The thesis at hand examines whether human capital endowment has significant explanatory power on foreign direct investments in OECD countries between 1985 and 2017. The author has applied quantitative as well as qualitative human capital variables in the spirit of the bilateral gravity approach, while applying a version of the gravity model which accounted for country- and year-fixed-effects. The persistence of the results has been analysed via subsamples of the periods 1985 to 2000 and 2001 to 2017. Additionally, both, FDI stock and flow data have been used to test whether the empirical findings are robust regarding the respective data choice.
In the past decades research produced plenty of papers, monographies and books regarding the analyses of the determinants of international financial capital flows (FDI). This strand of scientific literature is closely related to the activities of multinational companies/enterprises (MNCs/MNEs).
Table of Content
List of Tables
List of Illustrations
List of Abbreviations
1. Introduction
2. Definitions and theoretical Models of FDI
2.1 Early Theories and Models of FDI
2.2 FDI in the Spirit of the traditional International Trade Theory
2.3 FDI in the spirit of the Eclectic-Approach
2.3.1 The OLI Paradigm
2.3.2 Criticism and modem Relevance of the OLI Paradigm
2.4 FDI in the Spirit of the Gravity Approach
2.4.1 Motives for the Application of the Gravity Approach and early Shortcomings
2.4.2 Modern Gravity and Trade in Goods
2.4.3 From trade to FDI and the Proximity-Concentration Hypothesis
2.5 Summary and concluding Remarks
3. State and Strands of empirical FDI Literature
3.1 General Findings within empirical FDI Literature
3.2 Empirical Findings on the Relationship between FDI and HC
3.2.1 The empirical Relationship between FDI and HC before 2003
3.2.2 The empirical Relationship between FDI and HC since 2003
3.3 Interim Conclusion relating empirical Findings on FDI and HC
4. Motivation, Data and Model Specification
4.1. Conception and Interpretation of the FDI Data
4.2 Examination of the applied FDI Datasets
4.3 Measuring Human Capital
4.3.1 Key Concepts and Problems of Human Capital Measurement
4.3.2 Human Capital Measures applied
4.4 Control Variables and Data Sources
4.5 Econometric Approach and Model Specification
4.6 Hypotheses and expected Results
5. Empirical Analysis of the FDI Datasets
5.1 Descriptive Analysis of the independent and dependent Variables
5.2 Correlation Analysis of the Regression Variables
5.3 Discussion of empirical Results
5.3.1 Regression Results for FDI Flow Data
5.3.2 Regression Results for FDI Stock Data
5.4 Robustness Checks of the empirical Results
6. Summary of empirical Results and Policy Implications
7. Conclusion and Future Research
List ofReferences
List of Online Data Sources:
Appendix A: Figures
Appendix B: Tables
Appendix C: Regression Outputs
Acknowledgements
First and foremost, I would like to thank Prof. Paul J. J. Weifens for enabling me to write this thesis at his chair and for always supporting me in my academic and personal development. I also would like to thank Mr. Fabian Baier, without whom this work would not have been possible. In particular, his support with regard to the econometric framework and his tireless willingness to be available as a constructive dialogue partner have always been of great assistance to me in the preparation of this thesis.
Many thanks also go to Prof. Uta Pigorsch for co-refereeing this thesis and for giving a lecture on econometrics which provided me with a toolkit that reliably helped me in preparing this thesis.
Additionally, I would like to thank Dr. Marcus Wrede and Mr. André Bauermann, who have given me freedom within the scope of my professional activity and supported me to write this thesis in this form. I would also like to thank Mr. Martin Piechocki and Ms. Marie Seeck for correcting the content of this work, as well as Ms. Lisa Fischer and Ms. Lisa Timm for the linguistic correction of the final draft.
Beyond that I would like to especially thank Mr. Ruslan Abakumov und Mr. Jan Morgan for their untiring readiness to support and encourage me especially in those moments when I needed it most. Last but not least, I would like to thank all my friends and family who supported and motivated me unconditionally throughout my journey and without whom I would never have had graduated. This thesis is dedicated to you.
Abstract
The thesis at hand examines whether human capital endowment has significant explanatory power on foreign direct investments in OECD countries between 1985 and 2017. The author has applied quantitative as well as qualitative human capital variables in the spirit of the bilateral gravity approach, while applying a version of the gravity model which accounted for country- and year-fixed-effects. The persistence of the results has been analysed via subsamples of the periods 1985 to 2000 and 2001 to 2017. Additionally, both, FDI stock and flow data have been used to test whether the empirical findings are robust regarding the respective data choice.
The empirical results suggest that the use of stock data rather leads to meaningful results in terms of explanatory power of human capital on FDI. Nevertheless, significance, the effect magnitude as well as the causal relationship between foreign direct investment and human capital strongly depend on the proxy applied, with results varying substantially for different measures. The subsample analysis reveals that dynamics, describing the relationship between the variables of main interest, seem to fluctuate considerably for different time periods. However, irrespective of the total duration of the period examined and the multitude of HC variables tested, the present thesis does not provide a conclusive assessment of the relationship between human capital and FDI. Albeit it offers a good overview on why previous work has produced very divergent results regarding their relationship.
List of Tables
Table 1: Modern Empirical Hypotheses derived from the OLI Paradigm
Table 2: Levels of FDI Predictors
Table 3: Overview of Frictions in Trade and FDI
Table 4: Overview on merged OECD FDI Flow and Stock Datasets
Table 5: Human Capital Proxies and Data Sources
Table 6: Cross-Correlations of FDI Flows and economic control Variables
Table 7: Cross-Correlations ofFDI Flows and HC Variables
Table 8: Cross-Correlations between HC and economic control Variables in the FDI Flow Dataset
Table 9: 2nd Baseline Regression - FDI Flows 1985-2017 (Country-Fixed-Effects)
Table 10: 2nd Baseline Regression - FDI Stock 1985-2017 (Country-Fixed-Effects)
Table 11: Regression Output UNDP Education Index - FDI Flow 1985-2017
Table 12: Regression Output PWT Human Capital Index - FDI Flow 1985-2017
Table 13: Regression Output AHLOS - FDI Flow 1985-2017
Table 14: Regression Output censored AHLOS - FDI Flow 1985-2017
Table 15: Regression Output of remaining HC proxies - FDI Flow 1985-2017
Table 16: Regression Output UNDP Education Index - FDI Stock 1985-2017
Table 17: Regression Output PWT Human Capital Index - FDI Stock 1985-2017
Table 18: Regression Output AHLOS - FDI Stock 1985-2017
Table 19: Regression Output of remaining HC proxies - FDI Stock 1985-2017
Table 20: Regression Results on HC in Country-Fixed-Effects Regressions 1985-2017
Table 21: Regression Results on HC in Country-Pair-Fixed-Effects Regressions 1985-2017 89 Table 22: Comparison of HC Regression Coefficients
Table 23: Regression Results on HC in Country-Fixed-Effects Regressions of the Subsamples 1985-2000 (Sub S.l)and 2001-2017 (Sub S.2)
Table 24: Overview of Hypothesis in the Review of Nielsen et al. (2017)
Table 25: Comparison of mean average Inflows and Instocks
Table 26: Overview of Approaches to Measure HC in empirical Studies cited in this Thesis 119 Table 27: First Year of Availability of Advanced HC Variables Data
Table 28: First Year of Availability of remaining HC Variables Data
Table 29: Label of Variables in Stata
Table 30: Descriptive Statistics of FDI Inflows and economic Variables
Table 31: Descriptive Statistics of FDI Instock and economic Variables
Table 32: Descriptive Statistics of HC measures in the Flow Dataset
Table 33: Descriptive Statistics of HC measures in the Stock Dataset
Table 34: Country Rank according to UNDP EI vs. PWT HCI in 1990 and 2017
Table 35: Cross-Correlations of FDI Stock and economic control Variables
Table 36: Cross-Correlations between HC- and economic Variables in the FDI Stock Dataset 130 Table 37: Cross-Correlations of FDI Stocks and HC Variables
Table 38: Analysis of AHLOS and Aggregated FDI in 2001 and 2017
Table 39: Analysis of AHLOS Scores in Western Europe and CEE Countries 1995-2012
Table 40: 1st Baseline Regression - FDI Flows 1985-2017 (Country-Fixed-Effects)..
Table 41: 1st Baseline Regression - FDI Stock 1985-2017 (Country-Fixed-Effects)
Table 42: Regression of additional control Variables - FDI Flow 1985-2017 (Country -Fixed- Effects)
Table 43: Regression of additional control Variables - FDI Stock 1985-2017 (Country -Fixed- Effects)
Table 44: Regression Output UNDP Education Index - FDI Flow 1985-2000
Table 45: Regression Output UNDP Education Index - FDI Flow 2001-2017
Table 46: Regression Output PWT Human Capital Index - FDI Flow 1985-2000
Table 47: Regression Output PWT Human Capital Index - FDI Flow 2001-2017
Table 48: Regression Output AHLOS - FDI Flow 1985-2000
Table 49: Regression Output AHLOS - FDI Flow 2001-2017
Table 50: Regression Output UNDP Education Index - FDI Stock 1985-2000
Table 51: Regression Output UNDP Education Index - FDI Stock 2001-2017
Table 52: Regression Output PWT Human Capital Index - FDI Stock 1985-2000
Table 53: Regression Output PWT Human Capital Index - FDI Stock 2001-2017
Table 54: Regression Output AHLOS - FDI Stock 1985-2000
Table 55: Regression Output AHLOS - FDI Stock 2001-2017
Table 56: Regression Output UNDP Education Index - FDI Stock 1985-2017 (including ICT Patents Share)
Table 57: Regression Output PWT Human Capital Index - FDI Stock 1985-2017 (including ICT Patents Share)
Table 58: Regression Output AHLOS - FDI Stock 1985-2017 (including ICT Patents Share)
Table 59: Regression Output UNDP Education Index - FDI Flow 1985-2017 (Country-Pair- Fixed-Effects)
Table 60: Regression Output PWT Human Capital Index - FDI Flow 1985-2017 (Country-Pair- Fixed-Effects)
Table 61: Regression Output AHLOS - FDI Flow 1985-2017 (Country-Pair-Fixed-Effects)
Table 62: Regression Output UNDP Education Index - FDI Stock 1985-2017 (Country-Pair- Fixed-Effects)
Table 63: Regression Output PWT Human Capital Index - FDI Stock 1985-2017 (Country-Pair- Fixed-Effects)
Table 64: Regression Output AHLOS - FDI Stock 1985-2017 (Country-Pair-Fixed-Effects)
Table 65: Regression Output UNDP Education Index - FDI Flow 1985-2017 (Robustness)
Table 66: Regression Output PWT Human Capital Index - FDI Flow 1985-2017 (Robustness)
Table 67: Regression Output AHLOS - FDI Flow 1985-2017 (Robustness)
List of Illustrations
Figure 1: Global Inward FDI-Flows and FDI to GDP-Ratio - 1970-2017 in Million USD at current prices
Figure 2: Periodical Classification of the Number of Studies examined in Literature Reviews 21 Figure 3: FDI Inflows (RHS) and Instocks (LHS) - Global and by Economic Group 1980-2018 (Billions of current USD)
Figure 4: Scatterplot of Target AHLOS and FDI Inflows 1985-2017
Figure 5: Scatter Plot FDI Inflows and AHLOS 1985-2017
Figure 6: Asset/Liability Presentation of FDI Positions
Figure 7: Directional Presentation of FDI Positions
Figure 8: Scatterplot FDI Flow Data 1985-2017
Figure 9: Scatterplot FDI StockData 1985-2017
Figure 10: Boxplot of the FDI Flow Data
Figure 11: Boxplot of the FDI Stock Data
Figure 12: Histogram of the Openness in FDI Stock Dataset
Figure 13: Histogram of the Edu. Expenditures to GDP Ratio
Figure 14: Scatterplot of the Edu. Expenditures to GDP Ratio
Figure 15: Scatterplot of Patent Grants per Mil. Inhabitants
Figure 16: Scatterplot of the ICT Patents Share in the Target Country
Figure 17: UNDP El for all Countries 1985-2017
Figure 18: PWT HCI for all Countries 1985-2017
Figure 19: Scatterplot of censored AHLOS and Openness 1985-2017
Figure 20: AHLOS for all Countries 1985-2017
List of Abbreviations
Abbildung in dieser Leseprobe nicht enthalten
1. Introduction
In the past decades research produced plenty of papers, monographies and books regarding the analyses of the determinants of international financial capital flows (hereafter FDI). This strand of scientific literature is closely related to the activities of multinational companies/enterprises (hereafter MNCs/MNEs).
Early economic literature tended to associate the occurrence of FDI with the establishment of the first U.S. MNEs in the late nineteenth and early twentieth century. Those companies conducted market oriented FDI in the field of manufacturing (Jones and Bostock, 1996, p. 208f.). According to economic historians, FDI is by no means a phenomenon of the mid to late nineteenth century. Carlos and Nicholas (1988, p. 399f.) e.g. gave a comprehensive review on why English and Dutch East India companies should be recognized as the first MNEs, who conducted FDI as early as the sixteenth century. Their analysis revealed that historical activities of multinationals were driven by the factor human capital. They argued that by employing “Skilled traders, expert in trade negotiations, foreign languages, assessing the quality of goods and determining their proper handling, storage, and loading, allowed the trading companies to reduce their transacting costs below that of the market.” (Ibid., p. 411).
On the one hand, some aspects of the above-mentioned quotation have become meaningless from a modern point of view, since e.g. the handling, storage and loading of goods is mainly conducted by robots in cargo terminals. On the other hand, aspects like e.g. the knowledge of foreign languages is still recognized as an important factor influential on FDI and is subject to latest research in the field of possible FDI determinants.1
The general importance of foreign direct investment in the twenty-first century can be illustrated by the following quote from Mr. Mukhisa Kituyi2 '.
“We are at the dawn of a fourth industrial revolution, [...] [it] offers enormous opportunities for economic growth and sustainable development [...] governments around the globe have invigorated their industrial policies in recent years [...] wielding multiple instruments,from trade to education [emphasised by the author]. Central to these industrial policies is foreign investment, [emphasised by the author]”, (UNCTAD, 2018, p. 6).
The importance of FDI on a global scale can be further illustrated by taking a closer look at the quantitative development of its flows over the past 50 years, as shown in Figure 7:
Figure 1: Global InwardFDI-Flows andFDI to GDP-Ratio - 1970-2017 in Million USD at current prices
Abbildung in dieser Leseprobe nicht enthalten
Annual global FDI-Inflow (left hand side)
Annual FDI-Inflow to GDP Ratio (right hand side)
Source: UNCTAD; own calculation.
Figure 1 demonstrates that global financial flows saw an unprecedented growth in the late nineties, reaching a peak in 2015, as they nearly exceeded two trillion US-Dollar. Putting those figures in perspective, in 2000, the share of FDI inflows in global GDP amounted to 4.05%, equalling the contribution of a country like France to the global output. Admittedly since reaching those heights, FDI inflows reversed their direction and started declining - a trend which continued in 2018, as global inflows dipped further to 1.3 trillion USD. According to UNCTAD, this trend may persist in 2019 (2019, p. 2).
Most recent research by the OECD reveals that the diffusion of the digital economy might even lead to a general decrease in international investment, since the need for physical presence to service markets abroad will be declining, as a wide range of goods and services will be delivered in a digital form (Gestrin and Staudt, 2018, p. 9). Still, based on 2017 numbers, the share of FDI inflows in global GDP amounted to 1.78%, roughly exceeding the combined contribution of Switzerland and Saudi Arabia to the global GDP.3
The preceding remarks firstly clarify that FDI has a practical history of over 450 years, while secondly pointing out that human capital (commonly proxied via education) seems to play a constantly dominant role as its determinant. Thirdly, the developments described above hint towards the fact that the global competition to attract FDI will become even more challenging. The last fact makes it crucial for policy makers to understand the drivers behind foreign capital flows, enabling them to create an environment, which makes their country a favourable FDI destination. The aforementioned second fact motivated the author of this thesis to focus on the role of human capital as a determinant of FDI from a country-level perspective. Of course, this thesis is not the first to examine the relationship between FDI and HC, but the following four aspects will give it a raison d'etre:4
1. The literature review in the third section will firstly point out that research did not establish some kind of “best practice” approach, when it comes to modelling the relationship between human capital and FDI. Additionally, it will be shown that FDI related empirical studies, which explicitly introduced human capital as an explanatory variable, while econometrically applying a ppml-based gravity approach, are rather scarce.
2. With respect to the most current literature, this thesis directly addresses the findings of Konara and Wei (2019), who argued that HC is an insufficient condition in attracting FDI, as HC is contingent on language capital.
3. Regarding the aforementioned fundamental change in the global economic structure with its shift towards a digital economy, to the best knowledge of the author, this thesis is the first to analyse bilateral FDI data for the years 1985 to 2017, as figures for the years 2016 and 2017 have only recently been published by the OECD. This will make it possible to investigate whether the traditional determinants, like distance, market size, openness and especially human capital (in a quantitative and qualitative sense) are sufficient in explaining the distribution of foreign direct investment in the early digital economy age.
4. Last but not least this study will examine up to 33 years of cross-country FDI relations, allowing reliable conclusions on the persistence of certain factors as determinants of foreign direct investment in OECD countries.
The empirical findings in this thesis highlight that the relationship between human capital and FDI is neither stable over time, nor is it independent regarding the choice of the HC and FDI proxies. More precisely, empirical results suggest that the use of stock data rather leads to meaningful results in terms of the explanatory power of human capital on FDI. With respect to significance levels, the effect magnitude as well as the causal relationship between foreign 4 The term language capital will be explained in detail in subsection 3.2.2 direct investment and human capital, this thesis suggests that empirical results may vary substantially for different HC measures applied. The main objective of this thesis is therefore not to provide a conclusive assessment of the relationship between human capital and FDI. The author rather tries to give a comprehensive overview of the features of the FDI data available to researchers, as well as on the strengths and weaknesses of the HC proxies commonly used in FDI related research. Finally, this work partly helps to understand why previous research has produced very divergent results regarding the relationship of human capital and foreign direct investment.
The remainder of this thesis will be organized as follows: the second chapter provides some basic definitions and gives an overview on theories, trying to explain the occurrence and distribution of foreign direct investment in the context of multinational enterprises. The third section contains a literature review on FDI related publications thereby distinguishing between general findings in empirical FDI research, and subsequently reviewing empirical findings on the explanatory power of human capital, as well as the associated econometrical approaches in the context of FDI research. The following section precisely restates the authors motivation and presents the data, as well as the model specifications. Chapter five discusses the empirical results, while in the sixth section those empirical results will be embedded in the context of the literature previously presented and will be discussed along with corresponding policy implications. The last section will give an overview on possible future research questions and conclude.
2. Definitions and theoretical Models of FDI
Theoretical explanations on why capital pours to so many different corners in the world are closely bound to the activity of multinational enterprises, making it obligatory to explain the link via the most common definition of foreign direct investment, which is provided by the OECD'.
“FDI is defined as the establishment of a lasting interest in and significant degree of influence over the operations of an enterprise in one economy by an investor in another economy. Ownership of 10% or more of the -votingpower in an enterprise in one economy by an investor in another economy is evidence of such a relationship. ”, (2015, p. 5).
With respect to this definition, it is also necessary to point out that, if an international ownership investment construction does not reach the ten-percentage threshold, it is classified as an international portfolio investment (Alfaro et al., 2008, p. 6).5 Returning to FDI, it is crucial to distinguish between green- and brownfield investment, the former meaning that a company invests in physical capital (i.e. the construction of a new office or a new plant) and the latter describing the acquisition of an already existing enterprise (Faeth, 2009, p. 183).6
As already mentioned in the introduction, the above pronounced transnational financial relations are by no means a modern phenomenon. According to Agarwal (1980, p. 739f.), who wrote one of the first comprehensive reviews on theories of FDI, researchers historically were rather interested in the determinants of trade between nations than in FDI, a fact he uses explanation the lack of theoretically sound research in this field prior to the 1950s.7 The author came up with 13 theories, looking at FDI from a firm-, industry- and country-level perspective (Ibid., p. 763). A rather modern review of theoretical FDI models, which is often cited in current literature, was conducted by Faeth (2009) (see e.g. Nielsen et al., 2017, p. 64; Xie et al., 2017, p. 6; Harms and Knaze, 2018, p. 1). She distinguishes between research from a bilateral and unilateral perspective, while pointing out that researchers made use of time-series, cross-sectional as well as panel data and utilised micro- and macroeconomic factors and combinations of both to explain the occurrence of international financial flows. She concludes that there are nine different theoretical models of FDI (Faeth, 2009, p. 165f.).
It becomes apparent that a chronological discussion of all existing theoretical frameworks trying to explain FDI would be beyond the scope of this thesis, since its prime goal is to provide an empirical investigation of HC as a factor determining foreign direct investment. Therefore, the following subsections will only present the most important theoretical models, which emerged since the early 50s of the past century.
2.1 Early Theories and Models of FDI
According to Agarwal (1980, pp. 741-745), the beginning of the theoretical investigations on the determinants of FDI was marked by the application of the two following hypotheses:
1. Differential rate of return hypothesis, and
2. Portfolio hypothesis.
Hypothesis No. 1 was mainly driven by the idea of the profit maximizing firm and the application of the marginalist approach, according to which FDI was to stream out of those regions with low returns on capital, pouring to those countries, that yielded higher rates of return. This hypothesis was mainly driven by high FDI outflows from U.S. to the Western European countries. Though it did not turn out to be stable, as the ratio between the European and the U.S. rates of return reversed in the early sixties, but capital continued to flow towards the European direction (Agarwal, 1980, p. 741).8
Hypothesis No. 2 was derived from the prominent work of Tobin (1958) anâMarkowitz (1959), following a generalised approach and considering not only the return, but also introducing a risk correction element. Nevertheless, this theory was likewise not capable of explaining the foremost global FDI flows. The approach was rather characterised by a major drawback, namely its inability to explain the different commitments to FDI on an industry level (Agarwal, 1980, p. 745f). Besides, Dunning (1973, p. 299f.) criticized this approach because it was unable to account for the transfer of other input factors than financial capital. He concludes that the poor empirical outcomes were due to the fact that the portfolio theory did not consider the differences in behavioural characteristics, which were inherent to the actors conducting foreign direct investment.
2.2 FDI in the Spirit of the traditional International Trade Theory
A considerably more established strand of FDI theory is based on the neoclassical theory and makes use of a 2 x 2 x 2 general equilibrium model with two countries, two factor inputs - typically being capital and labour - and two goods. Such types of models assume no transport costs, factor endowments that eliminate the possibility of specialisation and suggest constant returns to scale production functions (Faeth, 2009, p. 166f.). Another important - but usually unrealistic - assumption of the neoclassical theory asks for perfect market competition, to allow for the equalization of factor prices thereby generating a stable market equilibrium (Agarwal, 1980, p. 740). Those types of models, like e.g. thsHeckscher-Samuelson-Ohlin-Model, predict capital to flow to countries being abundant of the factor labour, but scarce in capital. According to Dunning (1973, p. 305) the most famous attempts to apply such models were the approaches of Mundell (1957) and Johnson (1968). He described both approaches as unsuccessful, since they were static in the way that they fully neglected innovations or, when innovations were introduced in a comparative static situation, those specifications implied that innovations were directly transferable across the countries. The author describes those postulations as “7-7 totally unrealistic in a situation where information is costly to produce, is enterprise specific, and is sold under conditions of imperfect competition;”, (Dunning, 1973, p. 305). The predictions of the neoclassical model were famously empirically rejected by Lucas ’ seminal work, pointing out that the aforementioned assumptions of the model must be fundamentally wrong. This is due to those type of models being unable to explain the global foreign direct investment distribution, as well as the lack of the predicted equalisation in global investment flows (Lucas, 1990, p. 92).
Another famous approach in the spirit of the factor endowment theory is known as the ProductCycle-Theory, which is closely related to the work of Vernon (1966) (Faeth, 2009, p. 168).9 The author argued that the international trade literature of the fifties and sixties erroneously failed to consider the roles of uncertainty and inexperience, influencing international trade patterns, but also the timing of innovation and the effects of economies of scale (Vernon, 1966, p. 190). To account for those obstacles, he assumed knowledge to be by no means a globally available 9 Vernon himself credited many other authors for contributing to this stance of literature, especially dividing it into a group of authors focusing on the Product-Cycle from an (international) trade perspective and into a group that dealt with the patterns of FDI (Vernon, 1979, p. 255). free good, but rather an independent variable in the decision process on whether to invest or trade. This assumption implicated that producers, who were located in the home market, had a higher probability to recognize market needs and to introduce new products, than those producers, for whom the market under consideration was not the domestic market (Ibid., p. 191).
In a next step, Vernon examined the U.S. market, which by that time according to him, was characterised by the following facts: on the one hand it featured by far the highest average income in the world and on the other hand, the unit labour costs in the U.S. were pretty high and capital could move as free as in no other country in the world. The author followed that those market characteristics, which were accompanied by an “[...] effective communication between the potential market and the potential suppliers] of the market.”, (Ibid., p. 193) led to constantly higher spending on product development in the U.S.. Vernon used these characteristics of the U.S. market to generalize with regard to common aspects of a product innovation process, pointing out that when an innovative product or technology is developed, tested and launched, this tends to happen in a company’s home market. He attributed such outcomes to the fact that, on one hand home markets’ demand is one of the main drivers for innovation and on the other hand the demand for resources, necessary to develop a new product, as well as e.g. managerial and coordinating issues tend to favour the home market, where e.g. the company’s headquarter is located and therefore it has the most resources available (Ibid., pp. 192-196).
In the next stage the distribution process is determined by standardisation and the exploitation of economies of scale. This stage of the product-cycle has implications with respect to the choice of the location of production sites. According to Vernon, it will be rather likely that the locational choice will be determined by cost analysis plus transport considerations, leading to the establishment of foreign subsidiaries. Those subsidiaries can conclusively be used to not only serve third-party markets, but if labour costs can offset transportation costs, it is possible that foreign subsidiaries are used to serve the home market (Ibid., pp. 196-2OO).9
In the following decade other researchers picked up the Product-Cycle-Theory approach and refined it even further, but in the end it was not capable of continually explaining the distribution of global FDIs (Agarwal, 1980, p. 751f.). In a follow-up publication, Vernon himself argued that first of all, the post-World-War-II period was characterised by an increase in demand for mature U.S. products that were first exported and later manufactured directly in Europe - a process, which favoured empirical evidence with respect to the Product-Cycle-Theory.10 By the mid-seventies the European countries were catching up with the U.S. regarding the per capita income, thereby contradicting the assumption of the theory that different markets were described by different economic conditions. Both of the aforementioned aspects led to the crumbling of another central assumption: Vernon argued that by the end of the seventies the subsidiaries themselves were capable of developing and introducing new products, which relaxed the assumption that knowledge is by no means a globally free accessible good, with the U.S. being in the best favour to make use of it (Vernon, 1979, pp. 258-261).
Ultimately, this theoretical approach can be summed up as micro-orientated, rather investigating the behaviour of individual firms than those of countries. Moreover, it is an oversimplifying theory, lacking generalisation, as it can at best describe the behaviour within highly innovative industries (in the United States) (Dunning, 1973, p. 306; Agarwal, 1980, p. 752).
2.3 FDI in the spirit of the Eclectic-Approach
Although the Product-Cycle-Theory was ultimately not successful, the general approach of accounting for market imperfections paved the way for some major theoretical breakthroughs relating the development of theories on the global distribution of foreign direct investment. The most comprehensive approach, emerging from this strand of literature, was the eclectic paradigm, introduced by Dunning through the seventies.11 In a first step he suggests that the share of production abroad might be determined by “7—7 the extent of the market in each country as well as by “7—7 the competitiveness offoreign affiliates vis-a-vis indigenous and nonresident firms. ”, (Dunning, 1973, p. 313) thereby arguing for an integrated approach in analysing international economic activities. In two follow-up papers, which established themselves as classic in FDI research, Dunning claimed that such an integrated approach needed to take location-specific factor endowments of countries into account, while also considering the ownership-specific endowments of enterprises (Dunning, 1977, p. 395) as well as internalization incentive advantages (Dunning, 1979, p. 274). He justified this combination of previously separate literature strands relating the involvement into FDI, by the fact that none of those single hypotheses was able to explain the majority of empirical evidences (Dunning, 1977, p. 397).
2.3.1 The OLI Paradigm
Due to the importance of the OLI12 paradigm, the single factors shall be discussed in further detail. Dunning described the location-specific factor endowments as a combination of tangible and intangible assets (i.e. knowledge, as well as entrepreneurial and organizational skills), which might only be inherent to a specific country albeit accessible to all enterprises operating in that country.13 However such factor endowments might also be ownership-specific, meaning that only home country enterprises are able to access them, but they are capable of being used in combination with other resources in every possible country. He finally concluded that both endowment aspects are closely related to market competitiveness, thereby fundamentally driving company’s choice of production location (Ibid., p. 399).
The third aspect, namely internalization advantages, is formulated in the spirit of the property rights theory and is derived from the finding that such companies, who are most competitive in foreign markets, internalize their activities. The internalization aspect shall be distinguished in the sense of 'vertical and horizontal integration, usually associated with strategies in oligopolistic markets (Ibid., pp. 401, 407).
Dunning contributed to the idea on why MNEs conducted horizontal internalization (thereby becoming conglomerates) by taking structural and cognitive market imperfections into consideration. He argued that, if a company of a certain size has the funds to successfully engage in an internalization strategy, it may profit from product diversification or integration, which, in turn, would increase its opportunity to benefit from other effects like cross subsidization of costs or predatory competition (Ibid., p. 403). Especially these considerations can be summed up under what became famous as the Factor-Proportions hypothesis, utilised when trying to explain the existence of vertically integrated firms characterised by a geographically disintegrated chain of production (Brainard, 1997, p. 520f.; Faeth, 2009, p. 174f.).
The author also offered a more general reason for internalization activities by considering public interventions with respect to the resource allocation mechanism. Since interventions may lead to market failures, companies may seek opportunities to exploit them or protect themselves against possible consequences (Dunning, 1977, p. 403).
In summary the eclectic paradigm suggests that any cross-border activity, namely trade and FDI, can be explained via a single or a combination of the above-mentioned theoretical approaches.14 It seems worth mentioning that the OLI paradigm is by no means a normative approach, being able to predict future outcomes like e.g. which MNE will undertake foreign direct investments in a specific country or which countries will be most attractive for FDI. The paradigm rather accounts for positive outcomes. On the one hand, aspects driving the choices are not evenly distributed across enterprises, industries and countries, and on the other hand, determinants of the aspects may change over time (Dunning, 1979, p. 275f.). However, from current perspective, the hypotheses derived from the OLI paradigm may be summed up as follows:
Table 1: Modem Empirical Hypotheses derivedfrom the OLI Paradigm
Abbildung in dieser Leseprobe nicht enthalten
Source: following Dunning (2000, p. 164f).
2.3.2 Criticism and modern Relevance of the OLI Paradigm
Almost 50 years passed from the beginning of the construction of the OLI approach. Since then even the question on whether it should be labelled as a theory (like it initially was) or a paradigm, was extensively discussed by the scientific community (Dunning, 2001, pp. 175, 187). It becomes apparent that not only a general discussion of the OLI paradigm may very well be a graduation paper for itself, but that there also must be some (major) criticism from researchers (Macharzina and Engelhard, 1991, p. 27f.; Dunning, 2001, pp. 176-180). The author of this thesis therefor decided to limit the presentation of the criticism to the one fact, which seemingly emerged as the most longstanding and is also partly accepted by Dunning himself. The argumentation is cantered around the wide scope of the paradigm, as it includes so many variables that it seems unfeasible to operationalize them, making the paradigm a shopping list of variables (Dunning, 2001, p. 177).
The way how empirical researchers have handled the aforementioned wide scope can be illustrated together with the role of the OLI paradigm within the second decade of the 21st century. One possibility is to examine the comprehensive literature review with respect to FDI research provided by Nielsen et al. (2017). The authors developed a taxonomy of three levels, each representing certain characteristics, determining the likelihood on whether an MNE would conduct FDI or serve the market via other mechanisms.15 The following Table 2 presents those characteristics:
Table 2: Levels of FDI Predictors
Abbildung in dieser Leseprobe nicht enthalten
Source: Nielsen et al. (2017), p. 65.
What Nielsen et al. did not explain is why they chose a level classification, which clearly implies some kind of hierarchy. A possible explanation might be that they tried to follow the chronological development of the international trade theory. Nevertheless, this classification provides powerful insights regarding the previously stated operationalization problems within empirical research: as the findings of Nielsen et al. show, scholars seem to have split the OLI approach in single factors, picking those up, which may suit best for their empirical question of interest. Moreover, the review points out that since its inception in the seventies, the OLI paradigm established itself as one of the prevailing explanations to the occurrence of FDI. Although it must be concluded that the paradigm is rooted in the international business literature (Nielsen et al., 2017, p. 65), this approach and the classification in Table 1 will be picked up in section 3.1, where the general findings of empirical FDI research will be presented. Finally, Table 2 unveils that there is another approach, which established itself as very successful in explaining the distribution of FDI and will therefore be discussed in the following subsection.
2.4 FDI in the Spirit of the Gravity Approach
As became apparent from section 2.2, economists represent an important group of scientists, who extensively investigate the topic of international investment flows16. It is well documented that in the beginning of empirical FDI research an overwhelming majority of scholars applied international trade theory of comparative advantages, but another - “practically” driven approach - emerged in parallel within a small group of researchers, who mainly investigated bilateral trade flows between partner countries. The approach was directly derived from Isaac Newton ’s Law of Gravitation and is therefore called the Gravity approach (Bergstrand and Egger, 2013, p. 532), which emerged among economists in the early sixties, after being firstly introduced by Tinbergen (1962). But it was one of his students, namely Hans Linnemann (1966), who conducted the first extensive empirical study on international trade flows, while applying the first stage gravity equation (Benedictis, Taglioni and Bank, 2011, p. 56; Bergstrand and Egger, 2013, p. 431).17 Since the gravity approach is the empirical tool of choice for the author of this thesis and due to its somewhat remarkable path towards acceptance within the scientific community, the historical development of the gravity equation shall be discussed in more detail within the following subsections.
2.4.1 Motives for the Application of the Gravity Approach and early Shortcomings
As already described, the gravity approach received little attention across early researchers focusing on FDI and multinational enterprises. This is further pinpointed by the fact that the already cited comprehensive review of Agarwal (1980) neglected the discussion of any related literature. There are two equally important explanations to this finding: primarily, within the first stage, the gravity approach was only used to explain aggregate gross bilateral trade flows, because it represented an analysis of trade quantities valued at their respective prices (Bergstrand and Egger, 2013, p. 540f.). Secondly, the application of the gravity approach on the analysis of international trade flows lacked a consistent theoretical justification and was applied by analogy (Caves, 1996; Folfas, 2011, p. 1; Bergstrand and Egger, 2013, p. 532; Head and Mayer, 2014, p. 132). Although Linnemann’s (1966) seminal study was conducted using a Walrasian model, it came short in explaining the multiplicative form of the gravity equation (making it necessary to apply the natural logarithm), while assuming a general world price level and lacking a macroeconomically sound explanation for the inclusion of country’s GDPs (Bergstrand and Egger, 2013, p. 541f.).
After presenting early shortcomings of the gravity equation, it seems natural to ask why researchers started using it to analyse trade in the first place. One explanation can be derived after recalling subsection 2.2 which comprised the assumptions of the neoclassical theory, especially assuming no frictions. This must be recognized as a major drawback, as trade and FDI are clearly subject to a number of those. Bergstrand and Egger provided a discussion on specific frictions which impose costs on the relevant parties. The following Table 3 gives a brief overview on the relevant aspects:
Table 3: Overview of Frictions in Trade and FDI
Abbildung in dieser Leseprobe nicht enthalten
Source: Following Bergstrand and Egger (2013, pp. 534-537).
The authors argued that it is helpful to decompose frictions in natural and unnatural ones, since they are driven by different factors and demand for altering proxies. Within the natural trade frictions, the transport costs represent an important factor explaining trade activity in a bilateralcountry framework. Determinants of the transport costs might be geographic variables (most famously distance) as well as the quality of infrastructure. Besides, time (needed to deliver a good from its origin to its final destination) is also found to be a natural friction to trade (Ibid., p. 534f.). Regarding tariffs, the relationship appears self-evident, while nontariff barriers (NTBs) are described as custom procedures, specific national legislations and practices, which aim at altering domestic trade compared with international trade (Ibid., p. 536).
With respect to frictions to FDI, Bergstrand and Egger also distinguish between natural and unnatural frictions. Within the natural frictions to FDI, distance is recognized as a negative factor.18 Like already suggested in the introduction, language is expected to be a positive factor in bilateral FDI, if the parties involved share the same language. Ultimately policy variables are introduced as unnatural frictions to foreign direct investment. Possible examples for such variables might be a country’s tax policy or its import tariff rates in goods (Ibid., p. 537).
Returning to the above discussed./r/c/z^^x to trade, thefirst stage gravity equation as introduced by Tinbergen (1962), had the following form:
Abbildung in dieser Leseprobe nicht enthalten
With PXij representing the value of the merchandise trade flow from exporter / to importer j (in current prices), GDPi (GDPj) is the nominal value of gross domestic product in country i (/), DISTij represents the bilateral physical distance between the economic centres of countries i and./. ADJij is a dummy variable with the value of one, if country z and./ share a common land boarder. EIAlij and EIA2ij are similarly both dummy variables taking the value of one, if both countries are members of the British Commonwealth, or the Benelux free trade agreement respectively (and zero otherwise). Moreover, eij is supposed to be the log normally distributed error term. Finally, In refers to the natural logarithm.
It becomes evident that gravity equations allowed researchers to easily account for natural, as well as unnatural frictions to trade, by applying one of Newton’s fundamental contributions to science. Regarding the expected coefficients, it is suggested that countries with higher GDPs tend to trade more, which also applies to countries sharing a common border, identical historical backgrounds and/or social/cultural ties. Therefore, it is proposed that the coefficients ß\, ß2, ß<, ßs, ßt > 0, while ß3 is expected to be < 0 proxying the transport costs (Ibid., pp. 532, 541).
Ultimately, like mentioned before, there were two reasons why the gravity approach started to establish itself within the scientific community. Aside from the opportunity to operationalize frictions, analyses derived from Equation 1 delivered overwhelming results with respect to the explanatory power. Bergstrand and Egger pointed out that even models that werejust based on countries GDPs and distance produced R2 measures between 80 and 90 percent, which were well above the explanatory power of all models the previously discussed (Ibid., pp. 532, 562). The preceding paragraphs pointed out that the gravity equation must be understood as a very successful piece of positive theory explaining international (merchandise) trade. But its theoretical foundation took quite a while. Finally, it must be noted that albeit Table 3 already underpinned that (partly) the same frictions apply to foreign direct investment as to international trade, after its introduction by Tinbergen it took researchers almost thirty years to apply the gravity equation to the analysis of FDI - the main topic of this thesis. Therefore, the following subsection will provide a quick overview on what the author of this thesis calls second stage gravity models in trade and will then directly turn to the application of the model on FDI.
2.4.2 Modern Gravity and Trade in Goods
As indicated previously, the first stage gravity equations where missing a sound theoretical link. The point of introduction of such framework is up to debate, since e.g. the already cited work of Bergstrand and Egger (2015, p. 543) dates it back to the year 1979, while other authors like Head and Mayer argued that gravity “[...]lay[ed] outside of the mainstream of trade research until 1995.” (2014, p. 134).
However, this thesis follows the argumentation of Bergstrand and Egger, who describe the work of Anderson (1979) as a first step towards a more formally grounded approach towards gravity, which established one of two future paths for theoretical models: conditional general equilibrium (hereafter GE) models (Bergstrand and Egger, 2013, pp. 543-549). Anderson (1979, p. 106f.) firstly provided a coherent theoretical model for the application of the gravity approach on cross border trade in commodities, while secondly addressing the missing explanation for the multiplicative form of the gravity equation, one of the three drawbacks in the work of Linnemann (1966). It seems fair to point out that Anderson’s model had its own deficiencies. According to his statement, the model was significantly affected by the typical econometrical trade-off between efficiency and bias.19 Moreover, since the model was derived from CES utility functions, it became too complicated to be what other economists called their everyday toolkit (Head and Mayer, 2014, p. 134f.). Ultimately, his model used the convention of free trade prices, thereby assuming them to be unity - an assumption which applies irreconcilable regarding the frictions to trade described in Table 3. The lastly mentioned point is assumed to be the main driving force behind the further elaboration of gravity equations in the spirit of conditional GE models.
To complete the analysis on the theoretical approaches of trade analysis within a gravity framework, it is necessary to describe the alternative path to conditional GE models, namely the unconditional general equilibrium models. The main difference between the two approaches refers to the fact that modelling the relationship between (two) open economies asks for a two- stage budgeting process: in the first place one needs to deal with the production and consumption decisions on a domestic level, while in a second stage the decision is modelled which countries trade which goods with each other.
With respect to this distinction, the former approach is based on a trade separation assumption, neglecting the first stage and thereby implying an endowment economy (Bergstrand and Egger, 2013, p. 543). The latter approach asks for an explicit technology and market structure, thereby allowing for a combination of the gravity framework with traditional trade theories (Ibid., p. 549). Thereby GE models are distinguished between the New Trade Theory in the spirit of Kurgman’s monopolistic competition approach, assuming internal economies of scale and tastes for variety as an explanation for trade in homogeneous goods. Another strand is based on the Heckscher-Ohlin theory of differentials in factor endowments, while the most recent unconditional GE models, are based on the most ancient Ricardian approach which emphasised differentials in labour productivity (Ibid., p. 549.).20
The next major step rather referred to accurate empirical modelling of the previously described approaches and was marked the (theoretically grounded) introduction of multilateral resistance, a term coined by Anderson and van Wincoop (2003) in their classical contribution to gravity. Its inclusion allowed to better control for the bias caused by omitted variables. More precisely, the multilateral resistance terms account for the fact that import and export between country i and country j is notjust dependent on trade costs between those two countries, but also on the trade costs across all possible import and export markets (Head and Mayer, 2014, p. 138; Shepherd, 2016, p. 8).
2.4.3 From trade to FDI and the Proximity-Concentration Hypothesis
The extensive previous discussion of the gravity approach in trade was necessary, as its application to bilateral FDI data is also mostly carried out by analogy. Regarding the theoretical grounding of gravity based FDI modelling, it seems appropriate so start from scratch. As described in Table 3, FDI is also subject to natural as well as unnatural frictions. In the spirit of those insights, the first empirical examinations of FDI (as a dependent variable) that were conducted via the gravity approach, assumed a positive relationship between FDI, physical distance and trade barriers and a negative one regarding a country’s investment barriers and the size of economies of scale on a plant level. The hereby implied trade-off is summed up under the Proximity-Concentration hypothesis, seeking to explain horizontal expansion of MNEs across boarders (Brainard, 1997, p. 520). The work of Brainard (1997) is considered to be one of the first attempts that simultaneously analysed international trade and FDI (proxied via affiliate sales abroad) in a two country, two sector GE model (Blonigen et al., 2007, p. 395; Bergstrand and Egger, 2013, p. 561).21 Although it must be stated that her multiple equilibria model - formulated in the spirit of the New Trade Theory - only analysed the in- and outward relationship between the USA and 27 partner countries, while utilizing cross sectional data. This analysis therefore did not control for any fixed effects. Moreover, industry-level data was selected in a way that all industries were dropped from the data when service accounted for more than 50% of the revenues. Brainard argued that services must be understood as nontradeables and since her approach to measure transport costs required observable trade within a given sector, she was not able to include services into her industry-level analysis (1997, pp. 521-524).22 Her empirical analysis did find some support for the proximity-concentration hypothesis. More precisely, the shares of affiliate sales increased with regard to transport costs, trade barriers and corporate scale economies. As opposed, the sales were negatively dependent on investment barriers and production scale economies (Ibid., p. 538f.).
The aforementioned paper can be used to stress one important aspect of the early applications of the gravity approach on FDI: as the terms MNE and FDI were used fairly interchangeably, a bunch of early models directly picked up the approach in Brainard (1997), namely taking the affiliate sales as a dependent variable and a mirror image for FDI (Blonigen, 2005, p. 395f.).23 This rather implied an analysis of affiliate production and output than the analysis of investment stocks and flows (Markusen, 2002, p. 8).24 Moreover, early studies (see e.g. Eaton and Tamura, 1994, 1996; Brainard, 1997; Markusen, 2002) were also characterised by a simultaneous analysis of in- and outward FDI from and into the USA and Japan respectively, resulting from the fact that the availability of sales data was very limited.
It is important to point out that early gravity based bilateral FDI research also started to establish another direction. From a modern point of view this direction can be summed up under the term political economy of FDI25 This field of research relates to distinct topics trying to explain the occurrence of foreign direct investment like e.g. the role of bilateral investment treaties, a country’s tax policy, the influence of the quality of host country institutions, as well as the main variable of interest for this thesis, a country’s human capital endowment (Pandya, 2016, p. 456). The application of the gravity equation with respect to the above-mentioned issues is characterised by the use of aggregated FDI data on a national level. Early studies contributing to the topics described above while following a gravity approach were conducted by Blonigen and Davies (2004), who have analysed the effects of bilateral tax treaties on the U.S. FDI activity, and Levy Yeyati et al. (2003), who have studied the relationship between location of FDI and regional integration agreements (hereafter RIAs).26
In terms of the results of those analyses within both strands, one finding is mostly identical and controversial regarding the theoretical assumptions and therefore deserves some further discussion: like already described in the subsection 2.4.1, the gravity equation for trade yielded a negative sign for distance. It was expected that FDI would substitute trade where costs of transportation would be too high, thereby implying a positive sign for the distance variable (Bergstrand and Egger, 2013, p. 537). Unexpectedly, the empirical examinations of bilateral FDI via gravity nearly consistently generates highly significant negative signs for the distance variable. This finding is robust for stock and flow, as well as in a cross section and panel data analyses framework (see e.g. Eaton and Tamura, 1996, p. 63f; Markusen, 2002, pp. 230, 252; Levy Yeyati et al., 2003, p. 25; Blonigen and Davies, 2004, pp. 603, 610ff.).
However, much of this earlier work is limited to describing this circumstance without offering an obvious explanation. Blonigen and Davies (2004, p. 607) e.g. describe the relationship between FDI and distance as being ambiguous. Markusen (2002) comments this finding with respect to the substitution argument in the following way: “/(is possible that distance might affect production for export more negatively than production for local sale [..], but I am generally agnostic insofar as I do not understand the transaction costs of investing at a long distance. ” (Markusen, 2002, p. 249). From a modern perspective, Markusen comes closest to the explanation by naming the transaction costs. Now it is commonly assumed that distance captures information costs which constitute barriers to FDI (Bergstrand and Egger, 2013, p. 537).27
2.5 Summary and concluding Remarks
The previous subsections described the development of theoretical approaches aiming to explain the occurrence and the distribution of global foreign direct investment. Some approaches, like the Portfolio- and Product-Cycle-Theory were quickly rejected, while the Eclectic Approach still marks an important explanation, e.g. due to the ability of the associated theories to include market imperfections. Subsequently, the gravity approach was introduced, which marked a major breakthrough since it was very successful in explaining the empirically prevailing distribution of bilateral trade and later FDI. Moreover, it was concluded that early applications of the gravity approach lacked theoretical foundation and generated some theoretical controversy, but those aspects rather motivated researchers to close the theory gap - a task that has been successfully mastered for the gravity equation in trade, for simultaneous occurrence of FDI and trade flows and for FDI on a firm-level (Anderson, 2011, p. 155f.; Bergstrand and Egger, 2013, p. 559).
On the one hand, it is essential to point out that those theoretical frameworks form the foundation for the empirical investigations of FDI in the past, present and future. On the other hand, it is equally important to state that this field of research still did not provide one general theory which is capable of explaining all the different empirical findings. A circumstance that can probably never be corrected, as associated research went in different directions. The following section will further investigate those directions of FDI research and present the most relevant general findings thereby laying the foundation for the regression model applied in this thesis. Hereafter, the specific literature focusing on human capital as an explanatory variable to FDI will be presented to provide a motivation for the research focus of this thesis.
3. State and Strands of empirical FDI Literature
The introduction pointed out that FDI has a practical history of more than 450 years, although the history of empirical research related to FDI is not nearly as long. This seems obvious, since empirical (FDI) research requires reliable - or at least even some - data. Therefore scientists started to focus on the determinants of foreign direct investment in the 1950s, mainly focusing on the United States and the activities of U.S. MNEs (Jones and Bostock, 1996, p. 211). From a modern perspective, FDI research did not just grow rich in empirical literature, but, as indicated in section 2.4.3, it also developed different relevant strands. Both aspects can be described best by examining some recently published literature reviews.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2 presents a periodical overview of the number of studies examined in FDI literature reviews that were published within the last four years. Most commonly, all studies provide strong evidence that research interest within this topic skyrocketed in the past 20 years. Those reviews can also be used to underpin the different directions and focal points within empirical FDI literature.
Figure 2: Periodical Classification of the Number of Studies examined in Literature Reviews
Abbildung in dieser Leseprobe nicht enthalten
Source: own calculation.
For instance the review by Pandya pointed out that relevant literature can most commonly be distinguished into a more macroeconomically orientated strand, analysing factors attracting FDI on a country level, and a microeconomically focused strand looking at the FDI decision from a MNEs’ perspective (Pandya, 2016, p. 456). The latter especially applies to studies reviewed by Xie et al. (2017), who examined over 250 papers published within the past 30 years, focusing on country-specific determinants of cross border M&As. Most of the studies were published in international business and management journals, analysing data from equity transactions and thereby concentrating on a firm-level analysis (Xie et al., 2017, pp. 159-174). Focusing on M&A data has a key advantage, as one need not care about zero observations, a problem which is central to empirical research focusing on country-level aggregated bilateral FDI.
Zero observations are associated with the idea that a country did not reach a certain economic threshold to attract FDI, therefore zero observations contain important information for the economic analysis and policy conclusions (Daude and Stein, 2007b, p. 326; Anderson, 2011, p. 149). Albeit the handling of zero observation is by no means straight forward and is a possible source of seriously biased estimates. This topic will be discussed in full detail in subsection 4.1. At least this methodical problem does not seem to exist in the analysis of M&A data, since those type of studies explore the determinants of specific transactions as well as their economic consequences on firm- and industry-level (Nielsen et al., 2017, p. 128).
But, as described in Pandya (2016, p. 479), scholars lately seem to favour this more granular firm-level data, as it is helpful in providing insights about the modern structure of FDI. Pandya e.g. pickes up the previously described argument made by Brainard, that services should be seen as nontradeables, and points out that by 2012, this sector accounts for roughly 63% of global FDI stocks. Even more importantly, she analyses the current state of the relationship between FDI and trade and followes that modern structure of MNEs suggest a highly complementary link in the form of global production networks, which are characterised by internalisation in the spirit of the vertical integration already described in the eclectic approach. According to Pandya, the early arguments put forward by Dunning with respect to the efficiency and asset seeking FDI still hold true and are further supported by the massive declines in transport costs, allowing for even greater fragmentations of production processes (Ibid., p. 456f.). Finally, the firm- and industry-level analysis allows to assess the role of scale economies: enterprises that have firm- and plant-level scale economies are expected to prefer concentrated production and to engage in arms-length trade. Contrary, enterprises with only high firm-level scale economies and low plant-level scale economies are rather expected to engage in FDI (Ibid., p. 471).
Although Pandya directly picked up the different theoretical approaches explaining FDI and distinguished between the micro- and macroeconomic perspectives, she does not differentiate the results with respect to country-focus, the type of dependent variables, and ultimately the methodological choices within the studies reviewed. A more promising review is therefore the already mentioned work of Nielsen et al. (2017), which will be used to provide some insights into the most important general findings in empirical FDI research.
3.1 General Findings within empirical FDI Literature
A major advantage of the review provided by Nielsen et al. (2017), who reviewed 153 empirical studies between 1976 and 2015, is the fact that the authors classify their results with respect to the theoretical approaches to FDI as described in the previous section of this thesis. They started by distinguishing up to 17 hypotheses most commonly used in the context of FDI location choice. Table 24 in the appendix provides a list of these hypotheses, the corresponding variables, and how often these variables were included in the empirical studies reviewed by the authors. Some of the most commonly tested factors are the so-called economic factors, namely demand (115), tax rates (29), wages (83), physical infrastructure (69), human capital (61) and institutional quality (57). The results for those economic factors will be presented in detail, since some of them will be used in the empirical analysis of this thesis.
Demand, usually proxied via the target country’s GDP, is most commonly found to be a significant factor attracting FDI, which is perfectly in line with the market seeking hypothesis. The findings on tax rates and wages are not that straight forward, since the resource/efficiency seeking argumentation would predict that FDI is attracted by countries with low tax- and wagelevels. Three of the 27 studies including a tax variable did not find a significant relationship, while an equal share of studies found a significant positive, as well as a significant negative relationship between tax and FDI. 83 studies examined the relationship between the wage levels and FDI and 41 studies found a statistically significant negative relationship, although 28 studies found no significance while 14 studies found a positive association between those variables. Nielsen et al. concluded that the latter findings were strongly dependent on the inclusion of other variables. In particular the significant positive relationship was apparent in studies, which did not include the variables infrastructure and institutional quality28. Therefore, they associated the significance with an omitted variable bias and concluded that it is important to control for those factors. Moreover, the authors point out that endogeneity might drive the results here, as the casual direction of the link between wage-levels and FDI might be up to debate (Ibid., p. 72f.).29 Though one empirical outcome indicated by Pandya (2016, p. 458) may give reasonable arguments for a positive relation between FDI and wage-levels: a robust observed finding seems to be the fact that MNEs pay higher wages than their domestic counterparts. In the long run, this fact may explain the positive relationship - especially for countries where FDI has a great share in GDP.
Returning to the analysis of Nielsen et al., surprisingly the authors did not discuss the findings on physical infrastructure and human capital in depth. This may be due to the fact that both factors are found to be positively significant in about 70% of all studies. Only about 10% of the studies found them to be significantly negatively associated with the attractiveness of FDI. The lack of discussion comes with some surprise since the accurate measurement of human capital is a topic of core relevance for current empirical research (Angrist et al., 2019, p. 2ff). This thesis will therefore address this issue in the following subsections.
Studies that analysed the quality of governance and institutions find those aspects to be significantly positively associated with the attraction of FDI in about 75% of the studies, whilst 3% (or two studies) found a significant negative relationship. One of those two findings resulted from sample selection bias because the study focused on the location choice of Chinese MNEs, which is assumed to be primarily driven by market potential. It is expected that countries with high future market growth potential are associated with weaker intuitions, and that Chinese companies are less afraid of engaging in those markets. The other significant negative finding was characterised as not plausible, since it resulted from the survey of bilateral FDI relations between the U.S. and Japan, who are characterised by pretty much identical institutional designs (Nielsen et al., 2017, p. 73).
To conclude, economic factors play an important role in explaining FDI. Only 20 of the 153 studies reviewed did not employ any of those factors, with 12 of them focusing on industry specific aspects of FDI.30 However, hypotheses seven to 16 are of minor importance for the empirical analysis of this thesis, since they will not be tested in the regression analysis. Their empirical results will therefore be discussed to a lesser extent.
The following hypothesis suggests that special economic zones are a significant factor attracting FDI. This hypothesis was tested in 34 studies on a sub-national level and found to be true for nearly 80% of them. Nielsen et al. conclude that 21 of those positive findings must be attributed to an analysis of such constructs in China (Ibid., p. 73f.). The next hypothesis investigates whether it is true that industrial clusters attract FDI. This idea is sometimes cited as the agglomeration hypothesis (Faeth, 2009, p. 173), and empirical research finds strong evidence that this holds true, as 27 out of 32 studies proved that industrial clusters significantly help to attract FDI (Ibid., p. 74).
Hypotheses number nine to 13 are closely related to the idea of global cities31, which are expected to act as a magnet for FDI, due to their extraordinary degrees of interconnectedness to domestic and global markets, a cosmopolitan environment and great levels of superior producer services (Ibid., p. 67). 87 studies picked up at least one of the hypotheses related to the concept of global cities and an overwhelming majority of 80% of those studies investigated FDI on a sub-national level. Except for hypothesis 13, suggesting that congestion costs negatively impact FDI, empirical research strongly supports the idea that global cities attract FDI, since all other hypotheses were found to be significantly positive in 80 to 90 percent of the studies, which tested them (Ibid., p. 74).
The following three hypotheses broadly refer to parent firm characteristics and are among the ones most rarely considered. With respect to the idea of Krugman’s New Trade Theory and the ownership advantages postulated by Dunning, five of seven studies confirm that there is a significant positive relationship between companies’ intangible assets (which must be seen as a push factor) and such FDI locations, which must be recognized as attractive, by the previously presented hypotheses. Besides this, seven of ten studies find a positive association between a firms general international experience and the probability that such companies would undertake FDI in a region or country which is characterised by high risk and/or uncertainty. Moreover, with respect to hypothesis No. 16, 84% (16 of 19 studies) find that prior location-specific experience significantly drives subsequent location choices (Ibid., p. 74).
Finally, the last proposition discussed by Nielsen et al. is of major importance, since it directly addresses the gravity approach by examining what previous empirical research had to say about the influence of distance. It is worth mentioning that the physical distance variable is the sixth most included and is covered in 49 of the 153 studies reviewed. The findings are relatively unambiguous, seeing as about 80% confirm the previously discussed negative association between various distance measures and FDI location. Remarkably, the authors concluded that they found the number of studies which considered distance to be surprisingly small, since the role of this variable is undisputed. This conclusion favours the application of the gravity approach when analysing FDI from a macro-level perspective (Ibid., p. 74).32
The analysis provided by Nielsen et al. allows for interesting insights about the most important findings from empirical FDI research of the past decades. However, it must be stated that the review, despite its comprehensiveness, is not all-encompassing as it e.g. neglects the historical debate on the explanatory power of exchange rates. The reviews of Faeth (2009) and Blonigen (2005) consider this determinant. The latter discusses FDI from a firm-level perspective, concluding that literature on this topic is rich in theories but very scarce in robust empirical findings (Blonigen, 2005, p. 387). Faeth concludes that exchange rate (stability) is mostly found to be a positive factor, helping to attract FDI and that exchange rates should be considered in models aiming to explain FDI (2009, p. 181ff.).33
Finally, the previous amendment should make clear that a presentation of all possible relevant determinants of FDI does not appear feasible. So, after pointing out that macroeconomic factors as well as the gravity approach seem to provide great explanatory power to the distribution of global foreign direct investment, it seems appropriate to pick up the somewhat short-handed discussion on the relationship between human capital and FDI.
3.2 Empirical Findings on the Relationship between FDI and HC
In this subsection the focus will be entirely on the empirical findings and the explanatory power of human capital with respect to FDI. Moreover, the particular econometrical frameworks applied to investigate the relationship will be presented in further detail, as human capital is e.g. highly correlated with other variables like the per capita GDP, which is also commonly used as an independent variable in relevant studies. Moreover, endogeneity and omitted variables bias are some issues that might drive the results. Like in section 2, the findings will be presented in a chronological order ultimately leading to a division of this subsection into two parts: the first will investigate empirical findings until the year 2002, since on the one hand this was the year when Anderson and van Wincoop presented their seminal work on the Gravity Equation, thereby pointing out that previous papers seriously overestimated relevant findings (Anderson and Van Wincoop, 2003, p. 186). On the other hand the early 2000s were characterised by another methodological innovation because FDI scholars started to focus on the information incorporated in zero FDI flows and stocks (Levy Yeyati et al., 2003, p. 15f.), an aspect briefly discussed in section 3. Consequently, it is assumed that empirical results of the past 16 years presented in the second section yielded more consistent and unbiased estimates of the interaction between human capital and foreign direct investment.
3.2.1 The empirical Relationship between FDI and HC before 2003
One of the earliest modern reviews of empirical findings on FDI has been published by Blom- ström and Kokko (1997).34 Albeit the authors explicitly review home-country characteristics, their examination of literature investigating human capital exclusively focuses on firm-level analysis, proposing that in the eighties researchers only investigated the influence of training measures as a human capital proxy on firm-level while primarily investigating its explanatory power for developing countries (1997, p. 15ff.).
This conclusion is somewhat misleading, as the explanatory power of HC on a country-level was already tested in the late seventies, when Root and Ahmed (1979) studied the determinants of per capita FDI inflows of manufacturers into 58 developing countries between 1966 and 1970, while controlling for the literacy degree of the target countries as well as for school enrolmentrates (primary and secondary). The authors chose a multiple discriminant analysis over multiple regression analysis, since they regarded their dependent variable as categorical rather than continuous (Ibid., p. 756). Both variables turned out to be insignificant in the analysis, which the authors traced back to the very high correlation (over 0.5) with the highly significant GDP variable (Ibid., p. 762).
Another early analysis of HC was conducted by Schneider and Frey (1985), who also focused on per capita FDI inflows into developing countries and applied a multiple regression analysis, while analysing cross-sectional data for the years 1976, 1979 and 1980 (Ibid., p. 161). The authors used the number enrolled in secondary school as a percentage group of age group to proxy for skilled labour within a country (Ibid., p. 174).35 Their results are mixed, as on the one hand HC is stable and significantly positive at the 5% level for all the three years, and on the other hand the results are sensitive to the inclusion of political factors. More precisely HC turned insignificant in every period observed after controlling for political instability and the aid received by the respective country (Ibid., pp. 167, 169f.).
It must be concluded that Blomström and Kokko (1997) correctly pointed out that early empirical research focused on developing countries, but as described above, some scholars analysed the country-level perspective in depth. Overall, subsequent research has summarised the early findings on the relationship between foreign direct investment and human capital by stating that “[...] HC seem not to have a significant impact in attractingFDI. ” (Alsan et al., 2006, p. 623). Although this conclusion must be considered with caution, because those studies not just focused on subsamples of developing countries but they also investigated rather short periods of time. Moreover, since the study of Schneider a\v\ Frey (1985) did not provide any information on correlations between the independent variables, multicollinearity might have been a severe problem in their approach.
The last decade of the past century marked a period where the relationship between FDI and HC received much more attention and especially the gravity approach played an important role within empirical examinations. According to Bergstrand and Egger (2013, p. 561), a paper by Eaton and Tamura (1994) is among the first to analyse the link between FDI and HC via the gravity framework. Their FDI-data includes the years 1985 to 1990, while their data on HC only covers 1985, so they use these values for the whole sample period. Their proxy is the years of schooling of an average worker (Ibid., p. 489). The results are rather mixed and should be interpreted carefully, since the authors look at bilateral flows between USA and Japan respectively their trading partners, therefore results on the role of HC only refer to the in- and outward FDI for those two countries (Ibid., p. 496). While Eaton and Tamura find that HC does not contribute to the bilateral relations of Japan, their analysis shows a significant contribution of HC to the bilateral relations of the USA. The authors argue that this is due to the fact that U.S. multinationals are engaged in more technology intensive industries (Ibid., p. 497).
In a follow-up paper from 1996, Eaton and Tamura further investigate the role of HC and make more precise statements on its effect as a factor attracting FDI from Japan and the U.S.. They report elasticities for outward FDI of 1.1 respectively 1.4 for the former and latter country. According to their analysis, which is conducted with the same data as in the aforementioned paper, a high level of human capital is associated with larger FDI positions in the target countries. The authors explain this finding with the idea that a high level of HC decreases the costs of FDI (Ibid., pp. 52, 63f.).
Another early work applying the gravity approach while accounting for HC, is a paper by Wei (1995). He utilised a fixed effects cross country regression model while conducting a stock and flow analysis based on figures of the five largest FDI source countries in the period from 1987 to 1990. Human capital was measured via the literacy ratio in 1992 (Ibid., p. 193f.). The author reported that a 1% increase in the literacy rate is associated with a 1.57 percentage points increase in FDI - the results being similar for stock and flow analysis (Ibid., p. 194).36
A further important contribution of the nineties was the work of Kumar (1996), who was among the first to study FDI on a global scale, as his analysis covered up to 84 industrialised and developing countries. He examined U.S. outward FDI in a broader sense, since he analysed the Research and Development (hereafter R&D) expenditures of U.S. parent companies abroad in the years 1977, 1982 and 1989, thereby focusing on internalisation and location specific advantages as described by Dunning (Kumar, 1996, pp. 673f., 679). His analysis firstly revealed that between 1966 and 1989 U.S. MNEs spent about 95% of their R&D expenditures abroad in the industrialised countries. On the one hand this result supported the idea that empirical findings derived from FDI data for developing countries should not be seen as representative. On the other hand, it seems reasonable to argue that measuring FDI via R&D expenditures abroad is also accompanied by some restrictions, but it seems to be an appropriate approach to investigate target country characteristics helping to attract FDI.
In his empirical analysis, Kumar additionally tested whether technological resource factors significantly help to explain a country’s ability to attract R&D expenditures from the U.S.. To investigate this relationship, he used country’s total national R&D expenditures as a share of GNP, the number of patents granted to the residents of a specific country as a proportion of total patent grants, and finally total gross enrolment ratios for higher education (Ibid., pp. 679, 686). Regarding the econometrical framework, Kumar’s results must be seen as more reliable, since he pointed towards the correlation between the aforementioned variables. In addition, he made use of White’s consistent estimator approach to correct for possible heteroskedasticity in the pooled OLS regression (Ibid., p. 680).
From the three variables tested, only the R&D expenditures are statistically significant when analysing the full country sample, thereby confirming that target countries with high R&D expenditures are more likely to attract further inflows from U.S. MNEs (Ibid., p. 681).37 These findings are also confirmed for subsamples, with the provision that the share of residents in total patents granted in a country becomes highly significant for the subsample of developing countries. Kumar interpreted this finding as a strong hint towards the internalisation strategy, which seems to be applied most within countries with weak institutions and low levels of intellectual property rights protection (Ibid., p. 682f.).
Another major early contribution was the work of Borensztein, De Gregorio and Lee (1998), who investigated the relationship between FDI, HC and economic growth in FDI receiving countries. They analysed this relationship for 69 developing countries between 1970 and 1989, applying the seemingly unrelated regressions technique and using the initial-year level of average years of the male secondary schooling as a proxy for human capital (Ibid., p. 122f.). To account for the collinearity problems inherent in such models, the authors used instrumental variables technique and conducted two and three stage last square regressions as robustness checks (Ibid., p. 133f.). Borensztein et al. found out that FDI can have a significant positive influence on economic growth, but the effect is dependent on a sufficient stock auf human capital.38 No such interaction has been found for domestic capital, which the authors interpreted as “[...] a reflection of differences of technological nature between FDI and domestic investment. ” (Ibid., p. 134).
When looking at publications from the early 2000s, one publication stands out as it applied an innovative HC measure, which will also be applied in the empirical analysis of this thesis. Globerman and Shapiro (2002, p. 1901ff.) investigated in- and outward FDI in 144 developed and developing countries for the period average between 1995 and 1997 via an econometrical approach similar to the work of Kumar (1996). The authors were the first to use the Education Index incorporated in the Human Development Index (hereafter HDI) as a proxy for HC.39 Regarding incoming FDI, they found education to be generally significant and concluded that it helps encouraging FDI (Ibid., p. 1911). Furthermore, the analysis allowed for the following conclusions: on the one hand the influence of HC was identical for the full sample and a subsample of 115 countries (excluding OECD countries), with coefficients of 1.183 (significant at 1% level) for the former and 1.190 (significant at 5% level) for the latter. On the other hand, the results showed tremendous sensitivity with respect to the exclusion of other than the OECD countries, as a regression with a sample of 114 countries generated a coefficient of 2.068 (significant at the 1% level) (Ibid., p. 1909). Surprisingly, the authors did not comment on this finding. Therefore, it is not possible to conclude which specific country characteristics enhance the positive influence of human capital on FDI. This may be due to the fact that Globerman and Shapiro concluded that their findings on HC might suffer from multicollinearity and causality issues (Ibid., p. 1915). Ultimately, the authors did not embed their theoretical findings within one of the classical theoretical frameworks presented in section 2.
3.2.2 The empirical Relationship between FDI and HC since 2003
The first paper accounting for multilateral resistance while addressing the information incorporated in zero flows and focusing on the explanatory power of HC was the work of Shalz. who argued that “/C-7 the level of education in a host country should [positively] influence the level of inwardforeign direct investment [...]. ” (2003, p. 117). His argumentation focused on factor inputs, the assumption that MNEs serve the host market and the fact that multinationals are engaged in differentiated-product industries or technologically advanced industries, which appeal to more educated buyers. On the other side Shatz puts the resource seeking argument by underpinning that MNEs establish affiliates in countries with low level of education, as those countries are associated with low labour costs (Ibid., p. 117). According to his gravity-based analysis, a vast majority of FDI investors prefer educated labour. Due to the fact that his data on education (measured as the average years of schooling) incorporated three segregated levels, Shatz also draws the conclusion that education is rather a significant factor in countries with a heterogeneous level of education (Ibid., p. 144, 146). Still those findings should be interpreted very carefully, as he used cross-sectional data, which only covered the USA as origin country and instead of using FDI data he made use of the 1995 sales of affiliates of all non-banking U.S. multinationals in up to 192 target countries (Ibid., pp. 119, 124).
[...]
1 This statement revers to a paper published by Konara and Wei (2019) on the complementarity of human and language capital in FDI analysis. This paper will be discussed in detail in section 3.2.2 of this thesis.
2 Mr. Mukhisa Kituyi is the Secretary-General of UNCTAD.
3 The FDI inflow shares in global GDP were all calculated by the author on the basis of GDP in current USD published by UNCTAD.
4 Some economic analyses like for example in Alfaro et al. (2008) and Lane and Milesi-Ferretti (2001) do not distinguish between portfolio and foreign direct investment.
5 In the spirit of the analysis of Goldstein and Razin (2006, pp. 272-275), who point out that those types of investment are driven by diametrically different motives, and Anderson (2011, p. 156f.), who concludes that portfolio investment asks for different types of models, approaches mixing those types of investments are not further discussed in this thesis.
6 Xie et al. (2017, p. 137f.) provide a comprehensive discussion on the possible drivers and motives of companies when deciding about the type of entry mode into a foreign market. This discussion is not picked up in this thesis, as the authors point out that it mostly relates to managerial decisions and firm-level characteristics, while this thesis focuses on FDI from a country-level perspective. Although one argument made by Xie et al. is equally important from a country-level perspective: in cases when target countries limit the ability of foreign MNEs to acquire national enterprises, greenfield investment may be the only cost-effective form of market entry.
7 This is remarkable, since e.g. book IV of Adam Smith’s 1776 seminal work on the nature of the wealth of nations already discussed theories explaining the occurrence of international trade.
8 For a more comprehensive review on further empirical findings and why research neglected the differential rate of return hypothesis, see e.g. Caves (1996, p. 26f.).
9 It seems noteworthy that Vernon attributed this process specifically to the U.S. and focused on sectors that were labour saving. Moreover, he pointed out that this logic might not apply to certain industries, like e.g. defence, which might be driven by other considerations.
10 Dunning (1979, p. 269f.) went even further, as he argued that the foreign production of companies in the fifties and sixties was a field, which was completely dominated by U.S. MNEs, thereby research of those two decades simply reflected this positive outcome.
11 A full review of FDI theories based on the concept of market imperfections is e.g. provided by Agarwal (1980, pp. 749-754).
12 Ownership, Location and Internalization.
13 In an amendment, Dunning pointed out that such location-specific endowments might very well be understood as the proximity to the point of sale - being a positive aspect - and distance, which would be a proxy for transport and further transfer costs - being a negative aspect (Dunning, 1977, p. 414).
14 For a comprehensive classification of the single OLI aspects, see Dunning (1979, p. 276).
15 Such mechanisms are usually arms-length trade or licensing. A comprehensive discussion of the pros and cons of the different approaches is e.g. provided by Pandya (2016, p. 456ff.) or Macharzina and Engelhard (1991, pp. 34-38).
16 LOF stands for the liabilities of foreignness, which are described as the costs of doing business that are over and above those of domestic rivals (Nielsen et al., 2017, p. 65).
17 The term first stage gravity equation is introduced by the author and refers to the chronological distinction in Bergstrand and Egger (2015, pp. 540, 543), who distinguished between gravity models in the period 1962-1979 and from 1979 to present.
18 This conclusion is less self-evident and will be discussed in more detail in subsection 2.4.3.
19 Like already stated, the analysis of trade is less important within this thesis, therefore the econometrical implications of the model will not be presented, but a full discussion can be found in Anderson (1979, pp. 109-112).
20 Those models are not further described, as they primarily aim at explaining the occurrence and structure of international trade. A comprehensive discussion on implications and the implementation is e.g. provided in Bergstrand and Egger (2013, pp. 549-553.) as well as in Head and Mayer (2014, pp. 144-148).
21 Brainard s paper was firstly published as a NBER working paper (senes 4580) in 1993.
22 The approach is explained in full detail in Brainard (1997, p. 526f.).
23 Blonigen (2005), as well as Azémar and Desbordes (2010 provided a comprehensive review of early empirical studies that followed such an approach. Besides, Azémar and Desbordes (2010, pp. 929ff.) argued that U.S. foreign affiliate sales (FAS) data was the only one allowing for a precise investigation of FDI motives in the spirit of horizontal, vertical and export seeking FDI.
24 The literature discussion in section 3.2.2 will show that empirical analysis of FDI is rather sensitive to the measure applied.
25 This term refers to the title of a paper published by Pandya (2016).
26 The paper by Blonigen and Davis was first published as a NBER working Paper (series 7929) in 2000. Another paper of the early 2000s, which applied the gravity approach to investigated BITs, was a publication by (Hallward- Driemeier, 2003, p. 13f), but she used distance only as a control variable and did not report any regression results.
27 Since the information costs approach rather refers to the firm-level analysis of FDI and the specific information costs involved very significantly from firm- and industry-level perspective, the author will not discuss this topic.
28 A comprehensive discussion of those aspects is e.g. provided by (Alchian and Demsetz, 1972).
29 Although Nielsen et al. (2017, p. 71) discuss general methodological problems within FDI research and offer some promising solutions regarding the causality problem, like e.g. the use of instrumental variables, unfortunately only three of all the reviewed studies, that covered wages and taxation, methodologically addressed the causality issue. The study by Carstensen and Toubal (2004) investigated FDI inflows into the Central and Eastern European countries between 1993 and 1999, while Bobonis and Shatz (2007), as well as Rogers and Wu (2012), analysed FDI inflows into the U.S. on a state level. All of these studies seem to be too specific to allow for general conclusions regarding the relationship between FDI and wages as well as taxes, when causality is taken into account. Specific results are therefore not discussed.
30 The quantitative analysis of the papers and hypothesis covered by the papers which were examined by Nielsen et al. are provided in the technical appendix to this thesis.
31 A comprehensive discussion on the definition of this term and further characteristics of such areas is e.g. provided by Beaverstock, Smith and Taylor (1999).
32 Not all of the 49 studies applied the gravity approach directly, since ten publications did not control for the market size. Though it is not possible to investigate whether some of these ten studies were among those eleven that did not find distance to be significant.
33 Some recently published studies examining the relationship between FDI and exchange rates on a country-level are Barrell et al. (2017) and Harms and Knaze (2018). They are not presented in greater detail since exchange rates will not be applied as an explanatory variable within the empirical framework of this thesis.
34 This review was not considered in the previous section due to the authors not providing a separate list of the literature reviewed.
35 This definition seems somewhat incomplete as the authors did not explicitly name the age groups included.
36 The results were significant on a 10 percent level for flows and on a 1 percent level for stocks.
37 Kumar tested three different models, characterised by different configurations of the independent variables, to further account for the high collinearities. The results for R&D expenditures were all significant at the 1% and 5% level, although the comparability between the models is somewhat limited since the number of observations varied significantly (Kumar, 1996, p. 681).
38 It seems worth mentioning that the threshold was about 0.52 years of secondary schooling (Borensztein et al., 1998, p. 125), which makes the practical relevance somewhat negligible. Moreover, specific results are not reported here, since the authors argued that the fact that they used Balance of Payments data on FDI lead to a serious underestimation of the coefficients (Ibid., p. 134).
39 The Education Index combines adult literacy and primary, secondary and tertiary enrolment ratios for a certain country (Globerman and Shapiro, 2002, p. 1907).
- Citation du texte
- Lev Nazarov (Auteur), 2019, Bilateral Foreign Direct Investment in OECD-Countries. Exploring the Link between FDI and Human Capital via the Gravity Equation, Munich, GRIN Verlag, https://www.grin.com/document/1337515
-
-
-
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X.