Chinese FDI in Sub-Saharan Africa : theoretical research and an empirical analysis of the effects of Chinese and non-Chinese foreign direct investment on human and economic development

(1)

Chinese FDI in Sub-Saharan Africa

Theoretical research and an empirical analysis of the effects of Chinese and

non-Chinese foreign direct investment on human and economic development

University of Amsterdam

Faculty of Economics and Business Amsterdam, August 2015

T.H.L. Mosselman (10670130) thlmosselman@gmail.com

Master’s programme Economics; International Economics and Globalization

Master’s Thesis

Thesis supervisor: Dr. K.B.T. Thio

(2)

i

Abstract

This thesis focusses on Chinese foreign direct investments (FDI) in Sub-Saharan Africa (SSA). As the total amount of these investments has increased greatly in the last decade -from 68 million in 2003 to almost 3 billion USD in 2011- it could have certain effects on the development of African

economies. Following an extensive literature review on growth theory, FDI, country-of origin effects and Chinese FDI characteristics; an empirical analysis on the effects of Chinese and non-Chinese FDI on development is conducted. The most striking characteristics of Chinese FDI are the high level of involvement of state-owned enterprises and the long-term view, whereas negative effects might be caused by the lack of attention for human rights and the environment by Chinese investors. Based on the traditional Solow model and evidence from previous research, a growth formula is

constructed. This in turn is used in a regression to test the possible different effects from Chinese FDI and FDI originating in the rest of the world. Both the effect on the economic development,

measured by GDP per capita, and human development measured by the human development index (HDI) are estimated. Using a panel data set for 34 SSA countries during 2003-2012, the results indicate that Chinese and non-Chinese FDI have different influences on development in SSA. Surprisingly non-Chinese FDI has a significant negative effect on GDP through the system GMM method and on HDI through both the OLS fixed-effects and system GMM method. Estimates for the effects of Chinese FDI on GDP however, are significantly positive through both methods. Results for the coefficients from Chinese FDI on HDI are insignificant.

(3)

ii

Statement of Originality

This document is written by student Tunettus Hendrik Lucas Mosselman who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(4)

iii

Introduction

In the last fifteen to twenty years, one of the most spectacular changes in global economics and politics has been the economic rise of the People’s Republic of China (henceforth, China). China’s influence has grown worldwide and so has the impact of the demand of its economy and its foreign direct investments (FDI). Through these investments, new relationships with African economies have been born.

The reasons for China to invest in these fast-growing African economies are manifold. The most well-known reason is the fact that China does not have sufficient reserves of natural resources to sustain its impressive growth performance of the last decade; it is a net importer of metal ores and oil (Christensen, 2010). Another reason for China to invest in and build sustainable relationships with African countries, is that many economies in Africa have shown high growth rates recently and in total 1 billion people live in Africa (Allard, 2012). Next to these economic reasons, China also has political motives for providing aid to and investing in African countries. (Christensen, 2010)

The increase of FDI from China to sub-Saharan Africa -from 68 million in 2003 to almost 3 billion USD in 2011 (UNCTAD, 2014)- has led to some debate in recent years (Cheung, de Haan, Qian & Yu, 2012). The main reason for this debate is the possible different way of investing that China practices when compared to Western countries. Where Western firms are more reluctant to investing in countries with low levels of human rights, the Chinese government seems indifferent about these differences in African countries. An example of these different approaches is China’s commitment to its

investments in and arms deliveries to Sudan (Cheung et al., 2012). Another characteristic from Chinese FDI which makes it different from non-Chinese FDI is the strong influence of the Chinese government, through State Owned Enterprises (SOEs), on the outgoing FDI (Kragelund & Van Dijk, 2009).

Many papers (Cheung et al., 2012; Sanfilippo, 2010) have been written about the reasons and benefits for China from a stronger Sino-African relationship. However, one big aspect of this relationship has been neglected in the literature -as it has been in many discussions-: the effects on the African countries. Although there has been a discussion on the Chinese approach to this

relationship, not many empirical studies have been done on its effects in Africa. This is why this thesis will discuss and research these effects

In the last decades many studies have been conducted on the effects of FDI on the economic development of developing countries. However, not many of these studies have considered effects from different countries-of-origin on the development. The question this thesis will try to answer is:

What are the differences between Chinese and non-Chinese FDI and their effects on the human and economic development of Sub-Saharan African economies? This thesis will try to answer this

question through a literature review, which contains an overview of growth-theory, empirical research on FDI and an analysis of the different characteristics of Chinese FDI and its possible effects on development, and an empirical analysis based on panel data.

Possible positive effects following Chinese aid and investments for African countries can be large (UNCTAD, 2010). These effects can derive from infrastructure improvements, better access to the Chinese market and knowledge and the long-term view of the Chinese government. Potential

(6)

2

negative effects which are caused by the relationships with China are the crowding out of African industry, the Dutch disease and effects caused by the possible lack of attention for human rights, political reforms and protection of the environment by China.

Whether these effects are also measurable by the data will be tested in the empirical analysis. The model presented in this thesis is based on the Solow growth model and findings in the literature review. The benchmark for the effects that will be used in the thesis is the Human Development Index (HDI), which provides a broader measurement of development of Sub-Saharan Africa (SSA), for 2003 to 2012 in 34 SSA countries. Regressions will also be run on a model with GDP per capita as the dependent variable. Two different methods are used for estimating the regressions; OLS fixed-effects and GMM. The main independent variables will be the relative levels of FDI originating from China and the rest of the world. The results from both estimation methods indicate positive effects from Chinese FDI on HDI and GDP per capita and negative effects from non-Chinese FDI.

The thesis will be structured as follows. It starts with a literature review describing relative

theoretical and empirical papers. Based on this literature review, a growth model describing GDP per capita and HDI will be constructed and these will be tested through multiple regression analyses. The thesis will end with a conclusion and recommendations for further research and policies.

(7)

3

Literature review

In this literature review the discussed research is divided in several different sections. It contains a description of papers discussing the effects from FDI on economic growth from a theoretical point of view. Following this, empirical research on the FDI-growth nexus will be discussed. Stiglitz (2006) claims that GDP per capita is too narrow an indicator for human development, one also needs to take into account health and education. This is why research on the effects on development in a broader sense will be discussed as well. Next, literature on relationships between FDI’s country of origin and development will be discussed. Finally, the characteristics of Chinese FDI and its possible effects will be described.

Effects of FDI according to growth-theory

This section will present and discuss several economic theories on the relationship between economic growth and FDI. Most authors predict a positive effect through capital accumulation and/or productivity spillovers.

According to neoclassical theory formulated by Solow (1956), FDI promotes economic growth directly by raising the level of investment, a process called capital widening. This will in turn increase the capital stock, which would increase per capita income in the host economy. Especially for capital-scarce countries, this positive effect stemming from FDI can be important. However, the growth rate would only be higher for a limited amount of time, due to the assumed diminishing returns to capital. According to the neoclassical theory, there would be a very high amount of capital flowing from rich to poor countries due to possible higher returns to capital in least-endowed countries. In reality, it does not seem to work this way as the amount of capital from rich to poor countries is relatively low (Lucas, 1990). This puzzle is known as the Lucas Paradox.

Johnson (2006) makes the distinction between greenfield and brownfield FDI. The former means that the foreign investor constructs new production, distribution or research facilities in the host country, leading to an increase in physical capital stock as described above. The latter, on the other hand, implies that the foreign investor acquires pre-existing facilities in the host country. This will not lead to an increase in the stock of physical capital, so an increase in the level of economic growth through that channel is not to be expected.

Although brownfield investment might not affect the physical capital stock, it can have a positive effect on the technology and knowledge level of the host country. This capital deepening is predicted to have a positive effect on the economic growth of the host country through different mechanisms according to different growth theories. In the Solow (1956) model FDI can influence long-run growth through technological progress or labour force growth, both are considered exogenous in this model. The Solow-model was augmented by Mankiw, Romer and Weil (1992) by introducing human capital, thus explaining more than 80% of differences in growth levels. In the AK growth model of Romer (1986), FDI increases economic growth through technology transfer and technological growth. This theory implies that the capital stock can produce more efficiently, a process called capital deepening. Where Romer (1986) captured endogenous growth by introducing a technology parameter, Lucas (1988) introduces a human capital parameter. He puts more focus on the transfer of knowledge and learning by doing due to FDI. Romer (1990) argues that growth is driven by technological change and that this change is driven by market incentives from the research sector. In his model, foreign investors mainly contribute to economic growth by increasing the level

(8)

4

of human capital in the R&D sector through spillovers from FDI. Therefore, he expects

underdeveloped economies which are open to trade to grow faster than closed ones because they will attract more FDI. Blomström and Kokko (1998) examine spillover effects from multinational firms in the host countries. They argue that FDI will have a positive effect on long-run growth due to the positive impact on the level of knowledge through the introduction of new managerial styles and labour training. Other spillovers are mentioned by Görg and Greenaway (2004), namely imitation, competition and exports.

MacDougall (1960) models that incoming FDI will be profitable for providers of labour. He reasons that FDI will increase the stock of capital, decreasing marginal product of capital and increasing the marginal product of labour. However, the author believes the most important direct gains from a higher level of FDI are most likely to come through increased tax revenue.

Barro and Sala-i-Martin (1997) even believe that FDI is the only way out of the development trap for the least-developed countries (LDCs). Because these countries are unable to generate technological advances in their initial situation, they need to imitate or adapt foreign technologies. Kindleberger (1969) argues that incoming foreign capital always brings something useful to the host country because a foreign firm has a certain distinctive characteristic that was not present before. Empirical research on the effects of FDI on development of the host country

Many empirical papers on the effects of FDI on the economic growth or HDI in the host country find a positive relationship in development countries. However, this is often only positive under certain conditions (de Mello, 1997). Many papers investigate the influence of host country-specific factors as domestic (human) capital, trade openness, financial market development and the technology gap. This section starts by investigating previous research that focusses on economic growth and ends with research on the human development-FDI relationship.

Borensztein, de Gregorio and Lee (1998) show that FDI contributes to growth of the host country. Because FDI is a vehicle for the transfer of technology, this effect is larger than the effect of domestic investment. However, they find that this stronger relationship is only present after a certain threshold of human capital in the host country. Alfaro, Chanda, Kalemli-Azcan and Sayek (2010) also highlight that human capital has a critical role in reaping growth benefits from FDI. The paper also shows that an increase in FDI is linked with an increase in domestic investment, but this result is not robust. The authors thus conclude that the main reason for the positive effect on economic growth is the stimulation of the technological progress. Li and Liu (2005) use data from 84 - including 63 developing - countries over the period 1970-1999 to investigate whether FDI affects economic growth. They also test for endogeneity of both these variables and conclude that they are becoming increasingly interdetermined. They conclude that FDI and economic growth are strongly complementary connected. The authors also find that human capital has a positive influence on this relationship and the technology gap, whereas the difference in technological development between the source and host country has a downward influence.

Many studies have been conducted on the effects of FDI on capital flows in the host country.

Bosworth, Collins and Reinhart (1999) describe and study the different capital flows in 58 developing economies from 1978 to 1995. Both domestic and foreign investment and savings are considered. They conclude that FDI has a “highly beneficial effect on domestic investment” (p. 164) and a strong effect on the rise in domestic savings. On average one dollar of FDI leads to fifty cents of domestic

(9)

5

investment. De Mello (1999) also confirms that FDI is growth-enhancing. FDI seems to have a larger growth-effect in developing countries because the new technology introduced by FDI is often complimentary to old domestic technology whereas in developed countries it can often be substitutable to domestic technology. This in turn leads to a rise of domestic investments in

developing countries. Likewise, Mody and Murshid (2005) find there is a strong link between capital inflows and domestic investments. However, their data show that this link becomes weaker over time.

Several studies stress the importance of a developed financial market as a necessary precondition for growth. The line of reasoning is that the FDI can only improve growth when the host country’s financial market can treat the foreign capital efficiently so it can finance productive investments. Alfaro et al. (2004) research this link between the development of local financial markets and the FDI-growth nexus by using data on 71 countries from 1975 to 1995 for their model. They find that FDI promotes economic growth through financial markets and that a bad-functioning financial market can “limit an economy’s ability to take advantage of such potential FDI benefits.” Durham (2004) finds that the development of the stock market increases the positive effect of FDI on growth. Alfaro et al. (2010) show in their model that the growth-enhancing effect of a sufficiently developed financial market works through backward linkages between foreign and domestic firms.

Balasubramanyam, Salisu and Sapsford (1996) test the hypothesis by Bhagwhati (1978) which predicts that the volume and efficiency of FDI are different in an export promoting and an import substituting regime. They check the hypothesis through an analysis of data from 46 developing countries with different levels of trade openness. Their results support the hypotheses and show that the growth enhancing effects are larger in open countries than in countries pursuing an import substitution policy, confirming Romer’s (1990) expectations.

Zhang (2001) studies the link between FDI and economic growth in Latin America and East Asia and finds big differences in FDI-growth links in different countries. He points out that this is probably due to the enormous diversity in economic structures and provides evidence that FDI-led growth is strengthened by liberalized trade regimes and macroeconomic stability. Azman-Saini, Baharumshah and Law (2010) provide evidence for Zhang’s (2001) reasoning by testing the impact of FDI on economic growth and its link to economic freedom. They conclude that FDI by itself has no direct impact on growth, but when the level of economic freedom is sufficiently high the economy will profit from FDI.

UNCTAD (2001) argues that the manufacturing sector has the largest potential for FDI-led economic growth in developing countries as it contains a broad range of linkage-intensive activities. Aykut and Sayek (2007) discuss and examine the different effects of FDI in the primary, services and

manufacturing sector on economic growth. Through an empirical analysis of cross-country data between 1990 and 2003, they find that the positive effect of FDI on economic growth increases with the relative level of the manufacturing sector in FDI. And, vice versa, as the share of the primary and/or services sector in FDI flows increase, the effect on economic growth turns negative. As shown above, most previous studies investigated the relationship between FDI and economic growth. There are fewer studies that include the effects of FDI on development in a broader sense, HDI especially, these are described below.

(10)

6

Basu and Guariglia (2007) investigate the interactions between FDI, inequality and growth in 119 developing countries during the period 1970-1990. They also construct a model of a dual economy which’ main predictions are consistent with the conclusions from the data. They find that both growth and inequality (measured by the Gini-coefficient) are positively connected to FDI. The authors identify that FDI can amplify inequality when there is a large group with low initial human capital, because they cannot reap the profits from modern FDI-based technology.

Fortanier and Maher (2001) study many aspects of the connection between FDI and sustainable development. They confirm positive effects through capital accumulation and productivity growth, but state that all positive effects are dependent on institutional stability and sound macro-economic policies.

Sharma and Gani (2004) research the influence of FDI on HDI for 15 low-income and 19 middle-income countries. They conclude that FDI has a positive, though very small, effect on the human development of both groups of countries. They also find that in low-income countries an increase in HDI exerts a significant positive effect on FDI, showing a two-way relationship.

Reiter and Steensma (2010) examine the relationship between FDI, HDI, FDI policies and corruption. Their analysis consists of data from 49 developing countries during 1980-2005. Their results show a positive effect of per capita FDI inflow on the HDI. This effect is larger when policies that restrict FDI to certain sectors are in place, usually the sectors where human capital gain is largest. They also find that FDI inflow decreases year-to-year improvement in HDI when corruption is high.

Gohou and Soumaré (2012) empirically research whether FDI improves the welfare, measured by HDI and real per capita GDP, of citizens of 52 African countries for the years 1990-2007. They conclude that FDI has a significant positive effect on welfare in Africa as a whole. However, the effects are stronger for the poorest countries in Africa. These countries experience a higher poverty reduction through FDI according to the authors.

Country-of-origin effects

As shown above, most existing studies on the FDI-growth nexus focus on the role of host country factors and treat FDI as homogeneous flows of capital, not distinguishing by different origins. When researching the effects of Chinese FDI specifically however, the home country factors also play a role. The part of academic literature researching general country-of-origin effects on FDI-led growth is relatively small (Fortanier, 2007). However, this subject is worth researching because different sources of FDI might have different effects. These effects are expected to differ because market conditions, institutions, management systems and other factors can differ widely between source countries and even more between single investors. Several scholars have investigated the

determinants to invest for different countries-of origin, focussing on distance, language, culture and time (Búrcio, 2015). There is some literature on agglomeration effects of different country-of-origins from investors (Meyer & Tan, 2011). Another effect of the country-of-origin from FDI is the

perceived image that people have from certain investments (Búrcio, 2015).

Wang, Clegg and Kafouros (2009) present evidence which shows that the investors with different countries of origin have different ways of engaging in FDI. They investigate on an industry level in China the differences in behaviour between investors from Hong Kong, Macau and Taiwan (HMT) and those from non-Chinese Western (NCW) source countries. They find that HMT investors are

(11)

7

present in low-technology, labour intensive industries, while NCW investors dominate sectors with sophisticated technology. The authors explain that his can be seen by the different orientations from the investors. Whilst NCW investors are focussed on the Chinese market, HMT firms are more export-oriented. In their analysis they use a dataset of 157 sectors in China in 2004. Their results show that the NCW regression contains more significant results than the HMT regression. This is possibly due to the fact that HMT firms show ‘domestic investor’ behaviour, partly caused by ‘round-tripping’. This scheme involves Chinese firms sending capital to associated firms in Hong Kong that in turn invest this capital in China, which shows as FDI from Hong Kong in the statistics.

Monastiriotis (2014) researches the productivity spillovers from foreign ownership in 28 transition countries over the period 2002-2009. He finds that FDI originating from the EU raises productivity for the host country firms significantly more than FDI from non-EU countries. He also finds that

countries that are closer to or even a member of the EU and have more intense economic links with the EU, enjoy more positive spillovers than countries with weaker ties.

Elmslie, Ford and Rork (2008) research whether the country of origin of FDI matters to economic growth of individual states in the United States. They test this for investments from seven different countries for the years 1978-1997. They measure FDI as the relative amount of each state’s

employment stemming from a particular source country. The authors find that investments from Japan are more beneficial to growth than domestic investments, whereas Swiss and British investments are less effective in causing economic growth. After analysing the different capital-labour ratios of the source countries and host states, they conclude that the FDI is most beneficial to the hosts’ economic growth if the capital-labour ratio of the source country is close to the hosts’. The authors explain this by reasoning that the costs of technology transfer will be lower when the difference in capital-labour ratio is small.

Alegria and Monastiriotis (2011) also conclude that investments from a comparable country can bring more positive externalities; they find this effect as well for investment from nearby countries. They argue that this is the reason that in their data analysis of Greek and other EU-sourced

investments in Bulgaria, the investments originating in neighbouring Greece produced larger productivity spillovers.

This is not consistent with Waldkirch (2010), who presents a theory that links more positive externalities to a larger physical distance to the source country. The reasoning is that more inputs will be bought in the host country when the source country is further away. He also reasons that FDI originating from more technological advanced countries will have a greater positive impact on the host country than investments from countries that are further from the technology frontier. Gee and Karim (2011) investigate whether the effects of FDI on the growth of the manufacturing sector in Malaysia are dependent on the source of the investment. After analysing their data for the period 1991-2006 they find that FDI from the emerging economies Indonesia, Philippines and Thailand has a significant negative effect on the output growth in Malaysia’s manufacturing sector. The results for FDI from the European Union, however, are significantly positive. Reasons for these effects could be the differences in distance and technological development.

Fortanier (2007) investigates the role of the investor’s country of origin on the relationship between FDI and host country economic growth by using a dataset on FDI of the six major investor countries

(12)

8

in 71 host countries for the period 1989-2002. The FDI from these six countries accounts for 63% of global outward FDI stock. The author uses a panel data analysis to research the effect of different home country characteristics. She concludes that effects of FDI on economic growth depend on the country of origin, mostly its business structure, and that these country-of-origin effects differ by host country characteristics, especially trade openness and institutional quality.

Summarizing, most research find that it matters where FDI comes from. Although the researchers agree on the characteristics of the source country that affect the host country, they differ in whether these effects are positive or negative. The characteristics of a source country that might influence the development of a host country are distance, comparability and economic and/or political ties. Characteristics of Chinese FDI

As written in the introduction, the way that China conducts FDI might differ from other countries, although not all literature agrees on the magnitude of these differences. According to Kaplinsky, McCormick and Morris (2010) “Chinese FDI is qualitatively different” (p. 7). Kragelund and Van Dijk (2009) write that differences are smaller than expected and that “Chinese investments do not differ radically” (p. 90). The magnitude of FDI in SSA and its main sectors can be seen in the figures below. Figure 1 shows the total amount of FDI in SSA (□) and the relative level of Chinese FDI of this total amount (x). Figure 2 shows the yearly amount of Chinese FDI in SSA. From both figures it can be seen that there is a rising trend for the absolute and relative amount of Chinese FDI in SSA. It should be noted here that the peak in 2008 is due to a large acquisition of shares of a South-African bank by a Chinese bank (Economist, 2008). Table 1 shows the countries with the five largest Chinese FDI stocks in SSA in 2012 when the total stock invested amounted to approximately 19.5 billion USD. Figure 3 shows the division by sector of Chinese FDI in Africa, more than half of the amount of FDI was invested in natural resources. These numbers are for Africa as a whole, so they should be

interpreted with care. However, they can be quite reliable for SSA as 88% of Chinese FDI in Africa is directed there (World Bank, 2015).

Figure 1: Total FDI to SSA and Chinese FDI (% of total). Data by UNCTAD

0,0% 2,0% 4,0% 6,0% 8,0% 10,0% 12,0% 14,0% 16,0% 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 FD I ( m illio ns o f $ )

(13)

9 Figure 2: Chinese FDI to SSA. Data by UNCTAD

Country

FDI stock (millions of $)

% of total FDI

South Africa

4,775

_24.6%

Zambia

1,998

_10.3%

Nigeria

1,950

_10.0%

Angola

1,245

_6.4%

Sudan

1,237

_6.4%

Table 1: Chinese FDI stock in 2012. Data by UNCTAD

Figure 3: Chinese FDI to Africa by sector, 2011. (IOSC, 2013)

The most striking characteristic of Chinese FDI in SSA is to be found in the sort of companies that invest abroad. There is no disagreement in the fact that the share in Chinese FDI conducted by SOEs is very large: approximately 70 % (Huang & Wang, 2011; Cheung et al., 2012). The Chinese firms profit in many different ways from the political support of their government (Kaplinsky & Morris, 2009), of which the most important are discussed in the following paragraphs. Other remarkable

0 1000 2000 3000 4000 5000 6000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 FD I ( m illio ns o f $ )

Chinese FDI to Sub-Saharan Africa

Natural resources 53% Manufacturi ng 25% Telecommun ications 18% Others 4%

(14)

10

characteristics of Chinese investments in SSA are the focus on long-term ownership; many are focussed on resources and, again, the integration with the aims of the government in Beijing. The effects that are caused by these special characteristics are discussed in the next section.

The involvement and strategy of the Chinese government is often called the Beijing Consensus (McKinnon, 2010). This is a reference to the Washington Consensus (Williamson, 1990) which summarizes the rules that the IMF and World Bank impose on countries that request to receive aid. The main differences between these two consensuses are the level of interference in national policies and the final goal. Whereas the Washington consensus is based on some level of

interference in a country’s fiscal policy, the Beijing consensus is generally considered not to rely on this. However, the objective of the Beijing consensus is mutual benefit, in contrast to the -claimed- selflessness of the Washington consensus (McKinnon, 2010). As in all diplomatic relations, China adheres to its five principles of peaceful coexistence (Xi, 2006) in its relations with African countries. These five principles were also incorporated in the general principles and objectives of China’s African policy (FMPRC, 2006), which are: sincerity, friendship and equality; mutual benefit, reciprocity and common prosperity; mutual support and close co-ordination; learning from each other and seeking common development; and the one-China principle.

Chinese firms are stimulated by several government programmes and initiatives, for example the ‘go global’ strategy, the Forum on Chinese African Cooperation (FOCAC) and the creation of Special Economic Zones (SEZs). The ‘go global’ strategy was introduced in the tenth five-year plan (2001-2005) and was also included in subsequent five-year plans (Davies, 2013). The strategy consists of, among others, streamlining approval procedures and foreign exchange controls, identifying promising sectors and countries for OFDI and supporting outgoing firms through state-controlled banks and insurance agencies (OECD, 2008). Chinese companies are also able to obtain very low-cost loans from the policy bank EXIM (Kaplinsky et al., 2010; Sauvant & Chen, 2014). UNCTAD (2010) reports that because of (indirect) subsidies that Chinese companies receive from their government, domestic -African- companies have a competitive disadvantage.

In 2007 the China-Africa Development Fund (CAD Fund) was created by the China Development Bank (Renard, 2011). This fund, which has the goal of promoting Chinese investment in Africa, initially contained US$1 billion and is planned to eventually reach US$5 billion (www.cadfund.com). In 2009 this fund was responsible for 30% of Chinese investments in Africa (Weisbrod & Whalley, 2012). This is an example of the long-term view that the Chinese government has for the (investment)

relationships with Africa (Weisbrod & Whalley, 2012), which will be further discussed below. The FOCAC is a forum consisting of 49 African member states, China and the African Union, which holds triennial summits with heads-of-state (McDonald, 2012). In these meetings the development of the inter-governmental relations and certain goals are central. Many authors mention the bundling of trade, aid and investments (Kaplinsky & Morris, 2009; Weisbrod & Whalley, 2012) as typical for the Chinese relationship with Africa. Whereas most other source-countries do not use this method, the provision of Chinese aid for development projects is tied to giving investment

opportunities (Kolstad & Wiig, 2011). The projects for which aid-funding is provided are usually carried out by Chinese SOEs.

The creation of SEZs in China has been a big driver for Chinese economic growth in the last decades and now it ‘exports’ this experience through an overseas zone program, which is supposed to be

(15)

11

mutually beneficial (Bräutigam & Tang, 2014). They write that in 2013 the total number of government-approved zones was 19, of which many were financed by the CAD Fund. These

economic and trade cooperation zones are clustered forms of related industries and are sometimes created without the implicit approval of the Chinese government (Bräutigam, 2011).

Another main difference between Chinese and non-Chinese FDI is the timespan of the relationship. Whereas Western and Japanese investors, which are usually privately owned, have a short time-horizon, the Chinese SOEs operate with much longer time-horizons (Kaplinsky et al., 2010). In general, Western and Japanese investors in SSA are driven by shareholder-value; this can lead to a focus on short-term profit and highly risk-averse behaviour (Kaplinsky & Morris, 2009). Chinese firms -especially SOEs- have a different focus and are less risk-averse due to their access to cheap capital. According to many authors (Besada, Wang & Whalley, 2008; Kaplinsky & Morris, 2009; Kragelund & Van Dijk, 2009) these long-term relationships are mainly to secure strategic mineral resources reserves in Africa. Another difference between Western firms and their Chinese counterparts is that Chinese investors are less constrained by social and environmental constraints (Kaplinsky & Morris, 2009). The 2010 report of UNCTAD mentions that Chinese investments in Africa are characterized by a high level of inputs from China.

Many articles have been written about the determinants of Chinese FDI in SSA. Several of these articles conclude that Chinese FDI is attracted to countries with high political risk and weak

institutions (Kolstad & Wiig, 2011; Allard, 2012; Cheung et al., 2012). Some authors (De Haan, 2011; Allard, 2012) also find that the level of Chinese FDI to a country is higher when the country has natural resources. But Kolstad and Wiig (2011) note that this is probably a determinant for investment from every source country.

Gu (2009) and Weisbrod and Whalley (2012) both mention that Chinese firms manufacturing in Sub-Saharan Africa can be attracted by the possibility for ‘quota-hopping’ to take advantage of the African Growth and Opportunity Act and the Everything But Arms initiative.

Kaplinsky and Morris (2009) distinguish four types of Chinese investors in SSA, which have different times-of-entry. These are -in order of entrance and from large to small- central government SOEs, provincial government SOEs, private sector firms incorporated in China and Chinese private sector firms incorporated in SSA only. According to Farrell and Lin (2013) privately Chinese owned

enterprises act more alike to non-Chinese enterprises. Combining these two conclusions would lead to the expectation that the effects on HDI and GDP will become more equal as time passes.

Effects of Chinese FDI on Sub-Saharan Africa

The special characteristics of Chinese FDI in SSA are expected to have distinctive effects on the development of SSA economies. These effects, both positive and negative, will be discussed in this paragraph through an overview of both theoretical and empirical papers. The most discussed positive effects are the following. Through many big Chinese projects the infrastructure in many African countries has improved (The Economist, 2011). Because of closer ties to China, export to the Chinese market has increased, and one could expect the productivity to rise when cooperating with a big, successful economy. The long-term contracts for natural resources are a valuable source of financing for these poor African nations (Ahmed, Cheng & Messinis, 2011). Two different reasons for African governments to prefer Chinese investments over others are the following. African countries might be restrained to accept investments from their former colonizers. Also the involvement of

(16)

12

Western countries in local politics through the support of nongovernmental organisations is in big contrast to the principles of China -according to the Chinese government-: ‘unconditional aid’ and ‘refraining from intervening in local politics’ (Van Dijk, 2009).

Possible negative effects are summarized here. Because Chinese investments typically use Chinese workers, this might crowd out the African industry (Cheung et al., 2012). Another consequence of the use of Chinese workers is that the African workforce does not learn to produce more efficient. Because of the dependency on resource-exports, the Dutch disease is also a possible danger for some African economies; although this is not directly linked with FDI. Other negative effects could arise from the possible lack of attention for human rights, political reforms and protection of the environment by Chinese investors (Mol, 2011).

As with most investment decisions, the main objective of the Chinese investments in SSA is self-interest for China. However, this does not mean that SSA countries cannot profit from these investments. China’s official driver for its economic relations with Africa is that it pursues a win-win strategy and respects African countries as equal trading partners (SCIO, 2013). According to many authors (Kaplinsky & Morris, 2009; Kaplinsky et al., 2010; UNECA, 2010; Klaver & Trebilcock, 2011; Allard, 2012) Chinese investments in SSA can be profitable for both parties if African governments develop a strong policy framework. However, because of China’s non-interference policy corrupt African government can be held in power (Klaver & Trebilcock, 2011). Examples of such

governments are Sudan’s and Zimbabwe’s, both reigned by dictators. Mlachila and Takebe (2011) investigate the effects of Brazilian, Russian, Indian and Chinese (BRIC) FDI on low-income countries. They show that especially Sudan -which has been cut off from many sources of financing- has profited from Chinese FDI. Over half of Sudan’s government revenue came from the oil sector in 2008; this is mainly due to Chinese public investments.

As many Chinese investments take place through permanent entry modes, this long-term relationship can be profitable for both parties. The expected profitability lies in the fact that the investments can be taken into account for long-term planning (Schiere, 2011).

Klaver and Trebilcock (2011) conclude that Chinese FDI is not unambiguously advantageous for SSA. They reason that Chinese FDI helps the development of Africa by improving the infrastructure and the ability to extract resources, employing Africans and developing Africans. Nevertheless, Chinese FDI may not provide many positive spillovers to Africa, infrastructure projects can have high costs and the Dutch Disease can be caused. The authors write that several drawbacks can be redressed through good governance.

Weisbrod and Whalley (2012) test the contribution of Chinese FDI to Africa’s pre crisis growth surge. They use data on 13 countries, which account for 92% of Chinese FDI flows and 78% of SSA GDP between 2003 and 2009. Through a Solow growth accounting method, using capital stock, workforce and factor share data by country they conclude that without Chinese FDI, GDP growth in these countries would have been lower. Between 2003 and 2009 the GDP growth attributable to Chinese FDI varied between 1.9% in Zambia and 0.04% in Angola.

Renard (2011) analyses the different impacts that China has on Africa through several channels. Concerning FDI, the paper shows that there are many potential benefits to Chinese investments. Chinese investments in agriculture are very important to African countries with food shortages as

(17)

13

they can transfer technology and knowledge. Investments in natural resource sectors can have positive and negative effects caused by the accompanying service sector. However, the main negative effects of large investments in the natural resource sector are the often associated lack of transparency, corruption and rent-seeking. Renard (2011) recommends African governments to be aware of the negative effect of high dependence on resources, the Dutch disease. They can reap more benefits of the Chinese FDI by establishing strong, democratic institutions, cautious fiscal policies and by allocating a part of the revenues to the strengthening of non-tradable sectors. She also describes some cases of profitable projects in manufacturing which would not have been undertaken by Western investors, for example a cigarette-factory in Zimbabwe which improved the value added of its exports comparing with unprocessed tobacco. The author also expects that Chinese FDI will attract more FDI from developing countries to Africa.

Busse, Erdogan and Mühlen (2014) investigate the impact of Chinese activities in SSA through panel data and a Solow-type growth model. They do not find significant effects from FDI on GDP per capita growth in 43 SSA countries in their main regression. However, it should be noted here that the -insignificant- results were higher for Chinese FDI than FDI from the rest of the world. Through a model which uses three-year average data they find a positive effect, significant at the 10% level, of Chinese FDI.

The International Monetary Fund (2011) outlook finds that Chinese FDI in SSA is increasingly focussed on manufacturing, not only on resources. This shift is expected to turn out good for SSA as more positive spillovers can be expected from FDI in manufacturing, as shown by Aykut and Sayek (2007).

(18)

14

Empirical analysis

In this section an empirical analysis will be conducted and discussed. The chapter will be structured as follows. First the methodology will be described by creating the regression formula, then the data will be described as well as discussed why these data have been chosen and following that the regression results will be discussed and linked to previous research.

The methodology

To investigate the effects of Chinese FDI on the development of SSA, a Solow-type growth model will be used in this paper. As this paper is focused on both the human and economic development effect of FDI, the model is constructed for two different dependent variables. One regression will use HDI as its dependent variable whereas the other will use GDP per capita.

The choices of independent variables and control variables are based on evidence found in the literature review and will be specified in the data description.

The model used is based on the Solow-growth model from 1956 and then further analysed through Mankiw et al. (1992). Subsequently, following Islam (1995), the model is notated in such a way that allows a panel data regression.

(1) 𝑌𝑌𝑡𝑡 = 𝐾𝐾𝑡𝑡𝛼𝛼[𝐴𝐴𝑡𝑡𝐿𝐿𝑡𝑡]1−𝛼𝛼, 𝐿𝐿𝑡𝑡 = 𝐿𝐿0𝑒𝑒𝑛𝑛𝑡𝑡, 𝐴𝐴𝑡𝑡 = 𝐴𝐴0𝑒𝑒𝑔𝑔𝑡𝑡

This model is based on a Cobb-Douglas production function in which labour and technology grow by respectively population growth rate n and the constant rate g.

Now, define k as the stock of capital per effective unit of labour: 𝑘𝑘� =_{𝐴𝐴𝐴𝐴}𝐾𝐾 And y as the level of output per effective unit of labour: 𝑦𝑦� =_{𝐴𝐴𝐴𝐴}𝑌𝑌. Now production function (1) can be rewritten as

(2) 𝑦𝑦�𝑡𝑡= 𝑘𝑘�𝑡𝑡𝛼𝛼.

Because effective units of labour grow at the rate n+g, the depreciation rate of capital is δ and the economy has a savings rate of s, the change in capital per effective unit of labour can be described by the following relationship: 𝑘𝑘̇𝑡𝑡 = 𝑠𝑠𝑦𝑦�𝑡𝑡− (𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿)𝑘𝑘𝑡𝑡

Which, after substituting (2), can be rewritten as : 𝑘𝑘�̇𝑡𝑡 = 𝑠𝑠𝑘𝑘�𝑡𝑡𝛼𝛼− (𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿)𝑘𝑘𝑡𝑡.

In the steady-state, the relative capital stock is on a stable growth path so 𝑠𝑠𝑘𝑘�𝑡𝑡𝛼𝛼= (𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿)𝑘𝑘�𝑡𝑡

Solving this for 𝑘𝑘�, leads to a steady-state value of: 𝑘𝑘�∗_{= �} 𝑠𝑠 (𝑛𝑛+𝑔𝑔+𝛿𝛿)�

1 1−𝛼𝛼

After substituting this in production function (2), it becomes 𝑦𝑦�𝑡𝑡 = �_{(𝑛𝑛+𝑔𝑔+𝛿𝛿)}𝑠𝑠 �

𝛼𝛼 1−𝛼𝛼

Rewriting this as income per capita gives: 𝑌𝑌𝑡𝑡

𝐴𝐴𝑡𝑡= 𝐴𝐴𝑡𝑡�

𝑠𝑠 (𝑛𝑛+𝑔𝑔+𝛿𝛿)�

𝛼𝛼 1−𝛼𝛼

After taking logs: (3) ln �𝑌𝑌𝑡𝑡 𝐴𝐴𝑡𝑡� = ln(𝐴𝐴0) + 𝑔𝑔𝑔𝑔 + 𝛼𝛼 1−𝛼𝛼ln(𝑠𝑠) − 𝛼𝛼 1−𝛼𝛼ln(𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿)

Here, the term 𝐴𝐴0 is removed from the function by Mankiw et al. (1992) because it may reflect many

different endowments -e.g. technology, knowledge, climate, historical diplomatic relationships- which may differ across countries. They assume that ln(𝐴𝐴0) = 𝛼𝛼 + 𝜖𝜖 where α is some constant, also

(19)

15 (4) ln �𝑌𝑌_𝐴𝐴� = 𝛼𝛼 +_1−𝛼𝛼𝛼𝛼 ln(𝑠𝑠) −_1−𝛼𝛼𝛼𝛼 ln(𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿) + 𝜖𝜖

Although this formula already links income per capita to the savings rate and population growth, it cannot be assumed that s and n are independent of ϵ. Because this formula describes the steady-state of an economy, it should be adapted in such a way that it also considers the behaviour out of the steady-state. Following Islam (1995), 𝑦𝑦�∗_{depicts the steady state level of income per effective}

unit of labour and 𝑦𝑦�𝑡𝑡 its value in year t. According to Islam (1995) the rate of convergence to the

steady-state is equal to 𝜆𝜆 = (𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿)(1 − 𝛼𝛼) And the pace of convergence is given by d ln(𝑦𝑦�𝑡𝑡)

𝑑𝑑𝑡𝑡 = 𝜆𝜆[ln(𝑦𝑦�∗) − ln(𝑦𝑦�𝑡𝑡)].

This in turn implies that ln(𝑦𝑦�𝑡𝑡) = �1 − 𝑒𝑒−𝜆𝜆� ln(𝑦𝑦�∗) + 𝑒𝑒−𝜆𝜆ln(𝑦𝑦�𝑡𝑡−1)

After rearranging and substituting for 𝑦𝑦�∗_{it shows the growth of income per effective unit of labour:}

(5) ln(𝑦𝑦�𝑡𝑡) − ln(𝑦𝑦�𝑡𝑡−1) = �1 − 𝑒𝑒−𝜆𝜆�_1−𝛼𝛼𝛼𝛼 ln(𝑠𝑠) − �1 − 𝑒𝑒−𝜆𝜆�_1−𝛼𝛼𝛼𝛼 ln(𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿) − �1 − 𝑒𝑒−𝜆𝜆� ln(𝑦𝑦�𝑡𝑡−1).

However, this paper is interested in the income per capita, y, so the equation (4) needs to be reformulated to include y. After substituting ln(𝑦𝑦�𝑡𝑡) = ln(𝑦𝑦𝑡𝑡) − ln(𝐴𝐴0) − 𝑔𝑔𝑔𝑔 and moving ln(𝑦𝑦𝑡𝑡−1) to

the right-hand-side of (4), this gives:

(6) ln(𝑦𝑦𝑡𝑡) = �1 − 𝑒𝑒−𝜆𝜆�_1−𝛼𝛼𝛼𝛼 ln(𝑠𝑠) − �1 − 𝑒𝑒−𝜆𝜆�_1−𝛼𝛼𝛼𝛼 ln(𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿) + 𝑒𝑒−𝜆𝜆ln(𝑦𝑦𝑡𝑡−1) + �1 −

𝑒𝑒−𝜆𝜆_{� ln(𝐴𝐴}

0) + 𝑔𝑔�𝑔𝑔 − 𝑒𝑒−𝜆𝜆(𝑔𝑔 − 1)�

In this dynamic panel data model the terms �1 − 𝑒𝑒−𝜆𝜆_{� ln(𝐴𝐴}

0) and 𝑔𝑔�𝑔𝑔 − 𝑒𝑒−𝜆𝜆(𝑔𝑔 − 1)� are

respectively the country- and time-specific terms. Now it is possible to rewrite this model in a way that makes it possible to do a regression, based on the conventional notation used by Islam (1995): (7) 𝑦𝑦𝑖𝑖𝑡𝑡 = 𝛾𝛾𝛾𝛾𝑦𝑦𝑖𝑖,𝑡𝑡−1+ ∑𝑗𝑗=12 𝛾𝛾𝑗𝑗𝑥𝑥𝑖𝑖𝑡𝑡𝑗𝑗 + 𝜂𝜂𝑡𝑡+ 𝜇𝜇𝑖𝑖+ 𝜈𝜈𝑖𝑖𝑡𝑡.

Where 𝑦𝑦𝑖𝑖𝑡𝑡 = ln(𝑦𝑦𝑖𝑖𝑡𝑡), 𝑦𝑦𝑖𝑖,𝑡𝑡−1= ln�𝑦𝑦𝑖𝑖,𝑡𝑡−1�, 𝑥𝑥𝑖𝑖𝑡𝑡1 = ln(𝑠𝑠), 𝑥𝑥𝑖𝑖𝑡𝑡2 = ln(𝑛𝑛 + 𝑔𝑔 + 𝛿𝛿).

This gives us the regression formula:

(8) ln(𝑦𝑦𝑖𝑖𝑡𝑡) = 𝛼𝛼 ln�𝑦𝑦𝑖𝑖,𝑡𝑡−1� + 𝛾𝛾 ln(𝑠𝑠𝑖𝑖𝑡𝑡) + 𝛾𝛾 ln(𝑛𝑛𝑖𝑖𝑡𝑡+ 𝑔𝑔 + 𝛿𝛿) + 𝜂𝜂𝑡𝑡+ 𝜇𝜇𝑖𝑖+ 𝜈𝜈𝑖𝑖𝑡𝑡

This formula states that the logarithmic income per capita of country i in year t is determined by the logarithmic income per capita in year t-1; the logarithm of the savings rate; the logarithm of

population growth, technology growth and depreciation; time-(𝜂𝜂𝑡𝑡) and country-specific (𝜇𝜇𝑖𝑖) terms;

and at last other specific terms for country i in year t (𝜈𝜈𝑖𝑖𝑡𝑡). This last term can be replaced by

explanatory variables this paper is focussing on, namely FDI with different origins, and control variables. After adding the variables shown in the table, which are further explained in the next section, the complete regression formula looks like this:

(9) ln(𝑦𝑦𝑖𝑖𝑡𝑡) = 𝛼𝛼1ln�𝑦𝑦𝑖𝑖,𝑡𝑡−1� + 𝛼𝛼2ln(𝑠𝑠𝑖𝑖𝑡𝑡) + 𝛼𝛼3ln(𝑛𝑛𝑖𝑖𝑡𝑡+ 𝑔𝑔 + 𝛿𝛿) + 𝛼𝛼4𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖𝑡𝑡𝐶𝐶 + 𝛼𝛼5𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖𝑡𝑡𝑅𝑅𝑅𝑅𝑅𝑅+ 𝛼𝛼6𝜋𝜋𝑖𝑖𝑡𝑡+

𝛼𝛼7ln(𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑒𝑒𝑖𝑖𝑡𝑡) + 𝛼𝛼8𝐹𝐹𝐹𝐹𝑛𝑛𝑖𝑖𝑡𝑡+ 𝛼𝛼9𝐹𝐹𝑛𝑛𝑠𝑠𝑔𝑔𝑖𝑖𝑡𝑡+ 𝛼𝛼10𝑊𝑊𝑇𝑇𝑇𝑇𝑖𝑖𝑡𝑡+ 𝜂𝜂𝑡𝑡+ 𝜇𝜇𝑖𝑖+ 𝜀𝜀𝑖𝑖𝑡𝑡

Of the added variables, a quick analysis of the data reveals that trade openness is the only one that a logarithmic relation with the dependent variable, hence the ln(Trade) in the formula.

(20)

16

Dependent variables

𝑦𝑦𝑖𝑖𝑡𝑡 Gross Domestic Product per capita of country i in year t (constant 2005 $). World Bank

𝐻𝐻𝐹𝐹𝐹𝐹𝑖𝑖𝑡𝑡 Human Development Index. UNDP

Explanatory variables

𝑠𝑠𝑖𝑖𝑡𝑡 Gross Capital Formation, % of GDP. World Development Indicators (WDI) by World Bank

𝑛𝑛𝑖𝑖𝑡𝑡 Population growth rate in %. WDI

𝑔𝑔 Technology growth Constant over time and combined equal to

0.05 according to Mankiw et al. (1992) 𝛿𝛿 Depreciation rate

𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖𝑡𝑡𝐶𝐶 Inflows of FDI from China, % of GDP. UNCTAD and WDI

𝐹𝐹𝐹𝐹𝐹𝐹𝑖𝑖𝑡𝑡𝑅𝑅𝑅𝑅𝑅𝑅 Inflows of FDI not from China (rest of the world) Total FDI less Chinese FDI, % of GDP.

UNCTAD and WDI

Control variables

𝜋𝜋𝑖𝑖𝑡𝑡 Inflation, GDP deflator, annual change in %. WDI

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑒𝑒𝑖𝑖𝑡𝑡 Trade Openness: Import and Export of goods and services, % of GDP. World Bank

𝐹𝐹𝐹𝐹𝑛𝑛𝑖𝑖𝑡𝑡 Financial Market Development. Domestic credit provided by financial sector, % of GDP.

IMF

𝐹𝐹𝑛𝑛𝑠𝑠𝑔𝑔𝑖𝑖𝑡𝑡 Quality of Institutions: Rule-of-Law index. World Governance Indicators by World Bank

𝑊𝑊𝑇𝑇𝑇𝑇𝑖𝑖𝑡𝑡 Battle Deaths. WDI

𝜂𝜂𝑡𝑡 Unobserved period-specific effects affecting all countries

𝜇𝜇𝑖𝑖 Unobserved country-specific effects

𝜀𝜀𝑖𝑖𝑡𝑡 independent and identically distributed error term

To examine how the economic and human development in SSA countries is influenced by Chinese FDI and FDI originating from the rest of the world this paper uses different techniques to estimate the regression model. The main explanatory variables are first tested separately and the model is then extended by adding more possible explanatory and control variables.

Data description

The choices for the variables that are used in the empirical analysis of this thesis, listed in the table above, are all based on evidence found in the literature review. These variables have been chosen because they have been proven to influence economic or human development in empirical literature or according to growth theory. This section is built up as follows. First, the dependent variables are described; following this the main explanatory variables are discussed after which the control variables are explained. Tables 2 and 3 in the appendix show the correlations and a summary of the data.

Dependent variables

The variable this thesis is especially interested in is the coefficient that shows the effect of FDI with different origins on the human development of a sub-Sahara African country, the HDI. This Human Development Index has been developed by the United Nations Development Programme (UNDP) and measures the development of a country on multiple equally-weighted dimensions. These are: economic development measured by GDP per capita; development of knowledge measured by gross enrolment ratio and adult literacy rate; and health which is measured by life expectancy at birth (Anand & Sen, 1994). The HDI is widely used in development economics because it makes it possible to compare almost all countries in the world with a clear and objective measure of well-being (Lindauer, Perkins & Radelet, 2006). In the regression, the natural logarithm of HDI is used.

(21)

17

The other dependent variable which is tested in this thesis is the GDP per capita. This number is widely used in research papers as a measure of the standard of living (Sala-i-Martin, 2002) thus making it possible to compare results of this thesis with other papers. Data for GDP per capita are gathered from the World Development Indicators (WDI) database by the World Bank and are constant 2005 $. Following previous work on growth regressions, the natural logarithm of GDP per capita will be used.

Explanatory variables

The main explanatory variables this thesis is interested in are the value of FDI with different origins. Data on Chinese FDI are obtained from the UNCTAD bilateral FDI database, which are in turn based on data from the Chinese Ministry of Commerce (MOFCOM). These data are available from 2003 onward, before that year data were published by MOFCOM, as ‘Approved overseas investment data’. However, these data were released by the Chinese government, not approved of by and according to the IMF-OECD format (Cheung et al, 2012). This makes them less reliable and comparable to the data provided and checked by UNCTAD. This is why the analysis in this thesis only contains UNCTAD data from 2003 to 2012. A drawback of data on Chinese FDI which is still present in the data used is the investment-scheme through Hong Kong described by Wang et al. (2009). This probably leads to a negative bias on the Chinese FDI data. The variable on non-Chinese FDI flows is calculated by

subtracting the amount of Chinese FDI from the amount of total FDI to an SSA country. Data on the latter are obtained from the WDI database. Both data for FDI show the relative inflows of FDI in the host countries in SSA, measured in percentage of the country’s GDP. Both Chinese and FDI from the rest of the world are expected to have a positive effect on the HDI. This expectation is based on previous empirical work by, among others, Reiter and Steensma (2010). Because of the indications found in the literature review that Chinese FDI is regulated on a state-level, focussed on the long-term and is sometimes coupled with aid, it is expected that its effect on GDP per capita and HDI is possibly higher than the effect from non-Chinese FDI. It should be noted here that if the empirical results show that there is no difference in the effects between the FDI’s with different origin, that this is also a meaningful result.

The following explanatory variables are derived in the previous section that explains the regression formula and are very common in growth-theory. The savings rate is depicted by the gross capital formation (GCF) as a percentage of GDP which can be seen as the level of domestic investment. It is expected that the natural logarithm of the relative level of GCF has a positive effect on GDP per capita and HDI. The next natural logarithm from the Solow-model is n+g+δ. As technologic growth and capital depreciation are assumed to be constant and combined equal to 0.05 (Mankiw et. al, 1992; Islam, 1995), the only variable in this logarithm that is changing over time is population growth. This variable is depicted as a percentage and obtained from the WDI database. The effect from

n+g+δ on GDP per capita and HDI is expected to be negative. Both these expectations can be seen

from the signs in growth-formula (5).

Control variables

The control variables included in the analysis are chosen because of their effect in previous empirical studies. Inflation, measured as the annual change in GDP deflator, is an indicator for economic stability which is especially strong in times of hyperinflation. The next control variable, trade

(22)

18

openness, is depicted by the summed amount of imports and exports relative to GDP. The reasoning behind this variable, which is expected to have a positive effect, is that a country which is open to the world market has more possibilities for development. Another control variable which has been shown in the literature review to have a positive effect on development either direct or through FDI is financial development. Financial development in this analysis is measured by the domestic credit provided by the local financial sector as a percentage of GDP. The quality of institutions is another variable which can affect GDP per capita and HDI positively. This control variable is measured by the Rule of Law index from the Worldwide Governance Indicators database by the World Bank. The index is comprised of scores between -2.5 and 2.5 indicating the level of confidence and obedience to rules, contract enforcement, property rights et cetera. The final control variable is the amount of battle deaths. This cruel variable from the WDI database is included because it shows whether a country is part of a high-intensity civil or military conflict in that year, which is expected to have a large downward effect on GDP per capita and HDI.

According to the World Bank the Sub-Saharan Africa region consists of 46 countries. This dataset, however consist of 34 countries which are the countries for which UNCTAD has data on Chinese FDI. Not all data is available for all countries for every single year, this makes the panel unbalanced. All observations for the year 2012 have been dropped for Sudan as in the year before Sudan was split in Sudan and the new country South-Sudan. A list of the 34 countries is presented in the appendix. Regression results

In this section the results from different regression are discussed and linked to the research

discussed in the literature review. All regressions have been run in STATA 13, the regression outputs are shown in the appendix.

As mentioned before, this thesis will use a panel data analysis because this fits the data better than using a cross-section regression for the analysis. The main drawback of a cross-section regression is that it does not take into account that countries probably have different levels of technology and preferences that are unobservable (Islam, 1995), leading to omitted variable bias (Hoeffler, 2002). After performing a Hausman test, which compares fixed- and random-effects, it can be concluded that it is appropriate to use a fixed-effects estimation method for the dynamic panel data model. In tables 5 and 6 in the appendix, the results for several OLS fixed-effects estimations can be found. Table 5 shows the effect from different variables on the GDP per capita and table 6 on HDI. Because the fixed-effects method also has some possible weaknesses, results from a system GMM (Blundell & Bond, 1998) estimation are shown in tables 7 and 8.

Column 1 of table 5 shows the results of the regression estimation method on ln(y) containing just the basic Solow model variables. Both the coefficients for savings and the lagged dependent variable are highly significant and positive, just as expected and as theory predicts. This does not apply to the coefficient for the natural logarithm of (n+g+δ), which is positive but insignificant. The R-squared of 0.645 shows that the regression used in column 1 already explains a large part of the within country differences in GDP per capita.

In the next two columns, the main variables of interest are added alternately. In column 4 the model is extended by both variables of interest. From column 2, 3 and 4 it can be seen that FDI originating from China has a highly significant positive effect on GDP per capita. This cannot be said for

(23)

non-19

Chinese FDI, which has an insignificant, slightly positive result. It seems that the relative level of non-Chinese FDI has no effect on the economic development of SSA countries.

The final column shows the estimates of the model after extending it to control for macroeconomic stability, trade openness, development of financial markets, quality of institutions and conflicts. This extension from the model decreases the number of observations with 6% to 304. All but one of the coefficients have the expected signs, with only the coefficient showing the effect of battle deaths on economic development is significant. The coefficient for financial development is highly significant and negative, surprisingly indicating a negative relationship between relative domestic credit and economic growth. Another surprising aspect in table 5 is the R-squared line which does not change a lot between the columns. This indicates that the variables with which the model has been extended do not have a large explanatory effect on differences in growth rates between countries.

Table 6 shows the estimation results according to the same model as table 5, but this time the dependent variable is ln(HDI). As one of the components of HDI is GDP per capita, there are some similarities between the results, but there are also some differences. The most striking difference is that the coefficient for non-Chinese FDI is negative and highly significant in columns 3, 4 and 5. This implicates that a higher relative level of FDI originating in another country than China has a negative effect on the human development in SSA countries. The coefficient for Chinese FDI is positive but not significant.

In column 5 of table 6 the complete model is tested, this gives more significant results than in table 5. In this regression shows a highly significant positive coefficient for institutional quality. This

difference indicates that the level of the rule of law index affects development not measured by GDP, but the other components of HDI, namely education and health.

The R-squared of the model shows that the independent variables explain approximately 92% of the differences in HDI between countries, which is a good indicator of the quality of the model. However, as Nickell (1981) shows, the presence of the lagged dependent variable as an independent variable can lead to inconsistent and biased estimates. According to Hoeffler (2002) this estimated

coefficient of the lagged variable is likely to be biased downwards. In addition to this drawback, this method does not account for the possibility that some explanatory variable are endogenous. As described by Leefmans (2011) the lagged dependent variable is correlated with the error term which leads to dynamic panel bias.

The above concerns can be diminished by using the system generalized methods of moments (GMM) estimation method (Arellano & Bover, 1995; Blundell & Bond, 1998). This system GMM method and the method to implement it in STATA are further described in Roodman (2009). System GMM is an appropriate method to use with these panel data because the number of instruments is smaller than the number of countries.

The results for the system GMM estimation of the model concerning economic development are presented in table 7. In column 1 the basic Solow model is tested using ln(s) as an endogenous variable, following Hoeffler (2002), in the system GMM method. The coefficient for the lagged variable is highly significant and close to zero, indicating that the downward bias from the fixed-effects method has been mitigated. Similar to the findings in the previous method the coefficient for the savings rate proves to be highly significant and positive as expected. Another notable coefficient

(24)

20

throughout the table is the coefficient for the constant which is negative. This might point towards the idea of the Africa dummy which is considered by Barro (1991) and through which he tries to explain the lack of growth in Africa. Moving to the columns with the results for the explanatory variables it clearly shows negative significant coefficients for non-Chinese FDI positive coefficients for Chinese FDI; these are significant in column 2 and 4. The final column shows the estimates for the complete model, these do not differ much from the ones estimated through the OLS fixed-effects method. A remarkable coefficient is the one for inflation, which is positive and highly significant. This is the opposite of the expectation that macroeconomic stability and economic growth would be accompanied by low inflation rates. This is probably due to the twenty-four observations of deflation (negative inflation), which are usually indicators of an economic crisis. The validity of the model is tested by the test for the autocorrelation of first-difference equation

residuals, AR(1) and higher AR(2) which show the results for the null-hypothesis that autocorrelation is present. The system GMM estimator needs high first-order autocorrelation but not for the second-order (Busse et al., 2014). This can only be said for the estimates in column 4. Combining all these results, it can be concluded that the results of column 4 can be trusted. Meaning that there is evidence for significantly different effects from Chinese and non-Chinese FDI on the economic development of SSA countries.

Table 8 shows the results of the system GMM estimations for the natural logarithm of HDI. All coefficients for non-Chinese FDI are negative and significant at the one percent confidence level. The coefficients for FDI originating in China are all positive, but insignificant. When comparing these estimates to the ones derived with the fixed-effects method, it seems they are rather similar. The main differences are to be found in the coefficients for financial market development and

institutional quality; both turned insignificant and close to zero. The results from the AR(1) and AR(2) tests show that the estimations of all model specifications are valid because they clearly indicate high first-order but no second-order autocorrelation.

Previous research

Although the majority of the discussed papers predict or find a positive relationship between FDI and development, the analysis of FDI in SSA only finds this to be true for Chinese FDI. The majority of the FDI, originating from the rest of the world, seems to have no or even a negative effect. It is possible that FDI in general has no positive effect in SSA because certain conditions for a positive FDI-growth nexus are not met in the host countries. These conditions can be financial development (Alfaro et al., 2004), trade openness (Balasubramanyam, 1996) or institutional quality (Fortanier & Maher, 2001).

The expectation, based on the literature review, that Chinese FDI would have a positive effect on GDP is proven to be true. This confirms the results found by Busse et al. (2014) and Weisbrod and Whalley (2011) and the results in this analysis are more significant. According to the country-of-origin literature differences in effect can be caused by distance, comparability and

political/economic ties. As the distance to the rest of the world cannot be specified and to China varies per SSA country, the positive effect of Chinese FDI can be caused by the diplomatic links and maybe by comparability. China is closer, so perhaps more comparable, to SSA economies in terms of development than the largest Western source countries are.

(25)

21

Following the literature on characteristics and possible effects of Chinese FDI in SSA, the positive influence of Chinese FDI on the GDP per capita in SSA economies might be caused by the long-term vision and the high involvement of SOEs. Although many of the discussed papers stated that African governments should do more before they can reap the benefits of Chinese FDI, the estimates in this thesis indicate that these benefits are already earned.

Chinese FDI in Sub-Saharan Africa : theoretical research and an empirical analysis of the effects of Chinese and non-Chinese foreign direct investment on human and economic development