The relation between natural resources and income inequality in Sub-Saharan Africa: A difference between having resources and using resources

(1)

The relation between natural resources and income inequality in

Sub-Saharan Africa

A difference between having resources and using resources

Master Thesis International Relations Robin Aarts – S2100819

robin.aarts@hotmail.com Supervisor: Ruben Gonzalez Vicente

July 6th_{, 2018}

Abstract

This thesis focusses on the question of how the dependence on natural resources affects income inequality in Sub-Saharan Africa. There exists a huge debate in literature on the concept of a ‘natural resource curse’, whether this exists or not and what the main reasons behind this curse could be. However, most scholars seem to focus on how natural resources affect economic growth, whereas this thesis analyses the effect on income inequality. After a review of previous research, this thesis will bring forward its own empirical analysis with the application of regression models to investigate the effect of natural resource dependence on income inequality. Since no general link can be derived from this empirical analysis, the thesis focusses on three different countries (Mali, Angola and Zambia) to see how the relationship between natural resources and income inequality works for countries that are similar in their dependence on resources, but different in their level of income inequality. These countries share similarities in certain aspects, such as the importance of geography and economic isolation of natural resources, but they differ in other aspects, such as their colonial background and resource management.

(2)

1. Introduction

Several parts of the developing world are rich in natural resources and even though one might think that these resources form an asset for the development of a country, sometimes they appear to be the opposite. In several instances, the availability of natural resources can have a negative effect on the economic growth and institutional quality of a country (Venables et al., 2007). Moreover, natural resource wealth also seems to be causally linked to civil wars and conflict (Ross, 2004). In that sense, one can question the true effect of natural resources in the underdeveloped world. This paradox of resource-rich countries growing slower than resource-poor countries is known under the term of ‘the natural resource curse’. The processes behind the natural resource curse will be investigated in this thesis with a specific focus on income inequality in Sub-Saharan Africa. The main reason for choosing this geographic area is that Sub-Saharan Africa is the part of the world where a lot of natural resources are available, with 75 percent of its exports being derived from natural resources, but still the poorest countries are situated there (IBIS, 2015). How is this possible and how do natural resources play a role in their process of development?

First, the literature review will analyse the natural resource curse and since not many papers have been written on income inequality, it will mainly focus on economic growth instead. It appears to be the case that scholars hold different views on how natural resources can turn into a curse, rather than into a blessing. Some academics believe it is the type of resource that matters, others think it is mainly the institutional quality of a country that makes the difference, and some see the economic situation as the main factor behind the curse (Auty, 2007; Brunnschweiler and Bulte, 2008; Collier and Goderis, 2008; Diamond and Mosbacher, 2013; Frankel, 2010; Isham et al., 2005; Jensen and Wantchekon, 2005; Papyrakis and Gerlagh, 2004; Van der Ploeg and Poelhekke, 2009). These different perspectives will be brought to the fore and it will then be linked to the main topic of interest, which is that of income inequality. After focussing on the already existing literature, an empirical analysis will be conducted. This research will take all countries of Sub-Saharan Africa as its sample and the effect that will be measured is that of the dependence on natural resources on income inequality. ‘Natural Resources Rent (as a percentage of GDP)’ will be the main explanatory variable and three different measures of income inequality (Gini Index, Palma Ratio, Quintile Ratio) will function as the dependent variable. Several variables that can potentially act as confounding factors are added as control variables to the research, since they may have an effect on income

(4)

inequality. The thesis will then move on to a qualitative analysis, for which Mali, Angola and Zambia have been chosen to act as case studies. These countries have been chosen, since they are similar in their dependence on natural resources as a percentage of their GDP, but they perform differently in their level of income inequality. In the end, the information of the literature review will be combined with the information derived from the quantitative and qualitative analysis and the main research question of this thesis: “In what way does the dependence on natural resources affect income inequality in Sub-Saharan Africa?” will be answered. The hypothesis of this research is that natural resources have an effect on a country’s income inequality and that higher dependence on natural resources will lead to more income inequality. It is also expected that this negative impact is influenced by several factors, for the mere availability of natural resources would not necessarily lead to a negative effect on income inequality.

The answer to this question is relevant because income inequality is a major problem in the underdeveloped world and one that many try to tackle. This thesis contributes to already existing literature because it tries to find out how the dependence on natural resources affects income inequality. As has been stated before, most scholars focus on the effect of natural resources on economic growth, which is different from the focus of this thesis (Atkinson and Hamilton, 2003; Brunnschweiler, 2008; Collier and Goderis, 2008). Papyrakis and Gerlagh (2004) for example investigate how natural resources affect economic growth while using inequality as a transmission channel. Only very limited literature can be found on the relationship between natural resources and income inequality (Fum and Hodler, 2010; Goderis and Malone, 2011; Ross, 2007; Ross, Lujala and Rustad, 2012). Therefore, this thesis will focus on something different than most research that has been conducted on the topic of natural resources so far. This thesis also contributes to knowledge in that it combines both quantitative and qualitative research. One will see that no general conclusions can be derived for the whole region of Sub-Saharan Africa, and therefore, the thesis will introduce several case studies as well.

(5)

2. What are the main drivers behind the natural resource curse?

Scholars hold considerably different views on the topic of the natural resource curse; not only is there a debate on whether or not it exists, but also about the main drivers that could explain why natural resources form a curse, rather than a blessing. Several academics have written about the resource curse and do believe in its existence, but this does not hold for everyone. Brunnschweiler and Bulte (2008), for example, argue that it is the institution that affects the level of dependence, rather than the other way around. In this section of the thesis, different arguments and streams of thought on how natural resources affect a country will be outlined, with a narrow focus on Sub-Saharan Africa. First, scholars who believe the main reason behind the curse is the type of resource will be introduced, which will be followed up by scholars who rather think it is the institutional quality that matters in how resources are used (Jensen and Wantchekon, 2004; Brunnsweiler and Bulte, 2008; Diamond and Mosbacher, 2013; Isham et al., 2005). Lastly, some academics believe it is the economic situation of the country and the term ‘Dutch Disease’ will hereby be introduced (Auty, 2007; Collier and Goderis, 2008; Frankel, 2010; Van der Ploeg and Poelhekke, 2009; Papyrakis and Gerlagh, 2004). These three drivers are analysed the most and this will be the main point of focus in the literature review. Moreover, most scholars focus on the effect of natural resources on economic growth and in the last part of the literature review, it will be analysed what this could mean for our research interest of income inequality. This section will be closed off by focussing on scholars that have written on the specific link between natural resources and income inequality.

2.1. Resource type as the main driver behind the curse

Isham et al. (2005) argue that there is a difference between the type of resources; on the one hand there exist point-source natural resources, which are strongly associated with weak public institutions and lower levels of economic growth. These point-source natural resources are extracted from a narrow geographic area or economic base and are thus not spread out over the country. This is in contrast with diffuse resources, which are located in different parts of the country. Isham et al. (2005) argue that countries that own the former type of resource experience more severe economic slowdowns than countries whose exports are mainly based on the latter. First of all, it is stated that since a point-source resource is an easily controlled resource, it will lead to rentier effects, in the form of the government making use of the resources in an extractive manner. The state owns the resources, which makes it possible to suppress the population or to use violence against

(6)

it. Moreover, point-source natural resources lead to a process of delayed modernisation; the state that owns the resource would resist processes of modernisation, since this could lead to other forms of power that fall outside their hands. Lastly, there exist higher levels of inequality within countries that depend on point-source natural resources, for they experience more social divisions and less horizontal relationships of equality.

2.2. Institutional quality as the main driver behind the curse

Jensen and Wantchekon (2004) argue that democratic governments may break down or authoritarian governments may survive due to the discretion on the distribution of revenues coming from oil or minerals. Countries that are rich in natural resources are more likely to be authoritarian, have worse governance, or contain higher levels of government spending. More natural resources lead to more competition over who rules the state, which will lead to more political violence. Moreover, countries will use a large percentage of resource rents to maintain in power. This thus means that there is an effect of natural resource dependence on the type of political regime (Jensen and Wantchekon, 2004). Even though Brunnschweiler and Bulte (2008) also believe that it is mainly institutional quality that plays a role in the natural resource curse, they raise a critique concerning the direction of the relation between the two variables. It is not abundant resources that lead to bad institutions and slow growth with an increase in rent-seeking behaviour, but rather, bad institutions fail to industrialise and do not develop a non-resource sector within their economy, which makes the country highly dependent on the primary sector. Better institutional quality leads to less resource dependence, and not the other way around. Therefore, they argue that the curse does not exist; countries with poor institutions do not develop other economic sectors, and therefore do not reduce their dependence on natural resources. Causality would thus flow from institutions to dependence (Brunnschweiler and Bulte, 2008).

Diamond and Mosbacher (2013), who specifically focus on oil as a natural resource, argue that oil booms create negative prospects for development. They state that easy money coming from such booms leads to inflation, corruption, a distortion of exchange rates, a decline in the competitiveness of other export sectors, and a slowdown of growth in manufacturing. The reason behind this is that due to a fluctuation of oil prices on the world market, countries that are rich in oil can be harmed when booms go bust. Moreover, the wealth that is derived from oil leaves the state in a powerful position in comparison to civil society. It is weak institutions that make most of the resource curse possible, for

(7)

they are unable to prevent public officials from taking ownership of the revenue from oil and other resources (Diamond and Mosbacher, 2013).

2.3. Both type of natural resources as well as institutional quality matter

Stevens and Dietsche (2008) argue that the high price of resources should lead to large revenue inflows from exports, which will lead to opportunities for investment. However, most exporting developing countries have not been able to use the revenues from natural resources in a beneficial manner. When analysing why this is the case, they make a distinction between structural and agency-based arguments. The structural argument indicates that natural resources are often situated in a rentier state, in which the revenue generated by natural resource exploitation facilitates incumbent governments to diffuse pressures of democratisation. On the other hand, the agency-based argument indicates that politicians take personal advantage of natural resource wealth, in forms or rent-seeking or corrupt practices. Moreover, there exist different types of natural resources that can be linked to certain institutional outcomes; point-source resources for example lead to concentrated ownership, whereas this is not the case for diffuse resources. Connected to this difference in resources is the argument that point-source resources lead to vertical unequal relationships, while diffuse resources lead to horizontal equal relationships. This has as an effect that economic and political power are distributed in different ways and that social cohesion plays out differently as well (Stevens and Dietsche, 2008).

Basedau (2005) argues that natural resources play a detrimental role in the political and socio-economic development of a country, but it mostly depends on country-specific, as well as resource-specific conditions. First of all, a difference is made between several forms of natural resources and these differences can have an effect on how natural resources play out for the development of a country. Concerning the institutional aspect of the natural resource curse, Basedau (2005) argues that state elites do not have the incentive to improve the institutional quality of a country, for it could reduce resource rents from extraction. Just like other scholars, such as Stevens and Dietsche (2008), he uses the term ‘rentier state’, and explains that natural resource abundance is one of the main causes for a rentier state to exist. An abundance of resources thus leads to lower efforts to build a healthy institutional framework, and corruption and rent-seeking will happen more easily (Basedau, 2005).

(8)

Boschini et al. (2007) argue that it is exactly the combination of institutions and resource type that determines the effect of natural resources. One should focus on a combination of the type of institutions and the type of resource; the relationship between natural resources and economic growth is dependent on the institutional quality and differs per resource type. Concerning the aspect of institutions, Boschini et al. (2007) argue that more natural resources lead to a higher national income if institutions are production-friendly, but to a lower national income if institutions are grabber-friendly. This distinction is also made by Mehlum et al. (2007), who argue that in a producer-friendly state, rent-seeking and production are complementary activities, whereas in a grabber-friendly state, the two are competing activities. They believe that the resource curse exists in countries that are grabber-friendly, but not in countries that are production-friendly. This would means that natural resources are a blessing in the latter, but a curse in the former (Mehlum et al., 2007).Regarding the type of resource, it is argued that a lack of proper institutions is more disastrous for countries that own resources that are easy to appropriate. The negative effects of poor institutional quality are thus more severe when problematic types of resources are available and this is how Boschini et al. (2007) come to a combination of the two measurements.

2.4. Economy as the main driver behind the curse

Frankel (2010) focusses on several economic possibilities that could lead to the resource curse. First of all, he states that the volatility of commodity prices (and thus, natural resource prices) is high, which leads to higher risks and transaction costs. Moreover, when a country specialises in natural resources, this could have a negative effect on growth if the manufacturing sector is abandoned. Diversification in the manufacturing sector is thought to be better for economic growth than a commodity boom that erases manufacturing. This phenomenon is known as the ‘Dutch Disease’, which could lead to a real appreciation of the currency and increases in government spending. Governments tend to spend more during commodity booms, and part of the deficit is paid by borrowing from abroad. A last argument made by Frankel (2010) is that point-source resource revenues can lead to poor institutions and in that sense, natural resources will slow down the development of democracy. Auty (2007) argues in a similar vein and believes that the high rents that derive from resources will drive the government away from efficient wealth creation and into the redistribution of rents.

Collier and Goderis (2008) also focus on commodity prices and booms by stating that these booms have short-term positive effects, but long-term negative effects; higher

(9)

commodity export prices eventually lead to lower levels of real GDP. Countries with good governance are mostly able to turn commodity booms into higher sustainable outputs, whereas countries with bad governance are less likely to succeed in this. However, they argue that the resource curse does not work through governance, but is conditional on governance. Variables that are more important when explaining the curse are: exchange rate overvaluation; total investment; public consumption; private consumption; volatility; and growth in the service sector (Collier and Goderis, 2008).

Van der Ploeg and Poelhekke (2009) also focus on the topic of volatility and argue that the direct effect of natural resources on economic growth may be positive, but that the indirect effect between the two is negative via the channel of volatility. Primary commodity prices and resource revenues, and in that sense the real exchange rate, contain a high level of volatility in rich countries. Therefore, it is expected that resource-rich countries with poor financial systems will have lower levels of growth(Van der Ploeg and Poelhekke, 2009). Papyrakis and Gerlagh (2004) also focus on a transmission channel between natural resources and economic growth. They argue that natural resource abundance has a negative effect on economic growth; however, when other indices are taken into account, the effect becomes positive. They believe there exists a large negative effect on investments, in that resource wealth decreases the need for a country to save and invest. Moreover, natural resource abundance leads to less openness and a decline on terms of trade. It also weakens the manufacturing sector, for which human capital is one of the most important production factors (Papyrakis and Gerlagh, 2004).

2.5. Possible solutions to tackle the resource curse

Several scholars have posed possibilities on how to tackle the resource curse. Diamond and Mosbacher (2013) for example argue that the country should hand over a large share of the revenues made from resources to the people as a form of taxable income. Palley (2003) argues something similar with the introduction of an oil revenue distribution fund that would distribute revenues to citizens. He also argues that there should be more transparency in the form of compacts, monitoring, or requirements made by institutions such as the IMF and the World Bank (Palley, 2003). Moss (2011) also believes that one of the solutions for the resource curse could be the distribution of income that is derived from resource wealth to citizens in the form of cash transfers. Stiglitz (2005) argues that the main responsibility of the population is to ensure that the government is efficiently taking care of the resources and uses the funds to improve their well-being. When the country extracts resources and does not use the funds well, it becomes poorer instead of

(10)

richer. Therefore, it is important to take account of the behaviour of governments. Stiglitz (2005) proposes several possibilities with which this can be done: stabilisation funds, transparency, auctions, and interventions by developed countries.

(11)

3. How is previous literature linked to the measurement of income

inequality?

3.1. Link between economic growth and income inequality

Since the majority of papers that have analysed the natural resource curse have economic growth as their focal point, it will be discussed what this would indicate for income inequality. A link will thus be made between the outcomes on economic growth described above and possible outcomes on income inequality. One of the oldest and most traditional papers on the link between economic growth and income inequality is that by Simon Kuznets. He introduces the idea of the Kuznets curve, with which he argues that a country goes through a situation with an inverted U-shaped relation between income inequality and GNP per capita. Levels of inequality necessarily need to increase during the early stages of development and economic growth. However, as countries develop, levels of income inequality decrease again (Kuznets, 1955). Many scholars have criticised this hypothesis throughout the years and argued that this relationship between development and income inequality is not found in every country and that growth may affect income distribution in different ways. There is too much country specificity to make a generalised conclusion about this link (Bourguignon, 2004).

Thomas Piketty in his work ‘Capital in the 21st_{century’ (2014) argues that an} increase in income inequality that has been witnessed in the last few decades is a direct result of slower economic growth. A decrease in growth and an increasing capital to income ratio lead to a bigger share of income going to capital and a smaller share of income going to labour. This means that more money flows to the people who own the capital in the form of savings and less money flows to the working class in the form of wages, and thus, income inequality rises. All in all, Piketty (2014) argues that a decrease in economic growth leads to an increase in income inequality. However, Jackson and Victor (2016) state that rising inequality is not inevitable, even when growth rates are declining. This can mainly been avoided with policies that protect the wages against strategies that are in favour of the interests of capital. Policies need to make sure that employment levels remain high, even in times of a low economy (Jackson and Victor, 2016). Lundberg and Squire (2003) argue in a similar vein by stressing the importance of policies. They believe that the overall rate of growth cannot be separated from the distributional structure of growth. Variables that determine both growth and inequality are not mutually exclusive. This would mean that policies that are designed to improve one are likely to also improve the other. In general, they argue that promoting economic growth would lead to greater

(12)

inequality, but with the right set of policies both could be improved (Lundberg and Squire, 2003).

Bourguignon (2004) also states that there is a link between economic growth and income inequality. When a country makes a plan of development while focussing on poverty and growth on the one hand, and poverty and income inequality on the other hand, it is not taking the right approach. It should rather focus on the interactions that exist between distribution and growth. He states that case studies have shown that distributional changes in a country are affected by the pace and structural features of the country’s economy. He gives the example of Brazil, where one of the biggest contributors of rising inequality was the inability of poor households to incorporate themselves into the labour market as a consequence of slow growth (Bourguignon, 2004).

All in all, it can be stated that there exists a certain link between economic growth and income inequality, but the exact effects of the former on the latter are contested. Piketty (2014) for example argues that less economic growth leads to more inequality, whereas Lundberg and Squire (2003) state that promoting economic growth leads to greater inequity. Moreover, the Kuznets curve (1955) is highly contested by scholars that claim that the inverted U-shaped situation is not generalisable for all countries. Therefore, it is important to focus on the effect of natural resources on income inequality separately, for it cannot be derived from already existing literature on economic growth what natural resources mean for income inequality. Most literature on the natural resource curse argues that resource wealth has a negative effect on economic growth in a country, when taking into account the type of resource, the quality of institutions, and the economic situation of the country. However, since there is no clear link between economic growth and income inequality, it is important to conduct independent research on the effect of natural resources on income inequality.

3.2. Natural resources and income inequality

It has to be noted that even though most papers focus on the effect of natural resources on economic growth, several scholars have also mentioned the fact that natural resources could lead to higher numbers of corruption (Van der Ploeg, 2011; Diamond and Mosbacher, 2013; Basedau, 2005). Since corruption is likely to lead to a larger gap between the elitist groups and the rest of the population, it can be argued that even though the main research interest of these scholars is that of economic growth, they also say something about the indirect effect on income inequality. Van der Ploeg (2011), for

(13)

example, states that corruption is more viable to exist with natural resources dependence. The political elite can exploit and export resources and at the meantime, capture wealth and political power. However, the thesis will now focus on some papers that truly investigate the effect of natural resources on income inequality.

Goderis and Malone (2011) argue that in cases of resource booms, especially in oil and minerals, there exists an immediate diminishing effect on levels of income inequality. This effect eventually lowers throughout time until levels of inequality are the same as they were before the boom. Higher non-agricultural prices therefore lead to a decrease in inequality in the short term, whereas no significant effect is found in the long term. However, uncertainty about commodity export prices increases levels of long-term income inequality, which can also be derived from the fact that poor countries are less able to deal with resource booms than rich countries (Goderis and Malone, 2011).

Ross, Lujala and Rustad (2012) state that when the distribution of resources coincides with already existing divisions between groups, such as religious or ethnic divisions, real or perceived inequality can be a result, which in turn could create grievances. If the specific region where the resource is produced is relatively poor, resource wealth can help to close gaps between that area and the rest of the country. However, when the region is already relatively rich, resource wealth could be a factor that intensifies the gap. One possible way to mitigate existing inequality is to decentralise the resource revenues. Even though centralised distribution would probably be the best choice, it may not be possible because of lack of trust by the producing regions (Ross, Lujala and Rustad, 2012). Similarly, Fum and Hodler (2010) argue that resource rents may increase income inequality in societies that know large divisions. However, rents may reduce income inequality in more homogenous societies, for these rents tend to be distributed more equally or are even used to help the poor. Therefore, they come to the conclusion that resource rents may contribute to high income inequality in polarised societies, whereas they may contribute to low income inequality in homogenous societies (Fum and Hodler, 2010).

Ross (2007) elaborates on this argument, while making a distinction between vertical and horizontal inequality. With the former he means levels of inequality between the rich and poor within the same country, whereas the latter indicates the inequality that exists across different regions of the country. Concerning vertical inequality, he argues that governments can choose to directly distribute resource revenues. This could lead to a more equal distribution, allocation of rents in optimal ways, a reduction in

(14)

corruption and rent-seeking, and when it is conditional, it could even lead to achieving social goals such as school enrolment (Ross, 2007). Thus, if direct distribution works well, mineral rents could be allocated more equally. However, most governments in poor countries are not ready for this. Concerning horizontal inequality, a country could choose to divide mineral revenues between the central and subnational governments. However, central governments are usually better able to effectively absorb new investments, combat volatility and protect itself against a recession and inflation. Therefore, full centralisation may be better, because subnational governments are less able to deal with economic problems. The best approach is thus to collect revenues centrally and make decisions on allocations centrally, but also have some input from local or regional authorities (Ross, 2007).

(15)

4. How

does the dependence on natural resources affect income

inequality on the regional scale?

This part of the thesis will quantitatively analyse the effect of natural resources on income inequality. In order to make the research more robust, the focus will be on three different measures of income inequality (Gini Index, Palma Ratio, and Quintile Ratio), which will be explained in further detail below. This research is limited to the geographical area of Sub-Saharan Africa, which makes the dataset consist of 49 countries as a sample size. The variables that are used come from different datasets, which are the World Development Indicators, the Worldwide Governance Indicators and the Human Development Reports. The first two datasets are provided by the World Bank, whereas the latter is provided by the United Nations. The World Development Indicators (WDI) form the primary World Bank dataset on development, including global, national, and regional indices. It was first published in 2010 and it has been updated every year ever since, with current data covering the period from 1960 to 2017 (World Bank, 2018). The Worldwide Governance Indicators form a dataset that reports the governance performance according to six different dimensions, which will be explained later (World Bank, 2017). One has to remain aware of the fact that these sources are strongly shaped by the neoliberal agenda and its business interests. For example, the variable that measures ‘Regulatory Quality’ aims at the promotion of growth at the primary sector, which is driven by neoliberal values. However, one can assume these standards to be quite universal and there is not enough alternative data available that could be used.

For this research, a multiple regression analysis is conducted, which is a model that explains a given variable by a set of independent explanatory variables. In the literature review, one could notice that most scholars focus on three different factors that may have an influence on the relationship between natural resources and, in most cases, economic growth. This quantitative research will be based on these three different factors, and therefore, three different regression analyses are run, all including some standard control variables, but for each regression analysis, one of the three factors is added to analyse if they have an effect on the relationship between natural resources and income inequality. First of all, the disaggregate effects of different types of resources are tested for, together with some other control variables, focussing on oil, coal and minerals. Then the institutional quality of a country is tested for, together with the rest of the control variables. Lastly, the influence of the economic situation is being looked at and how that may affect the relationship between natural resources and income inequality. For the

(16)

independent variables, data from 2015 is taken, since most data is available for this year. However, for the dependent variables, the latest data available for all countries is taken, since availability is quite varied.

4.1. Dependent variables

 Gini Index: This variable measures the extent to which income distribution among households and individuals deviates from a perfectly equal distribution (World Bank, 2018). Since the distribution is highly skewed, a log function for the Gini Index is applied, since it better resembles a normal distribution. The Gini Index is calculated from a Lorenz Curve, with the latter visualising the distribution of a given quantity among the population. The Gini Index then summarises how much the actual Lorenz Curve deviates from perfect equitability (Farris, 2010). A histogram of the normal Gini Index, as well as the log function of the Gini Index can be found in the Appendix (Figure 1 and 2).

 Palma Ratio: This variable is based on the work of Palma and measures “the ratio of the richest 10 percent of the population’s share of gross national income divided by the poorest 40 percent’s share” (United Nations Development Programme, 2018). For this variable, a log function is applied, since it makes the distribution less skewed. A histogram of the normal Palma Ratio, as well as of its log function can be found in the Appendix (Figure 3 and 4).

 Quintile Ratio: This variable measures the “ratio of the average income of the richest 20 percent of the population to the average income of the poorest 20 percent of the population” (United Nations Development Programme, 2018). Again, a log function for this variable is applied. The histogram of the normal Quintile Ratio, as well as of its log function can be found in the Appendix (Figure 5 and 6). 4.2. Main explanatory variable of interest

 Natural Resources Rent: The main explanatory variable of interest of this research is that of natural resources rent. It measures how much revenue a country receives from its natural resources, measured as a percentage of GDP. Therefore, it can be considered to be a measure of a country’s economic dependence on its natural resources. The definition given by the World Bank is that it is the sum of forest rents, natural gas rents, oil rents, coal rents, and mineral rents (World Bank, 2018).

(17)

4.3. Main control variables

 Coal, Oil and Mineral rents: As has been analysed in the literature review, it could be possible that the type of resource has an effect on how natural resources are used by a country (Isham et al., 2005). Therefore, these three types of resources serve as some of the main control variables in the empirical analysis. This variable is made into a binary variable, in which 0 indicates a country that does not have the resource and 1 indicates a country that does have the resource. Moreover, an interaction effect is included, in which an interaction exists between the dummies for each resource and the total amount of resources. In this way, it is possible to analyse how much having that resource changes the effect of resources on income inequality. This allows the effect of natural resources to be different for countries with different types of resources. Of course one should realise that even though a distinction is made between these three different resources, there may not only exist different outcomes per different resource, but also for the same resource. For example, gold and copper are both mineral resources, but they may not have the same effect on income inequality. This will be further analysed in the qualitative analysis of this thesis.

 Manufacturing: This variable accounts for the net output of the manufacturing sector and is measured as a percentage of GDP (World Bank, 2018). As has been stated in the literature review, one of the ways in which natural resources may turn into a curse is by harming the manufacturing sector (Diamond and Mosbacher, 2013). The manufacturing sector is believed to be more beneficial for a country’s economic situation than the commodity sector (Frankel, 2010). However, one should keep in mind that it is not always the case that countries in which the manufacturing sector is thriving are doing well. Nevertheless, it has been analysed by several academics as one of the factors that drive the resource curse and therefore, this variable is included as one of the main control variables.

 Gross Domestic Savings: With this variable there is a focus on GDP minus the total consumption (World Bank, 2018). This variable is included in the empirical analysis, for it is another measure of the economic situation of the country through which natural resources may have an effect on income inequality. As already stated in the literature review, wealth deriving from resources may hamper the country in its processes of savings and investments (Papyrakis and Gerlagh, 2004).  Institutional Quality: The Worldwide Governance Indicators measure governance

(18)

items is high, and therefore the average of all of them is taken and turned into one index accounting for governance performance. This variable is being controlled for, since it is analysed in the literature review that the way in which natural resources have an effect may be influenced by the institutional quality of a country (Jensen and Wantchekon, 2004; Brunnschweiler and Bulte, 2008). The six different dimensions out of which this variable consists are taken from the dataset of the Worldwide Governance Indicators and are measured on a scale from -2.5 to 2.5 with low numbers indicating a weak performance of governance and high numbers indicating a strong performance of governance (World Bank, 2017). The six dimensions are the following:

- Voice and Accountability: This variable reflects the perception citizens have of their own participation in their government’s selection process, as well as freedom of expression, freedom of association, and free speech.

- Political Stability and Absence of Violence: This variable measures the perception of the possibility that violence will break out in the country, related to political instability.

- Government Effectiveness: This variable measures the perception of institutional quality, the quality of implementing different policies, and the credibility of the government.

- Regulatory Quality: This variable measures the perception of the government’s ability to implement policies towards the promotion of growth within the private sector.

- Rule of Law: This variable measures the perception of confidence agents have in rules and regulations set up by society and how much people abide by them. - Control of Corruption: This variable measures the perception of the level in which

public power is used for private interests. 4.4. Other control variables

Other variables that are used from the WDI are control variables that may also have an effect on income inequality:

 Total population of a country: This variable is described as all residents living in a country (World Bank, 2018). It is used as a control variable, for it could affect income inequality and it is important to take into account the size of a country. According to Odusola et al. (2017), there exists a relationship between population size and income inequality in Sub-Saharan countries. Countries that experience

(19)

higher levels of fertility rates have lower levels of income inequality. When fertility rates are high, a higher share of income goes to the bottom 40 percent of the population. This relationship also holds for other population-related variables; countries with higher rates of population growth tend to have a higher share of the income going to the bottom 40 percent (Odusola et al., 2017).

 GDP per capita, adjusted to the Purchasing Power Parity (PPP): This control variable is defined as the gross domestic product divided by midyear population (World Bank, 2018). The research will control for GDP per capita, rather than just GDP, for it is a better determination of life standards. Moreover, it appears to be a better explanatory variable for Gini than GDP growth (Luan and Zhou, 2017). It is also adjusted to PPP, since this makes it possible to compare the GDP of different countries. It controls for price level differences and variations of exchange rates (Vogel, 2008). According to Kuznets (1955), a rise in GDP per capita has a reducing or stabilising effect on income inequality. This is also stated by Luan and Zhou (2017), who argue that when a country has a higher level of GDP per capita, the country is likely to be more developed and in that sense, it has lower income inequality. For this research, a log function for GDP per capita (adjusted to PPP) is applied. A histogram of the log function, as well as of the normal GDP per capita (adjusted to PPP) can be found in the Appendix (Figure 7 and 8).

 Literacy rate of the adult population: This variable measures the percentage of the population older than fifteen who is able to write and read with understanding of a statement about everyday life (World Bank, 2018). This control variable is used, for a certain link can be found between literacy rate and income inequality, with an increase of the latter when inequality of educational attainment increases (Cameron and Cameron, 2006). This is also argued by Abdullah et al. (2011), who believe that the more unequal the distribution of education, the more unequal the distribution of income. However, in general, education has a reducing effect on income inequality, for it closes the gap between the rich and the poor (Abdullah et al., 2011).

(20)

5. Findings

This section will focus on the results of the regression models. First, the table with the Gini Index as the dependent variable will be introduced, then the table with the Palma Ratio as the dependent variable will be analysed, and the section will finish with the Quintile Ratio as the dependent variable. In all these three tables, the results of five different regression models will be looked at. In the first model, there is a focus on the effect of the main explanatory variable ‘Natural Resources Rent’ on one of the three versions of income inequality. In the second model, control variables ‘Total Population’, ‘GDP per Capita adjusted to PPP’, and ‘Literacy Rate’ are added. Controlling for these variables may affect the estimate of ‘Natural Resources Rent’ in the regression. In the third model, control variables that account for the difference in resource type are added. The dummies of ‘Oil’, ‘Coal’, and ‘Mineral’ are added, as well as the interaction between ‘Natural Resources Rent’ and the dummy. In the fourth model, the control variable of ‘Institutional Quality’ is introduced and in the last model, two control variables that account for the economic situation of a country are analysed, which are ‘Manufacturing’ and ‘Savings’. The thesis will now turn to the first table, in which the effect of ‘Natural Resources Rent’ on the Gini Index is analysed.

(21)

MODEL 1 2 3 4 5 NATURAL RESOURCES RENT -0.003**

(0.001) -0.001 (0.001) -0.002 (0.003) -0.002 (0.001) -0.003 (0.002) POPULATION SIZE -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) GDP PER CAPITA PPP 0.005 (0.038) -0.008 (0.045) 0.011 (0.041) -0.015 (0.045) LITERACY RATE 0.001* (0.001) 0.001 (0.001) 0.001* (0.001) 0.001 (0.001) COAL DUMMY 0.004 (0.053) OIL DUMMY 0.074 (0.045) MINERAL DUMMY -0.061 (0.048) COAL X RESOURCES 0.001 (0.005) OIL X RESOURCES -0.006* (0.003) MINERAL X RESOURCES 0.002 (0.003) INSTITUTIONAL QUALITY -0.010 (0.025) MANUFACTURING 0.001 (0.002) SAVINGS 0.000 (0.001) CONSTANT 1.669*** 1.566*** 1.678*** 1.540*** 1.655*** ADJUSTED R2 _0.079 _0.129 _0.108 _0.110 _0.211 F-VALUE 4.782** 2.590* 1.520 2.058* 2.386* a. *p<0.1; **p<0.05, ***p<0.01 b. Dependent variable: Gini Index

(22)

In this table, the effect of ‘Natural Resources Rent’ on the Gini Index is analysed. It can be stated that natural resources have a small negative, but significant effect on the Gini Index. This indicates that when the amount of natural resources increases, the Gini Index decreases. The adjusted R2 _{, which measures the explanatory power of the independent} variables on the variability of the response variable, is not high. This means that the explanatory power of this model is quite low. In the second model, the effect of natural resources remains low and negative, but becomes insignificant. The adjusted R2 _increases slightly, but stays relatively low. In Model 3, 4, and 5, no strong changes happen in the effect of natural resources on income inequality measured by the Gini Index; the effect is not statistically significant and therefore, no clear conclusions can be derived from this table. The only other variable that is somewhat significant is that of the interaction between oil and resources, which means that the effect of natural resources on income inequality is negative and significant for countries which have oil (i.e. the oil dummy is equal to one). For countries with oil, the natural resources effect is 0.002 + 0.006 = -0.008. The F-value of almost all models is different from 0 and significant, which indicates that the models have some explanatory power on the response variable.

(23)

MODEL 1 2 3 4 5 NATURAL RESOURCES RENT -0.005*

(0.003) -0.002 (0.004) -0.006 (0.007) -0.002 (0.004) -0.005 (0.005) POPULATION SIZE -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) GDP PER CAPITA PPP 0.067 (0.100) -0.027 (0.119) 0.066 (0.107) 0.057 (0.119) LITERACY RATE 0.003 (0.002) 0.002 (0.002) 0.003 (0.002) 0.003 (0.002) COAL DUMMY 0.078 (0.132) OIL DUMMY 0.230* (0.115) MINERAL DUMMY -0.202 (0.008) COAL X RESOURCES -0.008 (0.012) OIL X RESOURCES -0.017* (0.009) MINERAL X RESOURCES 0.007 (0.007) INSTITUTIONAL QUALITY 0.001 (0.063) MANUFACTURING 0.001 (0.006) SAVINGS 0.000 (0.002) CONSTANT 0.429*** 0.001 0.479 0.003 0.098 ADJUSTED R2 _0.047 _0.192 _0.115 _0.080 _0.133 F-VALUE 3.080* 2.196* 1.532 1.709 1.767 a. *p<0.1; **p<0.05, ***p<0.01 b. Dependent variable: Palma Ratio

(24)

In this table, the same regression models are being run, but this time income inequality is measured by taking the Palma Ratio as the dependent variable. One can see that there are no strong differences between this table and the previous one; in the first model, the effect of natural resources on income inequality is small, negative and significant, but becomes insignificant when adding control variables. The control variable that accounts for the interaction effect between oil and resources is significant in this table as well. The only difference is that now the oil dummy is significant as well, which was not the case in the previous table. This means that having oil tends to increase income inequality and since both the dummy and the interaction are significant, it means that having oil increases inequality, and also changes the effect of natural resources overall on income inequality. The adjusted R2_{is even lower for most models compared to the ones in the first} table, which means that the explanatory power of these models is lower than in the previous table. This can also be seen when looking at the F-value, which is only significant in Model 1 and 2.

(25)

MODEL 1 2 3 4 5 NATURAL RESOURCES RENT -0.005

(0.003) -0.001 (0.004) -0.007 (0.007) -0.002 (0.004) -0.005 (0.005) POPULATION SIZE -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) -0.000 (0.000) GDP PER CAPITA 0.086 (0.102) -0.029 (0.120) 0.101 (0.108) 0.091 (0.124) LITERACY RATE 0.002 (0.002) 0.002 (0.002) 0.002 (0.002) 0.002 (0.002) COAL DUMMY 0.052 (0.132) OIL DUMMY 0.225* (0.116) MINERAL DUMMY -0.226* (0.125) COAL X RESOURCES -0.008 (0.012) OIL X RESOURCES -0.017* (0.009) MINERAL X RESOURCES 0.007 (0.007) INSTITUTIONAL QUALITY -0.029 (0.064) MANUFACTURING 0.002 (0.007) SAVINGS -0.001 (0.002) CONSTANT 1.031*** 0.564 1.141** 0.492 0.603 ADJUSTED R2 _0.041 _0.076 _0.112 _0.056 _0.091 F-VALUE 2.805 1.848 1.519 1.487 1.503 a. *p<0.1; **p<0.05, ***p<0.01

(26)

The last table analyses the effect of natural resources on income inequality by using the measure of the Quintile Ratio. Again, the results show high similarities with the other two tables, despite the fact that this time the effect of natural resources is never significant, also not in the first model. The difference is that now both the oil dummy and the mineral dummy are significant, which was not the case in the previous table. This means that having oil tends to increase income inequality and having minerals tends to decrease it. One can see here that for minerals, it is only the dummy and not the interaction, which means that having minerals increases inequality, but does not change the effect of natural resources overall on income inequality. Again, the adjusted R2_{is low} for most models, which means that the explanatory power of these models is relatively low. Moreover, the F-value is not significant in any of the models.

All in all, it can be stated that all tables show similar results, irrespective of what measure of income inequality is used, which makes the quantitative analysis quite robust. What can be derived from this analysis is that no correlations can be found between natural resource dependence and income inequality. The variable of ‘Natural Resources Rent’ does not have any significant effect on any of the measures of income inequality, and neither do most of the control variables. However, since it is the case that several resource-based economies show high levels of income inequality, it is of importance to analyse other political economic factors on a more individual level. Therefore, the focus needs to shift to certain countries in specific, since no general argument can be made about the effect of natural resources on income inequality for the whole of Sub-Saharan Africa. This will be done in the next section, in which the qualitative analysis with a case study of three different Sub-Saharan countries is introduced.

(27)

6. How does the dependence on natural resources affect income

inequality on the national scale?

In this part of the thesis, the focus will be on three different countries in particular in order to see the link between natural resources and income inequality, since the conclusions from the quantitative analysis need to be complemented with some qualitative research. There is a focus on Mali, Angola and Zambia, which all have a completely different outcome in their Gini Index, with Mali having a low Gini Index, Angola having an average Gini Index and Zambia having a high Gini Index. However, they are similarly dependent on their natural resources when calculated as a percentage of their GDP. First, Mali will be analysed, which has a Gini of 33 percent and a natural resources rent percentage of 12.5 (World Bank, 2018). Then the attention will shift to Angola, which has a Gini of 42.7 percent and a natural resources rent percentage of 11.3 (World Bank, 2018). Eventually, Zambia is analysed with a Gini of 57.1 percent and a natural resources rent percentage of 14.3 (World Bank, 2018). It is interesting to know that the average Gini Index of all Sub-Saharan countries is 43.9 and the average percentage of natural resources rent is 11.8 (World Bank, 2018). One will be able to see how the colonial past has an influence on the relation between natural resources and income inequality, with Angola and Zambia inheriting a more hierarchical system than Mali. However, Mali and Angola score higher on the Corruption Perceptions Index than Zambia, while the Gini Index of Zambia is considerably higher, indicating that it is not only the perception of corruption that matters (Transparency International, 2018). Other political factors, such as the ability to implement policies of inclusive growth are also of importance, since Angola has shown to be better in that than Zambia. One will also see how the geography of the resource matters and how colonialism is linked to that. However, even though geography matters, it cannot be stated that all point-source resources lead to similar results. The way in which the country deals with the resource is also of importance, with Mali deciding to decentralise the resource sector and Angola and Zambia trying to diversify the economy away from the resource sector. A more elaborate comparison will be made after an individual analysis of the three particular countries. 6.1. Mali

Mali was colonised by the French in 1880 as the region then known as French Sudan. At first, the local population responded with confrontation and alliance, but in 1898 resistance against the French Empire came to an end (Guerrero, 2014). However, a cultural opposition remained by the Tuaregs who lived in the northern part of French

(28)

Sudan. In the south, the Blacks had already given the French their cultural acceptance, which led to the French privileging the south over the north for economic development. Moreover, most of the resources such as agricultural goods and gold were found in the south, which also steered French policy towards that area. Geography in that sense has had a negative effect on levels of stability and equality, since natural resources were unequally distributed throughout the country (Guerrero, 2014). The south thus became the focal point of infrastructure and trade for European companies. European people migrated to the south and brought along their capital, which stimulated the development of the south at the expense of the north (Guerrero, 2014). For example, investments were made in Bamako, the capital, in the form of infrastructure and new city design. The city was fundamentally changed with the appearance of roads and boulevards (Falola and Salm, 2004). Moreover, the French educated a ruling class that contained mostly people from the south (Chauzal and Van Damme, 2015). Investments in communications were also merely made in the south, which fostered its ability for trade and exchange (Guerrero, 2014). Wealth increased in the south, whereas the economic power of the north declined. The French ensured that they were able to exploit natural resources of the southern region, but at the same time they also tried to include the local population into processes of development (Guerrero, 2014).

When looking at more recent times, it can be seen that inequalities between the north and the south still exist, as in the period of colonialization. This difference between the rural north and the urban south leads to the issue of unequal access to resources of development (Cairnie and de Grasse, 2017). This again likely plays a significant role regarding income inequality, since development resources are an important factor in the stimulation of a country’s income. In that sense, colonialism has had an effect on income inequality and natural resources have played an important part, since they have attracted the French to the south, rather than the north. This is also argued by Chauzal and Van Damme (2015), who state that the French have shifted away the economic domination from the north to the south. The northern regions that once formed the most important component of Mali’s wealth before the French arrived, became a barren economic zone with instability and too little infrastructure. The south however turned into a well-functioning economy with the presence of agriculture and gold mining. Moreover, money that should have been spent on local development in the north was eventually used for military infrastructure by the government (Chauzal and Van Damme, 2015). This difference is still present nowadays, with poverty declining in the western part of the country, while remaining the same in the north (European Parliament, 2014). Besides,

(29)

the availability of gold in the south of Mali is further advantaging that region (Langer and Stewart, 2015).It can thus be stated that income inequality is taking place between different regions in Mali. However, when looking at data, it can be stated that the differences between regions are not high, with Bamako and the Kayes-Koulikoro region in the south showing the lowest poverty headcount (9%, 27%), but the North has a lower poverty headcount than the Mopti-Ségou and Sikasso regions, which are also located in the south (29% compared to 47% and 56%) (World Bank, 2013). Nevertheless, most gold is found in the south, and gold brings in more money when compared to agriculture (Imrich, 1999).

Not only do inequalities exist between different regions, but also between different sectors of the economy. The resource sector is completely separated from all other economic sectors and Mali has developed a gold monoculture (FIDH, 2007). Moreover, the majority of the people work in agriculture and they earn less than people working in the mining sector, who have a higher per-capita consumption. Therefore, redistribution among the population is important and should be promoted in Mali’s policy (World Bank, 2015). This is also stated in a report by the European Parliament (2014), in which it is argued that 80 percent of the population derives its income from agriculture. Development of the mining sector has only benefitted a few regarding local employment and content. Marouani and Raffinot (2004) argue in a similar vein, by stating that the increase of industrial growth due to the mining and gold sector has only benefitted a small part of the population. Most poor people work in food production and do not get a share of the higher growth rates of the industrial sector. Moreover, only the wealthier households are able to diversify and to shift away their main focus on food production (Crole-Rees, 2002). This difference between economic sectors is thus also part of how natural resources affect income inequality in Mali.

However, the current government of Mali seems to be aware of the fact that natural resources, especially gold mining, are leading to income inequality and it tries to tackle the problem. In the Poverty Reduction Strategy Paper of 2013, it is stated that more economic and social benefits coming from natural resources should go to the poor. Natural resource management should become more sustainable in order to reach the objectives of poverty reduction. One of the solutions that is brought forward is that of promoting small mines; community actions of mining are of importance and local communities should take over part of the responsibility (IMF, 2013). Moreover, mining sector development policies should align with employment and social policies in order to

(30)

include the vulnerable groups of the population. Natural resource management should be transferred to the territorial communities so that more responsibility is given to the population. It is believed that decentralisation would lead to a decrease in poverty and income inequality (IMF, 2013).

This approach is analysed by more institutions and scholars, for it is believed that gold mining should lead to the improvement of human rights, rather than to income inequality (FIDH, 2007). According to the Federation Internationale des Ligues des Droits de l’Homme, one of the reasons why mining does not improve the conditions of the poor is because it is mainly led by big companies. These companies try to exempt from social obligations and taxes and want to gain as much profit as possible. At the same time, the state does not fulfil its role as the supervisor of these companies and national revenue is not distributed equally, so that the Malian population does not reap most of the benefits. However, two of the main gold mine operators have set up programmes that are led by the community in order to support local development of the mining sector. Moreover, organisations of Malian civil society have started pressuring the mining companies to set up funds for the community-led development programmes (FIDH, 2007). Mining companies must thus finance a community fund that is partly led by the government and partly led by the local authorities (Mainguy, 2010). In 1999, a Malian mining code was introduced that imposed environmental and social responsibilities on large-scale companies (Mainguy, 2010). Teschner (2014) also analyses the positive outcomes of small-scale mining in Mali and how it has increased profitability and investment. Rural institutions have started to become able to capitalise on the increased revenues made by the mining sector. Social institutions and commune government have shifted their revenue streams from the state to the local artisanal mining economy. Not only people working in the mining sector benefit from this artisanal mining, but also people working in agriculture, with some people being able to finance metal fencing to protect their crops. Artisanal mining therefore improves the level of income of rural people and also funds social institutions in a bottom-up approach ((Teschner, 2014).

However, it is important to remain critical of the artisanal mining sector and decentralisation in Mali, since these processes do not always have positive results. In the country report of BTI (2018) for example, it is stated that Malian communities interact with national and international partners on the topic of natural resource management, and they thus have a say in how the resources are managed, used and owned. However, effective decentralisation remains a struggle and it can often lead to violence and conflict

(31)

(BTI, 2018). Moreover, according to Langer and Stewart (2015), power is not given to the local population, but rather exercised by appointees from the centre. Wing (2013) also argues how the process of decentralisation has been designed partly as a solution to the problem of the north, but that so far it has only had a weak capacity. The Malian population believed in the promise of more equality deriving from decentralisation and democratisation, but not much has been achieved (Wing, 2013). A report of SNV and CEDELO (2004) also raises some scepticism about decentralisation in Mali, for it may be likely that local authorities are interested in protecting their private interests by selling off natural resources or allocating land.

All in all, it can be stated that income inequality is affected by natural resources, and gold in particular. During the colonial period, the Tuaregs have driven the French to the southern part of the country and they remained there because of the resources that were present in that area. This has eventually led to the economic development of the south, whereas the north stayed behind. This difference between the two regions still exists nowadays, since still most of the resources can be found in this area, which only contributes to the economic disparities. Of course it has to be noted that it is not the case that everyone in the north is poor, nor everyone in the south is rich. The poorest region of the whole country, which is the Sikasso region, is situated in the most southern point of Mali (Yabi, 2017). This poverty is mainly found in the agricultural sector and thus it can be stated that certain income inequalities also exist within the southern region (World Bank, 2015). Moreover, income inequalities also occur between the different sectors of the Malian economy. The majority of the poor population works in the agricultural sector and does not benefit from the economic growth in the industrial sector of mining. However, it can be argued that Mali has an active policy towards the benefits deriving from the mining sector for all people. It tries to implement processes of decentralisation and more voice is given to small-scale artisanal mining, with a shift away from the large-scale mining companies. The local community benefits from the revenues and therefore, not all money is transported to the central government. The processes of artisanal small-scale mining and decentralisation could thus be an explanation of why the Gini Index of Mali is one of the lowest of Sub-Saharan Africa. Even though gold mining leads to income inequality, the country tries to solve its problems by actively implementing policy strategies. However, it is important to remain critical of these processes, for it has been shown that they do not always lead to positive results.

(32)

6.2. Angola

Angola was a colony under Portuguese rule from the 16th_{century until the 20}th_century, which started out as an important place for slave trade to Brazil and Portugal. The Atlantic slave trade ended in the 1860s and the Portuguese colonisers started to focus on other goods. Coffee, as well as diamonds, became two of the major export products during the 20th_{century. This changed in 1973, when oil became the most valuable export for} Angola (Ball, 2017). During this period of colonialism, Portugal and Brazil mainly focussed their attention on the capital city Luanda, for it was more easily connected than the inwards territories. It was located on the coast and in that sense, power moved towards the ocean (Ruigrok, 2010). This tension between the coast and the interior part of the country still exists today, with the replacement of the colonialist Portuguese by the coastal Portuguese speaking elites (Kyle, 2003). Concerning the way in which the Portuguese colonised the country, it can be stated that a certain caste system was created in which the small elite ruled the country, whereas the African underclass got denied from its rights of citizenship. There thus existed a political and social hierarchy in the Angolan colonial society (Ball, 2017). This social stratification still persists today, with a national class of owners that receives higher levels of income and has a higher status in the hierarchy (Gonçalves, 2010). Moreover, the system that was imposed on the country was weak and led to the creation of corruption and centralisation. Angola did not inherit any institutions or traditions on which it could build a social contract between the state and its citizens (Sogge, 2009).

After independence, many things remained unchanged when compared to the colonial period, especially the system of extraction and corruption that was created by the Portuguese. President José Eduardo dos Santos came into power in 1979 (and stayed in power until 2017), together with its party the MPLA (People’s Movement for the Liberation of Angola). They were highly authoritarian and based themselves upon a system of a centralised command economy (Hammond, 2011). The MPLA was supported by Socialist countries, such as the Soviet Union and Cuba, whereas the opposing party, UNITA, got help from the West with France, South Africa and the United States. Related to these opposing parties was the civil war that lasted from 1975 until 2002 (Ferreira, 2006). This war had a negative effect on the economic policy of Angola, for the government was unable to formulate good policies. The president and its clique pointed to the civil war as an excuse of the poor economic performance of the country and the war made it easier

The relation between natural resources and income inequality in Sub-Saharan Africa: A difference between having resources and using resources