to increase the number of potential investors in a country. Furthermore, Fabry and Zeghni (2006) mentioned that “the institutional pattern is an important part of the host country’s localization advantage. Stable, flexible and adaptable institutions contribute to build an endogenous attractiveness”.
Consequently, our research’s aim is to identify whether the low level of FDI inflows of 6 selected Balkan countries compared to countries from CE Europe1 is a result of their weak institutional environment. To answer that question we present an econometric model that tries to explain the differences of the level of FDI inflows between countries of the Balkan Peninsula and countries of the CE Europe, on the basis of their institutional environment2. Our choice to compare FDI inflows in the Balkans to countries from Central and Eastern Europe is a result of the following factors. Firstly, we wanted to include countries without vast differences in cultural characteristics and geographical distance. Secondly, we selected to compare FDI inflows of developing countries that were located at the European origin, since it is an area which is merely analyzed under the institutional point of view (Pournarakis and Varsakelis 2004). Thirdly acquisition of data for these countries was relatively straightforward.
Added value
1
Our dataset consists of the following countries: Balkans: Albania, Bulgaria, Croatia, Greece, Romania, Turkey.
Central and Eastern Europe: Belarus, Czech, Hungary, Latvia, Lithuania, Poland, Slovak Republic, Ukraine
2
6
In this part we outline the contribution of this paper to the current economic literature. Firstly, we examine the effect of institutional factors on FDI inflows which is a comparatively new field of interest. Secondly, we utilize the institutions dataset which was developed by Kaufman, Kraay and Mastruzzi (2006). The authors created six dimensions3 of governance taking into consideration hundreds of specific and disaggregated individual variables, measuring various dimensions of governance, taken from 35 data sources provided by 32 different organizations. Hence, we can argue that we include a complete dataset, which properly depicts the country’s institutional quality. Thirdly, differently to other research papers which focus only on institutional variables from a smaller number of countries (Pournarakis and Varsakelis, 2004; Fabry and Zeghni, 2006), we include a wider sample of SEE and CEE countries and a larger set of institutional and non-institutional variables. Fourthly, differently to Pournarakis and Varsakelis (2004) and Anghel (2005) we implement a more detailed econometric analysis; we employ factor analysis to reduce the dimensionality of our explanatory variables, since there is a high correlation not only between the variables which measure institutions but also between the variables which measure potential explanatory variables (such as the electronic infrastructures).
Literature review
The literature review in this paper consists of three parts. In the first part we present a brief introduction on the notion FDIs. In the second part we focus on the relationship between FDI and growth in order to support the belief that FDI is regarded as a positive instrument for the economic development of a country. Additionally, we summarize the determinants of FDI as these are identified in the economic literature. Finally, in the third part, there is an evaluation of the institutional quality and its connection to FDI as it was presented by other researchers.
Part 1: Theoretical background of FDI
3
7
FDI is an investment which entails the international investment made from an enterprise into a firm with different country of origin. The investor owns at least 10% of the ordinary shares, undertaken with the purpose of establishing a ‘lasting interest’ in the country (Navaretti and Venables, 2004). FDI is either horizontal (market-seeking investment where the firm aims to increase its influence in a host market, in order to strengthen its overall supply to that host market), or vertical (efficiency-oriented investment where a firm transfers a part of its production process to a host country with the purpose of taking advantage of the variety of natural resources or its cheap productivity costs) (Navaretti and Venables, 2004). The flows of FDI are either
inward where the foreign funds are invested in the local market or outward where the local firms
invest in a foreign country.
Additionally, Dunning (1977) while examining the function of FDI, emphasized the role that is played by ownership, location and internalization and, through that examination developed the OLI framework (Figure 1). More specifically, Dunning, through that framework, reported that a firm will invest abroad if it owns its specific assets, if a location advantage comes from investing in a host county and if internalization benefits appear from internalizing a part of its production process. Here it is important to note that in this paper we focus on location advantage in order to discover the reasons of the relatively low attractiveness of Balkan countries.
8
However, is the presence of FDI inflows always correlated to economic growth? Are there any country features that matter in selecting the ideal country to invest? Could country characteristics, such as fiscal and institutional indicators, explain the gap of FDI inflows between two countries? Is the institutional quality (measured by different indices) of a country a strong motive in order to attract foreign investors? We are trying to answer the previous questions in the next two parts of the literature review.
Part 2: Relationship between FDI and growth, FDI determinants
Relationship between FDI and growth
There is a widespread belief among policy makers that FDI creates productivity effects for host countries (Alfaro et al., 2006). According to the author there are some growth-enhancing mechanisms that could be appeared in a country as a result of FDI. Specifically, Alfaro mentioned that FDI contributes to adoption of foreign technology and know-how, creation of linkages between foreign and domestic firms and introduction of new processes. Moreover, he supported the opinion that an increase in FDI causes higher growth rates since the overall production of the host country increases. Similarly, Turkcan and Yetkiner (2008) analyzed the endogenous relationship between FDI and economic growth for 23 OECD countries and found that FDI and economic growth are significant determinants of each other. Additionally, Basu and Guariglia (2007) tested 119 developing countries for a period of 30 years finding that FDI enhances economic growth. Also, Golub (2009) referred to capital accumulation, reduction of the unemployment rate since there is a greater demand for labor and increased competition and de Mello (1997) emphasized the presence of spillovers combined with the presence of FDI.
Bhagwati (1978) found that FDI supports the economic growth of a nation since it strengthens economic efficiency. Additionally in 2007 Nourzad4 supported the opinion that FDI contributes towards reducing technical inefficiencies and this is higher the more open to trade the host economy is. Also, Nourzad (2007) indicated that the more FDI the more potential output in an economy since the FDI increases the overall national production, but this is basically observed in
9
developed countries compared to developing. Moreover, Hansen and Rand5 (2004) found not only a strong causal link between FDI and GDP, but also that FDI promotes growth in the same way as domestic investment. Likewise they referred to adaption of new technologies and knowledge as two basic instruments that enhance economic development in a host country and they confirmed a positive and significant correlation between FDI and GDP in a long term basis. Finally Hansen and Rand (2004) considered the association between FDI and economic progress to be positive, provided that the host countries have reached a minimum level of educational and/or infrastructure development.
On the other hand there are researchers which challenge the belief that FDI is positively correlated to economic growth. Specifically, Lyroudi et al., (2004) chose 17 economies and the results of this survey differ from others, suggesting a significant relationship between FDI and economic growth in developing countries. Moreover, they referred to how host countries shape their policy in order to attract or restrict FDI inflows. Specifically, tax incentives, import duty exemptions and infrastructure subsidies are used in order to attract FDI, and precautionary measures such as higher taxes, are used in order to ameliorate the negative impact and discourage the capital inflows. Similarly, Busse and Groizard6 (2006) advocated that the less regulated a country is the more advantages will receive with the presence of FDI inflows. Additionally, in the case of more regulated economies Busse and Groizard (2006) found that regulations could restrict growth through FDI. Finally, Herzer, et al,. (2006) who tested 28 developing countries found that FDI has no statistically significant impact on growth.
FDI determinants
Bevan and Estrin7 (2000) tried to explain the determinants of FDI inflows in Central and Eastern Europe. Apart from unit labor costs, host market size and country risk, the authors emphasized the positive impact of becoming a member of EU. Specifically, according to Bevan and Estrin (2000) when a country announces that a procedure to meet the entry requirements8 of the EU is initiated, its FDI inflows will increase with a relatively faster rate. Similarly, Sun (2002)
5 Examined a sample of 31 developing economies from 1970 to 2000 using fixed effects model in a panel cross country data analysis
6
Used panel data with the top 20, 30, 40 and 50 per cent of most regulated economies
7 Used both fixed and random effects models in a panel group of FDI flows from 18 developing to 11 transition economies over the period 1994–1998.
10
highlighted the role that is played by the Growth rate (in terms of GDP), total market demand and macroeconomic stability9, considering these three factors to raise the confidence of the potential investors. Moreover Sun (2002) stressed the impact of the regulatory environment combined with political instability; for example armed conflicts and assassinations presage an unstable environment. Finally the author emphasized the effect of investment promotion and the level of infrastructures in a host country, such as roads and the number of schools or universities or laboratories. Besides, Botric and Skuflic10 (2006) considered trade openness, GDP, GDP per capita, population and internet lines to have a strong statistical impact on FDI inflows in 7 SEE11 from 1996 to 2002. Finally it is worth to be mentioned that characteristics such as education level, GDP per capital, openness to trade and institutional development are extremely important for a country that desires to attract FDI (Nunnenkamp and Spatz, 2003).
Here, it is important to be referred that we follow the findings of Daniele and Marani (2006) and Anghel (2005) who consider the growth rate, the macroeconomic stability, the membership of EU and the electronic infrastructures as significant determinants of FDI.
As it is mentioned above, we are trying to explain the differences in the level of FDI inflows between countries of the Balkan Peninsula and countries of Central and Eastern Europe, on the basis of their institutional environment. Hence, in the next part we present an overall evaluation of the institutional quality and its connection to FDI as it was presented by other researchers.
Part 3: Institutional quality
In this part, we start with other papers which tried to explain the impact of institutions on a country and we continue with researches which focus on the relationship between institutions and FDI inflows.
To begin with Rodrik who argued that institutional quality “holds the key to prevailing patterns of prosperity around the world” (Rodrik,2004). The author distinguished rich from poor
9 Expressed by low inflation
10 Analyzed the FDI inflows in Southeast European countries (SEE)
11
countries advocating that rich ones (in terms of GDP) provide to their potential investors a stable business environment. Specifically, according to Rodrik (2004), rich countries have the following characteristics: “security regarding the property rights, private incentives associated with social purposes, monetary and fiscal policies which are grounded in solid macroeconomic institutions, idiosyncratic risks are appropriately mediated through social insurance, and people have resource to civil liberties and political representation” (Rodrik , 2004) The author argued that the abovementioned features contribute towards creating a desirable institutional environment for an investor and this atmosphere affects the economic development of a country positively. On the other hand, he claimed that poor countries have merely developed such rules. Moreover, Rodrik referred to the policy makers and their growing attention to institutions mentioning that governments are making reforms under the purpose of “1) reducing corruption, 2) improving the regulatory apparatus, 3)rendering monetary and fiscal institutions independent, 4)strengthening corporate governance and 5) enhancing the functioning of the judiciary system.” (Rodrik, 2004).
Similarly, Acemoglu and Robinson (2007) highlighted the importance of institutional characteristics on economic outcomes under specific conditions; such as the connection of legislation system or political stability to the types of business transactions. They argued that the political regime12 or economic institutions are fundamental factors that make institutions differ between two countries. According to the authors a striking example of this is the case of South and North Korea. South Korea relies on a Capitalist market while the political system in North Korea is Communism. In the first case the government enforced some rules in order to protect private property and production while in the second (North) there is no space for private property. Comparing these two countries it is easily observable the huge gap between their growths. Indeed, GDP in South Korea in 2009 was 834,060 million of US$, much greater from North Korean GDP that was 28.000 million of US $ (source: CIA). Therefore, it can be argued that the political system and economic institutions are two factors that explain the enormous difference of development between these countries (Acemoglu and Robinson., 2007).
12
Furthermore, Fabry and Zeghni (2008) who explored13 the impact of institutions on growth and human development, divided the institutions in two categories: formal and informal. According to the authors formal institutions could be distinguished between market and political institution. Specifically, the former includes enterprise reform, banking reform and interest rate liberalization, and expenditures on health and education. In contrast, political institutions contain democracy score ranking, general government debt, and federalism degree14. They considered as informal institutions the per cent of minorities in total population and the corruption perception index. They added the index of GDP per capita15 in their regression model and they checked the impact of all the aforementioned variables on Infant Mortality Rate (IMR) on both samples of countries. The authors found that IMR is more sensitive to institutions in CIS16 countries than in new EU members and they considered this to be a result of three factors: Firstly, there are some institutional requirements that must be satisfied from a country in order to be accepted in EU. Secondly, the CIS countries are less stable, less mature and less reliable in their institutional arrangements. Thirdly, there is a gap of cultural and ethnical specifications between CIS and new EU members (Fabry and Zeghni, 2008).
Additionally, Benassy-Quere et al, (2005) examined whether a positive relationship of institutional quality and FDI exists in a panel data analysis of 52 countries17. Benassy-Quere et al., (2005) confirmed the idea that public efficiency, which contains factors such as: property rights, tax systems, easiness to start business, rule of law, transparency and prudential standards are strong determinants of the attractiveness of FDI. They concluded that “weak capital concentration and employment protection tend to reduce inward FDI” (Benassy-Quere et al, 2005). In the same way, Anghel (2005) has stated that “countries whose governments have the following characteristics: 1) political stability, 2) a set of regulatory policies which encourage foreign trade and business development 3) high quality of bureaucracy 4) high degree of
13
The authors collected data from 1994 to 2006 creating two samples of host countries. Their sample consists of 10 European Union members (Estonia, Latvia, Lithuania, Poland, Hungary, Czech Republic, Slovenia, Slovakia, Romania and Bulgaria) and 11 CIS countries (Russia and Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Tajikistan, Ukraine, Uzbekistan)
14
Federalism degree (FE) is a dummy that takes the price 0 for unitary country and price 1 for a semi federalism or federalism country.
15 They chose this control variable since it reflects the average wealth of a nation. 16
CIS: Common Wealth Independent States: Armenia, Azerbaijan, Belarus, Georgia, Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine, Uzbekistan.
13
protection of property rights and 5) efficient control of corruption, tend to do better in attracting FDI inflows”. The author used the indices of two different institutional data sets18
and examined their impact on FDI inflows. The results from this study showed that poor institutions, negatively affect the amount of FDI that a country receives. (Brindusa Anghel, 2005).
Moreover, Bevan et el.,(2004) investigated whether there are institutions that have a relatively stronger impact on inflows. According to the authors “…institutions are a locational advantage of a country in order to increase its FDI inflows” so that institutional quality enhances FDI inflows; specifically in cases where countries focused on institutions like: banking sector, legislation system and private sector growth combined to trade liberalization it is more possible to observe a higher level of FDI inflows (Bevan et al.,2004). Also, Fabry and Zeghni (2006) who focused on SEE countries19 in order to find out “whether or not the weak inward-FDI are connected to non-reliable institutions and to a non EU membership”, confirmed the idea that local institutional arrangements affect the presence of FDI inflows20. In their results, they referred to uncertain institutional environment as an impediment to FDI and to the fact that the institutional quality has a stronger impact on FDI in non EU candidate countries comparing to the future members (Fabry and Zeghni 2006).
Also, Daniele and Marani (2006) focused on MENA21 countries and stressed the impact of quality institutions on FDI inflows suggesting that reforms on institutions and legislation system would increase the presence of foreign capital. They argued that MENA countries may exhibit poor performance because: 1) the size of local markets is relatively small 2) there is no strong international competition from MENA countries and this could be explained by the slow pace of
18
The first data set consists of the dimensions of governance from Kaufmann et al (2004). Specifically as it was mentioned this data set has the following variables: 1) Political Stability 2) Government Effectiveness 3) Regulatory quality Index 4)Rule of Law Index 5) Control of Corruption and 6)Voice and accountability index.
The second data set consists of the determinants of the quality of governments from La Porta (1999) data set. Specifically, this data set has the following indexes: 1) Property rights 2) Business Regulation Index 3) Corruption 4) Bureaucratic Delays.
19 SEE countries in their sample consists of: Albania, Bulgaria, Bosnia and Herzegovina, Croatia, Moldova, Romania, Serbia, Serbia & Montenegro, FYROM
20
Specifically, on the one hand they used the FDI per capital as dependent variable and on the other hand they took the following independent variables: GDP per capital, the index of competition policy, the corruption perception index, an index that measures the expenditure on health and education (as % of GDP), an index about enterprise reform and the freedom house index(which is a combination of 1) stability of governmental system and 2) legislative and executive transparency). Finally, they checked the impact of all the above-mentioned independent variables on FDI per capital
14
economic and trading reforms and 3) there is non-stable international competitiveness; for example the conflicts at these countries contribute towards creating an unstable environment. In order to build their research, Daniele and Marani (2006) used three controlling variables22 and the indices of Kaufmann, Kraay and Mastruzzi dataset23, finding a statistically significant impact of institutions on FDI inflows.
Similarly, Pournarakis and Varsakelis (2004) who chose 11 countries24 over the period 1997-2001 stated that while all the studied nations enjoy vigorous democratic institutions, there are still significant differences in their relative success in building quality institutions. Specifically, they took three indices which measure the quality of political institutions25, one index which is used as a measure of corruption26 and two more indices which measure the degree of internationalization27 under the purpose of describing whether the behavior of FDI inflows is explained from the level of its institutional quality. Pournarakis and Varsakelis (2004) firstly found that the market size and internationalization are significant determinants of the cross country variation of FDI inflows and secondly that the impact of all the institutional variables is strong enough in order to attract or restrict the FDI inflows that exist in a country.
Finally, Ali et al., (2010) argued that institutions arise to provide rules and procedures which decrease uncertainties involved in economic exchange. They used a panel set of 69 countries, from 1981 to 2005, advocating that institutional quality is a major determinant of FDI inflows. Specifically, among the variables they used in order to measure institutions, they stressed the importance of the property rights index, finding that it has a stronger effect on FDI inflows in comparison to corruption, democracy and political social tension and instability (Ali et al.,2010).
22
Specifically they took the following indexes: 1) real GDP growth,2)telephone mainlines for 1000 inhabitants in order to control for infrastructures and 3) a proxy of the development level, on the basis of energetic consumption per capital.
23 It contains the following six indicators:1) voice and accountability: which measures political and civil rights,2)political instability and violence: which measures the possibilities of violent threats to, or changes in government,3)government effectiveness: which measures the competence of the bureaucracy and the quality of public service delivery,4)regulatory quality index: which measures the incidence of market unfriendly
policies,5)rule of Law: which measures the quality of contract enforcement, the police and the courts, as well as the possibilities of crime and violence, and 6) control of corruption: which measures the exercise of public power for private gain, including both petty and grand corruption and state capture.
24 Their sample consists of two groups, the first has 8 countries(Czech Republic, Estonia, Hungary, Latvia,
Lithuania, Poland, Slovakia and Slovenia) which had been accepted to join the EU in 2004 and the second group has 3 countries(Bulgaria, Croatia, Romania) which were candidate to join the EU in the second enlargement phase. 25 Political Rights, Civil Liberties, Freedom of Press
26 Corruption Perception Index
15
Lastly, it is worth to be mentioned that the impact of institutional quality differs across sectors, since the authors confirmed that it is stronger the effect of institutions on manufacturing and services sectors while there is not a robust effect on primary sector.
To conclude, studying of the economic literature regarding institutional quality, revealed that institutional quality affects not only the FDI but also the overall growth in a country. Therefore in this paper we focus on institutions, making an attempt to investigate whether or not the poor institutional environment affects the distribution of FDI inflows in 14 selected economies. As we mentioned above, many authors28 who tried to investigate the relationship between FDI inflows and institutional quality in a country used the Kaufmann, Kraay and Mastruzzi dataset. Based on these attempts we chose this dataset in order to portray the institutional environment of the countries we include in our sample.
Description of the variables
In this part, we initially describe the variables which are used in our model and we secondly present figures which depict their performance. If we look at these figures in the appendix of this paper, it is clear that Balkan countries compared to CEE, are less stable under the institutional point of view and we expect this to be a significant factor of their lower distribution of FDI inflows.
The economic literature has identified a large number of factors which can possibly affect the distribution of FDI between countries. This makes the choice of the appropriate control variables for our model more difficult. In our paper we follow the findings of Sun (2002) and Bevan and Estrin (2004) who claimed that market size, growth rate, macroeconomic stability, infrastructures and the membership of the EU are strong determinants of FDI. We collected data related to these parameters, for the period 1996-2008, taking into account the availability of the data regarding
institutions. All data were extracted from the Worldbank database
(http://databank.worldbank.org).
16 Dependent Variable
Following the literature which regards the FDI per capita -FDIPC- as the ideal dependent variable for the analysis of FDI inflows (Fabry and Zeghni, 2006), and the skewness of such variable, we use the logarithmically transformed FDI inflows per capita - log( FDIinf / pop) - as dependent variable. Like Fabry and Zeghni (2006) we are making use of -the per- capita figures in order to stress the impact of the relative market size of the country since “market size is one of the strongest locational determinants of inward Foreign Investments” (Culem, 1988). Specifically, the economic literature documents that the greater the market size, the greater the total demand and as a result more goods are to be sold. Therefore, it can be argued that potential investors would prefer to locate their enterprises in countries which have a relatively larger market size, in order to serve the local market and as a result reduce their export costs (Navaretti and Venables, 2004). Moreover, it is important for the usage of the logarithm in our dependent variable to be justified. Standard econometric methods used further in this thesis make the assumption that the dependent variable is approximately normally distributed. Hence we have first assessed this assumption on the FDIPC variable, through the use of the Shapiro-Wilks (SW) normality test. Having tested for normality, we decided to make use of the logarithm of (FDIinf /
pop) in order to “render the distribution nearly normal” (Seyoum Belay, 2004) (See Figure 2)
Independent Variables
17
our dependent variable and we found the latter to have a statistically significant impact and more reliable results. Consequently, we decided to use the annual percentage growth rate of GDP (GDPgr) which allows us to control for the growth rate of the market. (See Figure 3)
Macroeconomic stability: In our model we choose to control for macroeconomic stability since it is a factor which directly affects the presence of foreign capital in a country (Sun, 2002). Specifically, countries which exhibit a non-stable macroeconomic environment increase the uncertainty of their market and as a result the investments became less attractive. Oppositely, it can be argued that a relatively stable macroeconomic environment increases the confidence of a market, making it more attractive for foreign investments. (Ali et al., 2010). There are many authors who investigated the FDI inflows and considered the inflation rate as a proxy for macroeconomic stability (Botric and Skuflic 2006; Ali et al, 2010) since the low inflation rate prevents a country from the negative impacts of economic recessions and in cases where governments will keep it at low levels, the investment risks will be reduced and as a result FDI will increase (Anghel 2005). Therefore, taking into account that inflation represents the yearly percentage increase in consumer prices we chose to control for macroeconomic stability through the inflation annual growth-Infl-. (See Figure 4)
Electronic Infrastructure: The host countries’ technological infrastructures have a significant impact on the distribution of FDI inflows on SEE countries (Botric and Skuflic, 2006). Therefore in this paper we decided to build a variable which controls for the level of electronic infrastructure in a country. This variable is made by combining two indices that serve as proxies of electronic infrastructures: 1) the internet users per 100 people and 2) the mobile cellular subscriptions per 100 people.
18
EU membership: Entrance of a country into the EU is expected to increase the foreign investors since it gives the opportunity to the firms which are located at the current EU members, to remove a part of their production process to countries with relatively low labor costs (Bevan and Estrin, 2004). Moreover there are authors who regard the membership of the EU as a catalyst for increasing the presence of foreign capital and decreasing the country risk (Fabry and Zeghni, 2010; Bevan and Estrin, 2004).Specifically, before entering to the EU, the candidate members must establish major reforms in order to meet the entering Copenhagen criteria and this improvement is regarded to positively affect the entrance of foreign capital (Fabry and Zeghni 2010). Similarly, as Bevan et al., (2004) stated: “the EU membership provides implicit guarantees with respect to future economic stability through the membership of the EU area”. Therefore, following the economic literature we regard the membership to EU as a mechanism which attracts the potential investors, so that we add this control variable in our model. Specifically, we create a dummy variable -EU- in order to distinguish the countries which are members of the EU. We construct this variable adding the number 1 from the year which countries joined EU onwards. (See Table 1)
Institutional Variables
We used the dataset of Kaufmann, Kraay and Mastruzzi, where the authors took data from 35 data sources provided by 32 different organizations and they created six variables in order to portray the quality of institutions in a country. It is important to note that each variable may take the number of -2.5 (worst performance) to +2.5 (best performance).
In the next part we present these indicators as they were defined by the authors (Kaufman, Kraay and Mastruzzi, 2009).
19
means, including domestic violence and terrorism” Kaufman, Kraay, Mastruzzi (2009), (See Figure 7)
Government effectiveness index –Goveff- which “measures the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies” Kaufman, Kraay, Mastruzzi (2009), (See Figure 8)
Regulatory quality index –Regql- which “measures the ability of a government to formulate and implement sound policies and regulations that permit and promote private sector development” Kaufman, Kraay, Mastruzzi (2009), (See Figure 9)
Control of corruption index –Corr- which “ measures the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as ‘capture’ of the state by elites and private interests” Kaufman, Kraay, Mastruzzi (2009), (See Figure 10)
Rule of law index –Law- which “measures the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, the police, and the courts, as well as the likelihood of crime and violence” Kaufman, Kraay, Mastruzzi (2009), (See Figure 11)
We checked the correlation of these six indices over time within the countries and identified that these indices are highly correlated. (See table 2)
Consequently, we could not use all of them in the same econometric model since we would confront the problem of multicollinearity. Therefore, we decided to implement factor analysis in order to construct a new variable which is named institutional quality –InstQual- and it combines the scores of the information included in the aforementioned six indices in order to check the impact of institutional quality on FDI inflows. (See Figure 12)
20
of (14*13*12) =2.184 observations29. Here it is important to be noted that there are two countries of our sample which have no data regarding their FDI inflows for one year30. Consequently we also calculated the average of the preceding and the following year in order to complete our dataset.
Factor analysis
This technique is used under the purpose of reducing the number of variables into a smaller number of latent variables (the factors) that still retain as much information as possible from the original variables (Cornish Rosie, 2007). All the variables which are used in factor analysis must be correlated to each other since “indices which are greatly correlated (either positively or negatively) are likely affected by the same factors, while those that are comparatively uncorrelated are likely influenced by different factors” (De Coster Jamie 1998, Overview of Factor Analysis).
Among the 6 types of Factor Analysis, we implement the Principal Component Analysis (PCA). Specifically, as Jamie De Coster (1998) stated “the purpose of PCA is to create a smaller number of components which account for the variability found in a relatively large number of measures. This procedure which is named data reduction, is implemented when a researcher does not want to contain all of the original measures in analysis but still wants to work with the information that they include”. Similarly, (PCA) entails the advantage of reducing the number of variables without losing much information (Lindsay I, 2002).
The function of PCA is the following: It identifies the variables which are highly correlated and recommends how they should be combined. Specifically, it firstly estimates the eigenvalues of the factors which were produced, suggesting keeping those with great proportions and eigenvalues similar or bigger to 1 (Kaiser criterion). The second step of PCA is to rotate these factors under the purpose of showing which variables must be included in each one of the factors which were retained. Having done the previous steps, the PCA gives the opportunity to the
29
Our dataset covers a period of 13 years (1996-2008), it consists of 14 countries and we take into account 12 variables in order to check our hypothesis.
21
researcher to explore separately what is the effect of each one of the variables on the factor that is built. In our model we use the PCA two times, firstly in order to build a factor which depicts the electronic infrastructure in each one of the countries of our sample, and secondly in order to build a factor which measures institutional quality. (See tables 3-8).
Econometrical Modeling Hypotheses
Here it is important to be noted that in the first six models we examine the impact of each one of the indices which measure institutions on FDI inflows separately. In the seventh model we are taking into account all the six dimensions of governance presented by Kaufmann et al (2006), and we synthesize them into a latent variable representing institutional quality. As it is referred earlier, to achieve this we are making use of PCA in order to create the new variable we are utilizing in this model.
1) Political instability has a significant positive impact on countries’ FDI inflows.
2) The Voice and accountability index has a significant positive impact on countries’ FDI inflows
3) The Regulatory Quality Index has a significant positive impact on countries’ FDI inflows 4) The Control of Corruption has a significant positive impact on countries’ FDI inflows 5) The Government effectiveness index has a significant positive impact on countries’ FDI
inflows
6) The Rule of Law index has a significant positive impact on countries’ FDI inflows 7) Institutional quality has a significant positive impact on countries’ FDI inflows
22 The models
Given the variable notation introduced above, and assuming that each one of our observations represents the logarithm of FDI inflows in current31 US $ per capita for each year t:1…T in each host country i:1..n., the seven models can be written as follows:
Model 1
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5 Pol stabit +eit
Model 2
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5 Voiceit +eit
Model 3
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5 Regql it +eit
Model 4
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5 Corrit + eit
Model 5
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5 Goveffit + eit
Model 6
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5Law it + eit
Model 7
LogFDI/popit = βο +β1(GDPgrit) +β2 Inflit + β3 EU + β4 EIQit +β5InstQual it + eit
23 Panel Data
In our study we apply a panel data estimation since not only our data are repeated measures of variables of 14 countries (repeated cross sectional time series) but also some variables of our sample came from panel surveys (Bruderl Josef, 2005). The implementation of panel data has some advantages comparing to cross-sectional or time series methods which are described in the following points. Firstly, panel data estimates contribute towards “increasing the degrees of freedom and reducing the collinearity among explanatory variables” and as a result improve the efficiency of econometric estimates since they are more informative (Hsiao, 2003). Furthermore, panel data could be used in order to analyze questions which cannot be addressed using time-series or cross-sectional datasets. Secondly, panel data “allow us to control for unobserved heterogeneity, study dynamics and test more complicated behavioral hypotheses than time series or cross-sectional datasets (Baltagi,2007).
When someone utilizes panel data, he has to decide which the most appropriate method to be used is. Specifically, it must be made a decision between fixed effects (FE), random effects (RE) and pooled OLS model. Regarding to this topic, it can be argued that the fixed effects have some
advantages in comparison to the other two aforementioned models.
Specifically, Kimino et al (2007) made a comparison between the three models which can be implemented in panel data analysis and he supported the view that pooled OLS is not so trustworthy like RE or FE. Furthermore, there are studies which exclude the OLS model in order to implement FE (Hansen and Rand, 2004) since the FE technique allows to decrease the problem of heteroskedasticity, which usually occurs in OLS models. On the other hand there are drawbacks of the FE model. “firstly the fixed effects estimator does not allow the estimation of the coefficients that are time-invariant and secondly, the number if unknown parameters increases with the number of sample obvservations”(Hsiao Cheng, 2006, Panel Data Analysis- Advantages and Challenges).
24
to 0.05 we will be able to reject the null hypothesis and therefore use the random effects (re) model. Otherwise (p-value > 0.05) we conclude that pooled OLS is the appropriate technique. However in case the aforementioned test rejects the pooled OLS model, we would also have to decide between using a RE or a FE model. For that purpose, we make use of the Hausman test; in this test the null hypothesis suggests that the coefficients which were estimated with the RE model have no differences to the coefficients of the FE model. If the null hypothesis is not rejected, then we can use the RE model and if it is rejected we should use the FE model. More specifically, if we observe an insignificant p-value of the Hausmann test (Prob from the χ2 test greater to 0.05) we can use the RE model. For P-values smaller or equal to 0.05 we conclude in favor of the FE model. (Baltagi et al, 2003) Finally, the F-test32 contributes towards making the correct choice between FE and pooled OLS model. The results of the tests mentioned above, are
presented in the next table.
Table 9: Choice of the ideal technique (FE vrs RE vrs Pooled OLS) we should implement
32 This test is used to decide between Fixed effects and pooled OLS model. Specifically, if the value of this test is lower than 0.05 we reject the null hypothesis and we conclude that the Fixed effects model is appropriate, otherwise we conclude the opposite. ( Parlow,A. 2010)
Model Hausman Test
Tested :Fixed/ Random p-value (selected)
Breusch-Pagan test Tested:Random/OLS p-value (selected)
F-test
Tested: Fixed/ OLS p-value (selected)
Selected
25 Diagnostic testing
After fitting the regression models, comparing between the type of models (FE, RE or pooled OLS), violations of the model assumptions have to be assessed. The main assumptions of all models presented here are :
Normality of the residuals: This assumption relates to the fact that in a regression model similar to the ones we fitted in this thesis, a standard assumption is that the error term is normally distributed. To test this assumption we have performed a Shapiro-Wilkis normality test to every fitted model. In this test the null hypothesis suggests that the data are normally distributed. In cases where the p-value of this test is higher than 0.05 the test does not reject the null hypothesis suggesting that there is no observed deviation from normality in the data. On the other hand in cases where the p-value is lower to 0.05, the test rejects the null hypothesis and therefore we can claim (with 95% confidence) that there is evidence of deviation from normality in our data. We checked the presence of non-normality in all models finding that the normality assumption for the residuals of the applied regressions could not be rejected since the p-values in all the 21 models we tested are higher to 0.05. At the appendix of this paper we present the results of the Shapiro-Wilks tests.(See table 10).
Multicollinearity: Multicollinearity occurs when there is high correlation between the explanatory variables. This implies that it is not possible to include all of the variables in the same econometric model and as a result we have to find a solution to remedy this problem33.Checking for the presence of multicollinearity, we identified that the variance inflation factors (VIF) are greater than 1034 in almost all the institutional variables and as a result multicollinearity exists. (See table 11) For that purpose the PCA analysis and, in consequence, the 7th model was fitted.
Heteroskedasticity: It exists when there is not a constant variance among the observations of a sample and when it is observable, the standard errors of the estimates are biased and as a result
33In our model we choose to implement principal factor analysis since all the variables which depict institutions (from Kaufmann et al,. dataset) are highly correlated.
34
26
the uncertainty increases in the regression model. We tested the plots of the residuals and we implement the Breusch-Pagan / Cook-Weisberg test under the purpose of checking for heteroskedasticity. In this test the null hypothesis suggests that the error variances are all equal and its purpose is to detect the presence of heteroskedasticity. The abovementioned testregresses the squared error term on a dependent variable in order to find out whether the p-value is less than the chosen level of significance. If it is less than 0.05 we support that heteroskedasticity occurs, if it is greater we claim that there were not enough evidence to reject homoscedasticity. The results from the heterogeneity tests are presented at the appendix of this paper. As it can be seen we present an extra table, since there is a different procedure to check for heteroskedastisticity in random effects model in comparison to fixed effects. In the models where heteroskedasticity occurs, we made use of the robust standard errors. (See tables 12, 13).
27 Results
Figures 2-12 present the time series of the variables analyzed. Moreover, in Table 15 we present the descriptive statistics. The results of the regressions are presented in Table 16.
Table 16: Regression results
Parameter Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Inflation annual growth -0.0003 -0.0003 -0.0002 -0.0003 -0.0003 -0.0003 -0.0002 (0.0002) (0.0003) (0.0003) (0.0003) (0.0002) (-0.0003) (0.0003) Joined EU (dummy) -0.233* -0.178* -0.177* -0.164* -0.183 -0.203** -0.192** (-0.108) (-0.092) (-0.095) (-0.094) (-0.114) (-0.094) -0.093) GDP growth annual 0.008 0.006 0.005 0.001 0.009 0.006 0.003 (0.008) (0.007) (0.007) (0.007) (0.008) (0.007) (0.007) Electronic Infrastructure 0.461*** 0.409*** 0.399*** 0.412*** 0.442*** 0.413*** 0.402*** (0.062) (0.040) (0.042) (0.041) (0.065) (0.041) (0.041) Political Stability 0.258* (-0.129) Voice and Accountability 0.323*** (-0.070) Regulatory Quality 0.241*** (-0.082) Control of Corruption 0.308*** (-0.087) Government Effectiveness 0.215 (-0.218) Rule of Law 0.323*** (-0.089) Institutional Quality 0.222*** (-0.050) Constant 2.265*** 2.050*** 2.096*** 2.227*** 2.216*** 2.165*** 2.195*** (-0.056) (-0.072) (-0.080) (-0.073) (-0.049) (-0.074) (-0.064) R2 0.572 0.565 0.553 0.563 0.564 0.564 0.553 Wald chi2 254.08 228.51 236.92 238.57 248.23 F-statistic 95.22 40.97 Prob F statistic 0.000 0.000 0.000 0.000 0.000 0.000 0.000
28
c)Models 1,5:Fixed Effects, Models 2,3,4,6,7: Random Effects significant at 10%,** significant at 5%, *** significant at 1%
With regard to the results, we can see that the coefficients of the inflation annual growth are negative at all the seven models that we have fitted, indicating a negative relationship with our dependent variable. However, none of the models shows a statistically significant connection with the logarithm of FDI inflows per capita neither in the 1%, nor in the 5% nor in the 10% level since the p-values of the coefficients of inflation are higher than 0.1. In the same direction are also the results of the GDP annual growth but in this case the coefficients have a positive sign demonstrating a positive non-significant relationship with our dependent variable. Subsequently, the effect of the year of joining to the EU onwards is significant in six out of our seven models but in the first four models it is important at the 10% level (0.05<p-value<0.1) whilst in the sixth and seventh at 5% level (0.01<p-value<0.05). On the other hand, at the fifth model the effect of the membership on EU does not seem to be significant at any level. Finally, our last control variable which is the quality of electronic infrastructures indicates a significant positive relationship with the natural logarithm of FDI inflows per capita. Specifically the p-values of the coefficients are lower to 0.01 in all the models, showing that the electronic infrastructure is an important motive for a foreign investor.
Nevertheless we should mention the impact of the variables which form the index of electronic infrastructures quality (EIQ). As we stated above, we are making use of PCA in order to create the EIQ index. Specifically according to the output of PCA, we decided to retain only the first of the produced factors in view of the fact that its eigenvalue is 1.85677 (higher than one, according to Kaiser criterion) and its proportion is 0.928. After having rotated the created factor we present the coefficients of the variables we used in order to build the EIQ. The results show that both variables35 not only have equal relevance (0.9635) and uniqueness (0.0716) but also coefficient (0.51893), thus we can argue that both variables have the same impact on EIQ index and they are positively related to that.
Concerning the variables related to our hypotheses, the first model indicates a positive relationship between political stability and our dependent variable which is statistically
35
29
significant at the 10% level. Therefore we can claim that there is evidence to reject our null hypothesis of no impact. In the second model, the coefficients sign of the voice and accountability index is positive with a p-value that is less than 0.01. There is also evidence pointing to a positive strong relationship between the regulatory quality index and the natural logarithm of FDI inflows per capita. In the fourth model where the added variable is the control of corruption we again observe that there is evidence of the existence of a relation. Indeed, the coefficient has a positive sign and its p-value is less than 0.01.
Differently to our first four models where there was evidence supporting the presence of a relation between FDI and the corresponding institutional variables, in the fifth model we can see a negative relationship between government effectiveness and our dependent variable. This could be justified by the fact that in that model, we are making use of the Fixed Effects procedure. On the other hand, our sixth model confirms our prior expectations of a positive statistically significant impact of the rule of law index on FDI inflows.
Finally, examining the seventh model we can observe that there is a significant (at 1% level) positive relationship between the quality of institutions and our dependent variable. However we should notice the impact of each one of the variables which form the institutional quality index, as the respective index was built after making use of the FA process. As it is presented in the appendix of this paper we can see that among the six created factors, we retained only the first36 since there are vast differences among the eigenvalues and the proportions of the produced factors. After having rotated the created factors we can see which are the relevance and the uniqueness of each variable on the factor we retained (institutional quality). Specifically, the
InstQual is basically defined by Rule of Law and Government Effectiveness as these two
variables have greater relevance on the produced factor, while the 42.4% of the variable political stability is not shared with the other indices. Finally, it is important to be referred that all the six abovementioned indices are positively related to the quality of institutions since their coefficients are positive. (See tables 3-5)
30
Conclusion and implication for the CEE and Balkan countries
In this paper we are trying to investigate whether the relative low attractiveness of FDI inflows to the Balkan countries in comparison to Central and Eastern Europe countries is a result of their weak institutional environment. To do so, we selected 6 countries which are located in the Balkan area and 8 countries which are located at the CEE since the unavailability of data did not let us study a wider sample of countries from the two abovementioned regions. Additionally, our data cover the period 1996-2008.
Following the economic literature (Culem, 1988; Botric Skuflic, 2006; Pournarakis and Varsakelis 2004; Fabry and Zeghni 2010) and statistical considerations on normality, we use the natural logarithm of FDI inflows per capita as dependent variable and the growth rate of inflation, the annual growth of GDP, the year of joining to the European Union onwards and a proxy which measures the electronic infrastructure as control variables. Additionally we took into account an institutional dataset which was developed by Kaufmann, Kraay and Mastruzzi and contains 6 different indicators37 regarding the quality of institutions of a nation. Specifically, we use these 6 indices as independent variables of our models under the purpose of examining the impact of each one of these indicators on our dependent variable separately. However, we found that they are highly correlated and as a result we could not contain all of them in one model since we will confront the problem of multicollinearity. Therefore, we decided to implement the factor analysis procedure so as to construct a new variable that combines the scores of the 6 aforementioned indices under the purpose of checking its impact on FDI inflows. It should be noted that we followed the same procedure in order to build the respective variable which measures the electronic infrastructures since we used two relative indices which are also highly correlated.
31
We employed panel data analysis since our data are repeated measures of variables, having tested which is the ideal technique to be used (Fixed effects, Random effects, Pooled OLS). Furthermore, under the purpose of having robust results, we checked for normality, multicollinearity, heteroskedasticity, autocorrelation and stationarity.
Our results indicate that the impact of the 6 institutions and of institutional quality on FDI is statistically significant. Specifically our hypothesis were to a large extent confirmed showing that well developed institutions play a key role in determining the presence of foreign investors in a country. In particular, institutions such as: political stability, voice and accountability, regulatory quality index, control of corruption and the rule of law have a positive and significant impact (at 1% level) on FDI inflows.
Moreover, as we stated in the literature review we found inflation to be negatively connected to FDI inflows per capita and that GDP growth and electronic infrastructures to be positively related to our dependent variable. Although we expected a significant relationship between GDP annual growth and FDI inflows, our results indicate the opposite. Thus we can argue that similar to others researchers38, there is a positive non-significant relationship between FDI and growth in developing countries. Likewise, our results are in the same direction to Niazi et al., (2011) where the authors mentioned that inflation has a negative but insignificant impact on FDI. In contrast, the electronic infrastructures were proven to be an important determinant of FDI inflows, statistically significant at 1% confidence interval.
On the other hand, while the impact of becoming an EU member onwards is statistically significant at 6 out 7 models39 , we see that it is negatively related to FDI inflows. This is in contrast to our expectations since in the literature we highlighted the positive impact of joining to EU. However, we could justify that finding mentioning that after the EU membership, a country may stop producing goods which cost a great amount of money and start importing these goods from other member countries, due to tax differences after the entrance on EU (Lui Eddie, 2009). This move may have a negative impact on the confidence of the potential investors since they may choose to export their products rather than planting a firm in a country.
38 Lyroudi et al.,2004; Busse and Groizard,2006; Herzer, Klasen and Nowak-Lehmann D, 2006; Nunnenkamp and Spatz, 2003
39
32
To conclude we can claim that we found evidence that the well-developed institutions play a key role in determining the presence of foreign investors in a country and the governments of the CEE and Balkan countries who seek for potential investors, should focus on that dimension.
Limitations and further research
The first limitation of this paper is about the number of countries and the number of years in our sample. Specifically, we could have used a wider sample of countries (both developed and developing) since our results cover only 14 nations which are located at the SEE and CEE. Additionally if we used more years we would have checked the impact of the EU on a country’s inflows from the year where the country announces its candidate to the Union and not from the year of joining onwards. Secondly we could have also increased the number of the control variables we used; for example, variables such as the level of education or the labor costs or the trade openness are regarded as motives for the foreign investors.
Another one limitation of this paper is the fact that we include all the three sectors40 on the FDI inflows, whilst we could have checked the impact of institutions on each one of these sectors separately. Similarly instead of inflows, we could have used FDI outflows from selected economies to developing countries since that choice gives the opportunity to the researcher to find include the impact of factor such as the cultural or the geographical distance. Finally, apart from Kaufmann et al dataset, there are other dataset which use different variables in to portray the quality of institutions of a nation41 and we could have used both datasets under the purpose of having more robust results.
40
Primary, Services, Manufacturing 41
33 References
Acemoglu, D. and Robinson, J. 2007. “The role of institutions in growth and development.” Università di Perugia, Dipartimento Economia, Finanza e Statistica, vol. 1(2).
Alfaro, L. Chanda, A. Kalemli-Oznan, S. and Sayek, S. 2006. “How does Foreign Direct Investment promote Economic growth? Exporing the effects of financial markets on linkages.” NBER Working Paper 12522.
Ali, F. Fiess, N. and MacDonald, R. 2010. “Do Institutions Matter for Foreign Direct Investment?” Open Economies Review, Springer, vol. 21(2), pp 201-219.
Ang, J. 2008. “Determinants of Foreign Direct Investment in Malaysia” Journal of Policy Modeling 30, pp.185-189.
Anghel, B. (2005). “Do Institutions Affect Foreign Direct Investment?” Universidad Autonoma de Barcelona.
Baltagi, B. Bresson, G. and Pirotte, A. 2003. “Fixed Effects, Random Effects or Hausman – Taylor?” Economics Letters 79, pp. 361-369.
Baltagi, B. 2007. “Comments on: Panel data analysis- Advantages and Challenges.” An Official Journal of the Spanish Society of Statistics and Operations Research, Springer, 16(1), pp. 28-30.
Basu, P. and Guariglia, A. 2007. “Foreign Direct Investment, inequality and growth.” Journal of Macroeconomics 29, pp. 824-839.
Benassy-Quere, A. Coupet, M. and Mayer, T. 2005. “Institutional determinants of Foreign Direct Investment.” Working Papers, CEPII research center.
34
Bevan, A. and Estrin, S. 2004. “The determinants of Foreign Direct Investment into European Transition Economies.” Journal of Comparative economics 32, pp. 775-787.
Bhagwati, J. 1978. “Anatomy and consequences of exchange control regimes”. Studies in International Economic Relations, 10. Vol. 1.
Botric, V. and Skuflic, L. 2006. “Main Determinants of Foreign Direct Investment in the
Southeast European Countries.” Transition Studies Review, Springer, vol. 13(2), pp 359-377, 07.
Brüderl, J.2005. “Panel Data Analysis.” University of Mannheim
Busse, M. and Groizard, J. 2006. “FDI, Regulations and Growth.” Hamburg Institute of International Economics (HWWA).
Busse, M. and Hefeker, C. 2007. “Political risk, institutions and Foreign Direct Investment”. European Journal of Political Economy 23, pp. 397-415.
Carkovic, M. and Levine, R. 2002. “Does FDI accelerate Economic Growth?” University of Minnesota.
Cheng, H 2006. “Panel data analysis - Advantages and challenges.” IEPR Working Papers 06.49, Institute of Economic Policy Research (IEPR).
Cheng, H. 2003. “Analysis of panel data.” Cambridge University Press.
Cornish, Rosie. 2007. “Principal Component Analysis.” University of Northern Colorado
Culem, G. 1998.”The locational determinants of Direct Investments among industrialized countries”. European economic review 32, pp 885-904.
Daniele, V. and Marani, U.2006. “Do institutions matter for Foreign Direct Investment? A comparative analysis for the MENA countries.”
De Coster, J.1998. “Overview of Factor Analysis.” University of Alabama.
35
Dunning, J. 1977. “ Trade, Location of Economic activity and he MNE: A search for an Eclectic Approach”.
Fabry, N. and Zeghni, S. 2006. “FDI in the New European Neighbours of Southern Europe: a quest of institutions-based attractiveness.” MPRA Paper 1109, University Library of Munich, Germany.
Fabry, N. and Zeghni, S. 2008. “Building institutions for growth and human development. An economic perspective applied to the transitional counties of Europe and CIS.” MPRA Paper
9171, University Library of Munich, Germany.
Fabry, N. and Zeghni, S. 2010. “Inward FDI in seven transitional countries of South-Eastern Europe: a quest of institution- based attractiveness.” Eastern Journal of European studies, Vol 1.
Foad, H. 2007 “Export oriented FDI, the Euro and the EU enlargement” San Diego State University
Golub, S. 2009. “Openness to Foreign Direct Investment in Services: An International Comparative Analysis.” World Economy, Vol. 32, Issue 8, pp. 1245-1268.
Haksoon, K. 2010. “Political Stability and Foreign Direct Investment.” International Journal of Economics and Finance, Vol 2, No3.
Hansen, H. and Rand, J. 2004. “On the causal links between FDI and growth in developing countries.” Discussion Papers, University of Copenhagen.
Herzer, D. Klasens, S. and Nowak-Lehmann, D. 2006. “In search of Foreign Direct Investment-led growth in developing countries.” Proceedings of the German Development Economics Conference, Gottingen, Verein für Socialpolitik, Research Committee Development Economics.
Hill, R. Griffiths, W. and Lim, G. 2001. “Principle of Econometrics. Third Edition”. John Wiley & Son, Inc.
