RIJKSUNIVERSITEIT GRONINGEN
FACULTY OF ECONOMICS AND BUSINESS
The Impact of the
Revealed Comparative
Advantage on Inward and
Outward FDI Intensities
of OECD Countries
[Master Thesis: International Economics and Business]
[5.07.2011]Franziska Korrmann s1535293
f.korrmann@student.rug.nl
Supervisor Master Thesis: dr. M.S.S. Krammer Supervisor Methodology: dr. P. Rao Sahib
Abstract
This paper examines the effect of the revealed comparative advantage (RCA), on inward and outward foreign direct investment (FDI) intensities. Industry based FDI determinants are not well understood. This research addresses the gap in the literature by analyzing 28 OECD countries and 34 ISIC Rev.3 industries of the years 1998, 2003 and 2008. While a positive impact of the RCA on FDI inflow is found the results of the effect on outward FDI flows are mixed.
Table of Content
Introduction 3
Literature Review 5
Linking the Comparative Advantage to FDI Flows 5
Most Related Literature and Research 9
Hypothesis Development 9
The Model 12
Data Collection and Variable Measurement 14
Dependent Variable 14 Explanatory Variable 16 Control Variables 16 Methodology 17 Results 17 Conclusion 22
Limitations and Suggestions for Future Research 23
References 25
Appendix 29
Table A1: Overview Cross-Country FDI Determinants 29 Table A2: FDI Industry Data – Industry Classification (ISIC Rev. 3) 30
Table A3: List of Countries 30
Table A4: Overview Variable Definition and Calculation 31 Table A5: Diagnostic Test Outcome: pooled OLS 32 Table A6: Fixed Effects Estimations (four year average FDI flows) 32 Table A7: Random Effects Estimations (four year average FDI flows)
33 Table A8: Random vs. Fixed Effects Estimations (four year average
of FDI flows) 34
Table A12: Random vs. Fixed Effects Estimations (single year FDI
Introduction
FDI has been an increasing phenomenon. As can be seen blow, FDI flows have been steadily increasing over time and have basically taken off in the late 1990’s; though the volatility of FDI flows has also been increasing. This trend has also led to an increasing importance of FDI flows for economic growth and development. Furthermore, with the increase in FDI flows other effects like technology diffusion and social impacts have become more prevalent.
Graph1: OECD FDI Flows
Source: OECD.Stat
Researchers, policy makers and managers alike have an interest in knowing what the driving forces behind the increasing FDI flows are and what impact they will have on their economies and competitive structure for their company. There is a need from all involved parties to know how their economies and companies can benefit best and take advantage of this trend.
The purpose of this paper is to find out why multinational companies (MNEs) would want to invest more in some industries in a certain country when compared with other industries. As country FDI analysis has shown, FDI takes place when MNEs can benefit from advantages that outweigh the disadvantages of moving operations abroad (Feath, 2009). Hence, the factor of interest of this paper will take up a measure of the advantage an industry holds. Specifically, the comparative advantage of a country will be analyzed with respect to FDI flows. The evidence of the link between a country’s comparative advantage in a certain industry and the FDI inflow or outflow of that industry is scarce (Palangkaryaya & Waldkirch, 2008).
The intuition is that an industry that holds an advantage should receive more FDI. This is because foreign companies might want to get access to the source of that underlying advantage. For example, if the underlying advantage is related to the skills of the work force, the foreign company could gain access to it by recruiting workers from that industry. On the other hand, if an industry holds a comparative advantage, a company might be more reluctant to move its operations abroad, because it might lose the access to the source of this advantage. For instance, if the advantage is caused by low factor costs, these low factor costs might not be present in another country and hence it will not be beneficial to move operations abroad.
The comparative advantage measure of this paper will be trade-related. Specifically, the Balassa index, which was developed by Balassa (1965), will be used. The Balassa index measures the export specialization of a country which expresses the revealed comparative advantage (RCA) of a country. A country holds a revealed comparative advantage in a certain industry if its export share of that industry is larger than that of a specified reference group. In other words, a country is said to be specialized in the production of an industry if its relative exports exceed that of the reference group in the same industry.
The dependent factor of this analysis is the inward and outward FDI intensity. Hence, the research of this paper concerns the impact that the RCA has on both the inward and outward FDI intensity. It will be asked if the RCA influences the relative inward FDI intensity. In other words, if a country holds a comparative advantage in a
industry? On the other hand, this paper will investigate if the RCA also has an effect on
the relative FDI outflows. Specifically, if a country holds a RCA in an industry, how does
this affect the relative outflows of FDI from that industry?
This paper will use OECD data of 28 OECD countries and 34 industries from the years 1998, 2003 and 2008 to determine the effect of the RCA on inward and outward FDI. A multiple regression analysis will be run with inward and outward FDI intensity as the dependent variable and the RCA as the explanatory factor. In addition, a fixed and random effects analysis is run. The models will be evaluated against each other and the results will be compared.
Following firstly, the existing literature will be discussed and then hypothesis will be developed from this. Next, the model will be put in an equation form and be explained. This is followed by the specification of the data collection and variable measurement. Then the methodology for testing the derived hypothesis will be developed. After this the results will be discussed and compared. Finally, a conclusion with limitations and ideas for further research will be given.
Literature Review
Linking the Comparative Advantage to FDI Flows
According to Ricardo, the comparative advantage results from relative endowment differences between countries in for example, resources, technology, entrepreneurial skills or factors of production (Husted & Melvin; 2004). The comparative advantage determines trade flows between countries. Holding a comparative advantage, a country can more efficiently and hence more cheaply produce a certain good. Countries are said to specialize in the production and the exports of the goods in which they hold a comparative advantage.
initial conditions (Grossman & Helpman, 1990; Morrow, 2010: Gourdon, 2009; Costinot, 2009; De Ferranti et al., 2002). Those determinants of the comparative advantage go beyond the measurement of factors intensities such as labor and capital in the production process. Hence, the analysis will use the RCA as a measure of comparative advantage, since it much better reflects the full set of comparative advantage determinants. Since the RCA is a ratio of the relative export share of a country in a specific industry compared to that of a reference group, it reflects the overall comparative advantage of a country in that industry.
Even though the literature directly linking the comparative advantage to FDI flows is scarce other links can be found between the two. The sources of the comparative advantage have also been found determinants of FDI flows. Hence, because the sources of the comparative advantage are determinants of FDI, there should be a direct relation between the comparative advantage, and FDI flows. Following, the individual sources of the comparative advantage will be linked to FDI flows.
Factor Endowments: Factor endowments and the resulting factor costs
differentials as a determinant of FDI have received much attention in the literature. Markusen (2002) explains in the vertical FDI model (VFDI) that firms seek to located in foreign countries, because of factor cost differentials, such as labor costs. In this relation, factor cost differentials are a form of comparative advantage. In addition, the factor-proportion hypothesis, which holds that MNEs should located where factor cost differentials are large (Markusen, 1984). Braconier, Norbäck and Urban (2005) estimate that approximately 20% of US affiliate sales are due to labor cost differentials. Furthermore, Culem (1988) found that host labor costs have a negative effect on FDI flows. Hence, the greater the advantage due to factor cost differentials, resulting from endowments, the higher the FDI inflows and the lower the outflows.
Technology: Technology differences also cause a comparative advantage to arise.
Culem (1988) found a positive impact of R&D on FDI flows. In other words, the better the advantage, due to technology, the more FDI inflows an industry should receive and the more FDI flows out from an industry.
Market Size: Another cause of the RCA, which can been linked to FDI flows, is
market size. The horizontal FDI (HFDI) model of Markusen (1997, 2002) linked market size to increasing inflows of FDI. Furthermore, the proximity-concentration hypothesis, developed by Krugman (1983), holds that increasing flows of FDI should go to larger countries, because economies of scale can be better exploited. Empirical studies taking up market size as a determining factor of FDI flows are Carr, Markusen, Maskus (2001) and Head and Ries (2008). Some GDP related measure is taken to express market size. The authors have found a positive effect of market size on FDI flows. Thus, the higher the advantage, resulting from market size, the greater the FDI inflows should be.
Location: Location has also been found as a source of the comparative advantage.
A location advantage is caused by for example low trade costs or preferential trade policies. In both, the proximity-concentration trade off theory (Markusen; 1984) and the vertical VFDI model (Markusen; 2002), are trade costs and trade policies determining factors of FDI flows. Dunning’s (1977; 1979) OLI model, is also build on a locational advantage, which is related to for example preferential tax structures. Both of those determinants can be found to have an impact on FDI flows (Carr, Markusen, Maskus; 2001, Miroudot, Ragoussis; 2009, Culem; 1988). The effect of tariffs and trade costs is negative on FDI inflows and outflows. Consequently, the higher the advantage, resulting from location, which means the lower the trade costs or tariffs resulting from FDI, the more inward or outward FDI takes place.
Initial conditions: Initial conditions are agglomeration effects or first-mover
the comparative advantage refers to the agglomeration effects. The larger the advantage, resulting from high agglomeration, the more FDI inflows can be expected.
Demand patterns: It is quite difficult to directly link demand patterns to FDI
flows. According to the horizontal FDI theory (HFDI) MNEs enter foreign markets through FDI in order to serve the foreign demand through local production (Markusen, 1997). In addition, Krugman (1991) also argues that firms should locate in the region with larger demand. However, his assessment is not based on FDI as such. From a theory perspective there seem to be some indications that it would be beneficial for MNEs to enter industries where the local demand is high rather than low. No direct empirical evidence can be found and hence this connection is somewhat uncertain. However, loosely interpreting, it seems that the higher the advantage, resulting from demand patterns, the more FDI should flow in such an industry.
Market Structure: Similar to demand patterns, it is also difficult to link market
structure to FDI flows. Although most FDI models assume a specific competitive structure, not much direct research taking market structure into account, has been done. For example, while the theoretical models of the HFDI and VFDI take the market structure implicitly into account, through the profit functions of a company, which are dependent on the competitive model and the number of companies (Markusen, 1997); the FDI determinant research has only scarcely taken market structure into account. The more companies compete in a certain industry, keeping demand constant, the lower the profit per firm (Nicholson et al., 2008). Campa, Donnenfeld and Weber (1998), find a negative effect of industry concentration on inward FDI. The higher the advantage from market structure, through e.g. a low industry concentration, the more FDI should flow into that industry.
Most Related Literature and Research
The existing literature and empirical research has so far only scarcely looked at the direct relation of the comparative advantage and FDI flows.
Qiu (2003) develops a mathematical model to explain the impact of the comparative advantage of a country on FDI flows. This author links the market opportunities, which result from the comparative advantage, to FDI inflows and outflows. Deriving the profit functions for companies in an industry holding a comparative advantage and for companies in industries of no comparative advantage leads the author to conclude that holding a comparative advantage leads to decreased FDI outflows and to increased FDI inflows. The author concludes that having a comparative advantage in an industry attracts inward FDI, while it discourages outward FDI.
Maskus and Webster (1995) and Palangkaraya and Waldkirch (2008) empirically research the effect of the comparative advantage on FDI flows. The outset of the economic model and the methodology, which is used by the authors, is the same. They measure the comparative advantage through the factor proportions hypothesis. This is the Heckscher-Ohlin Vanek (HOV) model1. Hence, the authors take factor endowments as the sole determinant of the comparative advantage. Although the results differ somewhat in magnitude of the coefficients between countries and manufacturing and service sectors, the authors find a positive correlation of net FDI (inward minus outward) with the comparative advantage. This implies that inward FDI (IFDI) is positively correlated with the comparative advantage. On the other hand, outward FDI (OFDI) is negatively correlated with the comparative advantage. Furthermore, because the authors find a less than perfect positive correlation between the comparative advantage and net FDI, which suggests that factor proportions are not the sole determinant behind FDI flows.
Hypothesis Development
The above discussed literature has established the link between the comparative advantage and FDI flows. Mainly the literature points in a specific direction with regard to the influence of the RCA on inward FDI. In particular, the general link between the
1
sources of the RCA and FDI inflows indicated that the larger the advantage, the higher the inflows. Furthermore, Qui (2003), Maskus and Webster (1995) and Palangkaraya and Waldkirch (2008) have all shown that the comparative advantage positively influences FDI inflows. Hence, the higher the comparative advantage the larger the FDI inflows of that industry should be.
So if a country is specialized in a certain industry, foreign companies are expected to invest more into that industry, as compared to other industries in the same country. Specifically, the higher the comparative advantage of a country in a certain industry, the higher should be the share of FDI inflows in that industry as compared to the total FDI inflows of that country. Thus, industry FDI flows are assessed in relation to the country total. FDI intensity is defined as the industry FDI flows divided by total FDI flows of a country.
H1a: The revealed comparative advantage has a positive impact
on the inward FDI intensity.
As the revealed comparative advantage is calculated with regard to a reference group, I wonder if FDI inflows also stand in relation to that reference group. If a country has a higher comparative advantage as the reference group, it is only reasonable to argue that it should also receive relatively more inflows of FDI in that industry, as the reference group. This means that the regional industry specialization measured by the RCA, might lead to a regional IFDI specialization. Hence, one could say that the regional distribution of IFDI flows should be in accordance with the revealed comparative advantage.
H1b: The revealed comparative advantage has a positive impact
on the inward FDI intensity as compared to the reference group.
concluded that the comparative advantage has a negative effect on FDI outflows. Since their research is the most related to the research of this paper, more weight is given to their conclusions, rather than to the indirect implications drawn from the link between the sources of the RCA and FDI flows.
Hence, if a country is specialized in a certain industry and thus holds a comparative advantage there, it would be unreasonable to move the production to a country which does not hold such an advantage. It only makes sense for companies to shift their production to countries where the industry comparative advantage is higher than in their own country, because otherwise they would move operations to a country where production is less efficient and hence more expensive. Following, it is expected that the higher the comparative advantage, the lower will be the outward FDI flows compared to the other industries of the country.
H2a: The revealed comparative advantage has a negative impact
on the outward FDI intensity.
Furthermore, if a country holds a higher comparative advantage in an industry as compared to the reference group, not only the should the outward FDI flows compared to the other industries of the country be lower, but it should also be less than that of the reference group in that industry and other industries.
H2b: The revealed comparative advantage has a negative impact
on the outward FDI intensity as compared to the reference group.
Figure 1: Relationship RCA Inward and Outward FDI
The Model
The revealed comparative advantage is a measure of the competitive advantage, which a country holds in a certain industry. Specifically, the comparative advantage is a measure of international competitiveness. It measures the competitiveness of a country’s industry as compared to that of other countries in the same industry in terms of exports. The more a country exports in a certain industry as compared to a reference group of countries, the higher the underlying advantage of the country’s industry must be. The RCA was first developed by Balassa (1965) and is in the literature also sometimes called the Balassa Index (BI). The revealed comparative advantage is calculated as follows:
RCAij = (Eij/Eil) / (Ekj/Ekl)
E are exports, i is country i, j is industry j, l is the reference group of industries
and k is the reference group of countries. Hence, the comparative advantage of a country in industry j is measured by country i’s relative exports in that industry as compared to the relative exports of the reference group k in industry j. The reference group of countries are 28 OECD countries (see appendix table A3). The reference group of industries, l, in this paper will be the total of all industries. A country is said to be relatively specialized in an industry if this index measure is above one. The same indices, as above, will be used in the rest of the paper.
The index of the RCA is then used to explain the inward and outward FDI intensities of a country on an industry basis. The proposed regressions, which will be tested, are:
1a: log(IFDIij/IFDIil) = b1 + b2 log(RCA)ij + b3 (home GDP per capita)i + b4 (openness)i + b5 (R&D)ij + b6 (employment share)ij + b7 log(average wage per employee)ij + b8 (fixed investment share)ij + eij
1b: log((IFDIij/IFDIil) / (IFDIkj/IFDIkl)) = b1 + b2 log(RCA)ij + b3 (home GDP per capita)i + b4 (openness)i + b5 (R&D)ij + b6 (employment share)ij + b7 log(average wage per employee)ij + b8 (fixed investment share)ij + eij
2b: log((OFDIij/OFDIil) / (OFDIkj/OFDIkl)) = b1 + b2 log(RCA)ij + b3 (home GDP growth)i + b4 (openness)i + b5 (R&D)ij + b6 (employment share)ij + b7 log(average wage per employee)ij + b8 (fixed investment share)ij + eij
The above listed regressions correspond to the hypothesis H1a-H2b. The positive sign in front of the variables is added arbitrarily, it does not indicate the sign of the expected influence on the dependent variable.
Equation 1a (corresponding H1a), expresses the relationship of relative inward FDI in industry j as compared to the total FDI, l, of all industries into the country and the revealed comparative advantage of industry j in country i. It is proposed that the RCA explains the intensity of FDI inflows. Specifically, the higher the RCA of industry j, the
higher the IFDI intensity in that industry should be. Hence, the expected sign of b2 in
equation 1a is positive. For all equations holds that exports in the calculation of the RCA are exports to the world and inward FDI into country i in industry j are total inflows from the world.
Equation 2a (corresponding to H2a), has a similar intuition behind it as 1a. Namely, the outward FDI intensity from country i and industry j is explained by the RCA of country i in industry j. The higher the RCA of industry j, the lower the outward FDI
intensity from country i in industry j should be. Hence, the expected sign of b2 in 2a is
negative. OFDIil is the total outflow of all industries, l, in country i, to the rest of the
world.
Equations 1b and 2b (corresponding to H1b and H2b respectively), are similar to 1a and 2a with the difference that inward and outward FDI are now not only a relative measure of country i, but are also taken in relation to the reference group k. The intuition behind this is that if country i is relatively more specialized in the exports of industry j as compared to the reference group k, it should also receive relatively more FDI inflows in industry j as the compared to the reference group k. Hence, the expected sign of b2 for
equation 1b is positive. On the other hand, if country i is relatively more specialized in industry j, as compared to the reference group k, it should have relatively lower FDI outflows of industry j, as compared to reference group k. Hence, the expected sign of b2
The controls, which are added in each regression, are two country control variables and four industry control variables. GDP per capita (spending power) and openness (ability to trade) are controls for inward FDI flows, while GDP growth (growth of the economy) and openness control outwards FDI flows. In addition, in each regression the industry controls, which are used, are the employment share per industry (industry size), R&D share per industry (technological progress), average wage per employee per industry (labor costs) and relative fixed investments per industry (investments). For further definitions and calculations of the control variables, please see appendix Table A4.
Data Collection and Variable Measurement
All data will be obtained from OECD.Stat, except openness is retrieved from the World Penn Tables from the University of Pennsylvania. In addition, all industries are classified at the ISIC rev.3 at the two-digit level. This is with the exception of agriculture, which is only provided at the one-digit level. All exports, FDI flows and control variables correspond to each other.
After matching the industries and countries of the variables, a sample of 28 OECD countries and 34 industries is obtained. Some countries are eliminated from the sample of the OECD, because they return only missing values, are classified in their FDI data, or cannot be matched for the FDI and RCA data. The list of countries and years can be found in the appendix in Table A3.
All data is collected for the years 1998, 2003 and 2008.
Dependent Variable
Graph 2a & 2b: Inward and Outward FDI Volatility over Time, G6 countries2
Source: OECD.Stat
FDI flows are measured in ratios. Specifically, the flow of an industry is taken to the total FDI in- or outflow of the country. The IFDI and OFDI observations for each country and industry (including the country total IFDI and OFDI) are used to calculate all the required measures to arrive at the dependent variables for equations 1a-2b.
Table 1: Summary Statistics Dependent Variables
Hypothesis Dependent Variable Years Mean St.Dev Min Max H1a Inward FDI intensity 4 year average -4.733708 2.040037
-15.45426 3.305439 H1b
Inward FDI intensity
to OECD 4 year average -0.142675 1.930926
-10.09374 5.091696 H2a
Outward FDI
intensity 4 year average -4.950896 2.257823
-14.45954 4.081942 H2b
Outward FDI
intensity to OECD 4 year average
-0.2663276 2.005815
-8.557876 6.291957 H1a Inward FDI intensity single years -4.238895 2.17353
-13.91051 5.732471 H1b
Inward FDI intensity
to OECD single years 0.3367838 2.211892
-8.345618 8.713877 H2a
Outward FDI
intensity single years -4.515059 2.2412
-13.69312 5.108274 H2b
Outward FDI
intensity to OECD single years
-0.3714644 2.130824
-8.277466 7.376011 The dependent variable of all equations has to be logged. That is because its distribution is not normal. Examining the summary statistics and the histogram for all dependent variables shows that the kurtosis far exceeds three, which is the value for the
2
normal distribution. Furthermore, the skewness of all dependent variables is very different from zero. After logging all dependent variables their kurtosis values are much closer to three and their skewness is much closer to zero.
Explanatory Variable
Hinloopen and Van Marrewijk (2000) have found that the RCA of industries is not changing significantly over a five year period. Hence, in this analysis, observations of all relevant variables are collected for the years 1998, 2003 and 2008. Because FDI flows from OCED.Stat generally have many missing values, three years of data are collected to have an ultimately reasonable sample size.
The RCA measure has received quite some criticism, because it does not follow a normal distribution (Laursen; 1998). Hence, in this paper the logarithm of the RCA will be used to overcome this shortcoming. This method was suggested by Vollrath (1991).
Table 2: Summary Statistics Explanatory Variable Variable Mean Std. Dev. Min Max Log(RCA) -0.4455707 1.113954 -4.60517 2.899221
Control Variables
The measures of country control variables, which will be used in this analysis, are openness of a country for both inward and outward FDI, home GDP per capita for inward FDI and home GDP growth for outward FDI. The advantage of taking these country controls is that they have been sufficiently tested in prior research.
Table 3: Summary Statistics Control Variables Variable Mean Std. Dev. Min Max
Home based country controls:
Openness 83.79524 53.98875 20.57 324.31 GDP growth 2.19381 2.31104 -5.71 8.43 Log( GDP per capita) 10.08167 0.415777 9.120677 11.08143
Home based industry controls
R&D per industry 4.50477 7.35608 0.01 68.69 Employment share per industry 2.157187 2.668018 0.01 18.42 Log( average wage per
employee per industry) 10.87228 1.520167 6.907796 16.03799 Fixed investment per industry 2.556467 5.206245 -0.6 47.57
(average wage per employee) and investments in the industry (relative fixed investment). The industry controls, except average wage per employee, are a relative measure based on the country total. A table of all controls, their definition, source and calculation is provided in the appendix in Table A4.
Methodology
Firstly, a pooled OLS regression is run in STATA. For all estimations the usual diagnostics will be analyzed. These include the fit of the model, statistical significance and residual analysis. In addition, for all estimations the usual checks for normality, heteroskedasticity, collinearity and outliers are performed.
As has been mentioned the one-year FDI measure will be used as a robustness check. Because three years of data were collected a fixed and random effects estimation will be done for the four year average measures of FDI and the single year observations. All calculations of the pooled OLS the fixed and random effects estimation are repeated for the one-year FDI measure and the significance of the explanatory variable, the sign of the coefficient and the goodness of fit are compared.
A Hausman test, of the fixed and random effects estimations, is done in order to evaluate which model better corresponds to the data. In addition, a Breusch-Pagan test is used to test for the presence of random effects.
Since the hypotheses of this analysis concern the signs of b2 in all four equations,
additional confirmation for the research question can be found if all estimations, show the same sign for one equation. Hence, all results of the explanatory variable for each regression will be compared.
Results
This part of the paper will report the results of the regression analysis of the pooled OLS estimations.
to far from the normal values. A table summarizing the results of the diagnostic checks can be found in the appendix (Table A5).
Table 4: Results Pooled OLS (four year average of FDI flows)
H1a H1b H2a H2b
VARIABLES Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD Log(RCA) 0.307* 0.392** 0.564*** 0.690*** (0.169) (0.184) (0.200) (0.208) Openness -0.00991*** -0.00356 -0.0101** -0.00297 (0.00370) (0.00404) (0.00393) (0.00412) Log(GDP per capita) 0.296 -0.428
(0.449) (0.490) R&D intensity per
industry 0.0395** 0.0374* 0.0417* 0.0183 (0.0176) (0.0192) (0.0222) (0.0227) Employment share per industry 0.226*** 0.00506 0.225** -0.0433 (0.0823) (0.0898) (0.0981) (0.102) Log(average wage
per employee per industry) 0.289*** 0.238** 0.270** 0.366*** (0.0884) (0.0965) (0.109) (0.113) Relative fixed investment per industry 0.372* -0.0808 0.533** 0.144 (0.195) (0.213) (0.231) (0.240) GDP growth -0.0202 -0.178** (0.0748) (0.0783) Constant -10.21** 2.006 -7.437*** -3.587*** (4.882) (5.326) (1.159) (1.205) Observations 209 209 202 200 R-squared 0.290 0.076 0.290 0.142
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
The results of hypothesis 1a are positive and significant at the 10% level. Hence, this confirms: the RCA has a positive effect on the intensity of inward FDI into a country. In other words, being specialized in a certain industry leads to a relatively higher inflow of FDI into that industry.
intensity of a country. The higher the comparative advantage that a country holds in a certain industry the more FDI flows out to the rest of the world from that industry.
In addition, hypothesis 2b is also rejected. The RCA shows a positive and at the 1% level significant sign. This means that the revealed comparative advantage of a country in a certain industry has a positive effect on the relative FDI outflow of that country in this industry as compared to the OECD reference group. Hence, the larger the RCA that a country holds in a certain industry, the higher the outward FDI flow of that industry, as compared to the OECD reference group.
Hypothesis H1a and H1b are both consistently confirmed. The sign of the explanatory variable is positive and significant in both regressions. On the other hand, hypothesis H2a and H2b are both rejected. However, they are both rejected on the same grounds. Namely, the sign of the explanatory variable is positive and highly significant. Thus, even though not all hypotheses can be confirmed, the results are nevertheless consistent in the sign of the explanatory variable for the effect on inward and outward FDI flows.
Furthermore, it should be mentioned that while some of the industry controls, which are significant in the regression estimations based on the FDI intensity per country (H1a and H2a), are not significant in the analysis based on the FDI intensity compared to the OECD reference group (H1b and H2b). This is also reflected in the goodness of fit measure, R², which is significantly lower for the regressions of H1b and H2b. A possible explanation for this might have to do with the variable measurement. Namely, the industry controls, except average wage per employee, are relative measures related to the country total. Hence, while these measures show an effect on the FDI intensity for each country (H1a and H2a), they are less useful in modeling the relationship of FDI intensities, which are calculated in relation to the OECD reference group.
The following table provides an overview of all results from the pooled OLS, fixed and random effects estimation for the four year average and single year calculation (which is used as a robustness check). In this table, only the results of the explanatory variable are reported. Specifically, the sign of the RCA, its significance and if the sign was as expected, are named. The complete results for the pooled OLS, the fixed and random effects estimation of the one year FDI flow calculations are provided in the appendix tables A9, A10 and A11 respectively. In addition, the results of the fixed and random effects for the four year average estimations can be found in table A6 and A7 respectively.
Table 5: Overview Results
Hypothesis
Dependent
Variable Years Pooled OLS Fixed Effects Random Effects
1a
Inward FDI
intensity 4 year average (+)* Y (+) Y (+)* Y
1b
Inward FDI intensity to
OECD 4 year average (+)** Y (+)* Y (+)** Y 2a
Outward
FDI intensity 4 year average (+)*** N (-) Y (+)** N
2b
Outward FDI intensity
to OECD 4 year average (+)*** N (+) N (+)*** N 1a Inward FDI intensity Single (+) Y (-) N (+) Y 1b Inward FDI intensity to OECD Single (+)** Y (-) N (+)** Y 2a Outward
FDI intensity Single (+) N (-) Y (+) N
2b
Outward FDI intensity
to OECD Single (+)*** N (-) Y (+)*** N *** p<0.01, ** p<0.05, * p<0.1
Table 5 shows an overview of the results of all estimations. The sign of the relationship of the explanatory variable is given in parenthesis and the significance is indicated by stars. If the sign of the RCA coefficient was found as expected is shown by a Y or N, in case of yes or no respectively.
results are preferred. The outcome of the Hausman and Breusch-Pagan test can be found in the appendix in Table A8 for the four year average measure of FDI.
As can be seen from the table above, the results of the pooled OLS regressions and the random effects estimations are similar in terms of the significance and the sign of the explanatory variable. As has already been mentioned above, for the four year average FDI measurement, the OLS estimation results are preferred, except for the regressions of hypothesis H1b and H2a. When the hypothesis of the fixed and random effects coefficient equality could not be rejected and no random effects could prove to be present, the OLS results are preferred. The same tests have been repeated for the single year measurements of the dependent variable. The results of the Hausman and Breusch-Pagan test of the single year estimations are reported in Table A12 in the appendix.
Even though the estimations of the OLS and random effects analysis show similar signs and significance for each of the run regressions, the results of the fixed effects analysis raise some questions. Only one of the results is significant and for the single year analysis the explanatory variable only shows negative coefficients. In addition, the sign of the explanatory variable, though not significant, for the regression of hypothesis H2a also shows a negative sign. On the one hand, does a negative coefficient of H2a and H2b confirm the proposed hypotheses, but on the other does it stand in contrast to the results from the pooled OLS and random effects analysis. However, the fixed effects estimations do not fit the data well for all run regressions, except H2a and H2b of the single year FDI measurements and H1b of the four year average FDI measurement.
Hence, while the results for H1a and H1b uniformly show the expected positive sign in the single and four year average estimation for the results which seem to have the most validation and are partially significant, the results for H2a and H2b are more conflicting. Namely, on the one hand the preferred estimation of H2a and H2b for the four year average estimation show both positive and significant signs for the explanatory variable. On the other hand, the estimation results for H2a and H2b of the single year estimations, which fit the data best, according to the Hausman test, show both negative, though not significant, signs.
time. In order to overcome this shortcoming somewhat data with five year gaps has been collected. However, as it seems from the data, even data with five year gaps does not provide enough within-cluster variation to work well for the fixed effects model. As a random example, see the graph below.
Graph 3: Revealed Comparative Advantage over Time
Source: OECD.Stat
Conclusion
The analysis of this paper has been directed at the industry determinants of FDI flows. Specifically, the impact of the revealed comparative advantage on FDI inflows and outflows in relation to a country’s total FDI flows and to the OECD reference group have been investigated.
was found, the negative results of the fixed effect estimations in the single year measure still make it difficult to come to a definite conclusion for H2a and H2b.
A possible explanation for a positive effect of the RCA on outward FDI flows could be that the sources of the comparative advantage are transferable. Namely, if a company is able to transfer the advantage it has over companies in the same industry from other countries, it could take the advantage to another country and produce there at least as efficient as in the home country. The phenomenon of the mobility of factors of production has already been quite a discussed topic in the existing literature. However, as mentioned above, other factors than just the traditional factors of production cause the underlying advantage an industry holds in its home country. For example, underlying advantages such as technological superiority or entrepreneurial skill might be transferred to another country. Hence, if a company is able to transfer the source of the comparative advantage from its industry to another country, it might be more prone to do so. This is because, the company might then not only benefit from its home source of the comparative advantage, but also gain access to foreign country’s industry advantage. Following, one could guess that companies from industries with a higher comparative advantage which is more transferable to other countries will invest more into outward FDI. However, this is only an assumption. More research is required to identify what causes the comparative advantage in each industry and then determine if this is a transferable source of advantage and how it influences outward FDI flows.
Limitations and Suggestions for Future Research
In addition, the model could most probably be improved if the industry control variables in the estimations in which the dependent variable is compared to the OECD, could also be based on a comparison measure with the OECD reference group and not only the country total FDI flows.
Furthermore, this research has been largely intended as a preliminary and introductory analysis for industry FDI determinants and specifically the effect of the comparative advantage on FDI flows. Future research should hence go on to include more detail in the analysis. Specifically, the sources of the comparative advantage of each industry should be identified and distinguished and it should then be tested what effect they have on the FDI flows. This could shed more light on the contradicting results which were found for FDI outflows.
References
Balassa, B. (1965), Trade Liberalisation and Revealed Comparative Advantage, The Manchester School, 33.
Bobonis, G.J., Schatz, H.J. (2006), Agglomeration, Adjustment, and State Policies in the
Location of Foreign Direct Investment in the United States, The Review of
Economics and Statistics, 89(1).
Braconier, H., Norbäck, P.J., Urban, D. (2005), Multinational Enterprises and Wage
Costs: Vertical FDI Revisited, Journal of International Economics, 67.
Bronzini, R. (2004), Foreign Direct Investment and Agglomeration: Evidence from Italy, Bank of Italy, Economic Research Department.
Campa, J., Donnenfeld, S., Weber, S. (1998), Market Structure and Foreign Direct
Investment, Review of International Economics, 6(3).
Carr, D., Markusen, J.R., Maskus, K.E. (2001), Estimating the Knowledge–Capital
Model of the Multinational Enterprise, American Economic Review, 91(3).
Chakrabarti, A (2001), The Determinants of Foreign Direct Investment: Sensitivity
Analyses of Cross-Country Regressions, Kyklos, 54.
Costinot, A. (2009), On the Origins of Comparative Advantage, Journal of International Economics, 77(2).
Choong, C.K., Liew, V.K.S. (2009), Impact of Foreign Direct Investment Volatility on
Economic Growth of ASEAN-5 Countries, Economics Bulletin, 29(3).
Chowdhury, A.R., Wheeler, M. (2008), Does Real Exchange Rate Volatility Affect
Foreign Direct Investment? Evidence from Four Developed Economies, International
Trade Journal, 22(2)
Culem, C.G. (1988), The Locational Determinants of Direct Foreign Investments among
Industrialized Countries, European Economic Review, 32.
De Ferranti, D., Guillermo, P.E., Lederman, D., Maloney, W.E. (2002), From Natural
Resources to the Knowledge Economy: Trade and Job Quality, The World Bank,
Working Paper, 23440.
Drake, T.A., Caves, R.E. (1992), Changing Determinants of Japanese Direct Investment
Dunning, J.H. (1977), Trade, Location of Economic Activity and the MNE: A Search for
and Eclectic Approach. In Ohlin, B., Hesselborn, P.O., Wijkman, P.M. (eds), The International Allocation of Economic Activity, London: Holmes and Meier.
Dunning, J.H. (1979), Explaining the Pattern of International Production: In Defiance of
Eclectic Theory, Oxford Bulletin of Economics and Statistics, 41.
Dunning, J.H. (1988), Explaining International Production, Allen and Unwin.
Fernandez-Arias, E., Hausmann, R. (2000), Is FDI a Safer Form of Financing, Inter-American Development Bank Research Department Working Paper, 416.
Feath, I. (2009), Determinants of Foreign Direct Investment- A Tale of Nine Theoretical
Models, Journal of Economic Surveys 23(1).
Gourdon, J. (2009), Explaining Trade Flows: Traditional and New Determinants of
Trade Patterns, Journal of Economic Integration, 24(1).
Grossman, G.M., Helpman, E. (1990), Comparative Advantage and Long-Run Growth, The American Economic Review, 80(4).
Hasnat, B. (2003), Labor Standards and Foreign Direct Investment, paper presented at the Eastern Economic Association Annual Meeting.
Head, K., Ries, J. (2008), FDI as an Outcome of the Market for Corporate Control:
Theory and Evidence, Journal of International Economics, 74(1).
Hinloopen, J., Van Marrewijk, C. (2000), On the Empirical Distribution of the Balassa
Index, Weltwirtschaftliches Archiv, 137(1).
Husted, S., Melvin, M. (2004), International Economics 6ed, Pearson Addison-Wesley. Hymer, S.H. (1976) The International Operations of National Firms: A Study of Direct
Investment, Cambridge, MA: MIT Press.
Kindleberger, C.P. (1969) American Business Abroad: Six Lectures on Foreign Direct
Investment, New Haven, CT: Yale University Press.
Kim, S.H., Pickton, T.S., Gerkin, S. (2003), Foreign Direct Investment: Agglomeration
Economies and Returns to Promotion Expenditures, The Review of Regional
Studies, 33(1).
Krugman, P.R. (1983), The ‘New Theories’ of International Trade and the Multinational
Krugman, P.R., (1991), Increasing Returns and Economic Geography, Journal of Political Economy, 99.
Laursen, K. (1998), Revealed Comparative Advantage and the Alternatives as Measures
of International Specialization, DRUID Working Paper, 98-30.
Lensink, R., Morrissey, O. (2006), Foreign Direct Investment: Flows, Volatility, and the
Impact on Growth, Review of International Economics, 14(3).
Markusen, J.R. (1984), Multinationals, Multi-Plant Economies, and the Gains from
Trade, Journal of International Economics, 16.
Markusen, J.R. (1997), Trade versus Investment Liberalisation, NBER Working Paper 6231, Cambridge, MA: National Bureau of Economic Research.
Markusen, J.R. (2002), Multinational Firms and the Theory of International Trade, Cambridge and London: MIT Press.
Markusen, J. R., Maskus, K.F. (2002), General-equilibrium Approaches to the
Multinational Firm: A Review of Theory and Evidence, Discussion Paper No
2002-15, Copenhagen: Center for Economic and Business Research.
Maskus, K., Webster, A. (1995), Comparative Advantage and the Location of Inward
Foreign Direct Investment: Evidence from the UK and South Korea, World
Economy, 18(2).
Miroudot, S., Ragoussis, A. (2009), Vertical Trade, Trace Costs and FDI, OECD Trade Policy Working Papers, 89.
Morrow, P.M. (2010), Ricardian-Heckscher-Ohlin Comparative Advantage: Theory and Evidence, Journal of International Economics, 82(2).
Newmayer, E., Spess, L. (2005), Do Bilateral Investment Treaties Increase FDI to
Developing Countries, World Development, 33(10).
Nicholson, W., Snyder, C., Luke, P., Wood, M. (2008), Intermediate Microeconomics, Cengage Learning EMEA.
Palangkaraya, A., Waldkirch, A. (2008), Relative Factor Abundance and FDI Factor
Intensity in Developed Countries, International Economic Journal, 22(4).
Qiu, L.D. (2003), Comparing Sectoral FDI Incentives: Comparative Advantage and
Vollrath, T.L. (1991), A Theoretical evaluation of Alternative Trade Intensity Measures
Appendix
Table A1: Overview Cross-Country FDI Determinants
Authors Y X Control Variables Methods Period/
Countries/ Industries Carr, Markusen, Maskus (2001)
Affiliate Sales GDP sum (+), GDP difference (-), Skill
difference (+), Interaction of skill and GDP difference (-), Host trade costs (+), Parent trade costs (+), Host trade costs times skill difference (-), Host FDI cost (-)
Distance (o), OLS, WLS, Tobit 1986-1994, US bilateral data for 36 countries Culem (1988)
FDI flow/GNP Lagged host country GNP (+), Growth rate host GNP (+), GNP growth rate differential home and host (+), Tariff rate (-), Host unit labor cost (-), Unit labor cost differential home and host (-), Exports home to host/ Lagged GNP home (+)
Nominal interest rate differential (+) GLS 1969-82, Germany, France, UK, Netherlands, Belgium Campa, Donnenfeld, Weber (1998)
FDI/Imports Tariffs (+),Concentration ratio*Tariff (-)
Concentration ratio (-),Plant size (+), Fixed investment, Bilateral exchange rates (-), Non-tariff barriers (+) Tobit 1984-91, 219 SIC industries, 19 countries Newmayer, Spess (2005)
Log FDI flows Number of bilateral investment treaties(+)
Log GDP per capita (+), Log population (+), GDP growth (+), Inflation (-), Income natural resources/GDP (+), Political constraint index (o), Composite political risk (-), Sub-indices (-) Panel, Fixed Effects, Random Effects 1970-2001, 120 countries Hasnat (2003)
Log FDI flows Labor standards- number of conventions (o) Log GDP (+), GDP growth (+), Log openness (+) OLS 1995-1999, 142 countries Chakrabarti (2001)
Net FDI per capita flows
Host per capita GDP (+), Host GDP (+), Host GDP growth (o), Relative labor-capital endowments (o), Home trade cost (o), Host trade cost (o), Bilateral trade flows (o), Home taxes (o),
Host inflation (o), Host political stability (o) Extreme bound analysis (EBA), OLS 1994, 135 countries Fernandez-Arias, Hausmann (2000) Log (FDI/ GDP+1)
GDP volatility (o), Suboil resources (o), Distance (+), Credit/GDP (o), 5 Kaufman indices (+)
Table A2: FDI Industry Data – Industry Classification (ISIC Rev. 3)
Sector Industry ISIC Rev. 3
Primary
1 AGRICULTURE AND FISHING A+B
2 MINING AND QUARRYING C
Manufacturing
3 Food products D15
4 Textiles and wearing apparel D17+18
5 Wood,publishing and printing D22
6 Refined petroleum & other treatments D23
7 Chemical products D24
8 Rubber and plastic products D25
9 Metal products D27
10 Mechanical products D29
11 Office machinery and computers D30
12 Radio,TV,communication equipments D32
13 Medical, precision and optical instruments, watches and clocks D33
14 Motor vehicles D34
15 Other transport equipments D35
16 ELECTRICITY,GAS AND WATER E
Services
17 CONSTRUCTION F
18
Sale, maintenance and repair of motor vehicles and motorcycles - retail sale of
automotive fuel G50
19 Wholesale, trade and commission excl. motor vehicles G51 20 Retail trade excl. motor vehicles - repair of household goods G52
21 Hotels and restaurants H55
22 Land transport - transport via pipelines I60
23 Water transport I61
24 Air transport I62
25 Supporting and auxiliary transport activities I63
26 Post and telecommunications I64
27 Financial intermediation, except insurance and pension funding J65 28 Insurance and pension funding, except compulsory social security J66 29 Activities auxiliary to financial intermediation J67
30 Real estate activities K70
31 Renting of machinery and equipment K71
32 Computer and related activities K72
33 Research and development K73
34 Other business activities K74
Table A3: List of Countries
Country Missing Years
1 Australia
3 Belgium 1996-2001 4 Canada 5 Czech Republic 6 Denmark 7 Finland 8 France 9 Germany 10 Greece 1996-2000 11 Hungary 1996-1998, 2000 12 Iceland 13 Ireland 1998-2002 14 Italy 15 Japan 16 Korea 17 Luxembourg 1996-2004 18 Mexico 19 Netherlands 20 Norway 21 Poland 22 Slovak Republic 1996-1999 23 Spain 24 Sweden 25 Switzerland 26 Turkey 1996-1998 27 United Kingdom 2009 28 United States
Table A4: Overview Variable Definition and Calculation
Name Definition Industry/
Country measure Calculation Source Inward or Ouward FDI flow
An investment in a foreign company where the investor owns at least 10% of the ordinary shares, undertaken with the objective of establishing a lasting interest in the country, a long-term relationship and significant influence on the management of the firm.
Industry OECD. Stat
Log(RCA) The revealed comparative advantage indicator shows a country's (or country group's) exports for an industry relative to total industries' exports, divided by OECD exports of the same industry relative to OECD total industries' exports.
Industry Log [ (expo c,i / expo c,total) / (expo OECD,i / expo OECD,total) ] OECD.Stat: STAN
Openness Openness at 2005 constant prices (%) Country Exports plus Imports divided by real GDP per capita.
Penn World Table Log(GDP per
capita)
Per head, US $, constant prices, constant PPPs, OECD base year
Country OECD.Stat
R&D intensity per industry
This indicator shows the R&D expenditures for an industry as a percentage of R&D expenditures for the total economy
Industry 100 * ( anberd c,i / anberd c,total ) OECD.Stat: STAN Employment share per industry
This indicator addresses the issue of employment structure and shows for each industry the total employment (i.e. total number of persons engaged) share in the total economy.
Industry 100 * ( empn c,i / empn c,total ) OECD.Stat: STAN Log(average wage per employee per industry)
Wages and salaries per industry divided by total employment per industry
Industry Log(total Wages per industry/ total employment per industry) OECD.Stat: STAN Relative fixed investment per industry
This indicator represents the investment composition of the total economy. It is calculated by dividing each industry's gross fixed capital formation by gross fixed capital formation for total economy.
Industry 100 * ( gfcf c,i / gfcf c,total )
OECD.Stat: STAN
Table A5: Diagnostic Test Outcome: pooled OLS
H1a H1b H2a H2b H1a H1b H2a H2b
4 year average 4 year average 4 year average 4 year average single year measure single year measure single year measure single year measure Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD Goodness of Fit 0.2899 0.0765 0.29 0.1422 0.2297 0.1063 0.1680 0.1466 Normality:
Jarque-Berap-value 1.254e-16 .17828149 7.881e-24 .04630015 4.743e-06 .00843323 33.658e-10 .04630015 Normality: Skewness Residuals .6612335 -.1315503 .9479742 .3901207 .4306519 -.5652904 .7334161 .2516032 Normality: Kurtosis Residuals 5.580765 3.571674 6.007722 3.358678 4.578374 3.293662 5.041945 2.817079 Heterostkedasticity 0.4623 0.7809 0.4825 0.4509 0.1296 0.3079 1 0.6022 Outliers - - - - Collinearity: dependent variable- log(RCA) 0.1249 0.1364 0.1675 0.244 0.0169 0.1608 0.1065 0.2324
Table A6: Fixed Effects Estimations (four year average of FDI flows)
H1a H1b H2a H2b
Openness -0.0696 -0.0494 -0.0434 -0.0241 (0.0456) (0.0469) (0.0318) (0.0419) Log(GDP per capita) 23.09** 33.98***
(10.76) (11.07) R&D intensity per
industry 0.0153 0.0221 -0.217 0.247 (0.240) (0.247) (0.227) (0.377) Employment share per industry 0.324 0.465 0.648 0.579 (0.755) (0.777) (0.931) (1.233) Log(average wage per employee per industry) -9.003 -8.387 -3.706 3.110 (5.582) (5.745) (2.929) (3.888) Relative fixed investment per industry 1.867 0.460 0.495 0.0182 (1.545) (1.590) (0.956) (1.377) GDP growth -0.195 -0.495** (0.142) (0.192) Constant -135.3* -248.3*** 39.02 -32.79 (75.30) (77.49) (31.22) (41.56) Observations 209 209 202 200 R-squared 0.244 0.321 0.228 0.318
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table A7: Results Random Effects Estimation (four year average of FDI flows)
H1a H1b H2a H2b
VARIABLES Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD LogRCA) 0.307* 0.392** 0.482** 0.664*** (0.169) (0.184) (0.217) (0.219) Openness -0.00991*** -0.00356 -0.0119*** -0.00248 (0.00370) (0.00404) (0.00439) (0.00439) Log(GDP per capita) 0.296 -0.428
(0.449) (0.490) R&D intensity per
Observations 209 209 202 200 Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table A8: Random vs. Fixed Effects Estimations (four year average of FDI flows)
H1a H1b H2a H2b Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD log(RCA) fixed effects
log(RCA) random effects * ** *** ***
Hausman Test: p-value 0.1234 0.0041 0.1266 0.0911 Breusch-Pagan: p-value 0.4921 0.9903 0.0089 0.5271
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table A9: Pooled OLS (single year FDI flows)
H1a H1b H2a H2b
VARIABLES Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD LogRCA) 0.208 0.629** 0.433 1.073*** (0.216) (0.268) (0.311) (0.355) Openness -0.00586 0.000600 -0.0135** -0.00247 (0.00476) (0.00548) (0.00537) (0.00576) Log(GDP per capita) 0.154 -0.552
(0.572) (0.674) R&D intensity per
industry 0.0681*** -0.00552 0.0445 0.0265 (0.0229) (0.0269) (0.0282) (0.0285) Employment share per industry 0.269** -0.0281 0.232* -0.0708 (0.108) (0.130) (0.128) (0.134) Log(average wage per employee per industry) 0.439*** 0.443*** 0.151 0.444** (0.116) (0.139) (0.149) (0.176) Relative fixed investment per industry 0.146 0.0580 0.316 0.187 (0.252) (0.310) (0.311) (0.318) GDP growth -0.0155 -0.302** (0.109) (0.120) Constant -10.35* 0.961 -5.307*** -4.128** (6.241) (7.377) (1.547) (1.825) Observations 182 168 165 150 R-squared 0.230 0.106 0.168 0.147
Table A10: Fixed Effects Estimations (single year FDI flows)
H1a H1b H2a H2b
VARIABLES Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD LogRCA) -0.500 -1.203 -2.758 -1.356 (2.149) (3.921) (2.257) (2.154) Openness -0.118** -0.102 -0.377*** -0.516*** (0.0522) (0.0857) (0.0845) (0.109) Log(GDP per capita) 34.92*** 57.80**
(12.50) (21.58) R&D intensity per
industry 0.102 -0.558 -0.332 -0.126 (0.325) (0.519) (0.371) (0.382) Employment share per industry -0.388 -0.254 -0.260 -0.704 (1.038) (1.593) (1.329) (1.430) Log(average wage per employee per industry) -8.603 -16.40 -25.59*** -13.87 (6.046) (11.30) (7.118) (8.199) Relative fixed investment per industry 3.827** 1.593 1.232 4.625** (1.471) (2.259) (2.114) (2.112) GDP growth -2.908*** -3.989*** (0.626) (0.772) Constant -257.2*** -396.0** 310.8*** 196.0** (92.51) (143.8) (82.04) (93.03) Observations 182 168 165 150 R-squared 0.292 0.289 0.547 0.680 Number of number 140 133 135 121
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table A11: Random Effects Estimation (single year FDI flows)
H1a H1b H2a H2b
VARIABLES Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD LogRCA) 0.195 0.629** 0.373 1.130*** (0.223) (0.268) (0.338) (0.395) Openness -0.00603 0.000600 -0.0153*** -0.00333 (0.00492) (0.00548) (0.00593) (0.00634) Log(GDP per capita) 0.107 -0.552
(0.585) (0.674) R&D intensity per
industry 0.0662*** -0.00552 0.0435 0.0243 (0.0237) (0.0269) (0.0305) (0.0318) Employment share per industry 0.241** -0.0281 0.228 -0.0737 (0.112) (0.130) (0.147) (0.153) Log(average wage per employee per
industry) (0.119) (0.139) (0.160) (0.189) Relative fixed investment per industry 0.195 0.0580 0.378 0.221 (0.260) (0.310) (0.339) (0.351) GDP growth 0.0175 -0.377*** (0.107) (0.120) Constant -9.546 0.961 -4.588*** -3.725* (6.377) (7.377) (1.670) (1.973) Observations 182 168 165 150 Number of number 140 133 135 121
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table A12: Random vs. Fixed Effects Estimations (single year FDI flows)
H1a H1b H2a H2b Inward FDI intensity Inward FDI intensity to OECD Outward FDI intensity Outward FDI intensity to OECD log(RCA) fixed effects
log(RCA) random effects ** ***
Hausman Test: p-value 0.0612 0.1082 0.0003 0.0000
