The influence of foreign direct investment on gross capital
formation in Sub Saharan African countries.
- A panel data analysis on the factors driving gross capital formation in 47 Sub Saharan African countries from 2010 until 2019-
MSc Thesis Economic Development and Globalization University of Groningen
Faculty of Economic and Business Anne Eline Nelis
Student number: 2970775 Lager der A 1 9718BJ Groningen Tel: +31(6)42586958 Email: a.e.nelis@student.rug.nl 29-01-2021
Abstract
In this research, the gross capital formation of 47 Sub Saharan African countries has been analyzed. This has been done by looking at the influence foreign direct investment has on the countries' gross capital formation. It can be concluded that the foreign direct investment received by the Sub Saharan African countries in the period between 2010 until 2019 had a positive influence on the countries' gross capital formation. Because corruption is playing a part in some Sub Saharan African countries, the influence of corruption on the effect of foreign investment on the gross capital formation has been taken into account. The level of corruption is indicated with the corruption perception index and used as an interaction variable. The lower the corruption perception index, the more corrupt a country is. A low score on the corruption perception index resulted in a lower effect of foreign direct investment on Sub Saharan African countries' gross capital formation. So, Sub Saharan African countries should have a higher level of corruption perception index. In this way, the influence of foreign direct investment on the gross capital formation would be bigger. In the end, a higher gross capital formation could lead to higher economic growth for the Sub Saharan African countries. The control variables used are: GDP per capita, GDP growth, and the inflation rate, consumer price (annual %).
Table of content
1. Introduction p. 4
2. Literature review p. 7
a. Gross capital formation p. 8
b. Foreign direct investment p. 9
c. The influence of corruption p. 11
3. Hypotheses p. 13
4. Methods and data p. 14
a. Dependent variable p. 15
b. Independent variables p. 16
c. Control variables p. 19
d. Scatterplot between gross capital formation and foreign direct investment p. 21
e. Methodology p. 22
f. Formulas for the research questions p. 23
g. Random effect model p. 25
h. Comparison with fixed effect model p. 25
i. R-squared estimator p. 25
5. Results p. 26
a. Hypothesis 1 p. 26
b. Hypothesis 2 p. 28
c. Cluster robust standard errors p. 29
d. Comparison with fixed effect model p. 30
e. R-squared estimator p. 31
f. Influence of control variables p. 31
6. Discussion p. 32
7. Conclusion p. 32
References p. 35
1. Introduction
The amount of gross capital formation (GCF) is of imperative importance for Sub Saharan African countries, according to Taiwo (2019). Taiwo (2019) claims that GCF is crucial in the Sub Saharan African countries’ economic growth. He stated a positive relationship between GCF and economic growth in a country. Uneze (2013) even argues that some Sub Saharan African countries’ low economic growth is the result of low capital formation. However, according to Oyejide (1999), the average gross domestic investment done by Sub Saharan African countries is lower than other developing countries. This can also be seen in the data provided by the World Bank (2020). It can be seen that the average GCF as a percentage of the gross domestic product (GDP), in the period from 2010 until 2019, in Sub Saharan African is 24.17 per cent compared to the world average of 24.32 per cent. Because the GCF is of imperative importance and is related to economic growth, according to Taiwo (2019), it is interesting to look at why the GCF in Sub Saharan African countries is lower than the world average and what influences this score. The GCF in a country is the fixed assets a county possesses plus the net changes in the level of inventories (Worldbank, 2020). The amount of capital formation in a country should be considered by deciding between the present and future consumption (Taiwo, 2019). According to Lensoy et al. (1975), today’s consumption should be lowered if you want to have more output in the future. This is because the resources should be put into the capital formation for its inhabitants’ future. Lensoy et al. (1975) claim that ‘the desirability of growth turns on the desirability of sacrificing present consumption for the benefit of future generations’. This statement is also in line with the arguments made by Ugwuabe (2014). Ugwuabe (2014) claims that GCF can be invested and used to increase welfare for a country’s inhabitants. This since new factories, equipment, and investments in infrastructure are examples of gross capital formation investments. Saleh (1997) also states that GCF could increase people’s welfare, including investments in people’s education, health, and skills.
reduction in the countries. On top of that, the GDP will grow as a result of the long-term foreign finance, according to Oyejide (1999). Loots et al.(2012) explain that African countries experience a financing gap because of their low current savings compared to their required investment ratio. Receiving FDI can be seen as foreign savings, and the financing gap can become smaller. For the Sub Saharan African countries, FDI is a big part of their GDP. Examples of countries where the FDI as a percentage of the GDP are big are; Congo, Rep. with a percentage of 31.1 in 2019, Mozambique with a percentage of 14.6 in 2019 and the Seychelles with a percentage of 13.9 in 2019 (Worldbank, 2020). It is interesting to see if this large part of FDI on GDP is also having a large impact on the GCF. For the Netherlands the percentage of FDI as percentage of GDP is for example 3.9 percent. This is a much lower amount, and the influence could be less.
Besides the FDI received, the level of democracy in the Sub Saharan African countries is taken into account. According to Taiwo (2019), the level of accountability in a democracy is positively related to the economic growth of a country. Oyejide (1999) claims that corruption could hold on investments due to an increase in costs of doing business in these countries. Foreign investors could also be uncertain about their future returns and therefore hold on their investments in Sub Saharan African countries (Oyejide, 1999). The choice was made to look at the corruption level in the Sub Saharan African countries because corruption has been a primary factor for holding back countries in their economic growth, according to Duri (2020). Duri (2020) claims that when the Sub Saharan African countries’ governments are as the world’s average, their GDP will increase. This can result in a GDP growth of 1 to 2 per cent per year per capita. It has been chosen to make an interaction variable of the level of corruption. This since, the level of corruption is affecting Sub Saharan African’s FDI according to (Oyejide 1999) and also their economic growth (Duri, 2020). It is interesting to see if therefore, the level of corruption is also affecting the influence of FDI on the GCF in the Sub Saharan African countries.
In this research, the first hypothesis claims that there is a positive influence of FDI on the GCF in the Sub Saharan African countries. On top of that, the level of democracy is very important for a country and its GCF. A high level of corruption results in lower levels of GCF. The level of corruption is also negatively affecting the FDI’s influence on the GCF, the second hypothesis claims this. The higher the level of corruption, the lower the influence of FDI on GCF.
Saharan African countries are not experiencing the total factor productivity and economic growth experienced by other countries around the world (Taiwo, 201). Comparing Sub Saharan Africa with other developing countries regarding their economic growth performance showed an outcome of poor performance for Sub Saharan African countries, according to Oyejide (1999). With this research, I have tried to find the reason for the lagging behind of the Sub Saharan African countries compared to the world level. This has been done by looking at the GCF as it can be seen as an investment in a country’s productivity (Taiwo, 2019). Oyejide (1999) mentioned that investments in capital formation could improve the welfare of the Sub Saharan African countries in the longer term. This will lead to an increase in the economic growth of the countries. He states that high investments result in growth in the country's overall economy (Oyejide, 1999). Two main factors that could influence the GCF in a country are researched as possible factors for the lagging behind. As explained in the above section, these factors are the FDI received by the Sub Saharan African countries and the level of corruption. It is expected that FDI positively impacts the GCF in Sub Saharan African countries. On top of this, it is expected that the level of corruption negatively influences FDI’s effect on the GCF.
With this research, I would like to add analysis and literature on the explanation of why Sub Saharan African countries are lagging behind on the level of productivity. Since one of the possible sources, explained by Taiwo (2019), for this lower level of productivity is the GCF, it is taken into account as the main variable. Possible factors influencing the GCF will be researched. This research study will add information to the existing literature in various ways. First of all, it will try to fill the research gap of why Sub Saharan countries lag behind the world regarding their productivity, which we will consider as the GCF. Second of all, it will extend the literature by analyzing the influence of FDI and the level of corruption on the GCF in the Sub Saharan countries. For policymakers in Sub Saharan African countries, this research could be interesting as it tries to find the results of the finance gap explained by Loots et al. (2012) and how this gap can be resolved. After reading this research, the policymakers can look at the factors influencing the GCF in their countries and tackle them. In the end, the Sub Saharan African countries’ welfare could rise due to this, as explained by Oyejide (1999). Uneze (2013) mentioned that it is essential for policymakers to look at the capital formation and promote this to have a higher economic growth.
if FDI received by a Sub Saharan African country positively or negatively influences the GCF in this country. In the second part, there will be looked at the influence corruption has on the effect FDI has on the GCF. This has been done with an interaction term.
Most of the data collected are received from the World Bank databank ‘World Development Indicators’ (Worldbank, 2020). After analyzing the data, different tests have been done to check for heteroskedasticity, the goodness of fit, and significance.
In the first part of this research, the literature overview will be given. In the literature review, the variables used in this research, GCF, FDI and corruption, will be explained. Their importance for the Sub Saharan African countries will be emphasized. The previous research done on these variables will be explained and the view of other authors will be given. Also, the research questions will be described in this part. After the literature review, the hypotheses will be formulated. How the data for the variables have been collected and from which data sources it comes will be explained in the methods and data part. In this part also the control variables used will be described. The control variables used are; GDP per capita, GDP growth rate and the information (consumer prices). Which model has been used to analyze the different hypotheses is explained, and the formulas are given. Furthermore, the outcome of the hypotheses will be given and analyzed in the results, followed up with a discussion where the outcomes of the different hypotheses will be explained. In the end, there will be an overall conclusion on the different research questions and hypotheses. Plus, the limitations of this research and ideas for future research will be discussed.
2. Literature review
Gross capital formation
As already mentioned in the introduction, the GCF in a country is the fixed asset plus the net changes in the level of inventories (Worldbank, 2020). The GCF is an important factor to look at for countries, because it could increase the welfare of its inhabitants (Ugwuabe, 2014). On top of that, it could benefit the productivity in a country as a result of the fixed investments done (Gossel, 2018). Akobeng (2017) states that GCF has a positive influence on poverty. He claims that GCF reduces poverty. For this reduction, an investment rate of 25 per cent over the total GDP is needed. However, he claims that only a couple of Sub Saharan African countries invest 25 per cent of their GDP as gross capital formation. Nonetheless, the GCF is very important for the Sub Saharan African countries as it affects the country's growth rate (Akonbeng, 2017). Perkins et al. (1987) is even claiming that the GCF should be 27 per cent or more to witness economic growth. The investments made due to higher gross capital can lead to a 7 per cent or more increase in the growth rate. He expressed that the total rate of GCF done by a country will determine the economic growth rate in this country. The benchmark of the medium to long term growth rate is 7 per cent, according to Akobeng (2017). Uneze (2013) explains a homogenous bidirectional causality between GCF and the economic growth of a country. This causality means that both variables, gross capital formation and economic growth, are influencing each other. Florin-Marius (2008) states that gross capital formation positively influences a country’s GDP and will result in a higher growth rate of a country. This positive influence of GCF on GDP will, in the end, result in better social cohesion and economic convergence. In line with this argument is Tiawo (2019), which claims that the GCF in Sub Saharan African countries is an important factor for the countries’ economic growth.
Looking at the GCF in Sub Saharan African countries is thus very important since it results in economic growth of the countries (Taiwo, 2019). On top of that, GCF will have a positive influence on the GDP, social cohesion and economic convergence according to Florin-Marius (2008).
Foreign direct investment
countries, including factor endowment differences, market access, and natural resources access. The decision to invest in Sub Saharan African countries is mostly driven by natural resources available in this country (Adegboye, 2020). According to Ugwuegbe (2014), FDI can happen in two different ways. The first way is when foreign firms have acquired domestic firms. The second way is that foreign firms are constructing new production facilities in the home country. The foreign country, which is investing, is looking for a long-term relationship with the country they are investing in (Baiashivili, 2019). Baiashivili (2019) argues that FDI can be defined as a ‘composite bundle’. With this ‘composite bundle’, he means that FDI adds capital stock, new production practices, managerial expertise, new technologies, and innovation skills to the home country. It will help the receiving countries with expanding their foreign market access (Krkoska, 2001). Oyejide (1999) is in line with this and claims that FDI positively influences the receiving country in various ways. Besides the economic influence foreign direct investment is having, FDI is also influencing better access to management techniques, marketing networks, and technology, resulting in raising productivity channels. These improvements could lead to higher incomes and employment and positively boost economic activity (Oyejide, 1999).
Hejazi et al. (2003) claims that FDI received by a country is having a positive impact on the gross fixed capital formation (GFCF) and also on the economic growth of the country. They clarify this argument by saying that FDI is bringing technology that will stimulate the productivity growth in the receiving country. Hejazi et al. (2003) claims that FDI could stimulate intermediate production in the domestic country. This could be seen as intra-firm export opportunities. When there is an increase in the production of intermediate products, it will positively affect domestic production. An increase in domestic production will stimulate the GFCF, according to Hejazi et al. (2003). However, the total impact FDI has on a country will depend on the motivation for making the investments (Hejazi et al. 2003).
formation in countries. If the FDI includes goods and services that are not available in the home country, it is reasonable to assume that the FDI will positively affect capital formation. It will have a positive effect since the goods and services are new in the country, and the inhabitants do not have enough knowledge to produce these goods and services by themselves. Hejazi et al. (2003) agrees with this argument and states that outward foreign direct investment in the service industry positively affects the domestic economy. However, when the FDI involves goods and services already provided by the domestic producers, it could have less effect on capital formation (Agosin et al. 2005). Agosin et al. (2005) stated that FDI and domestic investment complement each other if the FDI is in an undeveloped sector of goods and services. They are substitutes if the FDI is done in a developed sector of goods and services. The paper by Tang et al. (2008) explained that the FDI could outweigh the domestic investments done. According to Tang et al. (2008), this could be the case because multinational enterprises want to have a monopoly in the market and thus outweigh the domestic investments.
For Sub Saharan African countries, it is essential to look at the FDI received as it is a crucial capital source, according to Adjasi et al. (2012). Adjasi et al. (2012) claims that FDI is needed due to the low domestic savings and income levels in the Sub Saharan African countries. The FDI received by these countries will help them to grow economically and invest in themselves. As already explained above by Hejarzi et al. (2003), Adjasi et al. (2012) also confirm that FDI is transferring technology and processes, leading to productivity gains in the receiving country. According to Adjasi et al. (2012), the FDI done in Sub Saharan Africa will contribute to the countries’ economic growth. Because foreign investment is so important for Sub Saharan African countries, their influence on the GCF is interesting to look at. According to Oyejide (1999), Sub Saharan African countries have gone through different FDI stages since 1960. Oyejide (1999) argues the first stage, which was around 1960, was mainly focused on protecting import-substitution industries and exploiting natural resources. The second stage of the FDI inflow was around 1970. During this period, the influence was mostly driven by the commodity booms. The last phase argued by Oyejide (1999) is between 1980 and 1990, where FDI led to structural adjustments and policy reforms. Countries that are mostly investing FDI in Sub Saharan African countries are the European Union, Japan, and the United States of America (Oyejide, 1999).
“What influence does foreign direct investment have on the gross capital formation of Sub Saharan African countries in the period from 2010 until 2019?”
The influence of corruption
How the level of corruption is correlated with the influence of FDI on GCF is also a very interesting factor to look at in Sub Saharan Africa. This is because the level of corruption is relatively high in some Sub Saharan African countries. Also, corruption is one of the biggest challenges some Sub Saharan African countries experience. Therefore, an interaction term regarding corruption has been created. The interaction term of corruption indicates if the influence of FDI on the GCF is changing when the level of corruption is changing.
The challenges regarding corruption could influence different fields of the development of the Sub Saharan African countries. Corruption can affect the fields of economic, social, and political development for the Sub Saharan African countries. Corruption can also be seen as a primary factor for holding back countries in their economic growth (Duri, 202). Furthermore, Sunkanmi et al. (2014) states that there is a causal relationship between corruption and a country’s economic growth. It is even stated that when the governments of Sub Saharan African countries will be brought to the average of the world governments, the GDP of the countries will increase. The change in government standards could have a result of a GDP growth of 1 to 2 per cent per year per capita (Duri, 2020). That the level of corruption is a problem for Sub Saharan African countries is also explained by Gossel (2018). Gossel (2018) explains that the Millenium Development Goals set by the World Bank are hard and unlikely to be achieved as a result of high corruption levels.
are stated in Appendix 1. In this appendix, their scores on the global corruption barometer are stated.
Moreover, Sunkanmi (2014) claims that there is a significant relationship between corruption and GCF. Dao (2008) is in line with this argument and states the importance of measures of instability. One of these measures of instability is corruption. He states that there is a negative relationship between the investment outcome and instability measured. He also claims that uncertainty in a market gives a higher beginning threshold to start investing in this market by foreign firms. High corruption leads to a high uncertainty level in a country. On top of that, are the level of crime and corruption resulting in high losses in the market. These high losses will result in high costs of doing business. In the end, high costs of doing business will negatively affect the gross capital formation in a country. Dao (2008) concludes that governments should be transparent, and no groups of people should be favoured in a country over others. Being more transparent will positively affect the gross capital formation in a country (Dao, 2008). Akobeng (2016) argues that policymakers should incorporate GCF with their rule of law and their administration effectiveness.
Sunkanmi (2014) states that there is a significant relationship between FDI and corruption. He claims that corruption holds back economic growth and thus does not increase the GDP in countries.
Since the level of corruption influences both the GCF and the FDI done and received in Sub Saharan African countries, it is interesting to look at the influence that corruption has on the influence that FDI is having on the GCF. It could be that corruption will disturb this influence or enhance it. Leading to the establishment of the following research question:
“Does corruption have an effect on the influence of FDI on GCF in Sub Saharan African countries, while taking into account the corruption perception index (CPI) in the period from 2010 until 2019?”
3. Hypotheses
The research questions established in the literature review are provided in an overview in this chapter. As explained in the introduction, the GCF is important for Sub Saharan African countries as it is crucial for their economic growth (Taiwo, 2019). However, FDI is also an important factor for the Sub Saharan African countries (Baiahivili, 2019). How FDI influences the GCF in Sub Saharan African countries is being researched by the first research question. This research question is therefore as follows:
“What influence does foreign direct investment have on the gross capital formation of Sub Saharan African countries in the period from 2010 until 2019?”
For this question, a significant positive influence of FDI on the GCF in Sub Saharan African countries is expected. This means that a higher amount of FDI leads to a higher GCF in Sub Saharan African countries. This for a period from 2010 until 2019. As the outcome of this expectation, the following hypothesis has been established:
Hypothesis 1; There is a significant positive influence of foreign direct investment (FDI) on the
Since the level of corruption is high in some Sub Saharan African countries, it is interesting to look at how corruption affects the influence of FDI on GCF (Transparency International, 2020). Therefore, the research question is as follows:
“Does corruption have an effect on the influence of FDI on GCF in Sub Saharan African countries, while taking into account the corruption perception index (CPI) in the period from 2010 until 2019?”
For this research question, an interaction term between the Corruption Perception Index (CPI) and the FDI has been made. A significant negative effect is expected between the interaction variable existing of the FDI and the Corruption Perception Index (CPI) and the dependent variable GCF. This means that a lower score on the CPI leads to a lower influence of FDI on the GCF in Sub Saharan African countries. The lower the score of the CPI, the more corrupt a country is. Leading to the following hypothesis:
Hypothesis 2; Corruption, low CPI score, has a negative effect on the influence foreign direct
investment (FDI) has on the gross capital formation (GCF).
4. Methods and data
The data that has been used in this research is mostly received from the WorldBank (2020) database. All the variables’ data: GCF, FDI, GDP per capita, GDP growth rate, and inflation (consumer prices) are generated from the WorldBank database. Only the data used for the CPI is received from the Transparency International database.
For all the used variables, a period from 2010 until 2019 and all the 47 Sub Saharan African countries have been analyzed. The 47 Sub Saharan African countries that have been analyzed are stated in Appendix 2. A period from 2010 until 2019 has been chosen because the CPI data was available for this timeframe. The CPI data below the year of 2010 was missing data on too many countries and would not give a representative overview of the CPI.
The dependent variable
For this research, the variables GCF and FDI relative to GDP are used. Both variables are used relative to GDP because in this way the comparison with each other can be made. The monetary value that has been used to calculate the GDP is the US dollar for both variables. The data that has been used for the GCF is received from the data bank of the WorldBank and is called ‘Gross capital formation (% of GDP)’ (2020). This data set can be found under the database of the ‘World Development Indicators’. The countries were added under the name of ‘region; Sub- Saharan Africa’. The period that has been selected is the time range from 2010 until 2019. Only for Sao Tome and Principe, no data was available for their ‘Gross capital formation (% of GDP)’ (Worldbank, 2020).
The GCF can be seen as the investments made in fixed assets plus the adjustments made in the level of inventories (Worldbank, 2020). Examples of fixed assets include the manufacturing of railroads and roads, the enhancement of hospitals, schools, and buildings. Most of the time, the fixed assets are divided into two categories. These categories are construction and machinery, and equipment. Construction covers the manufacturing of, for example, roads and communication systems. Whereas machinery and equipment covers, for example, the power generating machines and furniture people are using (National Accounts Statistics, 2007). Inventories can be seen as stocks of goods that firms retain for possible shocks or fluctuations in the sales or production of goods (Worldbank, 2020).
GCF can be calculated with the following formula:
Gross capital formation (GCF) = total value of gross fixed capital formation (GFCF) + changes in inventories + acquisitions (less disposals of valuables) (Eurostat, 2017).
However, there are also some limitations regarding the data gathered for the calculation of GCF. It is hard to ensure that all the data is correct if a country has a corrupt government. It might be that this data is not reliable (Worldbank, 2020).
In the figure below (figure 1), you can see the evolution of the GCF (GCF_n) for all the different Sub Saharan African countries on the vertical axis, and the analyzed time period (Year_n) can be seen on the horizontal axis. Only for Somalia, there was no data available. For the different countries, different fluctuations can be seen. It can be observed that for some countries, a positive fluctuation of the GCF has been experienced. Whereas for others, a decrease in the GCF or stable level has been observed.
Figure 1; development of the gross capital formation in Sub Saharan African countries from 2010 until 2019.
The independent variables
Foreign direct investment
Investments done or shares purchased by a buyer in a foreign company can be called foreign direct investments when the percentage of voting stock and management interest is at least 10 percent in the company. The economy in which the foreign company operates should also be different from the economy in which the buyer is operating (Worldbank, 2020). The total net inflows coming from foreign buyers is divided by the GDP of the receiving country. The total net inflows looks at the new inflows coming into the country but also takes into account the disinvestments (Worldbank, 2020). According to the Worldbank (2020), FDI includes “equity investment, including investment associated with equity that gives rise to control or influence; investment in indirectly influenced or controlled enterprises; investment in fellow enterprises; debt (expect selected debt); and reverse investment” (Worldbank, 2020).
In figure 2, GCF (GCF_n) and the FDI (FDI_n) in the Sub Saharan African countries over the years (Years_n) 2010 until 2019 can be seen. For some countries, for example, Benin and Uganda, the two variables are following the same pattern. However, for other countries, the development of the two variables is very different from each other. For example, in Liberia, the FDI started very high and experienced a downward trend. In comparison, the GCF experienced an opposite direction and saw an increase in this variable.
Corruption perception index
The data set used to calculate the ‘Corruption Perception Index (CPI)’ can be found in the Transparency International database (Transparecy International, 2020). Each Sub Saharan African country has been analyzed individually for the years 2010 until 2019. This is a result of not having the availability of data sets for Sub Saharan Africa as a whole region for all the years. The data sets have been set in an excel sheet with all the countries (Sub Saharan African countries), years (2010-2019), and their CPI score. For some countries, not all the CPI scores for all the years were available. It has also been chosen to begin in 2010 since years below 2010 were missing too much data on the Sub Saharan African countries (Transparency International, 2020). The CPI score is ranking countries based on how 13 surveys and assessments of corruption view the level of corruption in a country. It is mostly looking at corruption in the public sector. Experts and business executives perceive these surveys and assessments. In total, 180 countries have been analyzed by the CPI. Scoring a 100 at the CPI means that the government is very clean and has no corruption. The lowest a country can score is a CPI of 0. This means that the country is extremely corrupt. So the higher a country scores on the CPI, the less corrupt a country is. The average score of the world in 2019 is a CPI score of 43 (Transparency International, 2020). However, Stephenson (2014) claims that the CPI score is not a good indicator of the level of corruption over time. This especially for the years before 2012. Stephenson (2014) explains that the CPI scores of the different countries are not comparable to each other. This is a result of various factors influencing the total CPI score. Different factors could have an influence for each country and in this way, a non-comparable view is created. Per year the factors could also differ he claims (Stephenson, 2014). As Stephenson (2014) claims that most of the unreliable data is from before 2012, the decision has been made to still use the CPI scores to measure the level of corruption. This is because, in this research, we are looking at the years between 2010 and 2019. Thus, most of the data used will be from post-20, which will probably be less problematic, according to Stephenson (2014).
Figure 3; development of the CPI level in Sub Saharan African countries from 2010 until 2019.
The control variables
Control variables are added to the formulas because they could have an influence on the dependent and independent variables. The control variables are held constant; however, it could influence the outcome when they change. It is important to add control variables to see the relationship between the dependent and independent variables. This is because if you do not add the control variables, it might be that your results are not representative. This is since other factors influence your dependent and independent variables that should be taken into account but are now left out (Helmenstine, 2020). For this research, there has been chosen to use three control variables: GDP per capita (current US$), GDP growth rate and the inflation, consumer price (annual %). These three control variables have been chosen because they could influence both the dependent variable ‘GCF’ and the independent variable ‘FDI’ and the interaction term ‘corruption’.
GDP per capita
The variable GDP per capita is calculated by the total amount of gross domestic product (GDP) for a country divided by the midyear population a country is counting. According to the Worldbank (2020), GDP is: GDP= gross value added by all resident producers in the economy +
product taxes - subsidies not included in the value of products.
The formula for GDP per capita is as follows: GDP per capita= total amount of gross
domestic product/ midyear population of a country. GDP per capita can thus be seen as the average
amount of GDP a person has in a specific country (Worldbank, 2020). The average GDP per capita in the Sub Saharan African countries used for this research is; $2504. There has only been an observation of 448 GDP per capita variables. This amount of observation is because not all the countries had data on their GDP per capita for all used years (Worldbank, 2020).
GDP growth rate
The GDP growth rate of a country is calculated by looking at the current year/ period and the previous year/ period. The following formula can be used to calculate the GDP growth rate:
GDP growth rate= (GDP in current period- GDP in previous period)/ GDP in previous period and multiplying this number with 100 (Pal, 2020).
The data for the GDP growth rate is received from the WorldBank data set. The data can be found under the world development indicator ‘GDP growth (% annual)’. All the Sub Saharan African countries were included as well as the years ranging from 2010 until 2019. The only country which had no data on the GDP growth indicator was Somalia (Worldbank, 2020). The average GDP growth rate of the Sub Saharan African countries in the period from 2010 until 2019 is 35.70 per cent. Observation of 448 variables on GDP growth rate has been done. This is because not all countries have data on all the years being observed.
Inflation, consumer prices (annual %)
The inflation rate consumer prices (annual %) looks at the differences between the changes in costs of purchases. It is calculating the changes in costs people experience in purchasing goods or services. These changes could be, for example, calculated yearly (Worldbank, 2020).
until 2019 were received. The countries with no data on their inflation rate were: The Central African Republic, Eritrea, and Somalia (Worldbank, 2020). The average inflation rate calculated by the consumer prices is 7.6 per cent over the Sub Saharan African countries in the time period of 2010 until 2019.
In table 1, the descriptive statistics of the used variables are stated. In this table, the mean of the gross capital formation (GCF) is 24.16 per cent, which is lower than the average of 25 per cent stated by Akobeng (2017). As told in the literature overview, Akobeng (2017) stated that a country should invest 25 per cent of its GDP as gross capital formation to influence the poverty rate positively. That the average gross capital formation in Sub Saharan African is lower compared to the world average level of gross capital formation (GCF) is als been shown in the data provided by the World Bank (2020). The average gross capital formation for the world in the period from 2010 until 2019 is 24.32 per cent (World Bank, 2020). For the foreign direct investment received by Sub Saharan African countries, we can see some enormous outliers. This can be seen when looking at the minimum and maximum level of foreign direct investments. The average level of CPI in Sub Saharan African countries is also lower than the world average of 43. This indicates that there is a higher level of corruption in Sub Saharan African countries compared to the world average.
Variable Obs. Mean Std. Dev. Min. Max.
GCF FDI CPI GDP (cap.) GDP growth Inflation 431 440 460 448 448 419 24.16908 5.383365 31.59565 2504.723 35.70109 7.63081 9.31758 10.40219 12.04483 3549.905 209.8901 22.09201 0 -11.63537 8 234.2356 -46.08212 -60.4964 53.98798 103.3374 66 21711.15 1542.62 379.848
Table 1; descriptive statistics of the used variables
Scatter plot between gross capital formation and foreign direct investment
there is an interesting correlation between the two variables. It is almost a vertical relationship with some outliers. But it is a positive relationship because they are moving a bit to the right. It can be stated that the higher the FDI, the higher the GCF will be. Since the FDI is almost one line, it indicates that the FDI ratio is very close to each other in the Sub Saharan African countries. It can be seen that most of the FDI observed is between +/- 0 and 10. However, for the dependent variable, there is more diversity in the amount of GCF. The highest GCF analyzed is the GCF of Mozambique in 2013. In 2013 Mozambique experienced a GCF of 53.9879. One of the outliers for the FDI variable is Liberia in 2010, which had an FDI of 103.3374.
Figure 4: scatterplot between gross capital formation and foreign direct investment
Methodology
Formulas for the research questions
To analyze the research questions, a random effect model has been used. For a random effect model, the formula is as follows:
𝑌!" = 𝛽1𝑋!"+. . . +𝛽#𝑋#!"+ 𝑢!"
In this formula the 𝑌!" explains the dependent variable. The i= 1,..., N is explaining the units used and t= period used. 𝛽#is defining the coefficient for the independent variables, interaction term, and the control variables. The 𝑋!" is representing the independent variables, interaction term, and control variables (Princeton, 2007). The error term 𝑢!" exists of two parts, namely the individual specific component 𝛼! and the regression random error 𝑒!" (Fingleton, 2020).
Research question 1:
The first research question analyzed the influence FDI has on the GCF in Sub Saharan African. The hypothesis stated that a higher amount of FDI leads to a higher GCF in Sub Saharan African countries in the period of 2010 until 2019. The control variables GDP per capita, GDP growth, and inflation have been used in this equation. It is expected that there is a significant positive effect from FDI on GCF in Sub Saharan African countries. For this research question, a random effect model has been used. To see if this hypothesis is statistically significant, the following equations have been made:
𝐺𝑟𝑜𝑠𝑠 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛!"
= 𝛽1𝐹𝐷𝐼!" + 𝛽2𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎!" + 𝛽3𝐺𝐷𝑃 𝑔𝑟𝑜𝑤𝑡ℎ!" + 𝛽4𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛!" + 𝑢!"
Research question 2:
positively or negatively. The interaction variable ‘FDICPI’ has a total of 432 observations. This number of observations results from not all countries having data on both the variables ‘FDI’ and ‘CPI’. If one country is missing values on one of the variables, no interaction term can be created. The hypothesis expects that CPI has a negative influence on the effect which FDI is having on GCF. The following equation has been made to analyze this relationship:
𝐺𝑟𝑜𝑠𝑠 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛!"
= 𝛽1𝐹𝐷𝐼!" + 𝛽2𝐶𝑃𝐼!" + 𝛽3𝐹𝐷𝐼𝐶𝑃𝐼!"+ 𝛽4𝐺𝐷𝑃 𝑝𝑒𝑟 𝑐𝑎𝑝𝑖𝑡𝑎!" + 𝛽5𝐺𝐷𝑃 𝑔𝑟𝑜𝑤𝑡ℎ!" + 𝛽$𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛!"+ 𝑢!"
In the conceptual model (figure 5), the interaction effect of the CPI on the influence of FDI on the GCF is explained.
Figure 5: Conceptual model influence of the corruption perception index
Random effect model
For the different hypotheses, a random effect model has been used. This type of panel data model has been chosen due to the outcome of a Hausman test done. When analyzing a panel data set, two options can be chosen. These options are a fixed effect model and a random effect model. A Hausman test has been performed to know if we should use a fixed or random effect model. For the random effect model, it is necessary that the error variable has no correlation with the explanatory variable. If the explanatory variable is related to the error term, we could better use a fixed effect model instead of a random effect model. According to Torres-Reyna (2007) the rationale behind the random effect model is ‘the variation across entities is assumed to be random and uncorrelated with the predictor or independent variable included in the model’. In the random effect model, it is possible that the differences between entities influence the dependent variable. Another difference between the fixed effect model and the random effect model is that in a random
effect model, the time invariant variables are able to be included in the model (Torres-Reyna). By the Hausman test, the null hypothesis stated that the random effect model is favored. So, the null hypothesis states that there is no correlation between the error variable and other explanatory variables (Statisticshowto, 2020). The alternative hypothesis is stating that the preferred model will be a fixed effect model. We can check the hypothesis by looking at the outcome of the p-value (prob>chi2). If the outcome of the p-value is lower than the significance level ( ≤0.05) we should reject the null hypothesis and use the fixed effect model. If the outcome of the p-value is larger or equal to the significance level (≥0.05) we have failed to reject the null hypothesis and should use the random effect model. The null hypothesis for a Hausman test will be as following:
H0; the preferred model is the random effect model H1; the preferred model is the fixed effect model
Comparison with fixed effect model
After running the Hausman test, a random effect model was chosen as the best option for both hypotheses. However, it is interesting to see the differences between the fixed effect model and the random effect model. The random effect model assumes no correlation between an individual effect and any regressor. For the fixed effect model, this correlation is possible. The fixed effect model assumes the same constant variances and slopes across the individuals. By the fixed effect model, there is an individual difference in the intercept. (Park, 2011). The formula for the fixed effect model is as follows:
𝑌!" = 𝛽1𝑋!"+. . . +𝛽#𝑋#!"+ 𝛼! + 𝑢!"
The same as in the random effect model is 𝑌!" representing the dependent variable in the formula. The i=1,...,N represents the units used and the t= representing the period used. The𝛽#is representing the coefficient for the independent variable, interaction term and the control variables. The 𝑋!" is explaining the independent variable, interaction term and the control variable. The 𝛼! is in this formula representing the unknown intercept for each unit. The 𝑢!"shows the error term (Princeton, 2007).
R-squared estimator
r-squared looks at the variation in the dependent variable that can be explained by the independent variable (UCLA, 2020). According to Adkins et al. (2011) “the r-squared measures the proportion of sample variation in the dependent variables accounted for by the regression”. Three different types of r-squared can be measured. These types are the overall r-squares, the within r-squared, and the between squared (Stata, 2021). For cross-sectional data, it is assumed that the value of r-squared is small. This, as the variation between the individuals, is difficult to explain, according to Hill et al. (2012). For the fixed effect model, the preferred r-squared measurement is the within r-squared. The within estimator is looking at the time-varying factors in the model. The random effect model can also use the within r-squared measurement as it uses both the within and between information (Stata, 2021). It has also been chosen to look at the overall r-squared for both models. It has been chosen to not use the between r-square since the fixed effect model is only looking at impacts happening within a variable and not between variables (Torres-Reyna, 2007).
5. Results
Hypothesis 1
For analyzing the first hypothesis and looking at the influence of FDI on the GCF in Sub Saharan African countries a total of 43 countries have been taken into account. The corr. gives a value of 0 which assumes that the differences across units are uncorrelated with the regressor. A random effect model has been used since the p-value .118 ≥.05. The null hypothesis of the Hausman test is thus accepted, and a random effect model should be used.
calculated by looking at the standard deviation of the FDI and multiplying this by the coefficient of the formula for the FDI. Since the mean of the GCF is 24.169 (table 1), the new mean would be 24.169 + 1.290 = 25.459. This is a small change and the effect of FDI on GCF is therefore small. The Breasch-Pagan Lagrange multiplier (LM) test has a value of .00 ≤.05.This indicates that there is an insignificant difference between the Sub Saharan African countries and that a random effect regression should be used. Looking at the goodness of fit of the model the rho has a variance value of .695. This means that 69.5 per cent of the total error variance can be explained by the individual specific error variance. The F-test gives also a value of .00 ≤.05.which indicates that the random effect model is an appropriate model to use.
Dependent variable: gross capital formation Without robustness With robustness Foreign direct investment 0.124*** 0.124
(0.0327) (0.0804) GDP per capita 0.00100*** 0.00100** (0.000209) (0.000245) GDP growth 0.00172 0.00172 (0.00539) (0.00161) Inflation -0.0576 -0.0576 (0.0429) (0.0339) _cons 21.25*** 21.25*** (1.382) (1.531) N 395 395 Standard error in parentheses * p<0.05, **p<0.01, ***p<0.001
Hypothesis 2
When testing the second hypothesis, a new interaction variable ‘corruption (FDICPI)’ has been added. This interaction variable looked at the effect corruption (CPI) has on the influence FDI has on the GCF. Hypothesis 2 tests whether the level of corruption has an influence on the impact FDI has on GCF in a Sub Saharan African country. For this hypothesis 43 countries have been analyzed. The corr. has a value of 0 which indicated that the difference across units is uncorrelated with the regressor. After running the Hausman test, a p-value of .253 gives the indication that a random effect model should be used. This since .253≥.05and the preferred model will be a random effect model.
The outcome of the coefficient -.0058 explains that the interaction term is having a negative influence on the effect FDI is having on the GCF in a country. This can be seen in table 3. An increase of one per cent in the interaction term is correlated with a decrease of 0.58 per cent on the influence which FDI is having on the GCF. This means that a higher level of corruption will make the influence of FDI on the GCF in a Sub Saharan African country weaker. However, the coefficient of the interaction term is not statistically significant. The influence of the CPI on the GCF has a significant positive sign at a 0.1 per cent significance level. The outcome of the test expects an increase in the GCF when the corruption in a country is decreasing. This is because the higher the CPI, the lower the level of corruption is. An increase of one per cent of the CPI is correlated with an increase by 27.2 per cent on the GCF. Since the interaction term is not statistically significant, we can not accept the second hypothesis.
The Breusch-Pagan Lagrange multiplier (LM) test gives a p-value of .00 which indicates that we should use a random effect regression instead of a simple OLS regression since .00≤.005. For the second hypothesis the rho value is .688 this indicates that 68.8 percent of the total error variance is explained by the individual specific error variance. The F-test gives a value of .00≤ .005 which confirms the usage of a random effect model.
Dependent variable: gross capital formation Without robustness With robustness Foreign direct investment 0.317** 0.317
(0.123) (0.205)
Corruption perception index 0.272*** 0.272** (0.0704) (0.0910) FDICPI -0.00581 -0.00581 (0.00354) (0.00521) GDP per capita 0.000770*** 0.000770* (0.000226) (0.000391) GDP growth 0.00253 0.00253 (0.00528) (0.00151) Inflation -0.0503 -0.0503 (0.0440) (0.0363) _cons 13.08*** 13.08*** (2.538) (3.142) N 387 387 Standard error in parentheses * p<0.05, **p<0.01, ***p<0.001
Table 3: Influence of corruption on the relationship between gross capital formation and foreign direct investment.
Cluster robust standard errors
robustness’ the coefficients and their significance are stated when there has been looked at the cluster robust standard errors. For the first hypothesis, a difference in significance can be seen at the independent variable ‘FDI’. After applying the cluster robust standard errors the independent variable ‘FDI’ is not significant anymore. When checking for heteroskedasticity the independent variable ‘FDI’ is thus not significant anymore, and the first hypothesis could not be accepted anymore.
For the second hypothesis, there are also some differences. In table 3, the difference between with and without checking for robustness are present. When checking for heteroskedasticity, we can see that the independent variable ‘FDI’ is not significant anymore. The significance of the variable ‘CPI’ changed from a significance at a 0.1 percent significance level to a significance at a 1 percent significance level.
Comparison with fixed effect model
In Appendix 3, the comparison between the fixed and random effect model analyzing hypothesis 1 is stated. It can be seen that the significance level of the variables is not changing. It can be concluded that the independent variable, FDI, still has a significant influence on the dependent variable, GCF, on a 0.1 per cent significance level. However, the level of influence FDI is having on the GCF has decreased. In the fixed effect model, the coefficient of the influence of FDI is .119 whereas in the random effect model, this influence had a coefficient of .124. For the fixed effect model, it can be stated that a one per cent increase in the FDI is correlated with an increase of 11.9 per cent in the GCF instead of the increase of 12.4 per cent in the random effect model. This means that for the fixed effect model the influence of the FDI received on the GCF in a Sub Saharan African country is still significant and that hypothesis 1 can be accepted. However, the influence of FDI on the GCF in the fixed effect model is less than in a random effect model.
random effect model. The coefficient of the CPI is higher in the fixed effect model than in the random effect model. For the fixed effect model an increase in the CPI by one per cent is correlated with an increase of 30.8 per cent on the GCF instead of a 27.2 per cent increase in the random effect model. However, still the interaction variable is not significant so the second hypothesis can not be assumed to be true.
R-squared estimator
In Appendix 5, there has been looked at the differences between the within and overall r-squared of the hypotheses, in both the fixed effect model and the random effect model. It can be seen that when we are looking at the ‘r-squared within’, the values for both the hypotheses are higher in a fixed effect model. For the first hypothesis the value of the ‘r-squared within’ in the fixed effect model is .101 compared to the random effect model with a value of .098. For the second hypothesis the value of the ‘r-squared within’ in a fixed model is also higher with a value of .128 compared to the value of the random effect model of .122. Looking at the ‘r-squared overall’ it is in contrast to the ‘r-squared within’ and the value of the random effect model is higher than the fixed effect model. For the first hypothesis, the value of the ‘r-squared overall’ of the random effect model is .063 compared to a value of .041 for the fixed effect model. It can be stated that for the first hypothesis, in the preferred random effect model, 6.3 per cent of the variation in the dependent variable ‘gross capital formation’ is explained by the independent variable ‘foreign direct investment’. Additionally, for the second hypothesis, the value of the ‘r-squared overall’ in the random effect model, which is .111, is higher than the fixed effect model with a value of .082. For the second hypothesis also the random effect model is the preferred model and 11.1 per cent of the variation in the effect of FDI on GCF can be explained by the interaction term ‘corruption’.
Influence of the control variables
influence which FDI is having on the GCF in Sub Saharan African countries will be lower. The coefficient for the interaction term (FDICPI) in the equation without control variables is -.0063 compared to the interaction term (FDICPI) with control variable which is -.0058. So, without the control variables added, the negative influence that the interaction term ‘corruption’ has on the influence FDI has on the GCF in a Sub Saharan African country is bigger. Corruption is thus disturbing the influence of FDI on the GCF in countries.
6. Discussion
In this chapter, the outcomes of the hypotheses and tests will be discussed. The first hypothesis is accepted after analyzing the random effect model on the variables ‘GCF’ and ‘FDI’. It can be stated that the influence of FDI in Sub Saharan African countries seems to have a positive effect on the level of GCF in these countries. This is also in line with the arguments made by Kroska (2001) and Amighini (2017) in the theoretical framework. Kroska (2001) stated that an increase in FDI would lead to a higher GCF. Amighini (2017) mentioned that FDI has an arm’s length relationship with the host country which in the end leads to an increase in domestic investments. The hypothesis, which states that there is a positive influence between FDI and GCF, can be assumed to be true.
For both hypotheses 1 and 2, the effect of the independent variable and the interaction term on the dependent variable were lower in the fixed effect model than the random effect model.
The outcome of the r-squared is different for the r-squared within compared to the r-squared overall. For the r-squared within, the values of the fixed effect model where higher in both the hypotheses. Whereas, for the r-squared overall, the random effect model’s values were higher in both the hypotheses. A conclusion for the higher r-squared within in the fixed effect model could be that fixed effect models are explaining the variance within a variable. The overall r-squared levels are low and conclude that the variation in the dependent variable ‘GCF’ is not explained much by the independent variable ‘FDI’. This conclusion is the same for the interaction term. The low overall r-squared shows that the interaction term ‘corruption’ does not explain much about the variation in the effect of ‘FDI’ on the ‘GCF’.
The influence of the control variable on the hypothesis is different for hypothesis 1 compared to hypothesis 2. It can be seen that for hypothesis 1, adding the control variables to the equation leads to a higher outcome and thus a higher influence of the independent variable on the dependent variable. This is opposite to hypothesis 2, where the control variables have a decreasing influence on the outcome. When the control variables are added, the influence which the interaction term is having on the relationship between the independent variable and dependent variable is lower than without control variables. It can be seen that the control variables have a bigger effect in hypothesis 1 as this coefficient is increasing.
7. Conclusion
significant so we can not accept the hypothesis regarding the influence of corruption on this relationship. However, for governments in Sub Saharan African countries it would be good if they can be as transparent as possible to attract more FDI. This is in line with the arguments made by Cleeve (2012) regarding investment climates.
With this research, a look at the influence of FDI on GCF in Sub Saharan African countries has been extended. From previous research not a lot of information was known. For further research, it will be interesting to compare Sub Saharan Africa with other parts of the world. In this way, there will be a comparison around the world and the difference between countries or continents will be visible. An example of a comparison could be between Sub Saharan Africa and other parts of the African continent or a comparison with other worldwide continents. Another interesting aspect to do further research in is mentioned by Oyejide (1999), who claims that African wealth owners are not investing in their own countries but in other regions than Sub Saharan Africa. It would be interesting to know their motivation to not invest in Sub Saharan African countries. If these motivations are known, Sub Saharan African countries could do something about this and attract more investors from their own countries and regions.
As with every research, this research has its problems and limitations. First of all, most previous research did not investigate this topic. Therefore, not a lot of literature was available. The available research mostly regarded the variables itself however not the relationship between the variables chosen for this research. Therefore, finding enough information was difficult. Secondly, it would have been interesting to look at a wider time range. However, this was not possible because there was not enough data on the corruption perception index before the year 2010. More data on the level of corruption and previous years would have been interesting. In this way, a wider overview of the Sub Saharan African countries’ development could have been given. As Stephenson (2014) already claimed, is the CPI score maybe not the most reliable indicator regarding the level of corruption in a country. For further research it would be interesting to look at more indicators of the level of corruption to get a better overview.
References:
Adegboye, F. B., Osabohien, R., Olokoyo, F. O., Matthew, O., & Adediran, O. (2020). Institutional quality, foreign direct investment, and economic development in sub-saharan africa. Humanities
and Social Sciences Communications, 7(1). https://doi.org/10.1057/s41599-020-0529-x
Adjasi, C. K. D., Abor, J., Osei, K. A., & Nyavor-Foli, E. E. (2012). Fdi and economic activity in africa: the role of local financial markets. Thunderbird International Business Review, 54(4), 429– 439. https://doi.org/10.1002/tie.21474
Adkins, L. C., & Hill, R. C. (2011). Using stata : for principles of econometrics (4th ed.). Wiley Agosin, M. R., & Machado, R. (2005). Foreign investment in developing countries: does it crowd in domestic investment? Oxford Development Studies, 33(2), 149–162. https://doi.org/10.1080/13600810500137749
Amighini (2017). FDI and capital formation in developing economies: New evidence form industry-level data. National Bureau of economic research, No. 23049, 2-20.
https://www.nber.org/papers/w23049
Akobeng, E. (2017). Gross capital formation, institutions and poverty in sub-saharan africa.
Journal of Economic Policy Reform, 20(2), 136–164. https://doi.org/10.1080/17487870.2015.1128833
Baiashvili, T. & Gattini, L. (2019). Impact of FDI on economic growth: the role of country income levels and institutional strengths. European Investment Bank. 1-25. https://www.eib.org/attachments/efs/economics_working_paper_2020_02_en.pdf
Bartels, F. L., Alladina, S. N., & Lederer, S. (2009). Foreign direct investment in sub-saharan africa: motivating factors and policy issues. Journal of African Business, 10(2), 141–162. https://doi.org/10.1080/15228910903187791
Cleeve, E. (2012). Political and institutional impediments to foreign direct investment inflows to sub-saharan africa. Thunderbird International Business Review, 54(4), 469–477. https://doi.org/10.1002/tie.21477
Duri, J. (2020). Sub-Saharan Africa: Overview of corruption and anti-corruption. Transparency
International. 1-23. https://www.u4.no/publications/sub-saharan-africa-overview-of-corruption-and-anti-corruption.pdf
Eurostat (2016). Glossary: Changes in inventories. Eurostat statistics explained.
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Changes_in_inventories
Eustat (2020). Gross Capital Formation. Eustat; Euskal Estatistika Erakundea- Instituto Vasco de
Estadística. https://en.eustat.eus/documentos/elem_2523/definicion.html
Florin-Marius, Pavelescu (2008). Gross capital formation and economic growth during early 2000’s in eu-member and candidates states. Revista Romana De Economie, 26(1), 166–179. http://www.revecon.ro/articles/2008-1/2008-1-9.pdf
Gossel, S. J. (2018). Fdi, democracy and corruption in sub-saharan africa. Journal of Policy
Modeling, 40(4), 647–662. https://doi.org/10.1016/j.jpolmod.2018.04.001
Hejazi, W., & Pauly, P. (2003). Motivations for fdi and domestic capital formation. Journal of
International Business Studies, 34(3), 282–289. https://link-springer-com.proxy- ub.rug.nl/search?dc.title=Motivations+for+FDI+and+Domestic+Capital+Formation&date-facet-mode=between&facet-start-year=2003&facet-end-year=2003&dc.creator=hejazi&showAll=true Helmenstine, A. M. (2020). The role of a controlled variable in an experiment. ThoughtCo. https://www.thoughtco.com/controlled-variable-definition-609094
Hill, R. C., Griffiths, W. E., & Lim, G. C. (2012). Principles of econometrics (4th ed., international student version). John Wiley & Sons.
Krkoska, L. (2001). Foreign direct investment financing of capital formation in central and eastern Europe. European Bank for reconstruction and development, No. 67. 1-18. https://www.ebrd.com/downloads/research/economics/workingpapers/wp0067.pdf
Loots, E., & Kabundi, A. (2012). Foreign direct investment to africa: trends, dynamics and challenges. South African Journal of Economic and Management Sciences, 15(2), 128–141. https://sajems.org/index.php/sajems/article/view/148
National Accounts Statistics (2007). Capital Formation. National Accounts Statistics- Sources and
Methods. 203-219.
http://mospi.nic.in/sites/default/files/reports_and_publication/statistical_manual/Chapter%2025.p df
Oyejide, T. A. (1999). Taking stock of sustainable development finance in Sub-Saharan Africa.
United Nations Department of Economic and Social Affairs Division for Sustainable Development.
29-38. https://sustainabledevelopment.un.org/content/documents/nairobi_meeting.pdf
Pal. T. (2020). GDP growth rate calculator. Omni calculator.
https://www.omnicalculator.com/finance/gdp-growth#how-to-calculate-economic-growth-rate Park, H. M. (2011). Practical guides to panel data modeling: a step by step analysis using stata. International University of Japan. Public Management & Policy Analysis Program. 2-52. https://www.iuj.ac.jp/faculty/kucc625/method/panel/panel_iuj.pdf
Perkins, D.W., M. Roemer Gillis, and D.R. Snodgrass (1987). Economics of Development, 2nd ed.
W.W. Norton.
Saleh, K (1997). The measurement of gross domestic fixed capital formation in Indonesia. Central
Bureau of Statistics. 2-8. http://www.oecd.org/sdd/na/2666669.pdf
Statisticshowto (2020). Hausman test for endogeneity (Hausman specification test). Statistics how
to. https://www.statisticshowto.com/hausman-test/
Statisticshowto. (2020). Heteroscedasticity: simple definition and examples. Statistics how to. https://www.statisticshowto.com/heteroscedasticity-simple-definition-examples/
Stephenson, M. (2014). Post-2012 CPI scores can’t be compared across time, either. The Global
Anticorruption Blog. https://globalanticorruptionblog.com/2014/10/07/post-2012-cpi-scores-cant-be-compared-across-time-either/
Stata (2021). Xtreg- Fixed, between, and random-effects and population-averaged linear models.
Stata.com https://www.stata.com/manuals13/xtxtreg.pdf
Sunkanmi, O. A., Isola, A. L. (2014) Corruption and Economic Growth in Nigeria. Department
of Economics, Lagos State University Journal of Economics and Sustainable Development. Vol.5,
Taiwo, A. (2019). The fourth industrial revolution and economic growth in Sub-Saharan Africa.
Adeyemi College of Education, Ondo. 109-117. hrcak.srce.hr
Tang, S., Selvanathan, E. A., & Selvanathan, S. (2008). Foreign direct investment, domestic investment and economic growth in china: a time series analysis. World Economy, 31(10), 1292– 1309. https://doi.org/10.1111/j.1467-9701.2008.01129.x
Torres-Reyna, O. (2007). Panel data analysis fixed and random effects using stata (v.4.2).
Princeton University; data and statistical services.
https://www.princeton.edu/~otorres/Panel101.pdf
Transparency International (2020). Global Corruption Barometer; Africa. Transparency
international. https://www.transparency.org/en/gcb/africa/africa-2019/results/nam
UCLA (2020). Regression analysis. Stata annotated output. UCLA Institute for digital research
and education, statistical consulting. https://stats.idre.ucla.edu/stata/output/regression-analysis/
Ugwuegbe, (2014). The Impact of Foreign Direct Investment on Capital Formation in Nigeria: A Co-Integration Approach. International Journal of Economics Finance and Management Science. https://www.researchgate.net/publication/270707557_The_Impact_of_Foreign_Direct_Investme
nt_on_Capital_Formation_in_Nigeria_A_Co-Integration_Approach/link/572d1fab08aeb1c73d11b822/download
Uneze, E. (2013). The Relation Between Capital Formation and Economic Growth: Evidence from Sub-saharan African Countries. Journal of Economic Policy Reform 16 (3): 272–286.
https://www-tandfonline-com.proxy-ub.rug.nl/doi/pdf/10.1080/17487870.2013.799916?needAccess=true
Worldbank (2020). World Development Indicators. World Bank.
https://databank.worldbank.org/source/world-development-indicators
Appendix
Appendix 1: Corruption perception index score and global corruption barometer
Lowest CPI score Global Corruption
Barometer
Highest CPI score Global Corruption
Barometer
Somalia (9) Seychelles (66)
South Sudan (12) + 82% Botswana (61) + 52%
Equatorial Guinea (16) Cabo Verde (58) + 39% Democratic Republic of Congo (18) + 85% Rwanda (53)
Guinea Bissau (18) Namibia (52) + 78%
Appendix 2: Sub Saharan African countries
Angola Congo
(Democratic Republic)
Guinea-Bissau Namibia South Sudan
Benin Congo
(Republic)
Kenya Niger Sudan
Botswana Côte d’Ivoire Lesotho Nigeria Tanzania Burkina Faso Equatorial
Guinea
Liberia Rwanda Togo
Burundi Eritrea Madagascar Sao Tome and Principe
Uganda
Cabo Verde Ethiopia Malawi Senegal Zambia
Cameroon Gabon Mali Seychelles Zimbabwe
Central African Republic
Gambia Mauritania Sierra Leone
Chad Ghana Mauritius Somalia
Appendix 3: Hypothesis 1 with fixed effect model
