Urbanisation & urban-rural disparity : how does urbanisation affect the urban-rural disparity in the developing world?

(1)

1

Urbanisation & Urban-Rural Disparity

How does urbanisation affect the urban-rural disparity in the developing

world?

Master Thesis Political Science - Political Economy

Growth & Inequality in the Developing World

Date: 29-06-2018

Author: Jacob Meijer (10386785)

Supervisor: Dr. Sijeong Lim

Second Reader: Dr. Lukas Linsi

(2)

2

Abstract:

All developing countries are experiencing urbanisation at the moment which has all kind of implications. One of these implications is the impact of urbanisation on the urban-rural disparity in a country. The academic literature has not produced a coherent answer to this problem yet. Different researches produce conflicting results and use different concepts for both the variables in this causal mechanism. The first part of this thesis consists of a large-N analysis on all the developing countries which shows that the urbanisation rate has an increasing effect on the urban-rural disparity. In the second part of the research a case study on Rwanda will be conducted to analyse the causal mechanisms that interfere in the relationship between urbanisation rate and urban-rural disparity. The results of this analysis show that the political mechanisms are the most decisive in the case of Rwanda. This leads to a mechanism where high rates of urbanisation lead to political conditions that favour the urban area which in turn will lead to higher urban-rural disparity.

(3)

3

1. Introduction

Imagine a world where more than half of the population is living in cities. For citizens of the developed world this is already reality for quite some time and it is only a matter of time before this also becomes reality for the developing world. It is now the turn for the developing countries to urbanise, and they are not doing this in a slow pace. Although they are all experiencing rapid urbanisation, both the urbanisation rates and levels still highly differ between the different countries. Most of the Asian countries will convert into countries with higher shares of urban citizens than rural citizens in the coming years, while for most of the African countries this will take up to another 30 years (Feng, 2015). So even though these countries are all urbanising fast, the majority of the citizens still live in the rural areas and will in some cases keep doing so for over three decades (Cali & Menon, 2012: 172). This changing distribution also has important effects for the divide between the growing urban area and the shrinking rural area. Different authors and reports (Kuznets, 1955; Liddle, 2017; World Bank, 2009; The World Economic Forum, 2015) argue that this changing distribution will lead to a higher urban-rural disparity in favour of the urban area. This higher disparity does not only have negative effects for the rural areas, but also for the country as a whole. It effects the willingness of the rural citizens to invest in human capital. This leads accordingly to less material capital investments which leads to economic stagnation (Chao & Tan, 2017: 256). This reduction of the economic growth is also caused by a declining efficiency in the developing countries that is caused by the rapid urbanisation (Bradshaw, 1987; 227). A high urban-rural disparity is thus harmful for the whole country. On the other hand, urbanisation is needed in order for a country to develop economically (Henderson, 2010: 515). It is even argued that sustained economic development does not occur without urbanisation (Ibid.). In this sense urbanisation is an unavoidable evil that developing countries have to get themselves through. But does the urbanisation always have to have negative effects on the urban-rural disparity? There are also authors (Cali & Menon, 2012; Hnatkovska & Lahiri, 2018; Ravallion e.a., 2007) who argue that urbanisation does not have to lead to a higher urban-rural disparity or can even lead to a lower urban-rural disparity. By understanding this mechanism of this causal inference, the negative effect of the urbanisation on the urban-rural disparity might be addressed and even solved. In this way this research will not only participate in the academic discussion around the urbanisation and the urban-rural disparity, but will also try to address the social problem that belongs to this causal relationship. Next to this social problem, this research will also provide an addition to the existing literature by using a different approach than used in other articles. Next to the different outcomes they produce, the different researches also use contrasting concepts for both urbanisation and urban-rural disparity which can have essential consequences for their outcomes. Urbanisation level is the most

(6)

6

used concept for urbanisation, while other authors argue that it is actually urbanisation rate that has an effect on the urban-rural disparity. Both these concepts capture quite a different dimension of the urbanisation and can have a different influence on the urban-rural disparity. Next to this discussion, there is also dispute on whether to use the urban-rural inequality or the urban-rural poverty gap as the concept for disparity, these two concepts will also have different outcomes and different implications. The academic relevance of this research lies not only in trying to dispute these discussions on concepts and whether they have an increasing or decreasing effect, it will go further than that. The most interesting addition is not testing in what way this mechanism works, but how this mechanism between urbanisation and urban-rural disparity works. This brings me straight to the main research question of this research: How does urbanisation affect the urban-rural disparity in the developing world? Before this question can be addressed, the literature review will present the different theories on this theme and the discussions that are held amongst their supporters. This will also provide room for discussion on some possible causal mechanisms that can interfere in the relationship between urbanisation and urban-rural disparity. These different political and economic mechanisms will be further developed in the theoretical framework where the different scenarios they can create will be introduced. In order to address this question of how the urbanisation affects urban-rural disparity a mixed method approach will be used. The first part will be a large-N cross-sectional analysis on developing countries where the relationship will be tested and explored. This will show the kind of effect the urbanisation has on the urban-rural disparity, whether it increases or decreases the urban-rural disparity and whether it is the urbanisation level or rather the urbanisation rate that has an impact on the urban-rural disparity. This analysis will also already test some parts of the mechanisms that were discussed in the theoretical framework. These will in turn be tested more in-depth in the case study on Rwanda. In this case study the different mechanism will be used to see whether they can explain how in Rwanda, and also possibly in other cases, the urbanisation does or does not lead to a widening urban-rural disparity. After these two different approaches, it will be clear that it is the urbanisation rate and not the urbanisation level that has an increasing effect on the urban-rural disparity. This can be explained for the greater part by the political mechanism and lesser by the economic mechanisms. The fast urbanisation leads to favourable political circumstances for the urban area which in turn will lead to a higher urban-rural disparity.

(7)

7

2. Literature Review

This thesis will start with the literature review that provides an overview of the existing studies on how urbanisation shapes urban-rural disparity. The first subsection focusses on conceptual issues. Here it will already become clear that urbanisation and urban-rural disparity can be interpreted in multiple ways. Urban-rural inequality and urban-rural poverty gap are both used as concept for the urban-rural disparity. These different concepts can often also be explained by different mechanisms. The same applies to urbanisation, where this can be either the level or the rate of urbanisation. The second subsection discusses different economic mechanisms that might explain this causal relationship between urbanisation and urban-rural disparity. The growth of the manufacturing sector and the agricultural sector for example is one of these mechanisms. Another mechanism discusses the different possible spill-overs from the urbanisation. The third subsection reviews the political mechanisms that might explain the causal relationships. These can either be based on the urban collective action potential that will create a higher urban-rural disparity or on the elections and the political allegiances that can create an urban bias or a rural bias. This literature review will end with an overview of the addition of this research to the existing literature.

2.1. Urbanisation & Urban-rural Disparity

This first part will discuss different important aspects of the existing literature. First, it will discuss how various studies have defined and operationalised both urban-rural disparity and urbanisation. Second, it will discuss how these choices affect their theoretical expectations and empirical findings. 2.1.1. Overall Inequality

The idea of a relationship between urbanisation and urban-rural disparity finds its base in a 1955 article by Kuznets. In this article he developed a theory on the relationship between urbanisation and overall inequality. This theory will be briefly introduced, as well as a reaction on this theory. After that the focus will be on the main interest: the urban-rural disparity. Kuznets (1955: 7) argued that urbanisation will in the first period of this process lead to a rising inequality until it reaches a turning point, after that higher levels of urbanisation will lead to lower inequality. This relationship in a bell-shape is now better known as the Kuznets curve. Urbanisation means an increasing share for the more unequal parts of the two, the urban. The turning point happens when the lower-income groups within the urban population who are often former rural-urban migrants can take advantage of the possibilities the urban life offers them (Kuznets, 1955: 17). This means a lower inequality in the urban which according to Kuznets will lead to a lower total inequality. What he does not address is that this rise of the urban area is also often leading to a bigger gap with the rural area, which leads to bigger overall inequality. Kuznets also brings up a political reason, as the urban lower-income groups acquire more political influence, policy will be implemented to create a more

(8)

8

adequate share of the growing income of the country (Kuznets, 1955: 17). This argument again only focusses on the urban area; if urban lower-income groups acquire more political influence, this means that the share of political influence for the rural area will further decline. This skewed distribution of political influence among urban and rural areas can lead to an overall higher inequality.

Different articles have since reacted on different aspects of this idea formed by Kuznets, I will focus here on an important critique by Henderson (2010). The research by Kuznets was done in the 1950s and thus focused on the urbanisation of todays’ developed countries like Germany, the United Kingdom and the United States. These countries took almost 100-150 years to experience full urbanisation (up to 60-85% urbanised). The same full urbanisation for current developing nations often takes around 30 years. Henderson (2010) talks about the effects of this quicker urbanisation. “Rapid urbanization is traumatic, requiring massive movement of population, replacement of traditional institutional and social structures with modern ones centred in a formal legal apparatus, and massive local and intercity infrastructure investments with the required financing mechanisms, all in a short time span of time (sic)” (Henderson, 2010: 516). In response to this rapid urbanisation todays’ developing countries have explicit national urbanisation policies to control and direct the urbanisation (Henderson, 2010: 515-516). This enhanced role of the national government in the urbanisation process, can also result in a bias, where either the urban or the rural areas are favoured in terms of resources (Henderson, 2010: 537). This links to the political aspect of urban or rural bias that will be discussed later on in this literature review. This rapid urbanisation can change the causal mechanism as Kuznets described, the effects of this urbanisation will most likely be more severe and this can lead to a higher inequality caused by the urbanisation. These remarks by Henderson (2010) serve as a good introduction to the next part. He argues that this rapid urbanisation will particularly have drastic effects on the differences between the urban and the rural areas. Subsections 2.1.2. and 2.1.3. will discuss two parts of this difference between urban and rural areas. The notion of the rapid urbanisation is also a good introduction to subsection 2.1.4., that discusses the urbanisation level and the urbanisation rate as potentially distinct forces in the causal mechanism. 2.1.2. Urban-Rural Inequality After above mentioned possible causal relationship between urbanisation and overall inequality, the next step of several articles was to focus on the relationship between urbanisation and urban-rural inequality. This is a logic step as these two concepts are inherently connected. Urban-rural inequality is also highly connected with inequality as it explains 40% of the overall inequality in a country and in time the urban-rural distortion is responsible for generating even more inequality within countries (Young, 2013). In countries that are in the early stages of the development process, this urban-rural

(9)

9

inequality can often even explain more than 70% of the overall inequality. Because these two often overlap it seems logic that a similar causal mechanism would be expected. Williamson (1965) adopted a same kind of reasoning when he argued that urban-rural inequality starts to rise once a country develops, at a certain point this inequality will peak and after that the development will lead to a declining urban-rural inequality. As explained before, Kuznets (1955) saw urbanisation as one of the main parts of development. With this in mind the urbanisation will have a causal relationship with the urban-rural inequality in the same way as the Kuznets curve. This theoretical idea is also supported by a report of the World Bank (2009:10), they found that urban-rural gaps in consumption rise with urbanisation until a certain threshold is reached, after that the urban-rural inequality gets smaller with rising urbanisation.

There are also several studies with conflicting results. Hnatkovska & Lahiri (2018) conducted a comparative study on China and India. They found that there is a linear relationship between urbanisation and urban-rural disparity. They argue that increased urban labour supply caused by urbanisation tends to reduce the relative price of urban labour, this results in lower urban wages and a declining urban-rural inequality. This finding goes against the hypothesis of Kuznets (1955) and the results of the World Bank (2009). Instead of rising urban-rural inequality, here the authors found a declining urban-rural inequality. They also did not find a change in this pattern where at a certain level of urbanisation this will lead to a bigger urban-rural inequality. There are also studies that found conflicting results for different areas in their research. Su e.a. (2015) observed big differences between provinces in a cross-regional study of China. In some regions the urbanisation goes hand in hand with excessive development of big cities and the lack of coordinated development of small- and medium-sized cities, which leads to an expanding urban-rural income inequality. In other regions, the transfer of rural surplus labour force has improved rural incomes which has narrowed the income gap between urban and rural populations. Dudwick e.a. (2011) show in their cross-national study also great differences between countries. Even if these countries are similar and have comparable levels of urbanisation and GDP per capita. 2.1.3. Urban-Rural Poverty Gap In the literature there are different views on what part of the urban-rural disparity should be studied as a consequence of urbanisation. In the paragraphs above the inequality between the urban and rural areas has been discussed. Another form of disparity that is often used in these studies is the difference in poverty incidence or intensity between urban and rural areas. Liddle (2017) wrote an article that tries to study the effect of urbanisation on both inequality and poverty. He found that initially increases in urbanisation lead to a lower urban-rural poverty gap and lower inequality, but when rapid or excessive urbanisation follows, this can lead to greater poverty and inequality (Liddle,

(10)

10

2017: 3-4, 6). Next to this incidence of poverty, he also looked at the differences on access to electricity between urban and rural areas. This is a secondary indicator for poverty as it also focusses on the lowest parts of the urban and rural populations and not so much about the overall inequality between the urban and rural. For this rural-urban electricity ratio Liddle (2017: 4) found a similar relationship as urbanisation has with poverty and inequality. Initially, higher urbanisation levels lead to a lower urban-rural gap in access to electricity, but after a rate of approximately 49%, higher levels of urbanisation will lead to a larger urban-rural gap in access to electricity. The author suggests that at higher levels of urbanisation, there is less incentive for governments to invest in rural infrastructure, which lead the rural populations to be left behind (Ibid.). This focus on both inequality and poverty make it a well-thought and convincing article.

Other studies focus more on just one side of the story. Cali & Menon (2012) find that urbanisation has a poverty reducing effect on the rural areas in Indian districts. Assuming that in the beginning the poverty incidence was higher in the rural areas than in the urban areas, this rural poverty reduction will lead to a lower urban-rural poverty gap. The authors did not find a similar pattern as Liddle (2017) did, where after a decrease of the poverty gap, eventually higher rates of urbanisation will widen this gap. This is not so unexpected as the urbanisation rate in India is relatively low compared to other developing countries (Cali & Menon, 2012: 198). They argue that this lower urban-rural poverty gap is not so much caused by the migration of the rural poor to the urban areas, but more by the spill-overs from the urban to the rural areas. This is a very interesting idea that will be discussed later on in this literature review when the different explanatory mechanism will be explored.

Opposite to this rural-aimed study is a research by Ravallion e.a. (2007) which focusses on the urbanisation of poverty. Their initial claim is that the incidence of absolute poverty is higher in rural areas than in urban areas. They then argue that urbanisation is a positive factor in reducing poverty and that the urban-rural poverty gap is moving towards the urban. This move towards the area where the poverty was lower means a lower urban-rural poverty gap. In this theory both the rural and the urban poverty incidence are declining, but the rural is declining quicker, this leads to overall poverty reduction, but also a smaller urban-rural gap. Interestingly enough the authors did not find this pattern in sub-Saharan Africa, here they only saw an effect on the rural poverty but not on the aggregate poverty. This means that the urban-rural poverty gap is still affected but that the change here only occurs on the rural side. It will be interesting to see whether such regional trends will be visible in this research. This research also is an interesting addition to the other studies as they focus only on one part of the disparity. A change within either the urban or the rural area will also have its

(11)

11

effect on the urban-rural disparity. In the analysis it will also be tested whether there is one area more affected and thus has a more decisive effect on the urban-rural disparity.

In the previous sections the studies that use either inequality or poverty as a measure of disparity were presented. It is interesting to see that the studies that use the same concept of urban-rural disparity already show quite diverse results. Some saw a decreasing effect of the urbanisation on the urban-rural disparity, while others saw an increasing effect of the urbanisation on the urban-rural disparity. This might be explained by the fact that they often use different units of analysis and different time frames which can produce different outcomes. These different outcomes were also noticeable between the studies that used different concepts of urban-rural disparity. These different results might be explained by the divergent focusses of the two indicators. The urban-rural poverty gap focusses solely on the poorer parts of the urban and rural areas, while the urban-rural inequality focusses on the whole population of both these areas. These different approaches can also result in different outcomes. Urbanisation can have very different effects on only the poorer parts than it has on the overall population. These different effects can also result in different outcomes on urban-rural disparity. In my opinion the urban-urban-rural poverty gap is a more burning problem in the developing world than the urban-rural inequality, as the poverty gap addresses the people that live in a really deprived situation. It is also encouraged by academics to focus research on poverty instead of inequality (Feldstein, 1999; Henderson, 2018). Later on in the analysis an operationalisation for urban-rural disparity will be used that focuses more on the poverty than the inequality but in a way also tries to capture both. In the method section this operationalisation will be introduced and more developed arguments for this decision will be presented. 2.1.4. Urbanisation Rate Next to the discussion on the different interpretations of urban-rural disparity, there is also dispute on the other important variable in this research: urbanisation. Most of the reviewed articles take the urbanisation level as their measure for urbanisation, instead of using urbanisation rate as a measure of urbanisation (Guo, 2009; Liddle, 2017; Lu & Chen, 2006; Su e.a., 2015; Yang, 2015). The urbanisation rate is something quite different than the urbanisation level and can also have different effects. Henderson (2010) already noted how the urbanisation rate instead of the level can actually play a more important role in the causal mechanism with urban-rural disparity. The World Economic Forum (2015: 31) also found that when urbanisation is rapid and poorly-planned socio-economic disparities can be exacerbated, which can have an effect on the urban rural inequality. Liddle (2017: 6) argued that this rapid urbanisation also has the same effect when focussing on poverty where it leads to a greater urban-rural poverty gap. But despite these different notions of the importance of the rate of urbanisation, all the articles discussed throughout this literature review use the

(12)

12 urbanisation level as an independent variable. In addition to the existing literature this research will focus both on the urbanisation level as the urbanisation rate as possible explanatory variables for the urban-rural disparity. Both these variables will be tested in the descriptive analysis to see what their influence is on the urban-rural disparity. The case study then will be used to delve deeper into the causal mechanisms of how these variables lead to urban-rural disparity. The different explanatory mechanisms that try to explain this causal relationship will be introduced and discussed in the following section.

2.2. Explanatory Mechanisms

What is clear from these several discussions is that these different studies, from cross-national, to cross-regional, and comparative studies all with different approaches create highly differing outcomes (World Bank, 2009; Dudwick e.a., 2011; Hnatkovska & Lahiri, 2018; Su e.a., 2015; Liddle, 2017; Cali & Menon, 2012; Ravallion e.a., 2007). These studies do not only show different outcomes, but often do also show variance between the different observations (countries or provinces). The question here is: which mechanisms can explain these differences? Lu & Chen (2006:48) argue that if the richer rural residents are the ones who migrate to the urban areas, urbanisation will increase urban-rural disparity. Guo (2009) discusses a different theory: if the urban-rural disparity is already high, the urbanisation will increase this gap, while if the disparity is relatively low, the urbanisation will reduce the disparity. The problem with the theory of Lu & Chen (2006) is that since this theory is focussing on the richer rural residents this will only tell us something about the urban-rural inequality and will not tell much about the urban-rural poverty gap. Apart from that, this theory is more a conditionality than a real intervening factor in the causal mechanism between urbanisation and urban-rural disparity. It can explain why in some cases urbanisation is leading to urban-rural disparity while is does not do so in other cases, but it cannot explain how the mechanism from urbanisation to urban-rural disparity works in these differing cases. This same problem is also apparent in the theory of Guo (2009). This is also more a condition for when urbanisation leads to urban-rural disparity than it is a mechanism that can explain how urbanisation is leading to urban-rural disparity in these different cases. Another problem with this theory is that it is quite static, as in that it has only one value at a certain time and that this does not change and can also not explain the possible change in the urban-rural disparity over time. It is also not really clear at what moment in the process this urban-rural disparity should be measured. Both these theories do a good job in identifying different conditions in the mechanism between urbanisation and urban-rural disparity, but they fail to identify the mechanisms that explain how urbanisation is leading to urban-rural disparity. In a sense they fail to grasp the dynamic characteristics. In response to this lack of

(13)

13

dynamism, this research will also focus on the more dynamic factor of urbanisation rate as an explanation for urban-rural disparity.

In the following parts of this literature review several possible explanatory mechanisms will be discussed, that might grasp this dynamism in a better way. First, two different economic mechanisms will be discussed, the growth of the sectors and the urban-rural spill-overs. Second, the influence of political mechanisms will be discussed. These are off course not as separated in real life, as they are presented here. In reality these different economic and political mechanisms interact with each other to affect the urban-rural disparity. The outcomes of the economic factors can influence how the political factors play out, and the political factors create the environment in which the economic factors act. For the sake of clarity in this research they are presented as two more or less loose entities.

2.2.1. Growth of Sectors

The idea behind the following argument is that the growth of the relevant sector might be an explanatory variable in the causal mechanism of urbanisation and urban-rural disparity. The first part of this argument will look at the growth of the manufacturing sector. The hypothesis is that if the growth of the manufacturing sector is high and goes hand in hand with the urbanisation rate this will lead to an increase in urban wages and employment and that this will lead to a higher urban-rural disparity. Industrialisation which often occurs with urbanisation will create a growth in the manufacturing sector. Many government strategies have focussed on this industrialisation, which has led to more rapid increases in the incomes in urban areas. This has resulted in a wider income disparity between urban and rural areas (Kniivilä, 2002: 308). A growth in the manufacturing sector can also mean less unemployment as there is an abundancy of jobs in the urban areas. Hnatkovska & Lahiri (2018) argue that this high employment rate can in turn reduce the urban wages which will tighten the urban-rural gap. Here comes an interesting parallel with the discussion on the difference between inequality and poverty. A change in income of the urban population will have an effect on the urban-rural inequality, but this will not automatically have an effect on the urban-rural poverty gap. The mechanism works in the same way for the agricultural sector. A high growth in the agricultural sector that goes together with the urbanisation will result in better wages for the rural population. It is argued that if the agricultural output per capita rises this will lead to a drop in urban-rural inequality (Cornia, 2017: 35). This is under the presumption that the income in the rural is lower than in the urban, otherwise a rise in agricultural income would mean a higher urban-rural inequality. A growth in the agricultural sector will not only affect the urban-rural inequality, but can

(14)

14

also influence the urban-rural poverty gap. If this growth in the agricultural sector also concerns the people who used to live in poverty, this growth can tighten the urban-rural poverty gap.

2.2.2. Urban-Rural Spill-overs

Next to these sector-centric factors, the economic linkage or spill-over is another important economic factor that can influence the causal mechanism between urbanisation and urban-rural disparity. Cali & Menon (2012: 172) argue that it is particularly important to understand this relationship in developing countries “because most of their population will continue to be rural for at least another decade- and for another three decades in the least developed countries”. Cali & Menon (Idem) identify six channels through which these spill-overs can occur, and in turn, reduce the urban-rural disparity. These will be shortly explained in this section.

The first of these mechanisms is the ‘consumption linkages’. A growing urban area means an increase in the demand for rural goods. This is caused by the growing income in the urban areas, but also because the new urban consumers demand more higher value-added products (Cali & Menon, 2012: 174). This increase in demand and demand for more valuable goods, will reduce the rural poverty which will lead to a lower urban-rural disparity. The second mechanism is the ‘rural nonfarm employment’. The expansion of urban areas can also lead to diversification of economic activity away from farming in the rural areas. This has an increasing effect on the rural income. This will lead to both a decline in the urban-rural poverty gap and the urban-rural inequality. Cali & Menon (Ibid.) argue that this effect is particularly important in the rural areas surrounding the urban areas. The third mechanism is the ‘remittances’. These remittances sent by the rural-urban migrants to their rural household of origin can have an important effect on the rural poverty and the urban-rural inequality. Ellis (1998) showed that the vast majority of these rural-urban migrants send a significant proportion of their income to their rural home. By reducing the rural poverty and tightening the urban-rural inequality, these remittances can lead to a smaller urban-rural disparity. The fourth mechanism is the ‘rural land/labour ratio’. The idea is that rural-urban migration reduces the rural labour supply, this reduction will lead to more land available per capita in rural areas. This higher proportion of land will increase productivity in agriculture and this will lead to higher rural wages (Cali & Menon, 2012: 175). Higher rural wages can result in lower rural poverty and lower urban-rural inequality, these two will lead to a lower urban-rural disparity. The fifth mechanism is the ‘rural land prices’. The growth of cities can increase the prices of agricultural land, which can generate a higher income for the landowners by selling or leasing this land. It depends on how the land is divided among the rural citizens whether it will reduce the urban-rural disparity. If almost all the land is in the hands of just one small group, this will lead to an increase in rural poverty, but if the land is spread out evenly through the community this can lead to a decrease in rural poverty. The

(15)

15

sixth mechanism is the ‘consumer prices’. A growing city is associated with lower consumer prices because of the increasing competition, surrounding rural consumers with access to the urban markets can also benefit from these lower prices (Jacoby, 2000). This might not have a direct effect on the rural poverty, but it can decrease the urban-rural inequality.

These urban-rural spill-over mechanisms can have different effects on both the urban-rural poverty gap and the urban-rural inequality. If the urban-rural spill-overs only benefit the richer rural residents the urban-rural inequality will be lowered but the urban-rural poverty gap will not be changed. On the other hand, if the spill-overs also benefit the poorest rural residents this can have a big effect on both the urban-rural poverty gap as the urban-rural inequality. Whether the poorest rural residents are affected depends on the kind of spill-over, but also to where these spill-overs are directed. In the case study it will be tested which spill-overs can explain how urbanisation is influencing urban-rural disparity and whether there are differences in the effects on inequality and poverty.

2.2.3. Political Mechanisms

Next to the economic factors, there are also political factors that can have an influence on the causal relationship between urbanisation and urban-rural disparity. These factors can either have a separate effect on this relationship or they affect this relationship in interaction with each other. The idea of the political factors is that either the urban or rural can for different reasons be dominant in the political sphere and that most policies will favour this group. This dominance of one of the two groups can in turn lead to a lower or a higher disparity. The theories around this idea will be discussed in this part.

The idea of a bias towards the urban area dates back to 1977 when Lipton created the so-called ‘urban bias theory’ in his famous book Why Poor People Stay Poor. This theory starts around the notion of equity and efficiency. The norm would be that resources would go to the areas where it is most efficient and where it creates the most equitable outcomes. These are often the rural areas, investments in these poor areas will raise the rural incomes and opportunities which will mean a higher total welfare (Lipton, 1977: 55). So why is it that often resources and policies are aimed at the urban areas? Lipton explains this with the idea that the urban citizens are in certain ways ahead of their rural counterparts: “the urban sector contains most of the articulateness, organisation and power. So the urban classes have been able to ‘win’ most of the rounds of the struggle with the countryside” (Lipton, 1977: 11). This political power of the urban population means that resources will be distributed both inefficient and inequitable towards the urban areas (Eastwoord & Lipton, 2000: 2). This urban political power will be categorised as urban collective action potential in the

(16)

16

theoretical framework. This urban collective action potential also has a more indirect way of creating an urban bias. A government will always try to stay in power and if possible prolong its term of office. Pierskalla (2016) argues that a way of securing this is to focus on the group of citizens that pose the biggest threat to their survival. The high urban collective action potential can make the urban area the biggest threat which would lead to a higher urban-rural disparity.

This idea by Pierskalla (2016) of securing their survival by supporting the group that poses the biggest threat to this survival, can also lead to a rural bias. In order for this to happen there have to be certain political factors in place that can overcome the urban collective action potential factors. These are the more direct political factors: if for instance a government risks losing an election because they are losing rural support this can lead to more resources and attention being directed towards the rural areas which will lead to a lower urban-rural disparity. These more direct political factors can also have an opposite effect where these different political institutions favour the urban area which will result in an urban bias which in turn will lead to a higher urban-rural disparity. In the theoretical framework this idea will be developed under the category of elections and political allegiances.

2.3. Addition to the Literature

To round off this literature review, the different additions that this research tries to provide to the existing literature, will be summarised. First, this literature review showed how there is dispute on whether the urban-rural inequality or the urban-rural poverty gap should be used as an indicator for urban-rural disparity. Where most of the other studies focussed on inequality as a measure of urban-rural disparity, this research will use the Multidimensional Poverty Index as a measure of urban-rural disparity. This index measures different forms of poverty and the difference on these between urban and rural areas. By using this measure that focusses on poverty this research will give new interesting insights into this field of study. Studies that only focus on income inequality, will not tell us much about the real disparity. Inequality is often highly affected by the highest incomes. If the highest incomes in the urban areas are earning more and more this will lead to a growing inequality, while it can be the case that the middle- and lower-income groups of both areas are moving towards each other, which creates a lower urban-rural disparity. In this case the inequality is misleading and does not tell us the whole story. In the case of urban-rural inequality it is often also that inequality is just a relative transfer of people from low- to high-income groups or sectors (Knight, 1976: 172-173). By focussing on poverty there will be a better view on the urban-rural disparity that really matters for developing countries. With this focus on poverty instead of inequality the real problem in the developing countries will be addressed. One of these problems is a declining economic growth caused by the high levels of rural poverty, that lead to a higher disparity (Chao & Tan, 2017: 245).

(17)

17

Second, there is also discussion on whether to use the absolute urbanisation level or the dynamic urbanisation rate for the concept of urbanisation. In my analysis both these concepts will be analysed to see which of these has an effect on urban-rural disparity and whether there are differences between the two. This has not yet been done in previous research, which makes this an interesting addition which can tell us more about the relationship between urbanisation and urban-rural disparity. Third, the many differences that were found in different regional and cross-national studies show that there are certain causal mechanisms that interfere in this relationship. In the second part of the analysis, the case-study, the focus will be on testing the different causal mechanisms that were introduced in this literature review and which will be elaborated on in the theoretical framework. These causal mechanisms will also possibly explain the different results that are observed between different countries. Fourth, by carrying out this combination of a large-N descriptive analysis and an explanatory case study, a more coherent view on this problem can be given. So far this is still lacking in the literature.

Next to these academic additions the approach of this research also tries to address some of the social implications. As discussed earlier, urban-rural disparity is a significant problem for developing countries and leads to a decline in economic growth and can also harm the social harmony. Urbanisation is an inevitable event for countries who want to develop (Henderson, 2010: 515), in order to prevent higher urban-rural disparity the answer should not be less urbanisation, but a more sustainable urbanisation. By focussing on the dynamic causal mechanisms that can explain why urbanisation is leading to urban-rural disparity, the real cause is revealed. If this causal mechanism is known, it is also easier to deal with the urbanisation in such a way that it will not lead to higher urban-rural disparity. In this way this approach of the research also focusses on the social implications and has a social addition.

(18)

18

3. Theoretical Framework

The literature review has showed the different mechanisms that might be capable of explaining the causal link between urbanisation and urban-rural disparity. In this theoretical framework the different scenarios that can be created by these mechanisms will be discussed. First, the scenarios created by the economic factors of the growth of sectors and urban-rural spill-overs will be discussed. Second, the way in which the political mechanisms can also affect this causal link will be discussed. Here a divide will be made between the exogenous and endogenous political factors.

3.1. Economic Factors

The economic factors, growth of sectors and urban-rural spill-overs, can both affect the causal link between urbanisation and urban-rural disparity in a different way. The growth of sectors creates four different scenarios that can then be altered by possible urban-rural spill-overs.

3.1.1. Growth of Sectors

For these four different scenarios the theory about the agricultural sector and the manufacturing sector will be combined. These four different combinations of the state of the agricultural sector and the manufacturing sector can have very different implications for the urban-rural disparity. In the first scenario both the manufacturing and the agricultural sector are growing alongside with the urbanisation. This is beneficial for both the urban and the rural areas, which means that the urban-rural disparity will be low. In the second scenario the manufacturing sector rises rapidly together with the urbanisation while the agricultural sector stays behind. This means that the urban area is profiting more from the urbanisation than the rural area which will lead to a higher urban-rural disparity. In the third mechanism the agricultural sector rises rapidly together with the urbanisation but the manufacturing sector stays behind. This means that the rural area is profiting more from the urbanisation than the urban area, under the presumption that the urban area was initially doing better than the rural area, this will lead to the lowest urban-rural disparity of the four scenarios. In the fourth and final mechanism both the manufacturing sector and the agricultural sector do not grow alongside the urbanisation. Since both areas experience the same effect of the urbanisation the disparity will be low. These four scenarios are summarised in figure 3.1.

(19)

19 Manufacturing Sector Agricultural Sector Grows Lags behind Grows Lags behind 1. Low disparity 2. High Disparity 3. Lowest Disparity 4. Low disparity Figure 3.1: Four scenarios of agricultural and manufacturing sector.

3.2. Urban-Rural Spill-Overs

In the literature review the possible influences of the urban-rural spill-overs are discussed. These spill-overs can change the characteristics of the four scenarios. Off course these spill-overs differ from each other and can have different effects on the causal mechanism, but the idea that the benefits of the urban spill-over to the rural will be the same for all the spill-overs. This means that their effect on the urban-rural disparity will also be relatively the same, which allows me to discuss the overall influence of the urban-rural spill-overs on the urban-rural disparity, instead of discussing all the spill-overs separately. In the first scenario (see figure 3.2), both areas are profiting from the urbanisation, but because of the urban-rural spill-overs the disparity between urban and rural can actually become lower. In the second scenario where the urban area is profiting but the agricultural sector is not, the urban-rural spill-overs can create a scenario where the urban-rural disparity is lower than it would be without the urban-rural spill-overs. In the third and fourth scenario, the urban-rural spill-overs will most likely have less of an influence since the urban area is not profiting that much in these scenarios. That means the urban area has not much to spill-over. In the case study it will be analysed whether these urban-rural spill-overs can be found, through which channels these spill-overs go, how they are directed and how they possibly influence the scenarios and thus the causal relationship between urbanisation and urban-rural disparity. Manufacturing Sector Agricultural Sector Grows Lags behind Grows Lags behind 1. Lower disparity 2. Lower Disparity 3. No Change 4. No Change Figure 3.2: Four scenarios of agricultural and manufacturing sector.

(20)

20

3.3. Political Factors

As discussed in the literature review the political factors can be divided in the urban collective action potential and the elections & political allegiances. These both can influence the relationship between urbanisation and urban-rural disparity in a different way.

3.3.1. Urban Collective Action Potential

The idea of the urban collective action potential factor is that the urban population contains most of the articulateness, organisation and political power (Lipton, 1977). In the struggle for policies and resources the urban tends to win it from the rural. It can also be that the government is reacting on this urban area organisational and collective action potential by directing more resources towards this area. They will do so because they have the feeling that this makes the urban area a bigger threat to their own survival (Pierskalla, 2016). Both these mechanisms mean that resources are unevenly distributed towards the urban areas, this can result in an urban bias which can in turn lead to a higher urban-rural disparity. If these mechanisms are not in place, it means the urban area has no advantage over the rural area in collective action, this will then not affect the urban-rural disparity.

3.3.2. Elections and Political Allegiances

Next to this mechanism that creates an urban bias there is also a set of factors that can create either an urban or a rural bias. Pierskalla argues that: “the direction of urban and rural bias is a function of the political threat that geographically distinct groups pose to the survival of the central government” (Pierskalla, 2016: 286). In the discussion of the first political factor it already became clear that the urban collective action potential often leads to the urban area be the most threatening group for governments’ survival which leads to an urban bias. So it must be different political factors that can steer this bias towards the rural area. In this part a few possible factors will be discussed that can be tested in the actual analysis. The first basic factor that can have an influence here is whether a country has a democratic system or not. In a democracy a government has to get as much votes as possible for their survival, this also includes the rural constituency, in a non-democratic system a government does not have to rely on votes and will thus most likely support the urban area because they possess the highest collective action potential. The focus of a democratic government on the rural area can possibly lead to a rural bias, but this depends on a few more political factors. The second factor is how big the margin is for the government, if they already have huge support there will be no real threat to their survival which means they also do not have to direct more resources towards a certain area because of this reason. If their margin is really small the threat to their survival can be a much bigger problem, which means that the government will be more inclined to direct resources towards this area that poses this political threat. The third factor that follows

(21)

21

from this is whether they rely on urban or rural support. It can either be that they always relied on a certain area for their support and that they will direct even more resources towards this area to make sure that they will keep their support in this area. It can also be that their potential lies in a different area than they used to focus on, this can lead to a shift of resources from one area towards the other. If a country has a democratic system where the government has to rely on, if their margin is not substantive yet, and if they have to rely on rural support for their survival, this can all lead to a rural bias. This rural bias is created by the different factors concerning elections and political allegiances. This rural bias will lead to a lower urban-rural disparity. If certain factors are not in place or if the government has to rely on the urban area for their survival this will create an urban bias. This will result in a higher urban-rural disparity. This theoretical framework has discussed four different sort of mechanisms that can influence the causal relationship between urbanisation and urban-rural disparity. These different mechanisms will be tested to see which can best explain the causal relationship between urbanisation and urban-rural disparity.

(22)

22

4. Methods

This research will make use of a mixed method design to approach the research question in the best manner. Johnson et al. (2007) asked 19 respected academics to come up with a definition of mixed methods research and conducted a content analysis on these definitions. This brought them to the following definition: ““Mixed methods research is the type of research in which a researcher or team of researchers combines elements of qualitative and quantitative research approaches (e.g. use of qualitative and quantitative viewpoints, data collection, analysis, inference techniques) for the broad purposes of breadth and depth of understanding and corroboration” (Johnson et al., 2007: 123). In this research this quantitative element will be carried out in the first large-N cross-sectional analysis and the qualitative part will be conducted in the case study on Rwanda. As Teddlie & Tashakkori (2009: 33) argue “a major advantage of mixed methods research is that it enables the researcher to simultaneously ask confirmatory and exploratory questions and therefore verify and generate theory in the same study”. This is exactly the goal that is envisioned by the research question of how the urbanisation rate affects the urban-rural disparity in the developing world. The first part of this research will focus on the confirmatory part and will test whether this causal relationship can be seen in this sample, while the second part will focus on the exploratory part and will examine how this causal mechanism goes. This design allows to simultaneously test a quantitatively focussed hypothesis and explore in greater depth the processes whereby the relationship occurred (Teddlie & Tashakkori, 2009: 34). The design that will be used in this research is called the sequential mixed design (Idem: 26-27). In this design the quantitative and qualitative strands of the study occur in chronological order. This research, as most of the sequential mixed designs, will start with the quantitative part and the qualitative part will follow on this. An advantage of this sequential mixed design is that it the latter analysis is dependent on the former which means that the questions and approaches of the case study can be adjusted according to the results of the quantitative analysis. This next part of the method section will discuss the method and the data collection of the first analysis. The discussion of the method of the second analysis will be discussed after the results of the first analysis are clear. As just mentioned these results will affect the exact approach in this second analysis.

4.1. Cross-Sectional Analysis

This first analysis will start the exploration of the relationship between urbanisation and urban-rural disparity. This analysis will already provide some useful answers to the main research question: How does urbanisation affect the urban-rural disparity in the developing world? To provide these useful answers some sub-questions will be used. The first is: Which concepts of urbanisation have a significant effect on the urban-rural disparity? If they do have a significant effect the follow-up

(23)

23

question will be: What effect do these concepts have on the urban-rural disparity? In other words: do they have a decreasing or an increasing effect on the urban-rural disparity? To answer these important questions, a large-N descriptive analysis is the most suitable. Using this method will give results on the patterns of this relationship over a large set of countries. The downside of a descriptive analysis is that it often does not give an explanation on how the causal mechanisms work. To solve this problem, this descriptive analysis will be combined with a case study to explore this causal mechanism. This descriptive analysis will also be used to check for different control variables, by doing so certain possible explanatory mechanisms can already be cancelled out before moving on to the case study. In the next part the data that has been used will be discussed. 4.2. Operationalisation Strategy In this part the focus will be on the data used in the analysis. As introduced earlier on, this research will focus on the developing world. First, this is done because the interest of this research lies on the developing world. Second, by focussing on only developing countries, the results will hopefully be more coherent. Since the Multidimensional Poverty Index, which will be introduced later on, is the main dependent variable, their focus will also be used in this analysis. The researchers, Alkire and Foster, applied this index to a total of 102 developing countries from different regions in the world. These 102 developing countries will also function as the population in this research. In the following part the different variables that will be used on this dataset to analyse the causal mechanism of interest will be introduced and discussed.

Urban-Rural Disparity: The first variable that will be discussed is the dependent variable of this research. As seen before, there exists quite some dispute over this concept. Some people argue that the focus should lie on the urban-rural inequality while other academics think the focus should lie on the urban-rural poverty gap. In this research the Multidimensional Poverty Index (MPI) will be used to operationalise this concept. The operationalisation for the urban-rural disparity is the difference between the urban MPI score and the rural MPI score. The MPI is a variable that assesses multidimensional poverty. It consists of three dimensions: health, education and living standards, and these three dimensions consist of ten indicators together (Alkire e.a., 2014). Individuals are scored on all these indicators and when they sit above the poverty cut-off of 33.33% they are identified as multidimensionally poor (Idem: 4). The MPI combines both the incidence of poverty as the intensity of the poverty in one score that covers all aspects of poverty. It is a deliberate choice to focus on this broad range of poverty instead of focussing just on income inequality. First, income in the developing world is quite hard to measure and does not tell us much about the real situation. In the developing world, a large share of the population still depends on the informal economy (Blades e.a., 2011: S1). Looking at the income differences can give misleading results since especially the

(24)

24

rural residents tend to earn their income via the informal way. Second, looking at this index of poverty, this will still tell a lot about the traditional urban-rural inequality. Cali (2008:4) argues that the MPI is inter-related with income inequality measures, so a high difference in MPI between urban and rural will also indicate a high urban-rural inequality. Third, to add to the first point, focussing on inequality can often even give conclusions that do not tell us much about actual urban-rural disparity. Inequality is often just a relative transfer of people from low- to high-income groups or sectors. This transfer of high-income rural citizens to urban areas will increase the inequality, but this does not automatically increase the urban-rural disparity. Inequality can actually imply that no one is made worse off and that total income is actually rising (Knight, 1976: 172-173). Fourth, by looking at poverty and the accompanying indicators the focus will be on the actual problem of urban-rural disparity. Rural poverty can for example have effect on the willingness of the rural areas to invest in human capital and this urban-rural poverty gap can also have negative effects on the economic growth (Chao & Tan, 2017: 245). By analysing how the urbanisation affects this problem, this research can possibly add something to the policy discussions concerned with poverty (Cali, 2008: 4). After this important discussion, the focus will now be on the actual operationalisation. For every country the urban and rural MPI score will be taken out of the most recent database (Alkire & Kanagaratnam, 2018). These scores will be used separately as MPI Urban and MPI Rural, and for the MPI difference, which is the operationalisation of the urban-rural disparity, the MPI Rural will be abstracted from the MPI Urban. This database comprises of the most recent observation for every country, unfortunately these observations come from different years. Although this is not optimal, I argue that this is still very useful data. It has consequences for the other variables which will be assigned to every single country depending on the year their MPI was measured. In the discussion, I will reflect on this and whether this has led to misleading results.

Urbanisation (level): The first important independent variable of this research is the urbanisation level. This is collected by the United Nations Population Division which measures the urban population as a percentage of the total. “Urban population refers to people living in urban areas as defined by national statistical offices” (World Bank, 2018). For every individual country the observation the year before the observation of the MPI (so t-1) has been used.

Urbanisation (rate): The second important independent variable of this research is the urbanisation rate. This is derived from the urbanisation level that is discussed above. For every country the average rate over five years before the MPI has been calculated. The rate here is the percentage change per year, not to be confused with the percentage-point change. The percentage change per year will be used because I have the idea that this captures the idea of urbanisation rate better than

(25)

25

the percentage-point change. A one percent change from ten to eleven percent will feel like faster urbanisation than a one percent change from eighty to eighty-one. I think this former will also have more effect on the urban-rural disparity than the latter. Using this percentage change as urbanisation rate captures these different dynamics. Per year this percentage change is calculated by the following formula: ((t2-t1)/t1)*100, the rates for the five different years are then added to each other and divided by five, this gives the average urbanisation rate over the five years before the observation of the MPI. The choice for a time-span of five years has been made because it is reasonable to argue that a rapid urbanisation rate that occurred five years ago can still have its influence on the urban-rural disparity. But on the other hand the urbanisation rate from the year before can also have its influence, to capture both this short-term and long-term effect an average over five years seems most useful.

GDP: To control for the size of the economy, the Gross Domestic Product (GDP) per capita of a country will be used. This gives a good view on the development of a country and the state of its’ economy. For every observation the GDP is used the year before the MPI. The log of this variable is used to convert it into a normally distributed variable, this leads to less disturbance of the results by the really big economies and the really small ones.

Inequality: Next to these central variables, there are also several control variables that will be used in this research. The first of these is the overall inequality in a country. To operationalise this variable, the GINI-index will be used. This index measures the extent to which the distribution of income among individuals deviates from a perfectly equal distribution. A Gini score of 0 represents perfect equality, while a score of 100 implies perfect inequality. For every country the Gini score from the year before the MPI is used.

Land Area: The next control variable is the surface of a land. This is taken from the World Bank (2018), the year before the observation of the MPI and is measured in square km. Here the log of this variable will again be used to convert it into a normally distributed variable, otherwise the results will be disturbed by the outliers that are either really big or really small.

Population: Another control variable that will be used is the population of a country. This is again taken from the World Bank (2018), and measured the year before the observation of the MPI. This variable is also not normally distributed. To make sure the results will not be disturbed by countries with either really large populations or really small populations, the log of this variable will be used. Aid: The last variable that will be controlled for is the amount of aid the countries receive. This variable is operationalised by using the Official Development Assistance, this is the amount in net

(26)

26

Millions in US$ of government aid received by a country (OECD, 2018). Unfortunately, these figures were not widely available over different years for the developing countries in the dataset. The only observation that was available is done in 2015. Even though this is after some of the MPI observations, this can still be useful in checking whether the aid a country receives can have an impact on the causal mechanism between urbanisation and urban-rural disparity.

Growth per sector: These variables will be used to test whether the mechanism with the growth per sectors can explain the causal mechanism. For both the manufacturing and the agricultural sector the annual % growth of the value added the year before the MPI will be used. This will show whether this sector is experiencing growth or decline. This data comes from the World Bank.

Electoral Process & Pluralism: For the control variable concerned with democracy, a part of the Economist Intelligence Unit’s Democracy Index will be used (The Economist Intelligence Unit, 2017). The electoral process and pluralism part is chosen because this captures the two different political mechanisms in the best way. For most of the observations the score of the year before the MPI is used, for some this is the score two years before the MPI and for just a few this is the score of the year the MPI was measured. A higher score on this index represents a more electoral and pluralist country, while a lower score represents a less electoral and pluralist country. 4.3. Analytical Strategy Now that the choice for this method and the different variables have been discussed, this last part will be used to give an outline of this analysis. Here the different steps and choices will be discussed. The analysis will start by looking at the urbanisation level and the influence on the urban-rural disparity. This relationship will then be tested by using the GDP per capita log. Afterwards it will also be checked whether there is an interaction between the GDP per capita log and the urbanisation level. After this the focus will shift to influence of the urbanisation rate on the urban-rural disparity. This causal relationship will then be tested using the different control variables that have been discussed throughout this method section.

Next to this analysis of urbanisation level and the urbanisation rate on the urban-rural disparity, there will also be attention to the influence on both the urban MPI and the rural MPI. Here the urbanisation rate will be tested on these two variables, as well as the log of the GDP per capita. Here the first causal mechanism from the theoretical framework will also be tested. The growth per sector will be used to test whether these can explain the levels of the urban and rural MPI and thus the difference between these two. Next to that the democracy index will also be tested to shed a first light on the possible working of the political mechanisms. This variable does not capture the full

(27)

27

mechanisms that are discussed in the theoretical framework, but it will be interesting to see whether this can already tell us something about this causal mechanism.

Urbanisation & urban-rural disparity : how does urbanisation affect the urban-rural disparity in the developing world?