Looking beyond Cape Town’s inequality
Income inequality across the poor areas of an extremely unequal cityBachelor thesis Human Geography (734301370Y)
Author: Hessel A. Moelijker
Student ID: 10259686
Guidance by: M.A. de Vries
Date: 20 June 2014
Address: Meester Treublaan 32Hs
1097DR Amsterdam
Contact: hesselmoelijker@hotmail.com
1
Table of Contents
1. Introduction ... 2
2. Literature review ... 4
2.1 Inequality across scales ... 4
2.2 Inequality in context ... 6
2.3 Measuring inequality ... 7
2.4 What causes inequality ... 8
2.4.1 Crime ... 8
2.4.2 Education ... 9
2.4.3 Housing & transport infrastructure... 11
3. Methodology ... 13
3.1 The Cape Flats and the townships ... 13
3.2 Research Questions ... 14 3.3 Units ... 15 3.4 Data ... 16 3.4.1 Income ... 17 3.4.2 Crime ... 18 3.4.3 Education ... 19 3.4.4 Housing ... 20 3.4.5 Transport infrastructure ... 21 3.5 Analytical methods ... 22 4. Analyses ... 23 4.1 Income inequality ... 23 4.2 Education ... 26 4.3 Housing ... 28 4.4 Crime ... 29 4.5 Transport infrastructure ... 32 5 Discussion ... 34
5.1 Income inequality in the Cape Flats ... 34
5.2 Causes of income inequality in the Cape Flats ... 35
5.3 The use of literature of a higher scale ... 36
5.4 Conclusion ... 37
Literature ... 38
Appendix ... 42
Cover picture: An aerial picture looking across the Cape Flats, towards the center of Cape Town (Sutherland, 2012).
2
1.
Introduction
For anyone visiting Cape Town, the abundance of diversity is something that is easily noticed. A population of various different ethnic groups as well as large numbers of tourists all speak many different languages and bring with them their own cultures, all the while seeing Table Mountain rise above skyscrapers and the harbor, the suburbs and the townships. But diversity comes in many ways and Cape Town is also a place of large economic diversity, better known as economic inequality. This less appealing form of diversity is most apparent when one compares the extremely wealthy Atlantic Seaboard areas (the western coast in figure 1), such as Camps Bay or Sea Point, with the Cape Flats, an area less than an hour drive away (the south-eastern area on figure 1). The Cape Flats is home to most of Cape Town‟s townships, areas where the black and colored population of Cape Town were forced to live during the Apartheid regime. These areas are generally seen as being home
to „the poor‟ of Cape Town, as historically they have been rifled with poverty.
I witnessed this situation myself while staying in Cape Town for half a year. Visiting different areas I saw many scenes one would expect; enormous villas with high electrical fences overlooking the ocean in the wealthy suburbs and small shacks housing whole families with electrical wiring dangling dangerously between them in the townships. What interested me most however was what I would not expect to see in these townships; expensive cars, large houses -sometimes with modern gates and electrical fences-, large commercial areas and surprisingly well organized and maintained public transport hubs. This led me to believe that the townships of Cape Town cannot be seen as one group, let alone be identified as „the poor‟. What I saw was a diverse area in which there is a division of low, middle and high income classes. This new insight has been of great interest to me since, providing the basis of my research.
South Africa‟s and Cape Town‟s history will not be prominently included in this research; only where truly relevant will it be mentioned. Instead, the focus will be on the current situation, using the most recent data from 2011. I chose to focus on the current situation to avoid influences from the Apartheid-era as much as possible. Even though it cannot and will not be fully neglected, the extreme effects of Apartheid on South Africa would severely interfere with the results if earlier data would be used. The main theme of this research is inequality in an urban context and using older data would change that into a post-Apartheid context.
Figure 1: Satellite image of Cape Town
3 This focus of this paper is specifically on income inequality, hence the economic focus in the previous examples. Income inequality should not be seen as the most important form of inequality, just as affluence is not a perfect indicator of development. I have chosen to research income inequality because I see prosperity not as the end but as a means to development, something Robert Lucas put to words as following: “By the problem of economic development I mean simply the problem of accounting for the observed pattern, across countries and across time, in levels and rates of growth of per capita income. This may seem too narrow a definition, and perhaps it is, but thinking about income patterns will necessarily involve us in thinking about many other aspects of societies too, so I would suggest that we withhold judgment on the scope of this definition until we have a clearer idea of where it leads us. ” (Lucas, 1988)
Besides my personal interest in this issue, it also carries with it certain social importance. By determining the causes of income inequality in the Cape Flats, that knowledge can be used to improve the livelihoods of those people that have remained in poverty. By increasing the income of the very poor, their living conditions may be improved. This could include improvements in for instance housing, education and health. Furthermore, as Calvó-Armengol et al. (2007) showed, income inequality can be a cause of crime. Therefore, a decline of income inequality may lead to a decline of crime in the area, further improving living conditions.
Next to the social importance of this research, this research aims to be a valuable addition to the current academic knowledge on urban inequality. More specifically, it will provide new knowledge on urban inequality on a small scale and income inequality across the Cape Flats, two topics on which current literature is lacking. Most research on inequality has been on a larger scale; however, with the high urbanization rates of today, inequality on a small scale could become more extreme and more widespread. To determine whether or not previous ideas based on research on larger scales are applicable to smaller scales, those ideas will be compared to the actual situation in the Cape Flats.
In this introductory chapter the main theme of the research has been introduced; income inequality across the Cape Flats. Based on that theme, the following research question has been formulated: Is there income inequality across the Cape Flats and if so, what could have led to this inequality?
The rest of this paper is divided into four chapters. In the second chapter, the current academic literature on the topics featured in this research will be discussed, as well as their relevance for this specific case. Next, the methods that have been used to conduct this research will be explained. That is followed by the analyses, featuring their results and discussing the implications of those results. The final chapter will present the conclusions of the research.
4
2.
Literature review
The goal of this paper is to complement the academic literature on urban inequalities. It will use the current literature on that theme, originating from various fields, as its foundations. Academic literature and data from various institutions and governments will be used to discuss the various theoretical findings featured in this research. Next to explaining the ideas of different authors, their relevance for this specific case will be discussed.
This chapter starts with two sections on the overarching themes of this research; inequality across scales and urban inequality. This will be followed by a discussion of income inequality, the central topic of this research. The final sections will be on the possible causes of income inequality; crime, education, housing and transport infrastructure.
2.1 Inequality across scales
Inequality is a global issue, existing all over the world in various forms and on different levels of scale. The United Nations Development Program categorizes inequality according to three different categories: Inequality across countries, inequality within countries and inequality across the world‟s people (UNDP, 2010). A distinction between different levels of scale is necessary in any research into inequality, as results may vary at different scales. Furthermore, different scales interact and inequality at one level can influence another level.
Chen and Crawford‟s (2012) research focuses on income inequality and health disparities, but their ideas on scale apply to any research on inequality. They claim that previous research often tried to find a relation between income inequality and health outcomes that would be consistent across different scales, which according to them is a wrong approach. Their research showed that income inequality measured at different scales may relate differently to health. This means that the results of research at one scale should not be regarded as being representative for all scales.
Furthermore, Chen and Crawford explain why income inequality measured at one scale is likely to be different from inequality measured at other scales. They provide four reasons for this: First, inequality at any scale is a component of the immediate upper scale. Second, a larger geographic area always results in a smaller measurement error since inequality is a measure of spread. Thirdly, greater homogeneity of population at a low scale will result in lower inequality than at a larger scale with a more heterogeneous population. Finally, income inequalities at different scales are associated with different contextual factors. Chen and Crawford (2012) illustrate these ideas with a comparison of income inequality at the state and the county level in the United States. This comparison shows that the spread of inequality across the US changes quite drastically when different geographical scales are used.
Peters (2013) did a similar research on income inequality in the US and found that when using a low geographic scale for measurement, such as counties, large urban areas look highly unequal and small rural areas appear more equal. He claims that the reasons for those results are primarily methodological, providing similar reasons as Chen and Crawford. Furthermore, both Peters (2013) and Chen and Crawford (2012) argue that states in the US with high inequality often include a number of large urban counties
5 with extreme inequality. These extremely unequal counties make those states appear extremely unequal, even though in most cases the majority of the counties would have average inequality. This explains why the spread of inequality changes so much when using different geographical scales.
Distinguishing between different scales is important in any research, but what is equally relevant is how scales are interconnected. Murray (2006) claims that we cannot explain one without the other, because different scales can be regarded as „two sides of the same coin‟. He argues that a high scale such as the global level is constructed from local action. Global processes are actually local-to-local flows that have become stretched across the world to become global. Murray emphasizes the interconnections of the system of scales by interpreting it as a “system of multi-fibred networks, characterized by multidirectional flows, nestled within a larger system, which creates at its nodes a mosaicking of space” (Murray, 2004, p. 52).
Both Cape Town and South Africa experience very high inequality, ranking among the most unequal cities and countries of the world. For this, the Gini coefficient is used as a measurement, which gives a score between 0, meaning complete equality, and 100, meaning complete inequality. Cape Town has a Gini coefficient of 67 (UN Habitat, 2012), making it even more unequal than South Africa which has a Gini coefficient of 63, placing it among the ten most unequal countries in the world (UNDP, 2013). Among the reasons for this inequality is the country‟s history: During the Apartheid regime the society was heavily segregated along racial lines. Apartheid and racial inequality will not feature prominently in this research, but both contribute strongly to the inequality measures above (Woolard, 2002). Interestingly, Özler (2007) shows through his research that poverty reduction in South Africa has mainly been spread over geographical areas instead of racial groups, suggesting that geographical location is a primary cause for income inequality in South Africa.
Cape Town is thus a very unequal city. However, high levels of inequality on one scale do not necessarily mean that inequality is high on other scales as well. This could very well be the case in Cape Town, as the inequality is very spatially allocated. De Swardt et al. (2005) show in their research that the inequality between the wealthy areas at the Atlantic seaboard and the poor townships in the Cape Flats is the most extreme in the city. However, this does not necessarily mean that inequality within the Atlantic Seaboard or the Cape Flats is particularly high. Since Cape Town is a very large city, with close to four million inhabitants (SDI and GIS Department, 2012), only stating that the city is very unequal and that some areas are rich and others are poor can severely overlook the situation within certain areas which could contain hundreds of thousands of people. That would be similar to the situation in the US as described in the research by Peters (2013) and Chen and Crawford (2012). An area such as Khayelitsha – a township in the Cape Flats - could very well be in such a situation, as it is roughly
38km2 and has a population of at least 400.000 (SDI and GIS Department, 2012). That
hypothesis is one of the main themes of this research, to assess the inequality within the Cape Flats. Due to the lack of previous research on inequality in only that area, there is a lack of knowledge on that specific topic. As it is generally introduced as an area of poverty, as in the research by De Swardt et al. (2005) or Rogerson (1999), one would expect inequality to be quite low since everyone would be equally poor. However,
6 during multiple visits in the area, inequality actually appeared to be of a significant level. This research therefore aims to assess the level of inequality within the Cape Flats, thereby clarifying whether inequality in Cape Town is due to a dichotomy of rich versus poor areas, or if inequality is spread throughout the city. Through the use of a previously unused geographical scale this research will thus attempt to provide new insights into inequality, which in turn might expand our knowledge on the issue as well as reveal new possibilities to counter inequality (Chen & Crawford, 2012).
2.2 Inequality in context
The following section will be on the context of inequality in this research, being urban inequality. I will explain what characterizes urban inequality as opposed to urban-rural inequality but also why these two types of inequality are so strongly related. Lastly, the specific characteristics of inequality in Cape Town will be discussed.
As mentioned in the previous section, income inequality in Cape Town is higher than it is in South Africa. The inequality within Cape Town, urban inequality, can however not be seen as separate from the rest of the country, which would be urban-rural inequality. De Swardt et al. (2005) and Poswa and Levy (2006) explain how poverty in other parts of the country, most notably the rural Eastern Cape, has an important role for poverty and inequality in Cape Town. Migrants from these areas come to the city, often settling in the Cape Flats area. Poswa and Levy claim that they are drawn to the city due to better job opportunities as well as a chance at better education and healthcare. According to Parnell and Mosdell (2003) the poorest among the South African urban population are generally the people that moved into a city most recently. This influx of poor people into certain areas combined with a lack of employment opportunities in Cape Town results in rising inequality. National inequality thus attracts poor people to the urban areas, thereby increasing urban poverty. Based on a different case, Glaeser (2009) claims that a decline of national inequality would lower the pull factor a city has on migrants. Subsequently, this may result in a reduction of urban inequality, indicating that urban and urban-rural inequalities are very directly related. Glaeser further argues that cities do not only attract poor, low skilled workers but also many highly skilled people. He regards this phenomenon of cities attracting both unskilled and highly skilled people to be one of the main reasons of urban inequality.
Burgers and Van der Waal (2008) and Bartlett et al. (2012) provide a universal discussion of urban inequality. They argue that globalization has led to countries focusing their economic policies on their cities. Growth numbers for South Africa and Cape Town comply with that idea, as average annual GDP growth in Cape Town in the period between 2005 and 2009 was 4,06% (City of Cape Town, 2014) while GDP growth in South Africa over the same period was 3,3% (World Bank, 2014). According to the researches by Burgers and Van der Waal and Bartlett et al., this generally leads to increasing urban-rural inequality: Urban dwellers are wealthier, healthier, better educated, better housed and have more livelihood opportunities. However, this does not mean that the living situation of most urban dwellers is better than that of most rural dwellers. Bartlett et al. claim that previous research has shown that inequality in urban areas is generally more extreme than in rural areas. Furthermore, particular features of living in urban areas, such as a higher cost of living, make the issue of urban inequality even more pressing.
7 The issue of urban inequality will probably continue to be very relevant in Cape Town. As Greig et al. (2007) explain, middle-income countries such as South Africa often experience strong urbanization. In accordance with De Swardt et al. (2005) and Poswa and Levy (2006), Greig et al. also regard rural poverty as one of the main factors in urbanization and subsequent urban inequality. Since urbanization seems unlikely to slow down considerably in the near future, it is important to expand our knowledge on urban inequality and on inequality within Cape Town. This research seeks to do so, to find out if inequality within the Cape Flats corresponds to ideas based on higher levels of scale.
2.3 Measuring inequality
Throughout the previous sections, inequality was mostly interpreted as an economic issue. This section will cover why an economic measure was chosen to indicate inequality.
Measures such as GDP, income and a Gini coefficient are all economic measures. These measures were used in the examples throughout the previous sections because this research will focus on a form of economic inequality; income inequality. The decision to use income inequality in this research is based upon the ideas explained by Ray (1998). He explains that inequality persists across a large range of dimensions, such as education or health. Income inequality will not cover any unequal situation completely, but it is a useful indicator of the intensity of overall inequality and has the potential of leading to revelations related to other dimensions of inequality. Previous research on poverty and inequality in South Africa, such as that by Woolard (2002), also regularly use economic measures. Woolard herself addresses the fact that these measures only tell a part of the story, but she agrees in them being the best proxy.
Earlier, income inequality was introduced as being a specific form of economic inequality. This statement is based upon the distinction made by Sen (1997). He explains that economic inequality is a more comprehensive term, which includes „causal influences on individual well-being and freedom that are economic in nature‟ as well as measures such as income and holdings. Since income is the only inequality measure used in this research, income inequality is the correct term.
An obstacle when using income as a measure, especially in the case of Cape Town, is the informal sector. As mentioned by De Swardt et al. (2005), a wide range of informal activities are essential to the livelihoods of millions of South Africans but these activities are almost entirely invisible. Therefore, income in this research will not include any source of informal activities.
Ray (1998) does criticize the use of income as a measure of inequality, as opposed to consumption. Especially in relatively poor areas, he claims that income is not only often invisible, but also very inconsistent. Since consumption remains relatively consistent even when income drops, which Ray claims is due to the necessity of specific consumption such as food combined with savings or loans, consumption would be a more accurate measure. Özler (2007) used consumption as a measure in his research on
8 poverty and inequality in South Africa. His own comparisons show that his results are consistent with research using income, but his are less pronounced.
2.4 What causes inequality
Apart from proving or disproving the existence of inequality, a large part of this and most research on inequality is determining what may cause it. An unequal distribution of income across people or places can be the result of a variety of reasons. Previous research has however determined a number of factors that tend to have a specifically large role in this. Deciding specifically which factors to include was based upon their relevance in the Cape Flats. In the next sections these factors and their role in the context of the Cape Flats will be discussed in the following order; crime, education, housing and transport infrastructure.
2.4.1 Crime
One of the most important causes of income inequality and urban inequality, according to numerous previous studies, is crime (Bartlett et al. (2012), Demombynes & Özler (2005), Parnell & Mosdell (2003)). Crime as an overarching concept comprises many different actions; the South African Police Services for instance distinguishes contact and property crimes, among others. The outcome of any type of crime for this research is however the same. They all have a negative relation with income in an area and have a positive relation with inequality of any kind. However, researching merely the intensity of crime in one area without a subdivision within that area will not provide information on inequality within that area. Furthermore, there is an important complexity when comparing crime with income inequality with the intent to research the causes of income inequality: Crime can be both a cause and a result of inequality. These different dynamics will be explained below.
Ellen and Regan (2010) discuss the universal ideas on the relation between crime and inequality in urban areas. They claim that one the main causes of that relation is that high crime rates serve as a push-factor. Due to the lack of safety in an area compared to other areas, people are pushed away from the unsafe area. Ellen and Regan argue that the first groups of people being affected by this are those with a high income and a high level of education. The fact that specifically these people are the first to abandon a crime-ridden area has an especially strong effect on income inequality. Firstly, that is because when they leave, the average income in that area drops while it rises in the area to which they move. Secondly, the presence of people with a high income and level of education generally has a positive effect on the development of an area. Therefore, if they leave the area, that will influence current and future inequality in the area.
High crime rates have a similar push effect on businesses and investment. Benyishay and Pearlman‟s (2014) research suggests that the subsequent absence of businesses in an area will result in low job opportunities and thus in low income levels.
As mentioned earlier, crime can also be a result of inequality; Calvó-Armengol et al. (2007) show that few job opportunities and low incomes are among the reasons for people to commit crimes. In turn, this can lead an area down a vicious circle, in which rising crime, decreasing job opportunities and income levels all influence each other in a downward spiral.
9 Demombynes and Özler (2005) offer an extensive discussion on the relations between crime and inequality in South Africa. They provide an interesting new perspective on that theme, suggesting that in a highly unequal city crime rates are highest in the areas that are wealthiest among their direct neighbors. This idea correlates with Becker‟s (1986) economic theory of crime. According to his theory, the incentives for anyone to commit a crime are determined by the different returns from either legal or illegal activities. This theory relates to crime for which return is an incentive, such as property crime. Demombynes and Özler use this theory to claim that travel costs from a low-income area to a high-low-income area are influential on crime rates. If those costs are low, the return from crime can be very high, creating incentive to commit these crimes for those whose legal income would be low. Demombynes and Özler show that crime is highest in the areas that are wealthiest among their direct neighbors, which they claim is due to low travel costs. Their research therefore suggests that high crime rates and low incomes are related, but not combined in the same area.
Another research on crime in Cape Town is that by Lemanski (2004), who studied the spatial implications crime has had in Cape Town. Her research shows that crime rates have been high and rising for a substantial period. Lemanski argues that this has led to a strict distribution of crime and especially vulnerability to crime along both spatial and social lines. These processes of exclusion, „ghettoization‟ and „enclaving‟ are regarded as typical urban problems by for instance Hubbard (2008) and Cochrane (2008). The most notable distribution of crime is that between the wealthy city center and suburbs on one side and the Cape Flats on the other side. Of these two different areas, the Cape Flats experiences far higher crime rates (Donaldson & Plessis, 2013), confirming the ideas posed by Ellen and Regan (2010) and Demombynes and Özler (2005). This research will use those theories to test whether or not income inequality on a smaller scale can also be explained by varying crime rates.
2.4.2 Education
Throughout the previous sections education has already been mentioned a few times, such as in the ideas of Poswa and Levy (2006) or Ellen and Regan (2010). But also in the works by De Swardt et al. (2005), Greig et al. (2007) and Sen (1997) education had a prominent role. Education is seen as an important feature of development issues such as inequality. This next section will discuss the literature on the relation between education and inequality, as well as placing it in a South African context.
Ansell (2005) discusses education extensively in her work, describing it as a central feature of societies. It affects not only the individual students, but also their families (current and future) and society in general. Also, its effects are present in various dimensions, such as economic, social and cultural. Similar ideas are expressed by writers focusing on the South African context, such as Letseka (2014), Spaull (2013), Van der Berg (2008), Woolard (2002) and De Swardt et al. (2005). All these writers claim that education has a positive relation with various socio-economic factors.
Education is sometimes criticized – mainly arguing that education serves to construct people as desired by the elite – but Ansell (2005) is convinced by its positive effects. She writes that school attendance facilitates cognitive development while providing valuable skills and knowledge for the future. Therefore, educated people generally have better economic prospects, better health and greater control over their lives. The research by Van der Berg (2008) and De Swardt et al. (2005) confirm the positive
10 relation between education and future income in South Africa. Parnell and Mosdell (2003) claim that that relation is due to high paid employment being unavailable to poorly educated people in South Africa.
Besides the influence education has on an individual, Ansell (2005) and Letseka (2014) argue that society benefits as well. They claim that education improves the access to information, allowing for better functioning democracy and enhancing social justice. When regarded as having such positive effects on society, education is often included in the term „human capital‟. Many governments and institution such as the World Bank and the International Monetary Fund (Ansell, 2005) use this idea of human capital, in which education is seen as an investment. Based upon this discussion it is fairly straightforward to realize the potential education can have for affecting inequality of many sorts, among them income inequality. As some people enjoy more or better education, they reap more of its benefits. This is why inequality is often seen as a major component of inequality on many scales.
There are several ways to measure educational inequality, the most common being enrollment rates and academic achievement. Enrollment is often focused on enrollment in primary schools, because in international spheres this is thought to provide the widest social benefits at the lowest costs (Ansell, 2005). This is reflected in the research by Spaull (2013), who uses numerous major cross-national comparisons of primary schooling. A problem with this indicator is that enrollment is often uneven across the different levels of education; in many developing countries enrollment at secondary and tertiary levels is much lower than at primary level (Ansell (2005), Gruber and Kosack (2014)). Another measure for education is academic achievement, indicating the highest accomplished level of education per person. This could tell whether or not a person completed a certain level, or in some cases if a person participated in but not completed a level. Farkas (2006) regards academic achievement as one of the most strongly and positively related personal characteristics to an individual‟s later occupational attainment, employment, income, housing, health and other measures of a successful life. The researches by Van der Berg (2008) and Letseka (2014) use academic achievement as a measure for education in South Africa and both show a positive relation with future economic prospects.
Gruber and Kosack (2014) argue that a common idea is that when the poor receive more education they will catch up to the elite, thus lowering inequality. Subsequently, neglecting education for the poor will cause them to fall even further behind. This idea has inspired numerous policies on spreading primary education across the developing world, such as in the United Nations‟ Millennial Campaign for universal primary education. Gruber and Kosack (2014) question this mechanism between education and inequality, criticizing the idea that high primary enrollment rates necessarily reduce income inequality. Their research shows that throughout the developing world high enrollment rates in primary education are generally combined with higher inequality in the future. They suggest that these results are due to what they call a “tertiary tilt”; the situation in which a country invests most of its educational resources on higher education. However, their research also showed that countries without a tertiary tilt, concentrating investments on lower education, do experience the commonly assumed positive relation between primary enrollment rates and inequality (Ansell, 2005).
11 The South African education system is often criticized. The country spends a large part of its national budget on education, around 20 percent (Visser, 2012). Despite that, the education system in the country is generally regarded as being of extremely low quality. According to Letseka (2014), the system is in a crisis and can be regarded as a national disaster, is inefficient and ineffective in its use of resources, making it essentially dysfunctional. Furthermore, the system lags behind that of various much poorer countries which spend less on education that South Africa. Van der Berg (2008) also heavily criticizes the education system, showing in his research that there is severe inequality between different schools. Students of poor and rural schools perform much worse than students of rich and urban schools, resulting in large labor market inequalities. This inequality between rich and poor schools also shows in Spaull‟s (2013) research, who claims that South Africa has developed two education sub-systems, of which only one is functional and actually able to educate students. The dysfunctional education for the poor entrenches them in poverty due to low labor market prospects. This could mean that going to a specific school has a stronger relation with income than educational achievement.
2.4.3 Housing & transport infrastructure
In the final section of this chapter I will discuss in what way housing and transport infrastructure relate to income inequality and why it is relevant in the case of the Cape Flats. These topics are discussed in one section since literature on these topics often combines them.
Housing and transport infrastructure both refer to certain services and physical structures necessary in a modern society. In this research, the specific services that will be covered by housing are water, sanitation, energy and the type of dwelling (indicating whether housing is formal or not); transport infrastructure covers roads and railways. These services are often included in the more comprehensive term „infrastructure‟, such as in Swilling‟s (2006) research on planning in Cape Town. In Swilling‟s research however, there are extra services included in the term infrastructure which are not included in this research.
Calderón and Servén (2008) did an extensive research on the relation between infrastructure and inequality in the context of Sub-Saharan Africa. Their results show that transport infrastructure development, if under the right conditions, has a positive impact on income levels and a negative impact on income inequality. They explain the negative relation between transport infrastructure development and income inequality using the notion of disproportionate benefits, as transport infrastructure provision may have a disproportionate effect on the income of the poor. This would be due to raising values of their assets, such as land or human capital, or by lowering transactions costs. Bond (1999) offers an additional explanation of the relation between transport infrastructure and income inequality. He claims that developing transport infrastructure increases overall work productivity, supports the growth of local businesses and raises the incentive for new businesses to be started in the area. As a result of these developments, incomes in the area will rise. Therefore, when these services are unequally distributed across different areas - which Swilling (2006) shows they are in Cape Town - this may cause income inequality to grow.
12 Concerning housing, Rogerson (1999) discusses how housing services are related to income inequality. His explanation is based on the idea of vulnerability and assets to counter it. Housing services are part of these assets, which can counter the vulnerability to loss of income due to events such as illness or general life-cycle changes. Rogerson claims that the urban poor are especially vulnerable to such events. Subsequently, improvement in the quality access to housing services can increase the assets of especially the poor, reducing their vulnerability to income loss. In turn, this may have a negative impact on income inequality.
According to De Swardt et al. (2005) and Swilling (2006) the current distribution of infrastructure in Cape Town, including housing and transport infrastructure, is unequal. The connection between the Cape Flats and the rest of the city, by roads and railways, has been neglected while the connection between the city center, the wealthy suburbs and the major national highways is far superior. Housing in the Cape Flats is generally of low quality, with large numbers of people living in informal housing and too many households without water, proper sanitation or electricity.
Finally, Özler (2007) refers to the former Apartheid regime in his research on inequalities in South Africa. He regards the inequalities in housing and transport infrastructure as an inheritance of that regime. When Apartheid led to the creation of the townships in the Cape Flats, the government did not provide proper housing or infrastructure. Özler claims that negligence of the area formed the foundation of current inequalities.
13
3.
Methodology
In the following the chapter the methods used to perform the analyses will be explained, including descriptions of the case, the units and the data. In the first section the case of the research will be described, including explanations of its history and the South African context in relation to the current situation. Secondly, the research questions will be introduced. After that, there will be a section describing the units of this research. That is followed by a section on the data, which is divided into four parts; income, crime, education, housing and lastly transport infrastructure. The final section of this chapter will explain the analytical methods.
3.1 The Cape Flats and the townships
The case for this research is the Cape Flats, an area in the south-east of Cape Town as can be seen in Figure 2. Cape Town itself is the second largest city by
population in South Africa, after
Johannesburg, and is one of three capital cities. The population of Cape Town in 2011 was about 3,7 million, of which roughly 1,4 million live in the Cape Flats (SDI and GIS Department, 2012). It being such a large and important city has made it into a quintessential example for various urban phenomena in developing countries. This is particularly the case when discussing spatial segregation, due to the city‟s townships. Most of these townships are located in the Cape Flats; in this section the formation of the current situation in the Cape Flats will be discussed.
There are few things that have left such an imprint on the South African society as Apartheid, lasting nearly 50 years before ending in 1994 (Rowntree et al., 2012). Apartheid refers to an era during which the white South African minority passed a series of Acts to circumscribe the mobility of colored, Asian and African people. These Acts had various implications throughout the society, such as laws that denied non-whites access to major cities unless they were in possession of a pass that indicated that they were employed in the city. Due to this regime, Apartheid-era South Africa is often presented as an extreme example of socio-spatial segregation (Daniels et al., 2008; Greig et al., 2007; Rowntree et al., 2012). As the history of Apartheid is extremely expansive, diverse and complex, it will not be covered extensively here. Its critical role in shaping the current Cape Flats can however not be overlooked.
In Cape Town, Apartheid led to a distinct pattern of spatial segregation. The white population settled at the Atlantic seaboard and the slopes of Table Mountain, the most desirable portions of the city. The black, colored and Asian populations were forced to move to the remaining, less desirable locations; mainly the Cape Flats (Rowntree et al., Figure 2: A simplified visualization of the
conglomerates in Cape Town
14 2012). The latter is an area in the south-east of the city, bordering on the Indian Ocean. Townships were established across the Cape Flats, each to be populated by one specific racial group. This racial segregation, both between the Cape Flats and the rest of city as well as within the Cape Flats, is still very relevant (SDI and GIS Department, 2012). For instance, the population of Langa, the oldest township in Cape Town, is still over 95% black whereas the population of Mitchell‟s Plain is over 90% colored. Since the white rulers during Apartheid deemed the non-whites inferior, the Cape Flats became an area of widespread underdevelopment (Daniels et al., 2008). Some common issues in the area are poverty, unemployment, crime, drugs, poor education and poor housing such as shacks (De Swardt et al., 2005). Adding to the history of the area, the post-Apartheid era brought new challenges, such as a large influx of poor migrants looking for work (De Swardt et al., 2005; Poswa & Levy, 2006). Migration has led to Khayelitsha becoming the largest township in the country; estimates range from 1 up to 1,6 million inhabitants (due to the informal nature of many townships, definitive numbers are not available) (Brunn & Wilson, 2013). The Cape Flats is thus an area whose history is based on racism and purposeful underdevelopment and where modern times brought new challenges. In this research I will seek to go past the notion of the Cape Flats versus Cape Town by looking at what is happening in the area itself. Modern times may have brought new challenges, but there has also been important progress (De Swardt et al., 2005). The current situation, formed by progress away from history on one hand and the development of modern challenges on the other, is central to this research.
The Cape Flats is not an administrative area nor does it have definite borders. It refers to the flat area in the south-east of the city where most of the city‟s townships are. Since there are no fixed borders, I had to decide what to include and what not; that will be discussed in one of the next sections.
3.2 Research Questions
In the introduction of this paper the main research question was introduced: Is there
income inequality across the Cape Flats and if so, what could have led to this inequality? To promote the clarity of the research, this question has been divided into a
number of sub-questions. The topics of these sub-questions have been discussed in the literature review and the questions themselves have been based on the literature.
The first sub-question refers to the first part of the main question: Is there income
inequality across the Cape Flats and of what level is that inequality? As it is practically
impossible for there to be absolutely no income inequality, determining the level of that inequality is the most important part of this question. Knowing the level of inequality, one can argue whether income inequality across the Cape Flats is significantly high or low.
The next sub-question refers to the second part of the main question and is based on the hypothesis that there is income inequality across the Cape Flats. This question is a combination of four very similar questions only differing in one variable: Is there a
relation or a causal relation between income and any of the following variables in the Cape Flats; crime, education, housing and transport infrastructure?
15 An important part of this question is the distinction between a relation and a causal relation: The former is “an association between two variables whereby the variation in one variable coincides with variation in another variable” (Bryman, 2008, p. 698). The latter includes causality, “a concern with establishing causal connections between variables, rather than mere relationships between them” (Bryman, 2008, p. 691). The aim of this research is to identify what causes income inequality across the Cape Flats, which can only be done by identifying causal relations. Mere relations would not provide an answer to the main question of this research.
The final sub-question relates to the issue of geographical levels of scale, since most ideas on urban inequality are based on research on a larger level of scale: Are the ideas
in current academic literature applicable to the small scale of the Cape Flats? For this
question, results of the analyses will be compared with the literature as presented in the previous chapter.
3.3 Units
This research will essentially consist of a comparison of units representing the population of the Cape Flats. As there is no data available at the household level, areas will be used as the unit of analysis. There are five different divisions of Cape Town in use which divide the city into administrative areas: Subcouncils, wards, suburbs, health districts and planning districts. Any data made available by the City of Cape Town is available for all these areas, so excluding any of them will not cause exclusion of data that would otherwise have been available. The borders of these areas do not correspond, except for wards and sub-councils; subcouncils are an agglomeration of multiple wards. Finally, all the borders are chosen politically and thus subject to change after elections. All the information on these areas is provided by the Strategic Development Information and Geographic Information System Department of the City of Cape Town, from now on referred to as SDI and GID Department.
Figure 3: The combined area of the wards
16 Figure 4: The selection of units representing the Cape Flats
Subouncils The units as a part of Cape Town Wards
Deciding on which areas were most suitable for this research was primarily based on scale. As the research is only on the Cape Flats, large areas would be unsuitable since they would provide only a few units. Using small areas means that there are more units for the analyses, increasing the significance of the results. For this reason, the units in this research will be wards, as those are the smallest areas available and thus provide the most units. As the Cape Flats is not a distinct area with solid borders, the wards which cover the Cape Flats best are included in the research. The subcouncils were also taken into account when making the selection by ensuring only complete subcouncils were included. This resulted in a total of 43 wards being included as units, which together form 10 subcouncils. Figure 3 shows the area of all the chosen units combined and Figure 4 shows the selection of the wards as a part of Cape Town, by themselves and as subcouncils. None of the wards has a name; they are only indicated by a number. Townships such as Khayelitsha, Langa and Nyanga are all divided into several wards, so these names will not be featured in the analysis.
3.4 Data
There will be one dependent variable and three independent variables in the analyses for this research. All these variables are indicated using data of varying natures and from different sources. This section will describe the data and how it was operationalized, using different subsections for each variable.
17 Because of practical reasons, this research is based upon secondary data, which brings both advantages and disadvantages. The main advantage is saving on costs and time due to not travelling to the Cape Flats. Secondly, since a lot of the data is from official sources, it is of high quality. Disadvantages of using secondary data would be not having control over the data quality as well as lacking certain data that could be of assistance to the research (Bryman, 2008).
3.4.1 Income
Income is treated as the dependent variable in this research. The data for this variable is retrieved from the most recent publication of the national South African census, which was undertaken in 2011 and published in 2012. The census data is supplied by Statistics South Africa and has subsequently been compiled for Cape Town by the SDI and GIS Department (2012). The census is publicly available online and includes various demographic, economic and social statistics, among them income. Income data is available as a distribution of income groups in a specific area (such as wards). As is often the case with South African data, these income groups have been further divided into five racial groups. However, as this research will not cover race, only the data of all those groups combined has been used.
There are ten income groups, including „no income‟ and „unspecified‟. As none of the wards in this research contained anyone within the „unspecified‟ group, that group can be ignored, resulting in nine effective groups. The data is presented as the monthly household income in South African Rand (ZAR); in 2011 one Rand was equal to roughly 0,10 Euro. For every group a percentage is given, indicating the relative amount of households in the area receiving a certain income. Table 1 shows the income data as it is presented in the census.
For the results of the analyses to be most significant, it was desirable to have one indicator for income per ward, instead of nine. The most desirable indicator would have been the true median income. Median is preferred over average income, as the latter is more strongly influenced by a few units with extreme measures, resulting in less realistic data (Bryman, 2008). However, as all households are already divided into groups, calculating a true median income is not possible because that would require the exact income of every household to be known. Therefore, a median has been calculated using the median of every group. This means that the group with an income of between Table 1: Income distribution in the Cape Town census (2011)
18 3201 up to 6400 ZAR was regarded as all having an income of 4800 ZAR. Using this method, a median income was calculated for every ward. In the case of ward 41 this resulted in a median household income of 5338 ZAR per month.
3.4.2 Crime
Since neither the census nor the City of Cape Town provides data on crime on the level of wards, the data measuring this variable will be retrieved from a different source. The South African Police Services (SAPS) supply crime rates of all their precincts throughout the country (SAPS, 2013). However, these precincts are relatively large areas which do not overlap with any of the administrative areas used by the City of Cape Town. Because this data is measured according to different units than the income data, it cannot be used in a statistical analysis. This variable will therefore be analyzed using spatial methods. Figure 5 shows the selected precincts and Figure 6 shows an overlay of the precincts and the wards. The latter shows that the areas do not overlap and that there are parts of the wards that are not covered by the precincts as well as the other way around.
The SAPS‟ data indicates the occurrences of specific criminal acts. These acts are all categorized according to seven categories: Contact crimes, contact-related crimes, property-related crimes, crimes heavily dependent on police action for detection, other serious crimes, subcategories of aggravated robbery and finally other crime categories. Per category there are a number of more specific subcategories, which can be seen in table 5 in the appendix. The data that is used in this research is that of the period between April 2011 and March 2012, since the census providing the income data was performed in October 2011.
Figure 5: The police precincts in the Cape Flats
19 For the analysis, the sum of the crime rates of every precinct was calculated, both per category as well as in total. Subsequently, the outcomes of those calculations were combined with spatial data of the precincts made available by the SDI and GIS Department (2014). This allows for the visualization of a spatial distribution showing the different levels of crime rates across the precincts. Subsequently, the distribution of crime can be compared to the distribution of income across the wards.
Source: SDI and GIS Department (2014)
3.4.3 Education
The 2011 census from which income data was retrieved also provides data on educational achievement. This data indicates the highest level of education people aged over 20 have enjoyed. There are seven groups, including „No schooling‟ and „Other‟. Since the „Other‟ group is always very small, never above 1%, and it cannot be ranked among the other levels of education, this group will be ignored. There are thus six effective groups in the analysis. This data is very similar to the income data; it shows the amount of people that reached a certain level of education relative to the whole population of the ward, as can be seen in Table 2.
20 As was the case with the income data, the educational data had to be manipulated so it would be suitable to use in a regression analysis. Again, the goal was to have just one measure of educational achievement per unit, instead of a distribution among six groups for every unit. The measure that was chosen to indicate educational achievement was the percentage of the population that completed grade 12, also known as matriculation in South Africa, which is similar to completing secondary education. In the case of ward 41, shown in table 2, this measure would read 90,8%.
There are two reasons to choose specifically secondary education as the measure. First, research by Woolard (2002) shows that poverty rates drop significantly with the completion of secondary education. Secondly, using the census data, this measure showed a relatively large range and standard deviation, which is particularly useful for researching inequality (Bryman, 2008).
Gruber and Kosack (2014) suggested the idea of a tertiary tilt, which is explained in the literature review. To test their idea in the case of the Cape Flats, part of the analysis will be to determine whether or not South Africa has a tertiary tilt. After that, the relation between primary enrollment rates and income inequality in the Cape Flats will be analyzed to determine whether or not that relation matches Gruber and Kosack‟s idea. The primary enrollment rate can be calculated by excluding only the groups „no schooling‟ and „other‟, as secondary and higher levels require a completed primary education.
3.4.4 Housing
Housing is the final variable for which the data is retrieved from the 2011 census of Cape Town. The services included within housing are water, sanitation, energy and the type of dwelling; the latter refers to a dwelling being formal or informal. As was the case with the income and education data, the census indicated how many households belonged to a certain group in percentages. Tables 6 up to 9 in the appendix show the data as it was presented in the census. As can be seen in those tables, all four indicators were measured using various groups. For instance, energy was divided across nine groups according to the energy source, such as electricity or coal, as well as being divided across three groups according to the application of energy; heating, cooking and lighting. After manipulation using averages, each service is measured according to just one value, such as the percentage of households with running water on the property. The manipulation as described above still leaves four indicators per unit. Using these four indicators all separately negatively affects the results of the regression analysis: Table 2: The distribution of educational achievement in the Cape Town census (2011)
21 Since there are relatively few units, a small amount of variables is preferred. If many separate variables are used in the analysis, the significance of the results is very low. Especially since the goal is to include housing, education and income in one multiple linear regression analysis, it is best to combine the separate indicators into one. Therefore, an average is calculated using the percentages indicating electricity, water, sanitation and the type of dwelling. That average is used in the analyses, referred to as housing.
3.4.5 Transport infrastructure
Quantitative data on levels of transport infrastructure per ward is not available, thereby rendering it impossible to include this variable in any statistical analysis. Spatial data, however, is available, allowing for spatial analysis similarly to the analysis of crime. The spatial data shows all major roads and railways as well as railway stations, as can be seen in figure 7. The data is available upon request at the City Maps Unit of the City of Cape Town.
Their dataset provides information on roads all across the city, including every single road and street. Since many of those streets are minor streets in residential areas, they have not been included in this research. As there are so many of those streets, they would negatively affect the legibility of the eventual map, thereby complicating the spatial analysis. Moreover, the major roads are more significant in determining the distribution of transport infrastructure. Because of this, only highways, major roads and main streets have been included in the analysis, as these are most significant for the level of transport infrastructure and do not interfere with the legibility of the map. Concerning railways, all the data has been used without manipulation. Combining both railways and roads provides a very suitable indicator of the level of transport infrastructure across the Cape Flats. The resulting image will be used in a similar way as in the spatial analysis on crime; by comparing it with the distribution of income across the area.
Figure 7: The railways and roads across the Cape Flats
22
3.5 Analytical methods
For the quantitative part of the research, regression analysis will be used to assess the relation between the dependent and independent variables. A regression analysis is the most suitable method for the scale of the variables in this research, which have all been codified to be of ratio scale. This analysis will result in several measures indicating for instance the strength of correlations between an independent and the dependent variable as well as the significance of that correlation (Bryman, 2008). Afterwards, if there are significant residues which could not be explained through this analysis, spatial analysis may be used to look for possible explanations.
When committing to regression analysis, a number of features of the data should be checked. First of all, the number of units influences the significance of any relation; generally, a higher number of units will result in a higher significance. As this research focuses on just the Cape Flats, there are 43 wards to be used as units. This is not an exceptionally high number of units, so the results may be insignificant. This would have been different if there was data available on a smaller scale, such as on the household level, or if the research was on a larger area, for instance all of Cape Town. As the former option is unavailable and the latter would result in an altogether different focus of the research, neither can be used. To prevent severely insignificant results, a third option is to use a small number of variables or few indicators per variable (Bryman, 2008). This has been achieved by manipulating the data of each variable, resulting in only one indicator for each variable.
Secondly, when performing a multiple regression analysis, it is important to be aware of any multicollinearity; referring to a relation between the independent variables (De Vos, 2012). If that correlation is very strong, it would negatively affect the reliability of the results. Concerning education and housing, the independent variables to be used in the multiple regression analysis, there is no worrying level of multicollinearity; they account for roughly 10% of the other‟s variance, meaning that the results of the analysis will be reliable.
Thirdly, skewness of the variables must be checked in order to determine whether any units have such extreme values that they influence the results. All the variables have an acceptable level of skewness; only the dependent variable‟s skewness is relatively high, but not to a level that it would severely affect the results. This is likely to be due to the low number of units and the fact that some wards have an exceptionally high income. Finally, all variables should have a linear distribution in order to perform a regression analysis. Since all the variables such a distribution, all the prerequisites for regression analysis are met (De Vos, 2012).
Spatial analysis consists mainly out of comparing spatial distributions. This method is not suitable to determine whether a causal relation is present; only a relation can be assessed. Apart from analysis, spatial methods serve a visualization purpose, as has been done throughout this and the previous chapters. Such images may support the outcomes of the quantitative research in a visual way. Lastly, spatial analysis may serve to analyze and explain any residues from the quantitative analysis (De Vos, 2012).
23
4.
Analyses
This next chapter aims to provide results with which answers to the sub-questions and ultimately the main research question can be formulated. The outline of this chapter is unlike in the previous chapters: First the statistical analyses will be covered, starting with income inequality, then education and finally housing. After this, the spatial analyses will be presented, first on crime and then on transport infrastructure.
4.1 Income inequality
The first analysis will be dealing with the dependent variable; income. For this research, income is indicated by the median monthly household income per ward in 2011. Before discussing income at the level of the wards in the Cape Flats, it is useful to know the median income in for instance South Africa or Cape Town as a point of reference. That would clarify what level of income is normal and what is exceptionally high or low. The data used in this research for the Cape Flats is derived from the national census of South Africa. Using that data, median incomes of the Cape Flats and Cape Town have been calculated. However, there is no data regarding income distributions available for the regional level (the province of the Western Cape) or for South Africa as a whole. Therefore, to make a comparison between regional, national and local income levels, the national and regional median income levels provided by official sources are used. The calculations of the local level showed that the median monthly household income in the Cape Flats area is 6252 ZAR and in the whole of Cape Town it is 12988 ZAR (SDI and GIS Department, 2012). The report based on the 2011 census show that this income in the Western Cape is 11955 ZAR and 10694 ZAR in South Africa (Statistics South Africa, 2012). These figures show the large inequality in the country as well as in Cape Town; the median income in the Cape Flats is much lower than at any of the higher scales.
An extra reference point for these income levels is the South African poverty line, also provided by Statistics South Africa (2014). They provide three lines, but they are all at the per capita level: The food poverty line (FPL) at 321 ZAR, the lower-bound poverty line (LBPL) at 443 ZAR and the upper-bound poverty line (UBPL) at 620 ZAR. The Cape Town census (2012) shows that the average number of household members is 3,5. Therefore, estimating the poverty lines at the household level results in an FPL of 1124 ZAR, a LBPL of 1551 ZAR and a UBPL of 2170 ZAR. This is an estimate, as the poverty lines are calculated based on the costs of the goods an average individual needs. For children these costs are generally lower, and since households generally include children the poverty line for households would be lower than these estimates (Ray, 1998). The median income of the Cape Flats is well over these poverty lines and only one ward has an income at roughly the level of the UBPL; 2176 ZAR. Overall, the above data shows that the median income in the Cape Flats is very low compared to income at other scales, but still much higher than even the UBPL.
Table 3 shows the income data of the different wards in the Cape Flats. A number of conclusions can be drawn from this table. Firstly, there appears to be very high income inequality based on the distribution across the wards; the income ranges from 2176 ZAR up to 25086 ZAR, the latter being over eleven times higher than the former. However, this result is somewhat misleading as there are a few wards with an
24 exponentially higher income than all of the other wards. Nonetheless, 2176 ZAR is roughly the UBPL of South Africa, suggesting that this ward is exceptionally poor, whereas 25086 ZAR is more than double the median income of the country, suggesting that this ward is exceptionally rich.
Table 3: Median monthly household income per ward in the Cape Flats (2011)
Secondly, the income distribution in table 3 can be used to calculate the Gini coefficient, which is the most widely used measure for income inequality. The Gini coefficient was previously mentioned in other chapters, indicating the high levels income inequality in both South Africa and Cape Town. The Gini coefficient relates to the distribution of the total income in an area across its population, resulting in a score between 0, meaning perfect equality, and 100, indicating total inequality. The Gini coefficient of the wards in the Cape Flats is 36, which is much lower than that of Cape Town, 67, and South Africa, 63. Inequality in the Cape Flats is thus clearly lower, but this was to be expected, since the selection of the Cape Flats was based on excluding the wealthy parts of Cape Town. The level of inequality in the Cape Flats is on a similar level to that in for instance Dakar (Senegal), Dar es Salaam (Tanzania), and Jakarta (Indonesia) (UN Habitat, 2012). These are large cities in developing countries, whereas the Cape Flats is only an area within a city. The fact that this single area experiences a similar level of inequality as those complete cities implies that the level of inequality in the Cape Flats is of a significantly high level level. The Cape Flats can therefore not be regarded as one homogenous, poor area. In turn, this suggests that inequality in Cape Town consists of more than a spatial dichotomy between the wealthy and poor areas; inequality runs deep in the city‟s fabric.
A visualization of the income data provides more information on exactly how income is distributed across the Cape Flats, as can be seen in Figure 8. The map shows the income distribution across the wards according to five levels of income. It however only shows the distribution of income; the level of inequality cannot be drawn from it. As can be seen in Figure 8, wards with a similar level of income are generally grouped together. Such a pattern suggests that there are relations between a ward‟s location and its level of income. These relations will be further investigated in the next sections.
Source: SDI and GIS Department (2012)
Ward Income Ward Income Ward Income Ward Income
18 4348 44 5866 75 8832 91 2874 33 2647 45 4810 76 10063 92 6963 34 2803 46 9110 78 10150 93 3171 35 3143 47 3998 79 6442 94 6437 36 2754 48 15432 80 3223 95 2581 37 2807 49 9088 81 9629 96 2809 38 4883 60 25083 82 5130 97 3907 39 2651 63 21346 87 3059 98 3335 40 3045 65 9721 88 4292 99 4361 41 5338 66 9468 89 2176 110 10123 42 3234 68 7074 90 2527
