Foreign direct investment affecting economic growth : developed countries vs. developing countries

FOREIGN DIRECT INVESTMENT AFFECTING

ECONOMIC GROWTH

Developed countries vs. developing countries

Bachelor thesis

By: Britt Dams

Student number: 10214364

Supervisor: Ron van Maurik

Date: 23-07-2014

Faculty of Economics and Business

Track: Economics and Finance

Field: Macroeconomics

Table of contents

Introduction

p. 3

Literature review

p. 5

What is foreign direct investment?

p. 5

- Definition

- FDI in general

- FDI in Latin America

Transmission channels of FDI

p. 7

- Human capital

- Trade

- Market development

Control variables

p. 12

Hypotheses

p. 13

Methodology

p. 13

Data

p. 13

- Developed countries – OECD

- Developing countries

Dependent variable

p. 14

- GDP per capita

Independent variables

p. 15

- Foreign direct investment

- Human Capital

- Trade

- Market development

Control variables

p. 16

Model

p. 17

Problems

p. 19

Results

p. 20

Conclusion

p. 26

References

p. 28

Appendix

p. 30

A - Countries included in the sample  

B - Variable descriptions

C -Tables

Introduction

There has been increasing interest in foreign direct investment (FDI) and what part it can play in the economic development of countries. This can be seen in the increasing levels of FDI in proportion to total capital flows. Alfaro et al. (2004) state that: “In 1998, FDI accounted for more than half of all private capital flows to developing countries” (p.90).

FDI doesn’t only provide direct capital financing to the host country, but many other externalities. According to UNCTAD (2000): “it facilitates the transfer of technology, organizational and managerial practices and skills as well as access to international markets” (p. 3).

These benefits have activated countries and its governments to increasing efforts in order to attract more FDI (Alfaro, Chanda, Kalemli-Ozcan & Sayek, 2004, p. 90).

Efforts made by countries include for example liberalization of FDI regimes and reduction of restrictions on FDI inflows (UNCTAD, 2000, p.3).

Many papers have studied the effect of FDI on economic growth in developing countries. Especially the amount of literature on FDI in China has increased a lot (Zhang, 2001, p. 680). Therefore I have decided to put the focus on developing countries in Latin America.

Throughout the global financial crises of 1997 to 1998, the level of FDI remained roughly the same in East Asian countries, whereas other types of private capital flows fluctuated. This was also the case for FDI in times of the Mexican crisis of 1994 to 1995 and the Latin American debt crisis of the 1980s (Loungani and Razin, 2001). According to Loungani and Razin: ‘This resilience could lead many developing countries to favour FDI over other forms of capital flows, furthering a trend that has been in evidence for many years.’

Just recently (29 May 2014), the Economic Commission for Latin America and the Caribbean (ECLAC) presented its report ‘Foreign Direct Investment in Latin America and the Caribbean 2013’. It stated that in 2013 Latin America and the Caribbean received 184.92 billion dollars in foreign direct investment (FDI), which is a 5% rise compared to 2012 in nominal terms (ECLAC, 2013). The part of FDI flows of Latin America and the Caribbean compared to the rest of the world remained at 13% in 2013. However, Global FDI flows have risen by 11% in 2013, according to the ‘Foreign Direct Investment in Latin America and the Caribbean 2013’ report. This shows the importance and the topicality on the subject of FDI.

Also, the existing literature focuses on the effects of FDI on economic growth in developing countries alone. This makes it interesting to compare the effect of FDI on economic growth in developed countries with the effect of FDI on economic growth in several Latin American countries. Comparing developing countries with developed countries will add to the existing literature.

This leads to the following research question:

“Is there a difference in the way FDI affects economic growth in developed vs. developing countries?”

To answer my research question I have done a literature study on the subject. It provides theory about the effect of FDI on economic growth and about its transmission channels. I also did empirical research in order to answer the research question. The empirical analysis is for a large part based on the research of Borensztein, De Gregorio & Lee (1998), but elements of the work form other researchers were also used. The dataset and estimation technique are different. A paneldataset of 34 developed countries (OECD members) and a paneldataset of 18 developing countries (Latin America) were used. The results from the fixed effect regressions are in contrast to the existing literature on transmission channels of FDI. My research first consists of a literature review, in which the term FDI is more clearly defined. Also an overview of the existing literature is given. Secondly, I discuss the hypotheses that were formed based on the existing literature. In the third section the

methodology of my empirical analysis is discussed. It describes the dataset, variables, model, estimation procedure and possible problems. The fourth section contains the results of my empirical analysis. In the last section a conclusion is given.

Literature review

This section covers the literature study. FDI will be explained in more detail and the effects of FDI on economic growth will be specified.

What is foreign direct investment?

Definition

Foreign direct investment is a type of investment where a resident entity of one economy engages in a cross-border investment in another economy. The intent of FDI is to start a lasting interest by a direct investor in a foreign entity, with the purpose to obtain a large degree of influence for the direct investor on the management of the enterprise. Usually, at least 10% of the voting power is used as a criterion (UNCTAD, 1999; “OECD Factbook 2013”, 2013).

An investing firm setting up a subsidiary or associate firm, mergers, joint ventures and acquisitions are examples of FDI (Investopedia, n.d.).

In recent years, the recognition of FDI as a factor for economic development has increased (UNCTAD, 2000, p.3). According to UNCTAD (1996), FDI brings in capital, technology, human capital and access to international markets. It is therefore seen as an instrument that enables economies to integrate into the world economy (as cited in UNCTAD, 1999, p.2).An increasing number of countries are creating an FDI attractive environment. As reported by UNCTAD (2000): “In addition to reducing restrictions on the entry of FDI, they are actively liberalizing their FDI regimes (p.3).

FDI  -­  in  general  

FDI is believed to bring positive effects to the host country, which leads to increased efforts from governments to attract more FDI. These are externalities like increased productivity, know-how, access to foreign markets and technology. Thus the benefits provided by FDI could mean increasing economic growth and modernization (Alfaro et al., 2004, p. 90). In particular for developing countries, receiving FDI inflows from developed countries can potentially be of great benefit. Wang (1990) speaks of other advantages in his article besides the well-known effects on income level and employment. For instance, the arrival of FDI promotes domestic technological change, increasing the rate of income growth (p.269). When a country allows foreign capital inflows and works on its domestic human capital

accumulations as well as technology adoptive efficiency of local firms, this might cause the GDP per capita gap between developing and developed countries to become smaller (Wang, 1990, p.269)

In general, most of the developed countries have a low volume of human capital, leaving them not capable enough to take advantage of technical progress (Balasubramanyam, Salisu & Sapsford, 1996, p. 95). The lack of human capital prevents developing countries from engaging in Research & Development investments and therefore in acquiring new knowledge. Balasubramanyam et al. (1996) call this a ‘gap in skills’ that can partly be overcome through FDI, due to its spill-over effects.

The general view on FDI is that it is an important channel for developing countries to access advanced technologies (Borensztein et al., 1998, p. 116). The increase in technologies will raise output. However, studies on this subject imply that the effect of FDI is not that

straightforward. As Durham (2004) describes, to what extent FDI enhances growth, depends on several other factors, commonly referred to as the absorptive capacity of the host country (p. 287).

On the other side, FDI can have a negative impact under some conditions. For example if there is too much of it in short time, small host economies might not have enough absorptive capacity (UNCTAD, 1999, p.2). For instance the level of education and the level of

development are insufficient.This can lead to the appreciation of the exchange rate, negatively affecting export and import substitution.

To improve the effect of FDI, the right policy is needed that strives for a better domestic productive and technological foundation; so competitive advantages can be fully exploited (UNCTAD, 1999, p.2).

FDI in Latin America

According to the report of the UNCTAD (1999): Argentina, Mexico, Brazil (since 1994), Chile, Peru (also since 1994) and Colombia have been receiving the largest inflows of FDI. But also smaller economies like Bolivia and Paraguay have had increasing inflows of FDI. The increase in inflows can be explained by the improved macroeconomic conditions

(UNCTAD, 1999, p.12). For example, the inflation in Argentina and later Brazil, has over all, become more stable. Thereby, increased privatization has made FDI more attractive.

In spite of the economic turmoil of 2008, inflows of FDI in Latin America rose to a new height, even though FDI growth was smaller than the year before. As surprising as these

results might seem, this was probably caused by pe-crisis market trends (ECLAC, 2008, p.1). Indeed, in 2009 the effects of the crisis became more visible; globally, FDI was sharply reduced. A year beforehand only FDI in developed countries was affected, in 2009 this was also the case for developing countries (ECLAC, 2009, p.10).

Transmission channels of FDI

Human Capital

Technological development plays an increasing role in growth rates in the literature. Borensztein et al. (1998) also believe in more recent economic growth theories, where technology diffusion is taken into account as an explanatory factor.

Growth rates in developing countries are subject to the ‘catch-up’ effect, meaning that it is easier and cheaper to imitate existing products, rather than creating new products. (p.119). How well a developing country can adapt and implement new technologies, is of impact on the growth rate (p.116). However, the level of human capital in a country needs to be high enough, to provide sufficient absorptive capacity in order to make use of these technologies (p.117).

Borensztein et al. (1998) focus on the interaction of FDI with human capital to affect the economic growth rate (p. 123).

They use the following formula for their empirical analysis:

FDI stands for foreign direct investment as a ratio of GDP. H indicates human capital, measured in years of the male secondary schooling, taken from Barro and Lee (1993). The study of Barro and Lee (1994) showed that this measure is the one most significantly correlated with growth (Borensztein et al., 1998, p.123).

Y0 is the initial GDP per capita and is used as a proxy for the ‘catch-up’ effect. Borensztein et

al. (1998) assume the existence of a ‘catch-up’ effect’, “ to reflect the fact that it is cheaper to imitate products already in existence for some time than to create new products at the frontier of innovation” (p. 119). A is a set of control variables that affect economic growth, also taken from Barro and Lee (1994).

For the regression, the authors use panel data for 1970 to 1989 from 69 developing countries. The Seemingly Unrelated Regressions technique (SUR) is applied. The interaction term between FDI and human capital is necessary to test in what way these variables affect economic growth; on itself or through interaction (Borensztein et al., 1998, p. 125). The regression results in a negative and non-significant estimate for FDI, but the interaction term is positive. Moreover, a threshold of 0.52 is found for secondary school attainment. Above this level, a country will benefit form FDI.

Borenszetein et al (1998) conclude from their results that FDI is an: “important vehicle for the transfer of technology” (p. 117). They find that the level of human capital enhances the effect of FDI on economic growth, but only when a certain level is met.

In contrast to the findings of Borensztein et al. (1998), Durham (2004) (who uses a sample of 62 non-OECD and 21 high-income countries and the Ordinary Least Squares regression technique) finds his estimates of FDI and human capital to be non-significant (p. 296).

The interaction effect of FDI and human capital on economic growth is also investigated by Kottaridi and Stengos (2010). They do so by taking a sample in which they split OECD and non-OECD countries. The sample was also split in high-income countries from middle-income countries and low-middle-income countries, according to the World Bank classification (Kottaridi and Stengos, 2010, p.862). From the Least Squares Dummy Variable (LSDV) regression, it turns out that for the sample as a whole, FDI estimates itself are positive, but

insignificant. To the surprise of the researchers, FDI estimates for OECD countries are negative and insignificant and high-income countries have positive and insignificant

estimates. However, positive and significant estimated were found for non-OECD countries as well as for middle-income countries (Kottaridi and Stengos, 2010, p.862). When looking at the estimates for human capital alone, they are positive and highly significant for OECD countries and high-income countries and significant for low-income countries. After including an interaction between FDI and human capital in the estimations, Kottaridi and Stengos (2010) find increased significance for human capital for the whole sample, but the interaction term is negative and insignificant (p.862). Considering the OECD countries, FDI is negative and highly significant, whereas human capital and the interaction term are positive and significant.

Apart from the human capital estimate that is negative and not significant, the opposite is true for non-OECD countries. For middle-income countries, FDI is highly significant and positive, though the estimates for human capital and the interaction term are not significant.

Unexpectedly, FDI and human capital are highly significant and the interaction term is negative and highly significant for low-income countries (Kottaridi and Stengos, 2010, pp. 862-863).

Trade

Bhaghwati (1978) hypothesised that the effects of FDI will be different depending on whether a country has an export promoting (EP), or import substituting (IS) strategy (as cited by Balasubramanyam et al., 1996, p.92). In their research, Balasubramanyam et al. (1996) build on the work of Bhagwati (1978). The EP strategy equates the average effective exchange rate on exports to the average effective exchange rate on imports, and is therefore trade neutral. The IS strategy is different in a way that the effective exchange rate on imports exceeds the effective exchange rate on exports (Balasubramanyam et al., 1996, pp.92-93).

Given its trade neutral characteristic, the EP strategy “allows for a free play of market forces and the allocation of resources on the basis of comparative advantage” (Balasubramanyam et al., 1996, p.94).  This type of resource allocation, supports specialization and economies of scale. Since competition between international trade and domestic sources is allowed for, this may lead to investments in technology and human capital by foreign- and locally owned firms (Balasubramanyam et al., 1996, p.96).

All of the above factors contribute to an ideal climate in which FDI can potentially be used to promote economic growth.

An IS policy on the contrary restricts competition and will probably not lead to such investments as stated above. Furthermore, an IS policy is inefficient because of its

dependency on tariffs and quotas on trade. Also, it induces rent seeking and underproductive profit seeking activities (Balasubramanyam, 1996, pp.93-96).

Balasubramanyam et al. (1996) test a part of Bhagwati’s hypothesis: “Other things equal, the partial derivative of the rate of economic growth with respect to FDI is higher in countries following an EP strategy than in countries following an IS one” (p.97). They expect that in EP countries, FDI enhances growth more than domestic investment does. Concerning FDI, a higher rate of technical innovation, spill-over effects and externalities related to human capital form their rationale behind this expectation (Balasubramanyam et al., 1996, p. 98).

For their empirical analysis, Balasubramanyam et al. (1996) classify countries as either EP or IS according to the ratio of their imports to GDP seen as a proxy for the complexion of trade policy (p.100).

Relatively high imports to GDP ratio points out to a low level of import protection, so countries with such a high ratio can be seen as EP countries. The authors find positive and significant results indicating FDI enhanced economic growth is higher in EP countries (Balasubramanyam et al., 1996, p.101).

As well as many other researchers, Durham et al. (2004), also include trade to GDP as an absorptive capacity variable in their OLS regression (p. 292). Other than the findings of Balasubramanyam et al. (1996), they find their estimates to be insignificant (p.296).

Openness to international trade is used by Alfaro et al. (2004) too. It is measured as the ratio of the sum of exports plus imports to total GDP (p.96).

Market development

“Holding constant the level of FDI, countries with higher legal standards likely channel foreign investment more efficiently” (p.288) is a statement made by Durham (2004). He has the opinion that in the existing literature, the existing level of domestic financial development is being ignored. The researcher believes that ‘deeper’ financial systems might absorb FDI in a more effective manner. Thus the capacity of domestic equity markets to effectively absorb FDI would apparently have positive relation with the development level of the domestic market (Durham, 2004, pp. 288-289).The proxy for market development used in their research is total stock market capitalisation relative to GDP. Durham (2004) claims that models should control for the initial level of financial development, and should include interaction terms between market development and FDI. For that reason, they use the

following regression equation:

Y is growth; FLOW is FDI; FID refers to some proxy for the level of financial development; and X is a set of control variables (p.289).

The financial development proxy: stock market capitalisation to GDP, is found to be

significant, implying the positive effect of FDI on growth. The higher the development of the stock market, the higher is the positive impact of FDI on growth (Durham, 2004, p. 296).

The development of local financial markets and its limitations to take advantage of FDI spill-overs, is underlined by Alfaro et al. (2004) (p.91). Even though FDI focuses on foreign capital, local financial markets are of great importance to take advantage of new knowledge. The more developed a financial market is, the higher the chance of backward linkages created by FDI. These linkages can be used by existing firms, but can also promote the formation of new companies (Alfaro et al., 2004, p. 92).

Alfaro et al. (2004) use four variables in their empirical analysis. First, they use liquid liabilities of the financial system, measured as M2/GDP (Alfaro et al., 2004, p.95). This measure is also used by Borensztein et al. (1998) as a proxy for financial development. As a second variable, commercial bank assets as a share of commercial bank plus central bank assets. This can be seen as a relative size indicator.

The third variable they use is private sector credit as a share of GDP. The last variable is bank credit as a share of GDP (Alfaro et al., 2004, p.95).

Stock market liquidity is also added by Alfaro et al. (2004) in their research. It is measured as the value of stock trading relative to the size of the economy. Furthermore, another variable is used: the relative size of the stock market, measured as the average value of listed domestic shares on domestic exchanges in a year as a share of the size of the economy (Alfaro et al., 2004, pp.95-96).

The following model is used, in which FINANCE stands for the different variables measuring financial market development.

After the regression, Alfaro et al. (2004) conclude that all the above financial market development variables are significant as an interaction term. However, the variables by themselves are insignificant. They assign these outcomes to the fact that the interaction term

is possibly of great importance as an allocation function (p.99).

Control variables

Barro and Lee (1994) have investigated the sources of economic growth. The researchers speak of a ‘conditional convergence effect’, meaning that “a country grows faster if it begins with lower real per-capita GDP relative to its initial level of human capital in the forms of educational attainment and health” (p.1). Following the neoclassical growth model, the growth rate is also affected by control and environmental variables. Those effects can be determined by looking at their impact on the steady state of the economy (Barro and Lee, 1994, p.12).

For instance, domestic investment as a ratio of GDP is such a variable. A higher value of this ratio leads to a rise of the steady-state ratio of output to effective worker. This in turn raises the growth rate (Barro and Lee, 1994, p.13).

In their research, Barro and Lee (1994) make the assumption that governmental consumption does not directly affect productivity. In spite of this, the ratio of government consumption to GDP does distort private decisions. These changes can be the result of the governmental activities themselves, or the adverse effects since this consumption is publicly financed (p.13). So a higher ratio leads to a lower growth rate.  

Political instability is also an influential factor in economic growth. The authors compare an increase in this variable with a decline in the security for propertyrights. Just like other

governmental distortions it is likely that this will lower the the steady-state level of output per effective worker. Therefore economic growth is reduced (for given values of the state

variables) (Barro and Lee, 1994, p.13).

Market distortions are expected to lower the growth rate for given values of the state variables as well. As a proxy, Barro and Lee (1994) use the black-market premium on foreign exchange (p.13).  

The control variables used by Borensztein et al. (1998) include variables frequently used in the literature. These consist of: government consumption, black market premium on foreign exchange, a measure of political instability, a measure of political rights, a proxy for financial development, the inflation rate, and a measure of quality of institutions (p. 122).

Hypotheses

With the findings of the different researchers described in the section above we can form some hypotheses about the impact of FDI on economic growth and trough which transmission channels this happens.

Based on the results of former literature, I expect the following hypotheses to be true:

1) There is a positive relation between the level of human capital and the effect of FDI in

economic growth.

A higher level of human capital increases the absorptive capacity of a country;

2) The level of openness to trade is positively related with the effect of FDI on economic

growth.

‘Export promoting’ countries are likely to attract more FDI;

3) A positive relation exists between market development and the contribution of FDI to

economic growth.

‘Deeper’ financial systems absorb FDI better and therefore countries with a more developed market can take more advantage of FDI spill-overs;

Methodology

An empirical analysis was constructed to examine the effect of FDI on economic growth. For this, several models were made, including the transmission channels described in the literature section above. The methodology differs from the existing literature since fixed effect panel data regressions were used on a dataset with developed countries as well as a dataset of developing countries.

In this section the dataset, empirical models and possible problems concerning the data are described.

Data

A comparative analysis will be made, to compare the effect of FDI and its externalities on economic growth (growth of GDP per capita) in developed countries with

developing/emerging countries. Two separate panel-datasets were composed. The first dataset consisting of 34 developing countries and the second consisting of 18 developing countries. The selection of developed countries is based on membership of the OECD.

Developed countries - The Organization for Economic Co-operation and Development

Initiated in 1960, an organization was created by 18 European countries, the United States and Canada, to enhance global development. Today, with 34 member countries from all around the globe, this organization is known as the Organization for Economic Co-operation and Development (OECD) (“About the OECD”, n.d.). The OECD aims at helping governments in making policy decisions to achieve wealth and fight poverty. To become a member, a country will be thoroughly examined on its ability to meet OECD standards (“About the OECD”, n.d.). Members of the OECD include a variety of countries; from advanced countries to emerging economies like Mexico. Apart from the member countries, the OECD also works together with emerging and developing countries. These countries are referred to as partners, for example: China, Brazil, South Africa and Latin America (“About the OECD”, n.d.).

Developing countries

The selection of developing countries was done within the region of Latin America. The countries included in the sample, are taken from the World Economic Outlook of the IMF (“WEO”, april 2014). However, a small adjustment of the sample was made.

For my empirical analysis not every developing and emerging Latin American Country will be used. Chile and Mexico will not be included since they are already part of the OECD countries. Chile became a member of the OECD on 7 may 2010 and Mexico on 18 may 1994 (“About the OECD”, n.d.). A list of the developed- and developing countries used in the sample can be found in Appendix A.

Both datasets, cover the period of 1980 to 2010. This time frame was chosen since it is in the recent past and also to create a sample size that was large enough. A large sample size better reflects the population and therefore increases the chance of significance in statistical testing.

Dependent variable

GDP per capita

To examine the effect of FDI on economic growth, GDP per capita (in current US Dollars) is taken as a measure. Instead of ‘regular’ GDP values, I look at GDP per capita since the population differs among the countries in the samples. As dependent variable in the model,

the log of GDP per capita is used, so that it becomes a growth rate of GDP per capita. GDP growth is approximately exponential, it tends to grow by a certain percentage per year, when looking in the long run. This means that transforming the variable to a logarithm, the growth will be approximately linear (Stock & Watson, 2012, p.562). Additionally, Stock and Watson (2012) state that: “the standard deviation of many economic time series is

approximately proportional to its level”. Meaning that: “the standard deviation of the logarithm is approximately constant” (p.562). For these reasons, logarithmic transformation of GDP per capita is useful, so that changes can be seen as percentage changes.

Also a stationary dependent variable is required. To check this I use the Im-Pesaran-Shin unit-root test for panel data. It can be used on an unbalanced panel, meaning that a panel can have “some missing data for at least one time period for at least one entity” (Stock and Watson, 2012, p. 390). This test assumes that all panels contain unit roots under the nul hypothesis and assumes some panels to be stationary under the alternative hypothesis. Using the OECD dataset, the test leads to a p-value of 0.0006, thus the nul hypothesis is rejected. Therefore I assume the log of GDP per capita to be stationary and use this measure as dependent variable.

Independent variables

Foreign direct investment

First, a measure needs to be defined for data on FDI. For this I agree with Borensztein et al. (1998) who use only FDI inflows. This is a more suitable measure since I want to study the effect of FDI and its spill-over effects for the receiving country. Other than that, FDI outflows are not expected to have a negative effect on the economic growth (Borensztein et al., 1998, p. 122). FDI is thus measured as FDI inflows as percentage of GDP.

FDI does probably not affect growth rates immediately, but in later periods (Durham, 2004, p. 291), so I take the lagged value of FDI as the right measure. Borensztein et al. (1998) also incorporated lagged FDI values in their empirical analysis. Yet their lagged FDI values represent an instrument to apply instrumental variable techniques (p.133).

Human capital

As discussed in the literature section above, many researchers (including Borensztein et al. (1998)) follow the human capital measure of Barro and Lee (1993). However I don’t find their data useful, since it is only measured every 5 years. Instead, I use the world development indicator: duration of secondary education, measured in years. This measure was taken from

the Worldbank (2013) and shows the length (in years) of secondary education per country.

Trade

Tade is measured as the share of total exports and imports to GDP, just like

Balasubramanyam et al. (1996) did. A higher value of this variable implies a higher openness to trade of the country.

Market development

To measure market development I use the same proxy as Borensztein et al. (1998), namely the ratio of the liquid liabilities of the financial system to GDP (M2/GDP) (p.125). This measure is also used because the corresponding data is widely available.

Control variables

Economic growth is not only affected by FDI, more factors contributing to growth need to be accounted for. Therefore control variables are added to the model.

Inflation

Inflation, is measured as the annual percentage change in the GDP deflator. This proxy was also used by Alfaro et al. (2004).

Population growth

Just like Barro and Lee (1994), I include annual population growth in my model, given its influence on economic growth.

Government consumption

Government consumption as a share of GDP is another control variable, also used by Borensztein et al. (1998), Barro and Lee (1994) and Alfaro et al. (2004). Government consumption is assumed to distort public decisions and as a consequence, GDP.

Domestic investment

Economic growth is also affected by the level domestic investment and therefore we need to control for it. It is measured as a share of GDP.

All variables are World Development Indicators and their corresponding data are taken from the World Bank. A precise definition of all variables can be found in Appendix B.

Model

The model used in the empirical analysis is based on the model Borensztein et al. (1998) used in their research. I use some of the same variables, but I also added other variables, found in the literature discussed above.

The general form of the growth formula is given below:

(1) G = α + β1DomInv + β2GovCon + β3FDI + β4HC + β5Infl +β6PopGr + ε1

Where G is the growth rate of GDP per capita, DomInv is the level of domestic investment as a share of GDP, GovCon is government consumption as a share of GDP, FDI is inflows of FDI as a share of GDP (lagged value, t-1), HC is duration (years) of secondary education, Infl is percentage change in GDP deflator, PopGr is the annual percentage of population growth and last, εi is the error term.

Several other models are constructed to assess the effect of the different transmission channels separately. This will also help in verifying the hypotheses stated earlier. In the models below, both the interaction term and the term itself are included. In this way I can better asses whether the independent variable affects the dependent variable on its own or through the interaction term.

Human capital

(2) G = α + β1DomInv + β2GovCon + β3FDI + β4HC + β5Infl +β6PopGr + β7FDIxHC

+ ε2

In this model the interaction term between FDI and human capital is included.

Trade

(3) G = α + β1DomInv + β2GovCon + β3FDI + β4HC + β5Infl +β6PopGr + β7FDIxTrade

+ β8Trade + ε3

The third model includes an interaction term between FDI and openness to trade. Trade is total export plus import as a share of GDP.

Market development

(4) G = α + β1DomInv + β2GovCon + β3FDI + β4HC + β5Infl +β6PopGr + β7FDIxMDev

+ β8MDev + ε4

In the last model, the interaction term between FDI and market development is included to evaluate whether, or to what extent, market development is a transmission channel. MDev is M2 as a share of GDP.

Since we deal with panel data that consists of observations on the same n entities (countries) for over two periods, fixed effect regression is used. This method controls for omitted

variables that are time consistent but vary across countries (Stock and Watson, 2012, p. 396). Fixed effect regression works in the following way, assuming the model:

(1.1) Yit = β0 + β1Xit + β2Zi + uit

The dependent variable is Yit, Xit is the regressor and Zi represents an unobserved omitted

variable that is time consistent but vary across entities. In a regression the effect of the regressor on the dependent variable is being estimated, holding constant Zi. Since Zi

corresponds to a variable that is constant over time, but varies for each entity, the model can be thought of as having n intercepts. Each intercept represents a different entity and can be described as: αi = β0 + β2Zi (Stock and Watson, 2012, p. 396). Now, rewriting the model

(1.1), it becomes a fixed effects regression model: (1.2) Yit = β1Xit + αi + uit

So αi can be seen as the intercept of a specific entity (the ith entity), whereas β1 is the slope

coefficient that is equal for each entity. According to Stock and Watson (2012), the idea behind αi is that it represents “the effect of being in entity i ” (p. 396). Therefore the term is

referred to as an entity fixed effect. Concluding, indicator variables signifying the entity specific intercepts, absorb the effect of unobserved omitted variables (Stock and Watson, 2012, p. 396).

In case of a fixed effect regression, four assumptions need to hold, as stated by Stock and Watson (2012, pp. 404-405).

1. The error term has conditional mean zero;

2. The variables for one entity are distributed identically to, but independent of the variables for another entity;

3. Large outliers are unlikely;

4. There is no perfect multicollinearity.

The first assumption was tested by making a residual-versus-fitted plot, a graph of the residuals against the fitted values. The graph shows no clear pattern for both datasets, indicating a well fitted model. Also, a kernel density plot, a standardize normal probability plot and a quintile-normal plot were graphed. For the OECD dataset, the plots predominantly indicate normality, though the latter plot shows that the distribution in the tails are somewhat off the normal distribution. The same can be said about the Latin America dataset. See the graphs in Appendix C.

The same variables for all entities (countries) were collected, however the variables of each entity do not depend on those of another entity. Therefore the second assumption holds. Using the robust option in the regression reduces the effect of outliers on the model. The regression shows no perfect multicollearity of the variables, none of the regressors is a perfect linear combination of the other regressors. Assumptions 3 and 4 are thus taken care of.

Problems

The results of my research must be carefully interpreted since it might suffer from several statistical problems.

One of the threats to the validity of my model is simultaneous causality. A ‘normal’ causal relation would be if economic growth depends on the level of FDI. But if the level of FDI is also determined by economic growth at the same time, there is simultaneous causality, leading to biased and inconsistent OLS estimators (Stock and Watson, 2012, p. 366). Of course, this could also be the case for the variables like human capital and population growth. A second problem is omitted variable bias. This arises when a variable is left out of the regression that is a determinant of the dependent variable and is correlated to one of the independent variables as well (Stock and Watson, 2012, p. 358). To overcome this problem, I

have also added some control variables to the model. Still there is a chance of omitted variable bias, as I omitted for example black market premium and political instability. These variables were included in the research in some of the existing literature on FDI though. Another solution would be instrumental variables regression (IV), as reported by Stock and Watson (2012, p. 360). However as wasl already stated by Borensztein et al. (1998), finding an adequate instrument is difficult, since a good instrumental variable must be highly correlated with an independent variable like FDI, but not correlated with the error term (p.133). The only variable Borensztein et al. (1998) could provide, approximately serving as an instrument, was a lagged value of FDI. Therefore I took one-year lagged FDI values as well.

Results

In the following section, the results from the empirical analysis will be discussed.

Significance was tested at the 5% level. The results of the regressions are shown in table 1a and 1b. Some summery statistics are presented in table 2a and 2b in Appendix D.

Latin  America-­sample  

The first regression is done on the general growth model (1). The variables domestic investment, FDI and inflation have the expected signs. Domestic investment and FDI have positive coefficients, whereas inflation has a negative coefficient.

What stands out is the negative coefficient of human capital, as one would expect that the level of human capital would contribute positively to GDP growth. The negative coefficient of population growth is likely to be explained by the fact that GDP divided by a larger population, leads to a lower GDP per capita.

As was stated in the section about control variables, government consumption was expected to be of negative influence to GDP, according to Barro and Lee (1994). However my

empirical results indicate otherwise, since the coefficient has a positive sign.

Only the variables FDI, human capital and population growth are significant in the first model. The variables domestic investment, government consumption and inflation are insignificant, meaning that within my dataset and model (1), there is not enough evidence to conclude that GDP growth depends on these variables.

Government consumption remains positive and insignificant, except in model (3), where it becomes significant after the variable trade and its interaction term with FDI is added. FDI on itself is positive in models (1), (3), and (4), but becomes negative in model (2) after the interaction term of FDI and human capital is included (with respect to the general model (1)). Also, FDI turns insignificant in the second, third and fourth regression. Inflation is negative and insignificant in each of the models. The variables human capital and population growth stay negative and significant in all of the models.

In regression 2, the second model (2) is estimated, including an interaction term of human capital and FDI. This was done to get a better understanding of human capital as a possible transmission channel for FDI.

FDI is negative and non-significant, just like Borensztein et al. (1998) found in their regression. Since Borensztein et al. (1998) expressed the importance of human capital as a transmission channel for the effects of FDI on economic growth, I expected the interaction term to be significant. On the contrary, I find that the interaction term of human capital and FDI is positive but insignificant. Therefore it can not be stated that GDP growth depends on the interaction of FDI and human capital. It has to be kept in mind though, that Borensztein et al. (1998) used a different measure of human capital.

The third model (3) includes the variables trade and the interaction term of trade and FDI. Now it can be tested whether these variables affect GDP growth on their own or through interaction. The third regression results in a significant and, surprisingly, negative coefficient of trade. The coefficient of the interaction term however is positive, yet insignificant. This means that the level of FDI doesn’t affect economic growth through the level of openness to trade, which is in contrast to the findings of Balasubramanyam et al. (1996).

The fourth regression on model (4) looks at the effects of market development as well as the interaction of market development and FDI on economic growth.

It gives a negative and insignificant estimation of the coefficient of market development. Alfaro et al. (2004) also found their M2/GDP variable to be insignificant. However, their interaction term with FDI was significant, in contrast to mine, which is positive but insignificant. Perhaps M2/GDP isn’t an adequate measure for market development. Using stock market capitalization to measure market development might have been better, like Durham (2004) did.

Table 1a.

Effect of FDI on economic growth – Latin America sample (dependent variable: Logarithm of GDP per capita)

Regression (1) Regression (2) Regression (3) Regression (4) Independent

variables Model 1 Model 2 Model 3 Model 4

Domestic Investment (0.00845) 0.00562 (0.00867) 0.00626 (0.00817) 0.0109 (0.00882) 0.00584 Government consumption 0.0132* (0.00666) 0.0123* (0.00676) 0.0144** (0.00630) 0.0134 (0.00837) FDI (1-year lagged

value) 0.0265** (0.0121) (0.147) -0.100 (0.0291) 0.00683 (0.0288) 0.00181 Human Capital -0.406*** (0.114) -0.451*** (0.124) -0.428** (0.154) -0.404*** (0.116) Inflation -4.04e-05 (2.89e-05) -4.18e-05 (3.01e-05) -3.88e-05 (2.32e-05) -4.20e-05 (2.90e-05) Population Growth -0.839*** (0.0844) -0.863*** (0.0856) -0.952*** (0.0939) -0.847*** (0.0921) Trade -0.00531*** (0.00160) Trade*FDI 0.000189 (0.000209) Human Capital*FDI 0.0216 (0.0247) Market Development -0.000829 (0.00380) Market Development* FDI 0.000500 (0.000448) Constant 11.17*** (0.735) 11.47*** (0.807) 11.71*** (0.913) 11.20*** (0.741) Observations 472 472 472 470 R-squared 0.560 0.563 0.577 0.561

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

OECD-sample

The same models are also estimated using data from the OECD sample, in order to compare developing countries with developed countries.

From the regression of the first general model (1) it follows that domestic investment, government consumption, FDI and population growth have positive signs. Human capital though has a negative coefficient, whereas I expected it to be positive. Inflation has a negative sign, as expected. Domestic investment, government consumption, human capital and

either of these variables. Yet, FDI and inflation are significant.

In the models 2-4, domestic investment remains insignificant and positively signed.

Government consumption remains the same in model (2), but becomes significant in model (3) when openness to trade and its interaction term with FDI are accounted for. In model (4), the coefficient of government consumption remains insignificant, but becomes negative. In comparison to the first model (1), the coefficient of population growth stays positive and insignificant, apart from the fourth model (4). Here the addition of market development and its interaction term with FDI turns the sign of population growth negative. Also human capital changes due to the composition of model (4). Its sign changes from negative to positive. The coefficient of FDI on itself remains positive and significant throughout all the models. Furthermore, inflation stays negative and significant in the models following the general model (1).

The regression of the second model (2), estimating the effect of the interaction term of human capital and FDI in particular, results in a negative and insignificant coefficient of human capital and a negative but significant estimate for the interaction term.

In the third model (3), trade and its interaction term are included. Both coefficients are significant, but trade on itself is positive, while the interaction term is negative.

In model (4) the variables market development and its interaction term with FDI are added to the general model (1). From the fourth regression it follows that market development is positive and significant. The interaction term is also significant, yet it is negative.

Table 1b.

Effect of FDI on economic growth – OECD sample (dependent variable: Logarithm of GDP per capita)

Regression (1) Regression (2) Regression (3) Regression (4) Independent

variables Model 1 Model 2 Model 3 Model 4

Domestic Investment (0.0154) 0.00139 0.000838 (0.0149) (0.0137) 0.00114 (0.0122) 0.00443 Government consumption 0.0576* (0.0303) 0.0590* (0.0292) 0.0604** (0.0281) -0.00503 (0.0175) FDI (1-year lagged

value) 0.0266** (0.0116) 0.133*** (0.0326) 0.0404*** (0.0130) 0.0225*** (0.00433) Human Capital -0.192 (0.305) -0.154 (0.315) -0.0987 (0.136) 0.189 (0.281) Inflation -0.0102*** (0.00356) -0.0101*** (0.00347) -0.00955*** (0.00220) -0.00744*** (0.00219) Population Growth 0.117 (0.143) (0.133) 0.0919 (0.105) 0.119 (0.0940) -0.0535 Trade 0.0180*** (0.00215) Trade*FDI 0.000131*** (4.49e-05) Human Capital*FDI 0.0155*** (0.00486) Market Development 0.0114*** (0.00107) Market Development* FDI -4.40e-05*** (7.20e-06) Constant 9.712*** (2.083) 9.445*** (2.125) 7.742*** (1.372) 7.524*** (1.897) Observations 899 899 899 781 R-squared 0.225 0.243 0.429 0.541

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Hypotheses

To evaluate the effect of FDI on economic growth through the discussed transmission channels, 3 hypotheses were formed, that will be discussed below.

When looking at the results from regression 2, I can conclude that there is a positive relation between human capital and the effect of FDI on economic growth in developing countries. However, due to insignificance, human capital can’t be seen as a transmission channel for FDI. As for the developed countries, the interaction term is negative but significant. From this we can reject the first hypothesis (1), both for developing and developed countries. So, there is no positive relation between the level of human capital and the effect of FDI in economic growth.

From the third model (3), which accounts for trade, the results indicate the following: for developed countries, there is significance of the interaction term but in a negative way, so reject hypothesis (2). For developing countries a positive relation does exist, however because of insignificance hypothesis (2) can be rejected as well. Hence, there is not enough evidence to conclude that the level of openness to trade is positively related with the effect of FDI on economic growth.

The results of the regression of model (4) show that for developed countries, FDI does affect economic growth through the level of market development (significant), but in a negative way. For developing countries however, a positive relation is found. Still the insignificance indicates that economic growth doesn’t depend on this relation. Hypothesis (3) can be rejected for developed- and developing countries. Consequently, no positive significant relation exists between market development and the contribution of FDI to economic growth.

Conclusion

This section concludes on the existing literature about the effects of FDI on economic growth, the findings of my own empirical research and problems that have to be kept in mind when doing research on this topic.

Since the end of the last decade, there has been an increasing interest for FDI. As was stated by Alfaro et al. (2004): “In 1998, FDI accounted for more than half of all private capital flows to developing countries” (p.90). Especially in Latin American countries, FDI inflows are very large and continue to grow. Even in times of crises, FDI was relatively little affected.

The reason behind the interest in FDI lies in the fact that FDI brings with it externalities, which contribute to economic growth and modernization. Among these externalities, also referred to as spill-over effects, are increased productivity, know-how, access to foreign markets and advanced technology.

To attract more FDI inflows, countries have been generating a beneficial climate for FDI, for example by liberalizing their FDI regimes and reducing restrictions on FDI inflows.

However the potential positive effect of FDI on economic growth is not that straightforward. How much a country is able to benefit from FDI is largely depending on its absorptive capacity. This absorptive capacity can be seen as the level of economic development that enables a country to exploit the benefits of FDI. The development of the financial market, the education level and the level of trade are some determinants of economic development. The existing literature treats these factors as transmission channels for FDI.

Now, returning to my research question:

“Is there a difference in the way FDI affects economic growth in developed vs. developing countries?”

As I expected after the literature study, the level of development of a country would be of great influence to the way FDI affects economic growth. Developed countries are expected to have more openness to trade and a better-developed financial market. Also, though the level of human capital is expected to be lower in developing countries, it can be an important transmission channel of FDI when it is above a certain level as stated by Borensztein et al. (1998).

In my empirical research I focused on the effect of FDI on economic growth through the transmission channels mentioned above. My research has led to surprising and ambiguous

results. All of the hypotheses were rejected, for the developing- as well as the developed countries. This means that there was not enough evidence that human capital, openness to trade and market development functioned as transmission channels for FDI, positively contributing to economic growth. My findings are in contrast to the existing literature, where these transmission channels are seen as very important. Therefore, from an empirical point of view, my results leave me unable to conclude properly on the research question.

There are several reasons I can assign these contrasting outcomes to. A different composition of the models, a different research method, different measures for the variables and different samples might have been the reason.

Also, this topic requires extra caution. In the existing literature even more variables are accounted for, but some of them were omitted from my models because collecting these variables was too difficult. Even though fixed effect regression was used, this only takes away part of the omitted variable problem. Another important aspect of growth theory is causality and with this the problem of simultaneous causality arises. For instance when FDI is not only a determinant of economic growth, but also vice versa. The use of Instrumental Variable regression techniques in subsequent research might lead to better estimates. However, as indicated by Borensztein et al. (1998), finding an instrument that is highly correlated with an independent variable like FDI, but not correlated with the error term, is rather difficult.

References

“About the OECD” [Information on a page]. (n.d.). Retrieved from

http://www.oecd.org/about/

Alfaro, L., Chanda, A., Kalemli-Ozcan, S. & Sayek, S. (2004). FDI and economic growth: The role of local financial markets. Journal of International Economics, 64, 89-112. doi:10.1016/S0022-1996(03)00081-3

Bhagwati, J. N. (1978). Anatomy and Consequences of Exchange Control Regimes, Studies in International Economic Relations, 1(10), New York: National Bureau of Economic Research

Balasubramanyam, V.N., Salisu, M. & Sapsford, D. (1996). Foreign direct investment and growth in EP and IS countries. The Economic Journal, 106, 92-105.

doi:10.2307/2234933

Barro, R., Lee, J-W. (1993). International comparisons of educational attainment. Journal of

Monetary Economics, 32, 361–394. doi: 10.1016/0304-3932(93)90023-9

Barro, R., Lee, J-W. (1994). Sources of economic growth. Carnegie Rochester Conference

Series on Public Policy, 40, 1-46. doi: 10.1016/0167-2231(94)90002-7

Borensztein, E., De Gregorio, J., Lee, J-W. (1998). How does foreign direct investment affect economic growth? Journal of International Economics, 45(1), 115–135.

doi: 10.1016/S0022-1996(97)00033-0

Foreign Direct Investment. (n.d). In Investopedia. Retrieved from http://www.investopedia.com/terms/f/fdi.asp

Foreign Direct Investment. (2013). In OECD Factbook 2013: Economic, Environmental and Social Statistics. Retrieved from http://www.oecd-ilibrary.org/sites/factbook-2013-en/04/02/01/index.html?itemId=/content/chapter/factbook-2013-34-en

ECLAC. (2014). Latin America and the Caribbean Received 184.92 Billion Dollars in Foreign Direct Investment in 2013 [Press release]. Retrieved from

http://www.eclac.cl/cgi-bin/getProd.asp?xml=/prensa/noticias/comunicados/4/52984/P52984.xml&xsl=/prensa /tpl-i/p6f.xsl&base=/prensa/tpl-i/top-bottom.xsl

Loungani, P. & Razin, F. (2001, June). How beneficial is foreign direct investment for developing countries? Finance and development: a quarterly magazine of the IMF.

Montero, A. P. (2008). Macroeconomic Deeds, Not Reform Words: The Determinants of Foreign Direct Investment in Latin America. Latin American Research Review, 43(1). 55-83. doi: 10.1353/lar.2008.0008

Stock, J. H. & Watson, M.M. (2012). Introduction to econometrics. Harlow: Pearson Education Limited.

United Nations Conference on Trade and Development. (1999). Foreign direct investment

and development. UNCTAD series on issues in international investment agreements.

Retrieved from http://unctad.org/en/Docs/psiteiitd10v1.en.pdf

United Nations Conference on Trade and Development. (2000). Tax Incentives and Foreign

Direct Investment: A Global Survey (ASIT Advisory Studies No. 16). Retrieved from

http://unctad.org/en/Docs/iteipcmisc3_en.pdf

United Nations, Economic Commission for Latin America and the Caribbean (2008).

Foreign direct investment in Latin America and the Caribbean. Retrieved from

http://www.cepal.org/publicaciones/xml/4/36094/LCG2406i.pdf

United Nations, Economic Commission for Latin America and the Caribbean (2009).

Foreign direct investment in Latin America and the Caribbean. Retrieved from

http://www.cepal.org/publicaciones/xml/2/39422/2010-414-LIEI-Book_WEB.pdf United Nations, Economic Commission for Latin America and the Caribbean (2013).

Foreign direct investment in Latin America. Retrieved from

http://www.cepal.org/publicaciones/xml/8/52978/ForeignDirectInvestment2013.pdf Wang J.Y. (1990). Growth, Technology Transfer And The Long-run Theory Of International

Capital Movements. Journal of International Economics, 29, 255-271. doi: 10.1016/0022-1996(90)90033-I

World Bank. (2013). World Bank Database Indicators. Retrieved from:

http://data.worldbank.org/indicator. World Economic and Financial Surveys, World Economic Outlook. (April 2014). Retrieved

from https://www.imf.org/external/pubs/ft/weo/2014/01/weodata/groups.htm Zhang, K.H. (2001). How does FDI affect economic growth in China? Economics of

Appendix

A - Countries included in the sample

Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israël, Italy, Japan, Korea, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, United Kingdom, United States

Developing countries

Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, El Salvador, Ecuador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay and Venezuela.

B - Variable descriptions

GDP per capita (current US$)

GDP per capita is gross domestic product divided by midyear population. GDP is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. Data are in current U.S. dollars.

Source: World Bank

Secondary education, duration (years)

Duration of secondary education is the number of grades (years) in secondary education (ISCED 2 & 3).

Source: World Bank

Domestic investment: Gross capital formation (% of GDP)

Gross capital formation (formerly gross domestic investment) consists of outlays on additions to the fixed assets of the economy plus net changes in the level of inventories. Fixed assets include land improvements (fences, ditches, drains, and so on); plant, machinery, and

equipment purchases; and the construction of roads, railways, and the like, including schools, offices, hospitals, private residential dwellings, and commercial and industrial buildings. Inventories are stocks of goods held by firms to meet temporary or unexpected fluctuations in production or sales, and "work in progress." According to the 1993 SNA, net acquisitions of valuables are also considered capital formation.

Population growth (annual %)

Population growth (annual %) is the exponential rate of growth of midyear population from year t-1 to t, expressed as a percentage.

Source: World Bank

Foreign direct investment, net inflows (% of GDP)

Foreign direct investment are the net inflows of investment to acquire a lasting management interest (10 percent or more of voting stock) in an enterprise operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long-term capital, and short-long-term capital as shown in the balance of payments. This series shows net inflows (new investment inflows less disinvestment) in the reporting economy from foreign investors, and is divided by GDP.

Source: World Bank

Trade (% of GDP)

Trade is the sum of exports and imports of goods and services measured as a share of gross domestic product.

Source: World Bank

Inflation, GDP deflator (annual %)

Inflation as measured by the annual growth rate of the GDP implicit deflator shows the rate of price change in the economy as a whole. The GDP implicit deflator is the ratio of GDP in current local currency to GDP in constant local currency.

Source: World Bank

General government final consumption expenditure (% of GDP)

General government final consumption expenditure (formerly general government consumption) includes all government current expenditures for purchases of goods and services (including compensation of employees). It also includes most expenditures on national defense and security, but excludes government military expenditures that are part of government capital formation.

Source: World Bank

Money and quasi money (M2) as % of GDP

Money and quasi money comprise the sum of currency outside banks, demand deposits other than those of the central government, and the time, savings, and foreign currency deposits of resident sectors other than the central government. This definition of money supply is frequently called M2; it corresponds to lines 34 and 35 in the International Monetary Fund's (IMF) International Financial Statistics (IFS).

C – Graphs

a) Latin America sample

 Residuals-­‐versus-­‐fitted  values  plot       Kernel density plot    

b) OECD sample

Residuals-­‐versus-­‐fitted  values  plot       Kernel density plot      

D – Tables

Summary statistics

Table 2a. Latin America sample

Descriptive  Statistics  all  variables  –  Latin  America  sample  

Variables   Number  of   observations  

Mean   Standard   Deviaton  

Min   Max  

Log  GDP  per  capita   547   7.602   0.7603   5.4781   9.5148   Domestic   investment   536   19.722   5.5029   5.2421   41.8595   Government   consumption   525   13.332   6.6450   2.9755   43.4792   FDI   499   2.081   2.4840   -­‐12.2084   17.1343   Human  capital   558   5.754   0.5861   5   7   Inflation   539   147.941   935.7982   -­‐26.3000   13611.63   Population  growth   558   1.784   0.7012   -­‐0.0667   3.14800   Trade   536   58.524   34.0000   11.5457   198.7668   Market   development   514   35.073   15.4573   10.0829   111.3253   Human  capital*FDI   558   10.677   14.1931   -­‐73.2506   102.8056   Trade*FDI   558   132.029   270.5863   -­‐1820.536   2491.0190   Market   development*FDI   558   77.837   139.0300   -­‐413.7004   1405.8800      

Table 2b. OECD sample

Descriptive  Statistics  all  variables  –  OECD  sample  

Variables   Number  of  

observations   Mean   Standard  Deviaton   Min   Max   Log  GDP  per  capita   1014   9.574   0.9149   7.0225   11.6265   Domestic   investment   1011   22.931   4.2287   9.8542   38.6969   Government   consumption   1006   18.904   4.9248   7.5156   41.4761   FDI   905   2.944   5.8160   -­‐55.0655   74.7106   Human  capital   1054   6.613   0.9327   5   9   Inflation   1002   9.385   22.6287   -­‐5.3903   390.6788   Population  growth   1053   0.661   0.6900   -­‐2.5743   6.0170   Trade   1004   76.889   44.6482   15.9240   333.5322   Market   development   833   86.012   77.7217   11.0369   669.8804   Human  capital*FDI   1054   16.566   37.7195   -­‐385.4582   522.9741   Trade*FDI   1054   273.984   1230.1300   -­‐18366.11   23872.07   Market   development*FDI   1054   296.902   2159.2720   -­‐32841.100   48933.03