Is FDI attracted by low costs or high quality?
An institutional approach to the determinants of FDI in developing countries
Doctoral thesis International Economics and Business
Abstract
In our work we focus on the effect of the institutional setting in general and formal education specifically on inward FDI flows in developing countries. Our analysis is based on a panel of 66 countries in the period 1995 – 2004, which we analyze in different ways; in addition to taking all countries together in one sample, we make a division by region and also allow for regimes, inspired by Hansen’s (2000) earlier work. This approach made us able to divide each sample into a high and low regime, characterized by above-median and below-median values for the independent variables, providing for additional insights into FDI determinants in the developing countries.
Our results were different than anticipated; most FDI towards developing countries is less focused on the availability of skilled labor, though supporting the fact that the availability of well developed institutions has a positive impact on FDI inflows.
Acknowledgements
First of all, I would like to thank my thesis supervisor, Dr. B. Los, not only for his guidance, patience and helpful advice in those situations that at first seemed hopeless, but also for his willingness to supervise my thesis even after an absence of six months.
Also, I am particularly grateful to Dr. G.H. Kuper and Dr. E.H. van Leeuwen. Dr. G.H. Kuper for his support concerning the statistical modeling, and answering my endless questions regarding the use of Eviews 5.1, and Dr. E.H. van Leeuwen for help with my program approval, which made it possible for me to graduate before the deadline of the 1st of September.
Finally, I would like to thank my parents and friends for their support during my studies in Groningen and abroad.
Table of Contents
1.0 INTRODUCTION
7
2.0 STYLIZED FACTS
8
2.1 FDI 8
2.2 Increased focus on quality of labour 9
2.3 Institutional reforms and inward FDI 11
2.4 Goal of research 12
3.0 LITERATURE OVERVIEW
14
3.1 Education and FDI 14
3.2 Institutions and FDI 16
3.3 Scope 17
4.0 ARGUMENTS AND HYPOTHESES
19
5.0 METHODOLOGY
22
5.1 Developing countries 22
5.2 Simple threshold approach 22
5.3 Data 23
5.4 The Econometric Model 27
6.0 TESTING THE MODEL
28
6.1 Descriptive statistics 28
6.2 Correlation matrix 28
6.3 Model specification 29
7.0 RESULTS
31
7.1 All Developing Countries 31
7.2 Regional level 34
7.3 Simple threshold approach 35
8.1 Education and FDI 36
8.2 Other determinants of FDI 37
8.3 Policy implications 39
9.0 CONCLUSION
40
APPENDIX
41
Tables, Figures and Graphs
Table 1: Average years of education developing regions... 10
Table 2: Descriptive statistics... 28
Table 3: Correlation matrix ... 29
Table 4: Redundant fixed effects test ... 29
Table 5: Hausman test ... 30
Table 6: Regression output... 32
Table 7: Manual threshold regression output ... 33
Table 8: Countries used in sample... 41
Table 9: Overview of countries used for high / low regimes ... 42
Table 10: Threshold estimation output with all variables included... 43
Figure 1: FDI inflow percentage of world into developing regions, 1978 – 2005 ... 8
1.0 Introduction
An interesting development in the last decades has been the increasing rate of globalization; particularly the increased levels of FDI can be seen as an important sign for this event. FDI (Foreign Direct Investment) is defined as a long term investment in a company abroad, and in that case the investing company needs to obtain a certain degree of control over its daughter firm; this is defined by holding more than 10% of the foreign firm’s ordinary shares or voting power (OECD1). Given that the FDI is beneficial for the growth of economies and it grew by almost 20% in the period 1978 – 2005 due to increased globalization, it makes an interesting subject for research, both for private and public sector.
In our work we focus solely on the developing countries, as they have seen a huge increase in the inflow of FDI over the last few decades which makes it an interesting object of research. We will test the significance of the location determinants that influence a MNE’s choice for a specific host market, seen through the lens of North’s Institutional theory. As we will observe in the next chapter, foreign direct investors regard the quality of labour as an increasingly more important determinant of FDI, given the shift of foreign direct investment flows towards more capital, knowledge and skill intensive industries. Although our interest is specifically focused on the effect of formal education in developing countries on inward FDI, we also asses the importance of other formal and informal institutions on foreign direct investment flows. The analysis is done initially at the aggregate and regional level, in order to take into account regional disparities. Our next step will be the application of a simple threshold model, which is inspired by the work of Hansen (2000). This different approach will divide our country sample in high / low regimes; each sub sample is characterized by above-median or below-median values for every independent variable. This will provide an alternative view on the determinants of FDI and result in a higher homogenised sub sample compared to the previous approaches.
The structure is as follows: First we will provide the necessary facts regarding Institutional theory, inward FDI in the developing regions, and the increasing importance of quality searching FDI; in particular the need of a well developed educational system.
Chapter three will contain a review of existing empirical literature, followed by chapter four, in which our hypotheses will be presented. Chapter five will introduce the research methodology, followed by the model specification tests (chapter 6) and the findings of our empirical research (chapter 7). The last two chapters consist of the discussion and policy implications and end with a conclusion.
1
2.0 Stylized facts
2.1 FDI
After a small downfall of inward FDI after the peak in 2000, according to the 2006 UNCTAD World Investment Report, FDI flows are again increasing with FDI inflows reaching US $ 916 Billion2 in 2005; an increase of 29% compared to 2004, which has already increased with 27% compared to the previous year. Currently the vast majority of FDI is still between developed countries, and only 35% (WIR 2006) of FDI over 2005 is targeted at the developing part of the world, with China as the world’s leading FDI recipient.
FDI inflow percentage of world into developing regions, 1978 - 2005 (average of two years)
-5.00% 0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 40.00% 1978 - 1980 1988 - 1990 1998 - 2000 2003 - 2005 Africa
Latin America and the Caribbean Asia and Oceania
West Asia
South, East and South-East Asia South-East Europe and CIS Total
Figure 1: FDI inflow percentage of world into developing regions, 1978 – 2005
Source: UNCTAD FDI/TNC database
Regarding the developing countries; they have witnessed a strong growth in FDI inflows over the last few years. Especially West-Asia experienced a high growth rate of 85% over 2005 and also Africa reached record levels of inward FDI, amounting US $31 billion. South, East and South-East Asia experienced a 20% increase in inward FDI, while in the Latin-American and Caribbean area, FDI growth was only a mere 3% in 2005. The share in global FDI inflow in South-East and the Commonwealth of Independent States (CIS) remained constant during 2004, totalling US $40 (WIR 2006). With respect to the developing countries as a whole; according to the World Investment Report 2006, their share rose from 20% in the period 1978-1980 to an average of 35% in 2003-2005. This indicates that the overall significance of the developing part of the global economy is increasing with respect to FDI inflow.
2
The current division of FDI is different from 30 years ago: FDI was mainly focused on obtaining cost advantages, meaning that the driver for investing in developing countries was based on lowering costs, especially those of labour (Caves, 1974). This is opposite to the situation nowadays, where we can observe a shift to the quality aspect of local FDI determinants, driven by the need of business process outsourcing of high-tech firms (Bunyaratavej, Hahn, doh, 2007). But although the shift to service oriented FDI and the need for a quality workforce is apparent, still the bulk of FDI towards the developing world is resource based (OECD 2003).
The future of FDI seems to be bright; in the light of increased openness, better government policies, less corruption and increased skill levels we can expect even a bigger role for the share of FDI in the world economy. And according to the World Investment Report 2006 especially the developing countries have seen huge progress in this area, signalling an increasingly more important place in the global economy.
2.2 Increased focus on quality of labour
According to the World Investment Report 2006, improving the country’s stock of human capital will have a positive outcome for a country’s enterprises and economy as a whole. A more developed workforce will attract attention from the capital, knowledge and skill intensive industry, consequently leading to technological spill-over effects making it possible for the local workforce to acquire even more knowledge and experience.
Taken together, improving the quality of the workforce will lead to the following benefits (Michie, 2002);
• It will result in increased productivity and profitability, as a consequence of workers being able to work harder and more efficient.
• Following is also an increase of the ability of the worker to absorb and to make better use of both experience and explicit knowledge.
• The overall increased quality of the workforce will lead to improved motivation, willingness and commitment.
available must be reached in order to maximize the positive effects associated with human capital enhancement.
Since the rising importance of human capital enhancement on the level of inward FDI, consequently leading to economic growth, countries have been through considerableinstitutional reforms in order to achieve a better educated workforce. Perhaps the biggest initiative in this area is UNESCO’s Education for All forum in which 164 countries worldwide agreed to enhance their formal education possibilities among children and (young) adults. This initiative stimulated governments in developing countries to invest more in education. Other initiatives were taken among several other countries: i.e. Indonesia, which with help of the Worldbank saw a significant increase of enrolment rates in three levels of education in the period 1970 – 2002.
Table 1 below provides an overview of the average years of education (since 1970) followed by the local labour force in each of the developing regions.
Source: Cohen and Soto (2001)
Overall, there have been significant improvements of the period spent in school. Especially the Middle East and North Africa saw the biggest increase with a more than four years higher average compared to the 1970’s, in contrary to Eastern Europe and Central Asia which saw only an increase of 2 years, but already started at a high rate of 5.8 years of average education.
Cohen and Soto (2001) provide us with the average years of education over a period of thirty years, though it does not tell us about the composition of the three levels of education. More recent data from UNESCO (2006) gives an accurate breakdown into the percentage that followed education at different levels. The following figure provides an overview of the Net Enrolment Rates3 in 2004, in the developing regions.
3 Enrolment of the official age group for a given level of education expressed as a percentage of the corresponding population
Middle-East & North Afrika
Sub-Saharan Africa Latin America & Caribbean
East Asia & Pacific
South Asia Eastern Europe & Central Asia
1970 1.6 1.7 4.5 3.2 1.9 5.8
1980 2.7 2.1 5.3 4.3 2.6 6.5
1990 4.3 3 6.7 5.4 3.1 7.1
2000 5.9 3.9 7.6 6.4 4.3 7.8
Net education enrolment 2004, developing countries 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%
Middle East & North Africa Sub-Saharan Africa Latin America and the Caribbean
East Asia & Pacific
South Asia Eastern Europe & Central Asia
Primary Secondary Tertiary
Figure 2: Net education enrolment 2004, developing countries
Source: UNESCO 2006
All less developed regions have significant levels of primary education available; at secondary level at least 50% of the total population is enrolled in this type of education. The only exception is the Sub-Saharan African region, of which a much lower part of the population is enrolled into the secondary level. Tertiary enrolment rates are in general low, except for Eastern Europe and Central Asia which has an average of 39.6% of the population enrolled into the highest level of education.
We can conclude that overall, the level of educated people in the developing part of the world has seen a steady growth due to institutional reforms. Although the increased education levels have in general positive effects on the attraction of FDI, not sure is yet the impact of each of the specific education type.
2.3 Institutional reforms and inward FDI
According to Bevan et al (2004) foreign investors see host institutions as a significant determinant of their location choice for foreign direct investment. In addition to factors such as low cost of labour or market size and growth, the institutional environment has increased in importance as a determinant of FDI since the late 1990s (Narula and Dunning, 2000, Meyer 2001). These studies prove that foreign direct investment is increasingly more attracted by the quality and availability of well developed institutions, motivating the choice of using Institutional theory as the foundation of this thesis.
According to North (1990), the institutional environment in the host country can be divided into a formal and informal area: formal institutions like government stability, openness to trade, formal education and informal institutions such as corruption.
developing formal education, resulting in an increase of survival rate by coping with competition. Incentives such as these will lead to organizational change, although this change will only happen in an incremental way and is mostly path dependent, given the specific nature of the institutional behaviour.
Especially acquiring competitive skills and knowledge through formal education gets particular attention in North’s work: “Given the objective function of institutional theory, a school or college, will enhance its survival possibilities and engage in acquiring the necessary skills and knowledge that will enhance its survival possibilities in the context of ubiquitous scarcity and hence competition”. And: “Organizations will induce public investments in those kinds of knowledge that they believe will enhance their survival prospects”. Hence, institutional theory explains for the change towards the situation of the highest pay-off, motivating institutional change in the area of formal education.
Furthermore, North (1990) states that the developing part of the world finds itself in its current state due to the fact that the institutional boundaries define a set of pay-offs to political or economical activity that do not support productive activity. In other words, the fundamental institutional framework of the developing world should be changed in order to meet the current demands; increasing the level of inward FDI. Reforms which result in increased inward FDI are the main contributor for economic growth in developing countries, (Borensztein et al 1998), as it is an important channel through which technological improvements are delivered (Bénassy-Quéré et al, 2005).
What makes the developing world an interesting object of research is that it is going through a series of institutional reforms, including in the area of formal education which is previously shown. This change is supported by the theory of institutions which motivates the move towards the situation of the highest pay-off, by improving the quality of host-country institutions.
2.4 Goal of research
The goal of this study is to define the location determinants that have the most significant influence on Foreign Direct Investments, seen from the institutional theory perspective. Our interest is specifically focused on the change in educational institutions, as previous research has defined a shift from quantitative to qualitative searching inward FDI in developing countries, defining an increased need for skilled labor. In this thesis we want to particularly determine the importance of improving formal educational institutions and therefore, start with asking the following question in order to capture the level of significance of formal education in attracting FDI;
Furthermore, we also want to capture the importance of institutional change in general in attracting FDI by asking the following question;
2. What is the impact of institutional change in the developing countries on attracting foreign direct investment?
The contribution to existing literature is as follows; by investigating whether FDI inflows are focused on quality or only on low cost advantages, we can define the importance of human capital enhancement through institutional reforms. We will make a distinction between the three different levels of education (primary, secondary and tertiary) which provides a better insight into the effectiveness of institutional reforms in each of these three sectors in the developing world. Simultaneously, we will asses the importance of well developed host country institutions in order to achieve higher FDI inflows, which are an important instrument for economic growth in developing countries. This will be done by selecting sub samples not only based on geographical proximity, but also by introducing an alternative data sorting technique: the simplified threshold approach, based on the work by Hansen (2000) which allows us to create more homogeneous sub samples leading to new, interesting insights.
3.0 Literature Overview
The following section consists of two parts; the first part provides an overview of the existing literature on the relation between education and FDI. We observed that the literature on the effect of formal education on FDI is rather limited. The reason could be the problem of constructing a database of quality variables of inward FDI; this is particularly difficult concerning human capital. According to Nunnenkamp & Spatz (2002), due to the fact that a variable like human capital is not widely used, there is a shortage of good data and consequently lack of research done. Data on traditional determinants like market growth and size, tax rates, trade openness and exchange rates are obtained much easier and therefore widely adopted.
This is followed by a second part on cross-country research on FDI in the developing world, specifically seen through the institutional lens. This part establishes the increased importance of local well developed institutions and highlights possible important determinants with respect to Foreign Direct Investments.
As stated earlier, we observed a shift from cost advantage to quality seeking FDI in the literature. Therefore, regarding education, we will make a distinction between cross country research done before the 1990s and after, defining two different categories.
We focus only on the developing part of the world, due to the observed change in determinants of FDI, which makes it an interesting object of research. Especially concerning the shift to quality seeking FDI, marking the need for institutional reforms.
3.1 Education and FDI
The following part provides an overview of all relevant and available cross-country studies in the developing regions on determining the importance of formal education in attracting FDI.
Hanson (1996) uses the adult literacy rate from 1960, 1965 and 1970 taken from UNESCO’s database as a proxy for human capital. In his cross country framework, focused on 105 countries in the developing world, he attempts to determine the effect of human capital on the attraction of FDI. The author concludes that the literacy rate, and thus human capital is of some significance but only when combined with life expectancy.
research that in the 22 developing countries the percentage of the population that followed a tertiary education have no clear relationship with inward FDI flows.
Other work from this period is a research by Schneider and Frey (1985); they conclude that the skill level of the employees, proxied with secondary education, has no significant effect on inward FDI. The focus is on explaining the flow of FDI in 80 developing countries for the years 1976, 1979 and 1980 by presenting a political-economic model, which associates real GNP per capita and a low balance of payments with higher FDI inflow. Political stability and bilateral-aid are of positive influence.
Finally, Root and Ahmed (1979) delivered a much quoted analysis of inward FDI in 70 developing countries in the period between 1966 and 1970. In their cross country analysis they test the significance of sixteen economic, and five social values including education. Especially when compared to previous studies, they take a wide variety of variables into account, although the economic ones are of main importance. The outcome for education, proxied with literacy rate and school enrollment is of no significance with respect to FDI, in contrast to government stability, GDP per capita & growth rate, infrastructure, rate of urbanization and economic integration.
Therefore, from the studies undertaken before 1990 we can determine that during that period there was no significant influence from education on the inflow of FDI. From this period, research on FDI determinants is extensive but mainly targets resourced-based FDI which is attracted by classic determinants as i.e. low cost of labor (Swedenborg 1979, Wheeler and Mody 1992), market size and growth (Dunning, 1980), exchange rates (Calderon-Rossel, 1985) openness to trade (Edwards 1990), tax rate (Hartman 1984) and trade barriers (Lunn 1980).
The second category of empirical research includes Bengoa et al (2003) who shows in his cross country research in determinants of FDI in 18 Latin American countries that the degree of economic freedom is an important economic variable in attracting FDI, but also signals the importance of other variables such as economic stability, liberalized markets and skilled labor that have to be available in order to achieve sustainable benefits. The dataset used in his research includes data from 1970 – 1999, and concludes that human capital represented by primary and secondary enrolment rates have significant influence on inward FDI.
growth rate and wage as independent variables and find that the labor force growth was of importance in attracting FDI, in contrast to labor wage.
Finally, the UNCTAD World Investment Report 2002 finds that there is a strong positive relationship between tertiary enrolment ratio and FDI inflows in 140 developed and developing countries for the periods 1988 – 1990 and 1998 - 2000. It states that investment in health, education and infrastructure are necessary to build up a knowledge based economy and to stay up with the global competition.
All together, we observe that especially human capital is an important determinant for skill seeking MNE’s, contrary to resource or market seeking FDI. Furthermore, there is a clear shift toward quality seeking FDI in the late 1980’s / early 1990s, which is driven by post-secondary education. This is consistent with the sub-Saharan African region that has the lowest level of education available and attracts FDI that is still mostly resource or market seeking (WIR 2006).
Primary education is of significance from the late 1980s and onwards, and it seems that this level of school is the minimum threshold in obtaining FDI inflows. Therefore, developing countries should provide at least this level of education for obtaining any inward FDI. As quality seeking FDI is becoming of more importance, countries should target this type of FDI by improving their educational institutions to provide for education at least at the secondary stage.
Given that an increase of primary education and consequently in the literacy rate has a clear positive relationship with respect to FDI inflows (Hanson 1989, Benevot 1989), the relationship between secondary and tertiary education and FDI is still unclear, according to Miyamoto (2003) there is no evidence of a specific type of education which has most effect on inward FDI. Most available studies use only literacy rate, primary, secondary and tertiary or a combination of at maximum two of these variables. This provides us with an interesting opportunity to asses the importance of these indicators for human capital.
3.2 Institutions and FDI
The following part provides an overview of all recent studies on determining the importance of well developed institutions in developing countries. We start by showing that although determinants based on cost advantages have a significant effect on FDI there is an increasing importance of a well developed institutional setting. After this the necessity of an efficient and stable political and regulatory system is introduced, followed by an explanation of the work done by Bevan et al (2004) who finds that especially sound macro-economic policies positively influence FDI.
developed human capital, low inflation and an efficient legal system promotes the inflow of FDI, in contrast to corruption and political instability. It also suggests that even without the availability of natural endowments, the “created assets” (Bevan et al, 2004) a country can still increase inward FDI levels by improving the institutional setting.
Concerning political institutional systems, Li and Resnick (2003) show that an improvement in democracy has a positive effect through increased property right protection, but at the other hand is negatively influenced by the increasing rules and regulations regarding the inflow of foreign capital. Furthermore, by making use of the international country risk guide index published by the PRS group4 as a measurement of institutional quality, Meon and Sekkat (2004) find a negative effect of political risk in general on FDI, but not for corruption, government effectiveness and rule of law specifically. This also supported by Bénassy-Quéré et al (2005) which examines the effect of institutional quality on bilateral FDI flows. They point out that low corruption, efficient tax systems, transparency, well established property rights, contract and justice efficiency have a positive effect on FDI. Other work in this area is done by Daude and Stein (2001), and Kaufman et al (2003) they find a significant relationship between political stability, government effectiveness, rule of law, voice and accountability, and regulatory quality. Also they find that corruption is a questionable variable of determining FDI, contradicting Wei (2000) who finds that corruption is negatively influencing inward FDI.
Bevan et al (2004) analyses the result of institutional change also on an aggregate level. They find that the improvement of formal institutions has a positive effect on attracting Foreign Direct Investment, and that foreign investors see host institution quality as a significant determinant of their investment location. Especially financial systems, foreign exchange, trade liberalization and legal development have a positive effect on FDI.
The preceding studies all use different datasets, from different periods and different methodological approaches in determining the significance of institutions in attracting FDI. In general they all agree on the importance of the availability of developed institutions. We conclude that not only traditional variables as market size and growth, low labor costs or the availability of natural endowments are the only determinant for investment. Factors concerning the financial and political systems, trade, corruption and legal development have also positive influence on FDI. Furthermore, of the studies above that take any form of human capital into account, find a significant relationship with respect to FDI.
3.3 Scope
The purpose of our work is to focus on the changes of the institutional setting in developing countries and the effect on inward FDI. We observed a shift from quantity seeking FDI, focused on traditional
4
determinants like market size and growth, or low cost of labor to quality seeking FDI, and consequently the need of a better educated workforce. Furthermore, we asses the importance of this need of skilled workers from the perspective of the institutional theory, and simultaneously we will capture the level of importance of well developed local institutions.
Earlier we defined regional differences in the FDI inflow and development level of educational institutions among the developing countries; therefore we will divide the data in different regions, in order to avoid regional dissimilarities to distort the outcomes. This is supported by Noorbaksh et al (2001) who warn for biased results when research is done at the aggregate level. Rather he encourages the application of analysis at the sectoral level in future research, which will provide for better and clearer insights. Also we will also make use of an alternative approach in defining the FDI determinants; this method is based on Hansen’s (2000) work, which consists of a data sorting technique that allows for high / low regimes, i.e. answering questions like; what are the differences in the effects of primary education on inward FDI in countries with on the one hand high and on the other low corruption?
As stated earlier, this thesis uses primary, secondary and tertiary level of education in order to grasp the institutional change in formal education affecting FDI inflows. This approach is unique compared to other studies as they only apply a maximum of two of these variables simultaneously. Also, the extent to which countries are covered is mostly limited, in contrast to our study, which represents all the different regions among the developing countries in order to provide a good overview of the current status of institutional reform.
4.0 Arguments and hypotheses
In the introduction to this thesis we described a movement among the developing countries towards institutional reform in the area of formal education, given the positive aspects of FDI and the increasing importance of quality searching foreign direct investments. This institutional reform in order to attract more FDI is motivated by the institutional theory as it will lead to enhanced survival in our global economy based on competition and scarcity (North, 1990). Given the objective function of the institutional framework, namely profit maximization, it will acquire the necessary skills and knowledge that induces the highest pay-off. Hence, from the institutional perspective, a host country government will adapt its policy on formal education in order to increase FDI flows, leading to institutional change through adaptation of the regulatory environment.
As explained earlier we will also use, next to the more traditional approach of dividing countries into sub samples of geographical proximity, a new technique of data sorting based on Hansen’s (2000) earlier work. Therefore we are not only referring to countries, but also to specific high/low regimes (high = above median of a sub sample based on an independent variable, vice-versa for a low regime). A detailed explanation of the adaptation of the original model to the simplified version and its use will be presented in the following chapter.
Improving the country’s stock of human capital will improve FDI inflows (Noorbaksh et al, 2001) by making the location factors of the host market more attractive to foreign investors. From the literature we concluded that skilled labor, able to handle more complex tasks is important in attracting quality searching FDI and according to the work of several scholars, we observed that in general, in order to attract any form of FDI at least primary level education is required. In the case of quality searching FDI there is at least the level of secondary education necessary (Borensztein et al, 1998). All together, human capital is since the 1990s an increasing more important determinant of FDI, especially for the growing demand of skill seeking foreign direct investments.
Previous research only used education at primary, secondary, tertiary level or any combination of two of these in order to grasp the institutional change in formal education affecting FDI inflows. In this study, we will use three of these education levels simultaneously. In order to grasp the importance of education in attracting FDI the following:
Hypothesis 1: Improving formal education has a positive effect on FDI inflows.
mostly multinational companies, are attracted by low transaction costs; costs associated with negotiation and enforcement (Bevan et al, 2004). The institutions provide the “rules” which can ultimately lead to a decrease of the transaction costs by shaping a well developed, efficient institutional setting. It is especially in the developing world where there are incomplete or inefficient institutional frameworks to be found. As most of these countries are putting effort in improving their institutional setting (i.e. lowering corruption, legal reforms, improving education etc), it makes an interesting opportunity to study this change and the effect on FDI inflows.
Previous empirical research showed us the importance of the formal and informal institutions in creating an attractive investment climate in order to increase inward FDI. Well developed institutions will have a positive effect on FDI, therefore we propose:
Hypothesis 2: Developing countries with an attractive investment climate shaped by institutions will attract more FDI.
Having already defined the importance of institutions in general and education specifically, we still have to look at a more detailed level into the subject. Therefore we have to determine the importance of other specific institutions. We will start by capturing the significance of the macro-economic situation in the host country. A high level of economic and financial risk is negatively associated with FDI inflows (Bevan et al 2004), due to increased instable financial and economic institutions. Risks associated with financial/economic institutions can be caused by high variability of exchange rates or non-availability of efficient working capital markets (De Soto, 2000). A developed financial infrastructure will lower transaction costs by making services as international fund transfer and payment more accessible. Also improved access to local funding possibilities makes it possible to avoid exchange rate risks and simultaneously providing credit opportunities for host market customers (Bevan et al, 2004).
Therefore we propose:
Hypothesis 3a: Well developed macro-economic policy will lead to higher FDI inflows.
is that these are also enforced by the government in order to improve credibility with Multinational Companies. If there would be any uncertainty regarding the enforcement of legal and regulatory institutions the perceived risk by investors will lead to a lower FDI inflow. Therefore:
Hypothesis 3b: The availability of effective legal and regulatory institutions will increase FDI inflows.
Another formal institution which influences FDI is the availability of a stable political system, if absent it may cause for political risks through changes in government policy that negatively affect investments, leading to less attention from potential investors. This is of importance given the long term nature of Foreign Direct Investments, the risks of (civil) war, nationalization, or even changes in tax laws. Hence, a stable political environment will have a positive effect on FDI inflows. Also shown in the literature on institutions and FDI, in general democracy has a stabilizing effect on political institutions and consequently leads to a higher degree of political and civil liberties (Hodgson, 2006). And from the work done by Harms and Ursprung (2002) we can derive a clear positive relationship between civil and political freedom and the inflow of FDI. Therefore we propose the following regarding the political environment:
Hypothesis 3c: A stable political environment will have a positive effect on FDI inflows.
“Corruption involves behavior on the part of officials in the public and private sectors in which they improperly and unlawfully enrich themselves and/or those close to them, or induce other to do so, by misusing the position in which they are placed” (Worldbank). As Wei (2000) already concluded in his study on FDI, corruption, as an informal institution, has a negative impact through increased uncertainty about the well functioning of the available host market institutions. Also, supported by the World Economic Outlook report 2003, it finds that the quality of institutions, of which corruption is a part, has significant influence on the inflow of FDI. And previously defined, the effective enforcement and credibility of host country institutions has a positive impact on inward FDI.
FDI is not always deterred by the existence of corruption in the host country; as MNE’s are often driven by low cost of labor, market size and growth, they will still enter the market in order to benefit from low costs. This is especially apparent in the case of a first mover advantage (Banerjee et al, 2006) where the MNE will enter even a weak institutional environment in order to gain a monopoly position. Nonetheless, although the effect of corruption on FDI is not fully clear, we still believe that corruption as a measure of institutional quality has a negative effect on FDI. Therefore:
5.0 Methodology
5.1 Developing countries
In order to grasp the full effect of institutional change and especially formal education on FDI, we will make use of a panel of 66 countries from the developing world over the period 1995 – 2004. In this sample the following regions are represented: Latin America and the Caribbean, Central and Eastern Europe, Asia, Sub-Saharan Africa and Middle East/North Africa (See appendix for a detailed per-region overview of all countries).
First, we selected the appropriate countries according to the United Nations overview of developing countries, and determined the time period from 1990 until 2006, in order to fully capture the effects of institutions which been subject to increased importance for FDI since the 1990s. Furthermore, the countries were grouped into specific regions, defined by UNCTAD and suggested by Noorbaksh et al (2001) who warns for biased result for research at the at aggregate level. In this way, we will avoid the problem of distorted outcomes due to country dissimilarities.
After collecting all the appropriate data, we decided to delete some countries due to missing numbers caused by shocks such as (civil) war, draughts, floods or any other external shocks. Some countries were not removed from the sample as linear interpolation (maximum of 10%) was applied in order to control for the missing values in education and corruption, especially pre-2000.
We finally settled with a sample of 66 countries for the 10 year period in between 1995 – 2004, divided into five regions. Although fewer countries than anticipated, we still believe by covering almost 77% of the developing countries’ FDI inflows (2004) to have a good overview of the determinants of FDI.
5.2 Simple threshold approach
An alternative approach to determining the significance of our variables is by applying Hansen’s (2000) data-sorting method. This technique is based on the division of data into several regimes; the programs5 provided by Hansen allow for two “groups” called regimes, within the range of an independent variable. The basic idea behind this is that a sample (i.e. a selection of countries) can be divided into a high and low group for every independent variable, by using Hansen’s data-sorting method. Each of these regimes has a different effect on the change of the dependent variable. For example; low primary education / high primary education or low corruption / high corruption, and the
specific effect of each regime on FDI inflows. The threshold size of a sample is selected endogenously by the program provided by Hansen, separating an independent variable into a high and low regime. This approach is supported by several scholars such as Papageorgiou, 2002 (threshold effect of trade on growth); Ho, 2005 (threshold effect of inflation on PPP) Borensztein et al, 1998 (human capital threshold for FDI inflow), who successfully applied threshold estimation on their country sample.
We believe that although FDI determinants can be grouped into regions which have similar traits the grouping method that selects countries according to their specific variable size is also a feasible way of determining the significance of determinants of FDI inflow.
Accordingly our complete data sample will be treated as by Hansen (2000) but, as the econometric approach falls beyond the scope of this thesis, we opt for the use of a “simple threshold application”: the sample will be split for every independent variable into a high and low regime, leaving other variables untouched, not selected by the programs provided by Hansen, but manually split into 50% high and 50% low regime6.
The sample split is based on the starting year 1995, as suggested by Hansen (2000). From this point we will run a panel data estimation in order to observe whether there are significant effects on the inflow of FDI, taking into account different regimes but leaving out the variable that determines the high and low regimes. Excluding this variable makes interpretation of the results easier, i.e. if a low-regime sample threshold variable has a coefficient of 0.5 it results into an increase of the dependent variable with 0.5 if the independent variable increases with 1, but this only holds if this increase does not result into the country being shifted to the high-regime of that specific country sample.
By using this approach in addition to our region panel data estimates we will also determine what the specific effect of the independent variables are of a country with i.e. low / high education or low / high corruption on FDI inflow. This will result in 20 separate panel data estimations, as every independent variable is grouped into a high and low regime.
5.3 Data
The following part provides an overview of all variables used in our regression analysis regarding institutional determinants of FDI.
The dependent variable is inward FDI flow as we focus on measuring the change in inward investment flows determined by institutional development. We considered taking absolute FDI inflows but finally settled with FDI divided by GDP, as in our data-pool the country sizes vary, especially regarding population and market size. Therefore we will use in our research FDI divided by GDP in order to take
into account the size of every country-specific economy. This will provide us with a better overview as this measure takes into account the relative size of FDI with respect to the economy as a whole.
The data for this variable has been extracted from the United Nations Conference on Trade and Development database7, which provides FDI data from 1970 and onwards.
FDI is defined in our thesis as a long term investment in a company abroad. This investment results in a certain degree of control of the foreign subsidiary; defined by holding at least 10% of the foreign entities’ ordinary shares or voting power. The data provided by UNCTAD is, in our opinion the most reliable source as it collects its FDI information directly from the concerned state-level agencies and furthermore is complemented with data from other supra-national organizations such as the International Monetary Fund, the World Bank and the Organization for Economic cooperation and Development.
Regarding formal education, improving the country’s stock of human capital will increase FDI inflows (Noorbaksh et al, 2001) by improving a country’s institutions of formal education. As defined earlier we opted to use three levels of education in order to define the importance of each of these specific types of education. The amount of the population who followed a specific type of education is represented by the Gross Enrollment Ratio (GER):
GER measures total enrollment in a specific level of education, which is expressed as the number of students enrolled into that type of education, (E) divided by the part of the population which forms the theoretical age group, (P) which is the group of people that is supposed to follow this level of education (Defined by UNESCO8). This measure of enrollment does not take into account the age of students, in contrary to the Net Enrollment Rate, which does take into account the official age group for a given level of education (Worldbank.org). The GER also includes i.e. adults or late entrants, which can results in a higher actual enrollment than the theoretical maximum enrollment of the age group, leading to the possibility of the GER score exceeding 100%. Limitations regarding the use of this data consist of only being able to measure the entrants into a particular level of education, it would be more appropriate to measure the number of students who actually finished their education, and instead of the flow, rather the stock of available human capital, as shown by Noorbaksh et al (2001).
7Available from: http://stats.unctad.org/FDI/ReportFolders/ReportFolders.aspx?CS_referer=&CS_ChosenLang=en
Although the Gross Enrollment Ratio is not the most ideal indicator for each of the three levels of education as we rather use the stock of graduated students, we opt for these variables given the extent of the availability of these data over the time period 1995 – 2004.
The data on enrollment rates is gathered from multiple sources within UNESCO, but mainly from UNESCO’s initiative “Education for All” for the period 2002 -2004 and earlier data from UNESCO’s institute of statistics9. Given the limited availability of tertiary enrollment data this variable was the main reason behind the specific size of our country sample.
From the work done by Bevan et al (2004) we explained that a sound macro-economic policy has a positive influence on inward FDI. Our indicator for macro-economic stability is inflation; we believe that low inflation as a result from sound macro-economic policy will lead towards increased FDI inflows, as macro-economic policy volatility results into higher risk for foreign investors and thus a less attractive investment climate.
We use instead of the Consumer Price Index (CPI), the GDP deflator (Nominal GDP divided by Real GDP), different from the CPI which is based on the price of a fixed basket of goods. Therefore, this measure is more accurate than using CPI as it measures the price change of an economy as a whole. Also, by using the GDP deflator, we avoid possible manipulation of the inflation rate through adjusting the prices of the goods in the basket by governments.
Furthermore, in order to cover the legal and regulatory institutions, a significant determinant of FDI as proven by Daude and Stein (2001) and Kaufman et al (2003), we added the index of economic freedom from the Heritage Foundation10, an US-based conservative think-tank that publishes research on several domestic and foreign policies. This index was first published in 1995 and covering 161 countries worldwide, defines 10 measures that are critical in the development of a country’s wealth. The indices provided by the Heritage Foundation consists of several indicators such as freedom from government (government expenditures and revenues as a percentage of GDP and total revenues consequently), investment freedom, fiscal freedom, property right enforcement, labor freedom, monetary freedom, trade and business freedom. All factors are equally weighted and form a part of the final score. As corruption is also included in the index, we decided to use the average of the nine other indicators, thus excluding corruption as it is already covered by the index from Transparency international.
The corruption Perceptions Index (CPI), first published in 1995 by Transparency International, is the most widely used indicator of corruption in research on FDI. This index covers more that 150
9 Available at: http://www.uis.unesco.org/statsen/centre.htm
countries, and is probably the most reliable measurement of corruption available. The index consists of a 10 point Likert scale (10=no corruption, 1=extremely corrupt) and reflects the perceived corruption among a country’s public officials and politicians. Corruption as an informal institution is expected to be of negative influence on inward FDI flows. We use the CPI, and complement missing values from the corruption index from the Heritage foundation, which uses the same methodology in capturing corruption as Transparency International.
Changes in government policy can affect FDI negatively through i.e. nationalization policies or changes in tax rates. Hence, a stable political system will positively influence FDI by offering reduced investment risks. We use the indicators provided by the Freedom House11, namely political rights and civil liberties as a proxy for stable government policy. Political rights covers the ability for a country’s population to participate in the political system, civil liberties is related to the right of expression and beliefs, associational and organizational rights, rule of law, and personal autonomy without interference from the state (Freedomhouse.org). This survey first started in 1972 offers an extensive and complete coverage of 151 countries worldwide. We expect a positive relationship between political rights, civil liberties and FDI.
Openness to trade, measured as exports minus imports divided by GDP, will be used as a proxy for an attractive investment climate in the host country. We believe that the amount of trade is the most appropriate measurement of the development of the host market, and defines the attractiveness of the local business climate taken into account the availability of the data over the time period and the number of countries. Data is provided by the World Bank Group’s, World Development Indicators (WDI CD-ROM 2006), the most extensive database available on macroeconomic variables.
Included in our analysis are two variables controlling for non institutional determinants of FDI12, namely market size and growth. Although shown previously, countries lacking a large market or natural resources are still able to attract FDI by providing an excellent investment climate through institutional development (Asiedu 2006), we believe in the importance of market size and growth. Market size is represented with GDP per capita in order to capture the economic potential of every individual within a country. This is expected to be of positive influence on FDI as Foreign Investors are not only attracted by well developed investment conditions but also by market size and especially market growth potential, in our analysis represented with GDP annual growth rate. Both variables are taken from the World Development Indicators CD-ROM 2006.
11 Available at: http://www.freedomhouse.org/uploads/fiw/FIWAllScores.xls
5.4 The Econometric Model
First, in order to capture the impact of the change of the independent variables on the difference of FDI inflow, the following model:
FDIGit = β1t + β2INFLit + β3ECFRit + β4CPIit + β5CVLit + β6POLit +β7OPENit + β8PRIit + β9SECit
+ β10TERit + β11GDPGit + β12GDPCit +
ε
itThe dependent variable is FDI divided by GDP, independent variables are inflation, economic freedom, corruption percentage index, civil liberties and political rights, openness to trade, three levels of education and the control variables GDP per capita and GDP growth, followed by the error term. The subscripts refers to country i and year t respectively.
Our data consists of 66 countries in the 10 year period 1995 – 2004, each country has 10 observations. Thus, given the time series and cross-section characteristics of the data, most appropriate would be a random or fixed effects model. Using ordinary least squares estimation could in our case lead to bias due to unobserved heterogeneity and omitted variables. Panel data enable correction of these biases, but at the cost of degrees of freedom, and the loss of explanatory variables that do not vary within the cross section unit (picked up by intercepts) (Carter Hill et al, 2001).
A random effects model, assumes that each specific effect is randomly distributed. Given that the assumptions for using the random effects model hold, it provides some extra degrees of freedom, and thus a more efficient model resulting in more accurate P-values. However, if not all the assumptions of the model are valid there will be inconsistency within the model.
A fixed effects model makes assumes that the data comes from normal populations, but differs in their means. In this case only the intercept parameter varies, although not over time. Hence, the fixed effects model treats the intercept β1t as a fixed unknown parameter, in contrast to the random effects model
which regards intercept β1t as a random drawing from the population distribution of countries (Carter
Hill et al, 2001). In the following chapter test specification will determine the appropriate econometric approach for our analysis.
6.0 Testing the model
6.1 Descriptive statistics
FDIG COR EFR GDPC GDPG INF OPENESS POL PRI SEC TER Mean 3.71 3.63 60.29 3006.11 4.26 16.12 83.59 3.40 99.01 64.05 21.92
Median 2.76 3.40 61.02 1754.73 4.39 6.69 73.86 3.00 103.56 70.77 18.30
Maximum 45.15 10.00 82.32 23408.15 38.20 948.53 213.33 7.00 155.22 111.80 89.90
Minimum -3.20 0.40 27.28 90.19 -13.13 -13.97 12.80 1.00 38.71 5.24 0.40
Std. Dev. 4.27 1.76 10.32 3649.28 4.20 56.38 40.30 1.96 18.34 26.87 17.89
Table 2: Descriptive statistics
Table two gives an overview of the descriptive statistics. Regarding primary enrolment rates, we observe a high mean, and a maximum value reaching 155% due to the fact that the Gross Enrolment Rate which can result in a higher actual enrolment rate than the theorerical maximum. Also we observe a low standard deviation, resulting from a low spread of the values in the data-set.
6.2 Correlation matrix
From the correlation matrix below, we observe a high correlation between political rights / civil Liberties and therefore remove CVL from the equation, leaving political rights to proxy for a stable political environment. Solving multicollinearity is possible through increasing the sample data (Cohen et al, 2003), but in our case this is not a feasible option due to limitations in data; hence we decided to exclude civil liberty from the equation.
The problem with multicollinearity is that if apparent, both variables don’t provide sufficient information to estimate the separate effect on the dependent variable (Carter Hill et al, 2001). It can result into wrong (high) P-values, even when the independant variable is a signifincant determinant of the dependant variable.
6.3 Model specification
Given the time-series and cross section characteristics of our data set, we opted for the use of a fixed or random effects model. In order tot test our assumptions we first determined whether fixed effects estimation is the appropriate technique compared to the common effects model. Eviews 5.1 provides the redundant fixed effects test, a way to asses the choice between the common and fixed effects model.
Redundant Fixed Effects Tests
Effects Test Statistic d.f. Prob. Cross-section F 8.620483 (65,583) 0.0000 Cross-section Chi-square 443.845957 65 0.0000 Table 4: Redundant fixed effects test
From table 4, we can read that both test statistics indicate significant differences from H0, which should therefore be rejected in favor of the alternative hypothesis, a fixed effects model.
Our next step is to determine whether it is possible to make use of the random effects model, which would provide us with extra degrees of freedom and thus a more precise estimation, given that the application of random effects estimator is justified.
A widely used method is the Hausman test; according to the table below the use of a random effect estimation is appropriate as H013 is not rejected at the 5% level of significance.
13 H
0: Random and Fixed effects model are both accepted H1: Only Fixed effects model is accepted Source: Eviews 5.1
COR CVL EFR FDIG GDPC GDPG INF OPENESS POL PRI TER SEC
COR 1 CVL -0.151 1 EFR 0.513 -0.481 1 FDIG -0.006 -0.109 0.074 1 GDPC 0.681 -0.087 0.458 -0.032 1 GDPG -0.016 0.058 -0.005 0.107 0.019 1 INF -0.150 0.157 -0.296 -0.052 -0.109 -0.32 1 OPENESS 0.280 -0.144 0.097 0.242 0.197 0.011 0.039 1 POL -0.099 0.911 -0.411 -0.063 -0.019 0.054 0.122 -0.109 1 PRI 0.015 -0.15 0.117 0.062 0.024 -0.074 0.005 0.084 -0.125 1 TER 0.190 -0.401 0.303 0.084 0.368 -0.021 0.037 0.16 -0.377 0.196 1 SEC 0.229 -0.377 0.179 0.106 0.343 -0.066 0.054 0.226 -0.341 0.449 0.694 1
Although the random effect estimator is accepted, we opt for fixed effects, as we are using fixed effects estimation at the regional level, and prefer to use a homogeneous approach to all regions, taking the loss of degrees of freedom for granted.
Correlated Random Effects - Hausman Test Test Summary
Chi-Sq.
7.0 Results
Table 6 and 7 present the results of our empirical analysis on the institutional determinants of FDI. We applied the fixed effects estimator on our dataset from 1995 – 2004, first on the developing countries all together then followed by a breakdown by regional level and sub samples defined by high / low regimes. The purpose of dividing our country sample into separate regions and regimes defined by country similarities is to determine whether institutional determinants are significantly different per region or regime. The results turned out to be indeed different at regional, global level and for the different regimes.
7.1 All Developing Countries
In our sample of all developing countries together, for education, none of our hypotheses are supported. All three indicators for level of formal education are not significant determinants of FDI at the aggregate level, although at the regional level there are different results.
Also there are no significant roles for corruption and political freedom, neither for market size and growth.
Economic freedom is positive and significant at the 1% level, determining the importance of effective legal and regulatory institutions. This outcome is consistent with Bénassy-Quéré et al (2005), Daude and Stein (2001) and Kaufman et al (2003), although not throughout the rest of our analysis, as only one region (Central and Eastern Europe) shows a significant relationship with our FDI variable.
Inflation, our proxy for macro-economic policy is significant at the 1% level, indicating a negative relationship with respect to inward FDI. Low inflation will lead to increased FDI inflows, and thereby supporting our hypothesis on macro-economic policy on a global scale, by providing lower risks for potential investors.
Dependent variable: FDI divided by GDP
Variable All Countries Asia CEE LAC MENA SSA
Constant -4.798030 (0.0838) -4.668894 (0.6727) -20.83292 (0.0142) 1.390692 (0.7958) -8.997887 (0.6595) -6.525327 (0.1513) Corruption -0.069101 (0.6927) 0.296706 (0.7199) -0.372823 (0.2265) -0.416968 (0.1489) 0.698953 (0.2265) 0.694918 (0.0307**) Economic Freedom 0.094662 (0.0039***) -0.044983 (0.6536) 0.082308 (0.1154) 0.119405 (0.0487**) (0.059563) (0.6915) 0.048591 (0.5398) GDP per Capita -0.000399 (0.3288) 0.000229 (0.8951) -0.001582 (0.0600*) -0.000688 (0.2735) -0.001983 (0.1455) 0.002286 (2399) GDP Growth -0.021671 (0.5205) 0.068434 (0.4566) 0.004054 (0.9545) 0.043394 (0.5206) 0.026593 (0.8759) -0.000956 (0.9895) Inflation -0.005411 (0.0372**) -0.023960 (0.0078***) -0.001686 (0.5207) -0.054318 (0.0517*) -0.022076 (0.7036) 0.002942 (0.7197) Openness to Trade 0.051834 (0.0000***) 0.164949 (0.0001***) 0.064550 (0.0077***) 0.045527 (0.0562*) 0.079844 (0.2074) 0.014541 (0.4999) Political Freedom -0.131805 (0.5571) 0.154255 (0.7795) -1.013163 (0.0400**) -0.236397 (0.4059) -0.956607 (0.6292) -0.331354 (0.4230) Primary Enrolment 2.053374 (0.3336) 15.84664 (0.7795*) 18.39015 (0.0048***) -1.606689 (0.6817) 6.874126 (0.4500) 1.231974 (0.6470) Secondary Enrolment -0.024681 (0.2572) -0.258928 (0.0056***) 0.039041 (0.3935) -0.020323 (0.4990) 0.146862 (0.2489) 0.032668 (0.7256) Tertiary Enrolment 0.003634 (0.8750) -0.212507 (0.0924*) 0.016415 (0.5719) -0.016147 (0.6960) 0.050693 (0.6927) -0.111176 (0.6131) R2 0.54 0.70 0.48 0.61 0.39 0.42 Observations 669 139 140 160 100 120
Table 6: Regression output
Simplified threshold regression output14, dependent variable: FDI divided by GDP
Regime PRI SEC TER EFR GDPC INF OPENESS GDPG POL COR R2 Observations
PRI high -0.046569 (0.1427) 0.012933 (0.7871) 0.036923 (0.5188) -0.000252 (0.7676) -0.017209 (0.0110**) 0.056445 (0.0057***) -0.070221 (0.2404) 0.141449 (0.6953) 0.105048 (0.7497) 0.55 329 low 0.018569 (0.5079) -0.007706 (0.7232) 0.116634 (0.0013***) -0.000298 (0.4380) -0.002245 (0.3073) 0.049024 (0.0010***) 0.015649 (0.6567) -0.469316 (0.0813*) -0.185774 (0.2937) 0.56 330 SEC high -2.668545 (0.5362) 3.89E-05 (0.9989) 0.084001 (0.0899*) -0.000171 (0.7669) -0.006655 (0.0349**) 0.084196 (0.0000***) -0.081355 (0.1232) -0.321407 (0.4371) -0.238150 (0.4086) 0.59 330 low 4.268385 (0.0521*) -0.049668 (0.3244) 0.107024 (0.0227**) -0.002257 (0.0092***) -0.002373 (0.7436) 0.018107 (0.2583) 0.020694 (0.6368) -0.216179 (0.4028) -0.157132 (0.4647) 0.42 329 TER high 6.923466 (0.1718) -0.028197 (0.4311) 0.102124 (0.0347**) -0.000805 (0.1186) -0.006231 (0.0559*) 0.095289 (0.0000***) -0.061589 (0.2250) -0.064111 (0.8778) -0.035148 (0.9019) 0.56 330 low 1.310297 (0.5103) -0.014161 (0.5738) 0.126332 (0.0049***) -6.97E-05 (0.9016) -0.000923 (0.8919) 0.003905 (0.7888) 0.007815 (0.8532) -0.288337 (0.2253) -0.220292 (0.2635) 0.52 329 EFR high 3.463394 (0.2673) -0.048453 (0.0807*) 0.027907 (0.3639) -0.000671 (0.1279) -0.037672 (0.0910*) 0.039888 (0.0222**) -0.079184 (0.0567*) 0.048061 (0.8649) 0.037723 (0.8819) 0.47 330 low 2.626646 (0.3755) -0.008299 (0.8211) -0.056951 (0.1459) 0.001484 (0.1262) -0.004193 (0.1482) 0.070746 (0.0001***) 0.042367 (0.4599) -0.332051 (0.3742) -0.081966 (0.7402) 0.59 329 GDPC high 4.305549 (0.1880) -0.014424 (0.5716) 0.004592 (0.8269) -0.003763 (0.9330) -0.019888 (0.2261) 0.038092 (0.0273**) -0.116078 (0.0059***) 0.227392 (0.4616) 0.085403 (0.7153) 0.44 330 low 1.082473 (0.7027) -0.057812 (0.1586) -0.016153 (0.7537) 0.141002 (0.0032***) -0.002553 (0.3813) 0.061754 (0.0014***) 0.089813 (0.0956*) -0.373426 (0.2445) -0.202406 (0.4262) 0.62 329 INF high 4.580056 (0.1649) -0.006255 (0.7991) 0.020902 (0.3858) 0.009797 (0.8293) -0.000374 (0.3516) 0.039795 (0.0217**) -0.103759 (0.0127**) 0.205992 (0.5058) 0.042551 (0.8584) 0.44 330 low -2.478899 (0.4280) -0.003319 (0.9308) 0.033789 (0.4545) 0.107669 (0.0425**) -0.000240 (0.6261) 0.020412 (0.2480) -0.051863 (0.2435) -0.295285 (0.2997) 0.203959 (0.4273) 0.46 329 OPENESS high -3.345732 (0.4476) 0.014096 (0.7704) 0.010157 (0.7862) 0.145887 (0.0117**) -0.000467 (0.5065) -0.004732 (0.1868) -0.050646 (0.3933) -0.533148 (0.3286) 0.178299 (0.5992) 0.49 330 low 6.420033 (0.0001***) -0.034619 (0.0393**) 0.016620 (0.4798) 0.063855 (0.0314**) -0.000447 (0.2786) -0.018066 (0.0072***) 0.014819 (0.6148) -0.008437 (0.9549) -0.144867 (0.2798) 0.63 329 GDPG high 2.103504 (0.3736) -0.048043 (0.0486**) 0.115391 (0.0000***) 0.012106 (0.7534) -0.000694 (0.1246) -0.000276 (0.9784) 0.009800 (0.5148) 0.136225 (0.5062) -0.491748 (0.0057***) 0.57 330 low 0.664689 (0.8455) 0.025744 (0.4837) -0.070390 (0.0493**) 0.130506 (0.0136**) -0.000128 (0.8390) -0.003035 (0.3251) 0.067568 (0.0008***) -0.590265 (0.1810) 0.384232 (0.2036) 0.56 329 POL high 2.971322 (0.3901) -0.046505 (0.2904) 0.011660 (0.8344) -0.092771 (0.0814*) -0.000691 (0.4225) -0.007567 (0.1013) 0.075564 (0.0009***) -0.030519 (0.5634) -0.092771 (0.7326) 0.56 330 low 1.765347 (0.4844) -0.008468 (0.7108) 0.000229 (0.9918) 0.087113 (0.0388**) -0.000197 (0.6352) -0.003406 (0.2311) 0.034364 (0.0144**) -0.029228 (0.5035) -0.176252 (0.4350) 0.51 329 COR high 2.527292 (0.4124) 0.012016 (0.6756) 0.004714 (0.8563) 0.070337 (0.1505) -0.000779 (0.0561*) -0.002088 (0.5160) 0.066836 (0.0001***) 0.045570 (0.3186) -0.326435 (0.3705) 0.46 330 low 1.851603 (0.5330) -0.056350 (0.0908*) -0.037809 (0.4325) 0.109573 (0.0141**) 0.001435 (0.2505) -0.009267 (0.0263**) 0.037578 (0.0508*) -0.089810 (0.0788*) -0.125177 (0.6720) 0.60 329
Table 7: Manual threshold regression output
***significant at 1% level, **significant at 5% level, *significant at 10% level
7.2 Regional level
We obtained mixed results from our regression at the disaggregate level. The region with the highest number of significant variables, although not always with the expected sign is Asia. Furthermore, we obtained significant results from the Central / Eastern European region and Latin America / Caribbean area. Middle East / North Africa and the Sub Saharan Africa region show that our institutional indicators provide for almost no explanation of the determinants of FDI.
With respect to our variables representing education, after non-significant results at the aggregate level, we found significant relations between primary education and FDI inflows in Asia (10% significance level) and Central and Eastern Europe (1% significance level) both with the expected signs. The outcome for secondary and tertiary education only finds a significant relation in Asia (1% and 10% significance respectively) although both without the expected sign.
Corruption, not significant on the aggregate level turned out to be significant at the 10% level in the Sub Saharan Africa region, although with the wrong sign, being the only variable determining FDI in that region, according to our results. Therefore, our hypothesis on the negative effects of corruption on FDI is not supported in our analysis.
On the aggregate level, our variable explaining for the relationship between FDI and the legal/regulatory institutions finds a positive correlation, but at regional level this is only supported in the Latin America and Caribbean area.
GDP per capita, a proxy for market size and together with GDP growth (market growth), don’t prove to be significant in most regions. GDP per capita does have a significant effect on FDI in the Central and Eastern European region, although with the wrong sign.
We achieved more significant results for inflation, determining FDI at the global level, it also proves to determine FDI inflows in Asia (1% significance) and Central and Eastern Europe (1% significance) both with negative signs, supporting our hypothesis on macro-economic policy.
Our indicator for an attractive investment climate (hypothesis 2), openness to trade, finds a positive and significant relation with inward FDI in Asia (1% significance), Central and Eastern Europe (1% significance) and Latin America and the Caribbean area (10% significance).
7.3 Simple threshold approach
Our alternative method in identifying the determinants of FDI by using the manual threshold approach indeed revealed different results. To complete the analysis we added to the appendix also the output which takes into account in the regression the variable that is being split into a high and low regime. Although the results are not much different, we prefer to exclude the variable that is being divided into regimes given our argument on easier interpretation earlier.
Our hypothesis on primary education is supported in only a few situations: it has a significant relationship regarding inward FDI in countries with low secondary education (10% significance) and low degree of openness (1% significance). With respect to secondary education its effect on FDI is in overall negative in the case of a high economic freedom regime (10%), low openness (5% significance), high GDP growth (5%) and low corruption (10%), hence rejecting the hypothesis on the positive effects of secondary education on inward FDI. Our tertiary education hypothesis is only supported in case of the high GDP growth regime, low GDP growth shows a significant relation with FDI, though negative.
Economic freedom, our variable representing legal and regulatory institutions finds a significant effect in al regimes, except for high primary education, high inflation high GDP growth and high corruption. Furthermore, our hypothesis on the importance of legal and regulatory institutions is not supported in case of high political freedom which proves to be of negative effect (10% significance level) on FDI, but of positive effect in case of low political freedom (5%).
Market size, proxied with GDP per capita, only is of significance in case of low secondary education (1%) and the high corruption regime (10%). Our other proxy variable in the model; namely GDP growth, finds mostly negative relationships with respect to FDI, except in case of low GDP per capita (10%, positive), contrary to high GDP per capita (1%, negative), hence, another threshold in our model. The outcome of GDP growth in the manual threshold approach reveals multiple significant outcomes, contrary to the global and regional level analyses in which no significant results were obtained.
