Loading...

The Effects of Diversity on Economic and Political Stability

Master's Thesis 2017 79 Pages

Economics - Other

Excerpt

Content

List of Figures

List of Tables

List of Abbreviations

1. Introduction

2. Literature Overview

3. Historic Origins of Ethnic Diversity
3.1 Emergence of Ethnic Variety
3.2 The Slave Trades
3.3 Colonisation and Border Drawing

4. Economic Development and Governance in Diverse Societies
4.1 Different Measures of Diversity
4.2 Effects of Ethnic and Linguistic Diversity
4.3 Effects of Religious Diversity
4.4 Endogeneity and Dynamics of Diversity
4.5 Effects of Social and Political Fragmentation
4.5.1 Ethnic Voting and Favouritism
4.5.2 Party System Fractionalisation

5. Diversity and Stability
5.1 Effects of Instability
5.2 Influence of Diversity on Conflict
5.3 Economic Causes of Civil Conflict
5.3.1 Financial Opportunities of Civil Conflict
5.3.2 Poverty and Inequality as Drivers of Instability
5.4 The Role of Institutions in Civil Conflict and Instability

6. Indices of Instability
6.1 Assessment of Instability Indicators
6.2 Towards a new Composite Indicator of Instability

7. Conclusion

8. Appendix

List of References

List of Figures

Figure 1: Ethnic and national borders

Figure 2: Conflicts by region

Figure 3: Split ethnicities

List of Tables

Table 1: Parameters of Instability

List of Abbreviations

Abbildung in dieser Leseprobe nicht enthalten

1. Introduction

Many developing countries do not only suffer from weak institutions, poor economic performance and corruption, but also from separatist movements and violent civil conflicts. The question arises why some countries could achieve economic growth and development, while others never experienced considerable economic development and are trapped in a vicious circle of re-occurring violent conflicts and economic deterioration. Since developing countries tend to be more diverse in terms of ethnicity, language and religion and many civil conflicts appear to have an ethnic or religious component, diversity is regarded as a main cause of economic and political instability. Furthermore, many scholars consider higher levels of diversity in the least developed countries to be the crucial factor that leads to inefficient policy decisions and impedes growth and development.

This master thesis addresses the question how diversity affects economic and political stability and elaborates appropriate parameters which are further used in a composite indicator (CI) to quantify a country’s stability, respectively instability. The thesis is structured as follows:

After a review of the literature on the relations between diversity and economic development and civil conflicts in chapter two, the historic and environmental conditions under which different ethnic and linguistic groups emerged are described. Further, the effects of external shocks which shaped ethnic development in the special case of Africa are assessed.

Chapter four introduces the most common measurements of ethnic, linguistic and religious diversity and describes the differences between fractionalisation and polarisation. Several studies describing the effects of different aspects of diversity on various economic and political outcomes are discussed.

This is followed by chapter five which is addressing the causes of instability and civil conflict. It is examined how conflict, as the major outcome of instability, is related to and can be driven by diversity and which roles economic and institutional aspects play in explaining civil conflicts.

In the subsequent part, several indicators which capture different aspects of stability are critically assessed. Further, parameters and their respective weights towards a new composite indicator of instability are elaborated. Subsequently, chapter seven concludes.

2. Literature Overview

In 1997 William Easterly and Ross Levine published an influential article called “Africa’s growth tragedy: Policies and ethnic divisions”. In an empirical cross-country study, they found that a high level of ethnolinguistic diversity has a direct negative effect on economic growth and an indirect effect on public policy choices which influence long- run growth rates negatively. They argued that large shares of the East Asia-Africa growth differential can be explained by ethnic diversity differences (Easterly and Levine 1997). Many scholars have since worked on the topic how ethnolinguistic fractionalization impacts economic variables, often assuming an inverse relationship. La Porta et al. (1999) studied the influence of various determinants on government performance and found that more ethnolinguistic fractionalised countries have an inferior public goods provision, but that governance and economic development is mainly driven by legal origin instead of ethnic fractionalisation. Alesina et al. (2003) assessed the effects of ethnic, linguistic and religious fractionalisation on the determinants of economic growth, social-political quality and the quality of institutions. Others studied the impact of religious fractionalisation and polarisation on investment and development (Kodila-Tedika and Agbor 2014). Most of the cross-country studies find a negative impact of diversity on development, but studies on the city level find also positive effects on wages and productivity and a trade-off between private and public goods provision (Alesina and La Ferrara 2005). Besides the sheer number of different ethnic groups, the cultural distance between them is important and depending on the distance, Desmet et al. (2012) examine the effects on public goods provision, redistribution, economic growth and the onset of civil war.

The literature on the economic effects of an ethnic, linguistic or religious diverse society, often related to the poor economic performance of African countries, is abundant. Further, a vast literature is concerned with the interconnection of political stability and economic growth (Aisen and Veiga 2012) and which aspects of instability are closest related to low growth (Jong-A-Pin 2009), while other studies discuss whether the causality should rather run from economic growth to political stability or the other way round (Miljkovic and Rimal 2008).

Whether ethnic and religious diverse countries are more prone to civil war is an often addressed research question (Fearon and Laitin 2003). Several studies examine the effects of ethnic (Montalvo and Reynal-Querol 2010) and religious (Gomes 2013) polarisation and fractionalisation on civil conflict as well as the economic causes of conflict (Collier and Hoeffler 2004) and their specific transmission channels (Caselli and Coleman 2013). Many scholars carried out research on the relationship between diversity and economic performance as well as on the relation of political stability and economic outcomes and how diversity may make violent civil conflicts more likely or severer.

Further empirical research has to be done on how fractionalisation or polarisation of a society affects broader aspects of a country’s overall political and economic stability and not solely focus either on the impacts on the economic performance or on the worst outcome of instability, which is civil war. Especially for foreign investors who want to invest into a diverse developing country or for policy makers who are concerned with deepen international relations, civil war may not be the only concern but already smaller events of instability, or the likelihood that they arise due to a country’s societal structure, may make actors reconsider their engagement. There are other important aspects of stability which’s interactions with diversity should be addressed such like inflation, social security, governance, crime and unemployment just to name a few.

Recent research suggests that the impact of diversity on economic freedom is conditional on the level of pre-existent political freedom - positive in democratic and negative in autocratic regimes (Heckelman and Wilson 2016). Based on a similar argumentation, it could be hypothesised that the effects of diversity on the stability of a country is conditional on the degree of political freedom. Thus, autocratic countries will be adversely affected, while democratic nations will suffer less or may even benefit. This is due to the presumption that more democratic societies and institutions are better equipped to deal with different political tendencies and tolerate minorities. Conflicts between different groups and attempts of the government to suppress minorities in the name of the majority are mitigate in liberal democracies by the rule of law. Whereas in autocratic nations problems associated with diversity could lead to an over-reaction of political institutions which worsens the struggle.

3. Historic Origins of Ethnic Diversity

In recent history, some countries which have been relatively poor, not only compared with Europe and North America but also with medium developed countries, could achieve to grow very fast and outperform others. While on the other side some countries had never experienced considerable economic growth or even suffer from a decreasing per capita Gross Domestic Product (GDP). As mentioned above, some believe that differences in development can be due to different degrees of ethnic or linguistic diversity. In fact, many developing countries share the same traits of larger heterogeneity and instability compared to developed countries which are often characterised by more homogeneous societies. Most of the most ethnic and linguistic fractionalised countries are located in Africa (Alesina et al. 2003, pp. 184-189). Africa has always been a diverse continent but the improper border design by Europeans in the 19th and 20th century fostered conflict between different ethnicities and is one of the longest lasting influences of Europe on Africa. Other historical events that have significantly shaped the poor economic, social and political development are the slave trades, the legacy of the colonial institutions and the pre-colonial structure of ethnic institutions (Michalopoulos and Papaioannou 2016, pp. 1802-1804).

3.1 Emergence of Ethnic Variety

One reason for higher ethnic diversity can be found in geographic variability. Several studies have shown a positive correlation between diversity in species and diversity in linguistics. For example, countries with higher mammal and bird diversity or greater endemism in vertebrate and flowering plant species also show a higher linguistic diversity. Sub-Sahara African regional distribution of vertebrate diversity and cultural diversity is quite similar (Maffi 2005). Latitude influences linguistic diversity through temperature and climatic variability. With increasing latitude, i.e. greater distance from the equator and thus higher variability of the climate, people were forced to establish bigger social networks for a solid subsistence, which homogenised them linguistically. Living closer to the equator, a continuous food production was possible during the whole year and therefore also groups of smaller size were able to sustain self-sufficiency and thus gave rise to many small languages (Nettle 1998, p. 354).

Empirical findings confirm that mean precipitation and mean temperature - which is strongly correlated with latitude - have a positive impact on the number of languages. Larger geographic heterogeneity, defined by varying agricultural land quality and elevation, is increasing the number of languages spoken as well. Contrary, people residing on a more homogeneous land are more likely to merge as an ethnic identity in order to protect the land from hostile outsiders. Assuming that homogeneous areas are more attractive to invaders, the common thread of being conquered will have a homogenising effect on the groups in those areas. Whereas groups in geographical variable regions are facing higher costs of migration which makes them more isolated from each other and over time the process of cultural drift will lead to different linguistic and cultural characteristics. Historically, land endowment was the main factor for production decisions and people gained knowledge on their specific type of land, those skills could not easily be transferred to a different type of land. This accumulation of location-specific human capital reduced mobility and contributed to the emergence of different ethnicities (Michalopoulos 2012, pp. 1518-1534).

3.2 The Slave Trades

An important event that deepened the cleavages between different ethnicities in Africa was the slave trade. Even before the first Europeans arrived, slave trading was a widespread phenomenon within Africa. Based on existing structures, European and Arab slave traders exacerbated and commercialised the slave trade in Africa. Between the 15th and the 20th century four different slave trades shook the continent. In the trans-Atlantic slave trade, people from East, West and West-Central Africa were traded to the European colonies in North and South America. The slavery of the other three routes can be dated back even earlier. In the trans-Saharan slave trade, slaves from Sub-Sahara Africa were brought to North Africa. In the Red Sea slave trade, slaves were taken from inner Africa to the Arab regions in the Middle East. During the Indian Ocean slave trade, slaves from East Africa were also brought to the Middle East as well as to India and islands in the Indian Ocean to serve on plantations. The most comprehensive one was the Atlantic slave trade with a total of about twelve million slaves taken out of Africa. The other three trade lines account for six million traded slaves (Nunn 2008, pp. 141f.). Adding the number of people who died during slave raids or on the way before they could be shipped, the slave trades had a tremendous effect on the African population. Without the European-driven slave trades, it is estimated that the population of Africa could have been double of what it actually was by 1850 (Manning 1990, p. 171).

An important characteristic of the slave trades was that the local African people enslaved each other and sold the slaves to the foreigners. Normally slaves were captured by raiding villages. The threat of being raided was countered by an increased arming, but modern weapons could only be acquired from Europeans in exchange for slaves. This unsafe environment led to a vicious cycle where increasing enslavement of others was necessary to obtain more weapons and to be able to defend oneself against enslavement. Pre-existing communities such like village federations broke up and raided each other but also within the own community. This prevented the development of broader ethnic identities and increased ethnic fractionalisation. The internal mistrust and conflict undermined political stability. In many African regions several state systems and broader federations between kingdoms were evolving before the arrival of the professional, mainly Portuguese slave traders. Soon after they arrived and established their business, the slave trades led to the collapse of governments and disintegration. Many ancient states disappeared and their governance structures were replaced by slave raiders and warlords who were incapable of developing stable states or to organise succession of leadership. Further, pre- established legal institutions deteriorated, because many penalties were transformed into enslavement which gave rise to false accusations in order to obtain more slaves (Nunn 2008, pp. 142-144).

The empirical picture confirms that countries where larger numbers of people have been extracted as slaves showed lower levels of indigenous political complexity and state centralisation in the 19th century and today are more ethnically fractionalised. During colonial rule, low slave export countries showed on average a higher GDP per capita than high slave export countries. This income gap increased significantly after African countries gained independence in the 1960s and 70s. One reason for this development could be that after the colonial rule had ended, the pre-colonial ethnic political structures became more important and thus the roots of the post-colonial state failures lie in the pre- colonial deterioration of local institutions due to the slave trades (Nunn 2008, pp. 164- 168).

There is empirical evidence that the slave trades led to mistrust that still exists today. Individuals whose ancestors were strongly affected by enslavement have less trust in their relatives, neighbours, co-ethnics and governments. In over 400 years of insecurity and fear of slave raids, a general mistrust was forwarded through the generations and still manifests nowadays, 100 years after the end of large-scale slave trades. Furthermore, due to the long lasting adverse influence on legal and political institutions, also today people feel less under compulsion of acting trustworthy, which leads to lower trust among people (Nunn and Wantchekon 2011, pp. 3249f.). The negative relationship between slave exports and state formation and economic development as well as the persisting mistrust in African societies shows the long term impact of how slavery impeded to overcome ethnic fractionalisation.

3.3 Colonisation and Border Drawing

Other events that could have fostered ethnic fractionalisation in Africa are the colonisation and the arbitrary drawing of borders by European powers.

In the second half of the 19th century Great Britain and France started to systematically explore Africa. More European powers started to engage in the exploration of the continent and in order to avoid conflict they signed many bilateral agreements that divided regions of Africa into protectorates, colonies or free trade areas. The Berlin Conference of 1884-85 determined the rules how the colonisers divided the largely unknown continent according to their strategic and political interests, not considering local geographic conditions or ethnic borders and were largely unwilling to change the borders even though they gained new information over time. Because drawing partitioned maps in Europe had at first no significant effect for the local African people, they regarded the colonial borders as dividing the foreigners and not them. Thus the African leaders did not form a broad opposition to the new borders. After many countries reached independence in the 1950s and 60s, the new African leaders feared that border realignment could undermine their positions. It was hoped that during the process of nation building and industrialisation ethnic cleavages would be mitigated anyway. The former colonisers were reluctant to the redesign of borders as well, because they did not want to risk their special trade rights and corporate deals with the new states (Michalopoulos and Papaioannou 2016, pp. 1807-1809). The following map shows the spatial distribution of ethnic groups in the late 19th century in contemporary national borders. It illustrates how the largely straight-drawn borders do not take the great ethnic diversity of Africa into account.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1: Ethnic and national borders

Source: Michalopoulos and Papaioannou 2014, p. 153.

When the colonial powers designed the borders of African states, cost saving was one important factor. The attempt for exploiting economies of scale and to minimise the costs of bureaucracy led to the creation of geographically large states with a low population density, but which encompassed various ethnicities. African ethnicities tend to be more regionally concentrated than for example the Indian cast groups and therefore larger African states have higher levels of ethnic diversity (Green 2012, p. 237).

Besides partitioning ethnicities in several countries colonial rulers and missionaries even created tribes or deepened cleavages between them when it was of use for ruling. The British colonisers for example followed a distinct divide and rule strategy in which they would choose a smaller minority group to get British education and to dominate the civil service as well as the police and military. This policy should enforce division and resentments to avoid the formation of anti-colonial alliances (Blanton et al. 2001, p. 480). The infamous division of Hutus and Tutsi in Rwanda was intensified first by the German and later by the Belgian rulers. Belgian colonisers classified the population following racial traits. Taller and lighter coloured were formally classified as Tutsi, darker and shorter people as Hutu or Twa. The Tutsi minority of 14 % was chosen to rule over the Hutu and Twa which made up 85 % and 1 % of the population. The Tutsi were granted educational, social and economic advantages, because the colonisers attributed to them a higher intelligence and a racial superiority. This ethnic discrimination was one of the factors leading to the 1994 Rwanda genocide against the Tutsi population (Sarkin 1999, pp. 772-773).

A way how many countries overcame ethnic and linguistic fractionalisation was by urbanisation. Through urbanisation, different but related ethnic groups would get in contact with each other and notice their common interests and create a common broader ethnic identity. By this broader ethnicity the state could facilitate a national ideology and identity. A rural population instead remains more isolated from each other. Empirical findings confirm that urbanisation is positively correlated with higher identification with the nation state and homogenisation within Africa. However, the colonial rulers tried to prevent urbanisation within their colonies as it could lead to the formation of an urban working class or other nationalist movements opposing foreign rule (Green 2012, pp. 242- 248).

4. Economic Development and Governance in Diverse Societies

When people of different ethnicities, who follow different religions or speak different languages live together in the same country, it can be difficult to compromise on public policies. There are two fundamental different ways to measure the diversity of a country’s population: fractionalisation which measures the sheer amount of different sub-groups i.e. the dispersion of the population and polarisation which measures the distance to a distribution with two equally sized groups. Fractionalisation and polarisation will be of different importance for various economic, political and social outcomes. Diversity is not static, therefore it is important to take into account different fault lines that can go through a society and how these impact different aspects.

4.1 Different Measures of Diversity

The measure of ethnolinguistic fractionalisation (ELF), which was used in many papers to assess the impact of fractionalisation, was constructed by Soviet researchers and based on data that they compiled in the early 1960s. The ELF measure calculates the probability that two randomly chosen persons in a country will belong to different ethnolinguistic groups. A theoretical maximum of fractionalisation is reached, when every person belongs to a different group (Easterly and Levine 1997, p. 1218). The most widespread way to calculate fractionalisation is by use of the Herfindahl index, which in other contexts is often used to measure the market concentration of firms in an industry:

Abbildung in dieser Leseprobe nicht enthalten

In case of ELF, ݏ௜௝ is the share of the ethnolinguistic group i in country j (Alesina et al. 2003, pp. 158f.).

Another way to measure diversity is by polarisation. Different to fractionalisation, polarisation is emphasising the extreme ends of a distribution. A polarisation index will reach its maximum with the existence of two equally sized groups. To measure diversity as a source of conflict it is important to also consider the polarisation. A polarised society is marked by intra-group homogeneity and inter-group heterogeneity (Esteban and Ray 1999, p. 401). A frequently used polarisation index is:

Abbildung in dieser Leseprobe nicht enthalten

where N is the number of groups and ߨ௜ the proportion of each group. This index captures the distance between the distribution of groups and the bipolar distribution with π = 0.5, which is assumed to be the level of maximum conflict. The conflict level depends on the size of the groups (Montalvo and Reynal-Querol 2005a, p. 301).

Assume three groups distributed to (0.5, 0.25, 0.25). According to the formulas above, the level of polarisation would be 0.875 and the level of fractionalisation would be lower with 0.625. If now the distribution changes to (0.5, 0.45, 0.05) polarisation increases to 0.955 while the fractionalisation index would decrease to 0.545. This example shows the different underlying concepts of fractionalisation and polarisation and that it is important to consider both when measuring diversity.

According to rent seeking models, social costs are higher when two equally sized groups compete for rents (Reynal-Querol 2002, p. 33). The problem of a highly fractionalised country is that the clash of many different interests could lead to sub-optimal public policy decisions (Easterly and Levine 1997, p. 1230).

4.2 Effects of Ethnic and Linguistic Diversity

The empirical evidence how ethnic and linguistic diversity affects the economy is mixed. A channel through which diversity can influence economic outcomes is for example via corruption. It was shown that with increasing ELF corruption increases, which has a negative impact on private investments, that in turn lowers economic growth (Mauro 1995, p. 695). In ethnic diverse countries the governing of regions or ministries is often allocated according to ethnicity, which produces more independent bribe takers. An independent bribe taker is not taking the effects of his bribe on the other bribe taker’s revenues into account. Uncoordinated bribing leads to more bribe per unit of output and less output in general. For instance, uncoordinated groups e.g. ministries follow rent- seeking strategies, where one group enforces an overvalued exchange rate and currency controls in order to gain rents from selling foreign currency on the black market, assume another ministry is imposing low interest rates on saving to facilitate cheap loans for their ethnic supporters. The overvalued exchange rate incentivises smuggling of money out of the country in fear of devaluation and lowers the savings which the other group could take as loans. The low domestic interest rate gives additional incentive to keep money in other countries. The group controlling the exchange rate can implicitly tax when it lowers the amount of foreign exchange. The groups do not internalise the effects of their rent- seeking on the whole society which in this case leads to high black market premiums and financial repression (Easterly and Levine 1997, pp. 1214f.).

Further, ethnicity based power separation leads common pool problems where everyone is seizing from the pool of rents until it is exhausted. This problem can be observed when economic activities are highly overtaxed. An example of conflict over economic rents among different ethnicities are the cocoa policies in Ghana after independence. Production of cocoa, which was a main export good, was largely controlled by the dominating Ashanti ethnic minority. A leader of the larger Akan ethnicity could enforce a freezing of the cocoa producer price and an overvalued currency. On the one side, various ethnic coalitions continued the damaging cocoa export taxation via increasing overvaluation. On the other side, ethnic and political supporters were granted importing goods at the official exchange rate, which generated rents because the black market exchange rate was much higher. The profitable cocoa export, controlled by one ethnic group, could have been beneficial for the whole country, but it was destroyed through uncoordinated rent-seeking and growth-retarding policies due to various competing ethnicities (Easterly and Levine 1997, pp. 1217f.).

The various interests of an ethnic fractionalised society may also lead to lower public goods provision, like schools or infrastructure. Because the more parties that have to compromise on different preferences regarding language and curriculum for schools or location of infrastructure projects, the lower is the satisfaction for each party and thus makes it harder to agree on growth enhancing policies. As a consequence, less public goods will be supplied. ELF was found to be negatively correlated with school attainment, financial depth and infrastructure indicators like paved roads, number of telephones per worker and electrical system stability, but positively correlated with black market premium. The high ethnic diversity of Africa helps to explain the widespread poor public policy decisions. Even when excluding African countries, the study finds that ethnic fractionalisation is affecting black market premium and the infrastructure in unfavourable ways. In general, various ELF measures are significantly related to various public policy indicators, which in turn are correlated to economic growth. Moreover, the authors modelled that countries with a higher level of ELF in 1960 will be more corrupt and less likely to follow the principal of the rule of law as well as less democratic in 1990. Comparing the growth rates of Africa and East Asia, the study finds that the higher levels of ethnolinguistic fractionalisation in Africa has a strong indirect effect on the growth difference through policy indicators and a strong direct impact, which explains between 25 % and 40 % of the growth differential (Easterly and Levine 1997, pp. 1230-1237).

However, the underlying data of the above mentioned study mixes ethnic and linguistic criteria but relies mainly on linguistic differences, which is downplaying the role of other important aspects of ethnicity such like race or skin colour. This means that North and South America appear as quite homogeneous in terms of language, but in fact there were obvious differences between blacks and whites in the US of the 1960s. Latin America is as well more divided by racial origins e.g. Afro-Latino, European, Indigenous or Mestizo than by the language, which is mainly the one of the former coloniser. To address this problem, more disaggregated variables of fractionalisation are necessary (Alesina et al. 2003, pp. 156-159).

Because the concept of ethnicity is not clearly defined and is based on various characteristics, Alesina et al. (2003) constructed an updated measure of ethnic fractionalisation, which includes linguistic and racial traits, but with a stronger emphasis on race. The variable includes 650 ethnic groups and 190 countries and dependencies. Additionally, they also employed an isolated linguistic measure which is based on the numbers of languages spoken as mother tongue and comprises 1,055 major language groups and 201 countries and dependencies. The former used Soviet ELF index is only available for 112 countries. Due to the different concepts of diversity, the influence of ELF is different to the influence of pure linguistic fractionalisation on various economic indicators (Alesina et al. 2003, pp. 159-161).

Because of its stronger emphasis on racial traits, the fractionalisation index shows much higher degrees of fractionalisation for Latin America and the Middle East, but lower for East and South East Asia than the Soviet ELF. The linguistic fractionalisation is lowest for Latin America as well as West and South Europe. Sub-Saharan Africa remains to be the most fractionalised region for both ethnic and the pure linguistic measures. Results using the modified measure show that ethnic fractionalisation is negatively related to real per capita GDP growth, by its negative correlation with schooling, financial depth, fiscal surplus and telephones per worker. These variables were shown to be significantly correlated to growth and thus can be seen as channels through which ethnic fractionalisation has influence. These findings confirm the results of the earlier study from Easterly and Levine (1997). Moving from a fractionalisation index of 0 to maximum ethnic heterogeneity of 1, would decrease yearly economic growth by 1.9 percentage points. Applied to the least and most fractionalised countries of the sample - South Korea 0.002 (99.9 % Koreans and 0.1 % others) and Uganda 0.93 (more than 40 ethnicities where the Ganda as the largest group make up only 17.8 %) - this would mean that 1.77 percentage points of the growth rate differential could be explained by different levels of ethnic diversity. The outcome of every 0.1 increase in ethnic fractionalisation is a decreased steady state income level by 14 %. If South Korea and Uganda got the same level of ethnic fractionalisation, the income difference would have only been half (Alesina et al. 2003, pp. 162-167).

Linguistic fractionalisation is as well strongly adverse correlated with growth. A move from total homogeneity to maximum fractionalisation would be associated with a decrease of the real GDP per capita growth rate by 2.5 %. Again, the channels through which linguistic diversity affects growth are schooling, financial depth and telephones per worker (Alesina et al. 2003, pp. 168f.).

To assess the impact that diversity could have on economic and political development it is not enough to focus on the sheer number of groups. It should also be taken into account the cultural distances between these groups. There is further a fundamental problem of how to categorise people into what kind of ethnic, racial or linguistic groups. The United States census data for example categorises into white, African American, Asian and Hispanic. The question arises, why for instance Hispanics are not disaggregated into Cuban Americans, Mexican Americans etc. While Mexico is divided into Whites, Mestizo and Indigenous, is it right to group them all as Hispanics after migrating to the US? Fearon (2003) suggests to survey the people to identify the most relevant ethnic groups in their country. This also reveals a problem of endogeneity of ethnic distinctions, where for example ethnic lines become more important when the economy performs worse and increases distributional struggles1 (Fearon 2003, pp. 196-199).

Fearon (2003) defines that an ethnic group is constituted when the membership is primarily reached by descent and members regard the common traits like language and customs as important and have a collective history. In this way the study defines 822 groups in 160 countries and defines a measure of cultural distance by making use of language trees. For instance, according to simple fractionalisation Belarus scores a level of 0.37 and Cyprus of 0.36. Belarusian society is divided into 78 % Byelorussian, 13 % Russians, 4% Poles and 3 % Ukrainians and Cyprus into 78 % Greek and 18 % Turkish. The fractionalisation is nearly the same, but the distance between the groups within the countries is much larger in Cyprus. The ethnicities in Belarus are similar in terms of religion, language and customs except the Poles who are Catholic instead of Orthodox Christians and speak a west Slavic instead of an east Slavic language. Whereas in Cyprus the groups are already divided on the first level. Greeks are mainly denominated to Christianity, Turks to Islam and their languages come from different language families (Indo-European and Altaic). Measuring cultural distance lowers Belarus score to 0.23 whereas Cyprus stays at 0.36. It is therefore important to check whether the effects of fractionalisation on political or economic variables depend on the specification of the used measure (Fearon 2003, pp. 201-215).

Language trees help to analyse the different effects of different levels of language heterogeneity and the cleavages between languages. A fundamental division like the one between Indo-European and non-Indo-European languages is estimated to have occurred 8,700 years ago, while finer divisions like German and Dutch happened more recent. Linguistic divisions occurred because people got more isolated from each other. The divisions increase over time but can be mitigated by interactions between the groups e.g. the spread of Latin words through commercial and political interactions helped to keep Germanic and Romanic languages more similar. Thus the depth of the linguistic division reflects the depth of the cleavages of different people. Empirical results show, that already fine distinctions can influence economic growth, but to influence redistribution or civil conflict the distinctions need to be broad. There is a significant relation between civil conflict and linguistic fractionalisation. When the level of aggregation is high i.e. when many groups belonging to different language families are residing in the same country - which is in fact rarely the case - with decreasing levels of aggregation the effect vanishes. Linguistic diversity has an adverse impact on redistribution on high levels of linguistic aggregation but the magnitude of the effect decreases with the level of aggregation and loses significance. This indicates that solidarity is possible between diverse groups as long as the cleavages are not too deep. For an effective provision of public goods, people need to coordinate their different interests and compromise, which can be harder in linguistic diverse societies. Linguistic fractionalisation affects school attainment and public health negatively and child mortality and illiteracy positively. These unfavourable effects already occur on low levels of aggregation, meaning that a relatively superficial separation matters. The effects on economic growth show as well that shallow cleavages are relevant. Even small linguistic differences can hinder the market integration (Desmet et al. 2012, pp. 324-335).

An important factor of economic performance and through which diversity has influence is the quality of government. Desired characteristics of a good government for economic development are the protection of property rights and noninterventionist behaviour. Taxation is an ambiguous feature of intervention because high taxation could either be seen as very interventionist or as a way to finance demanded public goods. The efficiency of government and quality of bureaucracy are further important aspects for good government, because governments could intervene in a reasonable way or the intervention could lead to distortions and corruption. Equipping civil servants with more regulatory power fosters corruption and thus on average interventionism should be linked to less efficiency. Good government performance is also identifiable through the quality of public goods like schooling, infant mortality, literacy and infrastructure (La Porta et al. 1999, pp. 225f.).

According to La Porta et al. (1999) these indicators of good governance which are crucial for economic growth are rather influenced by legal traditions than by ethnic fractionalisation. The three main groups of legal origins are socialist law, civil law and common law. Socialist law was designed to keep the socialist party in power and extract resources regardless the economic interests of large parts of the population. The intentions of civil law, as introduced in France and Germany, were nation building and increasing the power of the state, but in a more constrained way than under socialist law. The common law that originated in England is quite different. It aims to protect the individual and its property against expropriation by limiting the power of the state (La Porta et al. 1999, pp. 231f.).

In their study of government performance, La Porta et al. (1999) show that in general per capita income but also latitude have a positive impact on government performance. The regression including legal origin and the ELF measure from Easterly and Levine (1997) but excluding latitude and per capita income, shows that ELF has a large negative impact on government performance. Higher ELF is related to worse property rights protection and regulation, more corruption, longer delays, less tax compliance, higher infant mortality, worse infrastructure quality, less years of schooling and illiteracy as well as a smaller public sector with less subsidies and transfers and governmental consumption, less political freedom but more state enterprises. If including latitude and per capita income, which is negatively correlated with ELF, the adverse effects of ethnolinguistic fractionalisation become insignificant. This holds true for most good government indicators, except for the inferior public goods provision and the state ownership of companies which is an important instrument of favouritism. The changing effects of ELF after controlling for income indicate that fractionalisation worsens government performance which in turn reduces per capita income (La Porta et al. 1999, pp. 244f.).

Taking into account the different legal traditions makes ethnolinguistic fractionalisation mostly insignificant. Mainly socialist legal origins and to a lesser extend the French version of civil law are adversely related to indicators of government quality. A socialist law tradition is related to interventionism, larger transfers and less efficiency and democracy. Countries with a French legal tradition are as well more interventionist, less efficient and have a worse provision of public goods and are less democratic compared with countries of common law origin. Countries of German legal origin are similar to those of common law in terms of better government performance. Scandinavian law countries are very interventionist but not less efficient than common law countries and have a better public goods provision. These influences of the legal origins are significant for a lot of government quality indicators, even when controlling for latitude and per capita income, where the ELF measure loses its significance (La Porta et al. 1999, pp. 261f.).

Many weak governed countries are highly fractionalised and have been French colonies, ethnic fractionalisation is correlated with a French legal system. Measures of ethnic and linguistic fractionalisation are correlated with latitude and income per capita, but according to Alesina et al. (2003) income is likely to be endogenous and therefore it is unclear if it should enter the regression as a control variable. Even though latitude and ethnic fractionalisation are highly correlated, they see no causal relationship between them, which makes it not easy to disentangle the effects of the variables (Alesina et al. 2003, p. 169). It was shown in Chapter 2 that latitude is linked with ethnic fractionalisation through the opportunities of continuous self-sufficient food production (Nettle 1998).

Using the more disaggregated ethnicity measure and excluding latitude but including the legal origin and GDP per capita variables of La Porta’s et al. (1999) model specification shows that ethnic fractionalisation has an adverse impact on corruption, infant mortality, illiteracy and further decreases transfers and subsidies. A political rights and democracy index are as well negatively correlated with ethnic fractionalisation. This gives rise to the argument, that in more diverse societies a leading fraction enforces political restrictions in order to assert control on the other groups. Whereas less fractionalised societies are less prone to intense conflict and thus can be governed in a more democratic way. Including the independent variable latitude, confirms the largely insignificant effects of fractionalisation as already reported by La Porta et al. (1999) which could confirm the view that institutional quality is affected by geography. On many indicators of good government, the effects of latitude and ethnic fractionalisation are only significant when introduced separately. The effects of linguistic fractionalisation are less strong and less often significant. Applied to the model specification including latitude, linguistic diversity only increases the top marginal tax rate, infant mortality, illiteracy and decreases political rights. Excluding latitude, only corruption and illiteracy increase (Alesina et al. 2003, pp. 169-177).

As mentioned above, polarisation is another way of measuring diversity. Calculating the effects of linguistic and ethnic polarisation with the highly disaggregated data produces similar outcomes like the fractionalisation index but it is typically not significant and thus less appropriate (Alesina et al. 2003, pp. 169-179). A study based on a slightly different dataset shows that when the level of fractionalisation is low, it has a positive almost linear relation with ethnic polarisation, but with increasing fractionalisation the correlation decreases and finally turns negative. Contrary to ethnolinguistic fractionalisation, polarisation has no significant direct effect on economic growth but a negative effect on investment and a positive on government spending. Which could be due to higher expenses to mitigate social conflicts, since polarisation is often regarded to increase civil conflicts. Reducing polarisation from its theoretical maximum to the minimum of complete homogeneity, shows a total effect of a 0.91 percentage points higher per capita growth rate (Montalvo and Reynal-Querol 2005a, pp. 306-316).

[...]


1 Endogeneity problems are further examined in chapter 4.4.

Details

Pages
79
Year
2017
ISBN (eBook)
9783668756397
ISBN (Book)
9783668756403
File size
1 MB
Language
English
Catalog Number
v431010
Institution / College
http://www.uni-jena.de/
Grade
1,7
Tags
civil conflict ethnic conflict ethnic diversity linguistic diversity economic development ethnische Konflikte colonisation slave trade Kolonisierung Sklavenhandel

Author

Share

Previous

Title: The Effects of Diversity on Economic and Political Stability