Chapter 13
Chapter 13
Chapter 13
MULTIETHNIC SOCIETIES
Chapter 13
Major findings and conclusions on the relationship between horizontal inequalities and
conflict
1. Introduction
In this book we set out to explore the relationship between HIs and conflict: whether indeed such
a relationship pertains in recent conflicts; which type of inequality is most important; and in which
conditions conflicts are more likely to emerge. We did so by using case studies of countries in
three regions of the world, West Africa, South East Asia and Latin America, and also through
approaches. Throughout we have aimed to contrast countries (and areas within them) that have
managed to avoid serious conflict with those countries or areas that have experienced severe
violent conflict in recent decades. The aim of this chapter is to bring together the main
conclusions that emerge from these case studies. The next and final chapter of the book reviews
policy conclusions that arise. Before coming to general findings the chapter first briefly reviews
the nature of HIs and, drawing on material discussed in Chapter 1, explains why we believe they
are likely to be important, and, in particular, to predispose countries to violent conflict. This and
the following chapter thus contain the main ideas, findings and policy conclusions of the book as
a whole.
408
2. What are horizontal inequalities?
Horizontal inequalities (HIs) are inequalities between culturally defined groups, such as ethnic,
religious, racial or caste-based groups. The concept of horizontal inequality differs from the
‘normal’ definition of inequality (which we term ‘vertical inequality’) in that the latter type lines
individuals or households up vertically and measures inequality over the range of individuals
rather than groups, whereas HIs are multidimensional and encompass economic, social, cultural
opportunities.
• Social HIs cover inequalities in access to a range of services – education, health and
• Political HIs consist in inequalities in the group distribution of political opportunities and
• Cultural status HIs refer to differences in recognition and (de facto) hierarchical status of
Each of these dimensions is important in itself, but most are also instrumental for improving
performance on other dimensions. For example, political power is both an end and a means as
people desire power in itself, and also for the economic and social benefits that it permits;
control over economic assets is primarily a means to secure income but it is also an end;
similarly, education is wanted in itself (as an important basic human right) and also as a means
409
By way of shorthand, we often merge the first two categories and refer to ‘socioeconomic’
inequalities, but there are important differences here. The social HIs are usually the outcome of
the activities of the government (and NGOs), while the economic HIs arise from inherited
While outcomes such as health, income or political power are relevant across all societies,
inputs can vary, so the relevance of some elements varies across societies, depending on their
role in each society. For example, access to primary education is critical in very poor societies,
but in more developed countries where this level of education is universally available, access to
higher levels becomes more important. Equally, access to and ownership of land is of huge
importance in regions where agriculture accounts for most output and employment, but becomes
less important as development proceeds, when ownership of financial assets may become more
important.
• Unequal access to political, economic, and social resources and inequalities of cultural
status can have a serious negative impact on the welfare of members of poorer groups
whose well-being is affected by their own relative position and that of their group.
• Severe horizontal inequalities may reduce the growth potential of a society, because they
mean that some people, because of the group to which they belong, do not have access
employment for members of the more privileged groups is too easy for the purposes of
• Horizontal inequalities can make it difficult to eliminate poverty because it is often difficult
410
especially so because deprived groups face multiple disadvantages and discrimination
• Finally, this book has explored the hypothesis that high HIs make violent group
mobilization and ethnic conflicts more likely, by providing powerful grievances which
leaders can use to mobilize people, by calling on cultural markers (a common ethnicity or
religion) and pointing to group exploitation. We return to this in the next section.
As analyzed in Chapter 5, it is important to monitor HIs in order to identify where they are
measures, and different groupings, to see how robust the results are.
Some orders of magnitude of socioeconomic HIs in the particular countries studied were shown
in earlier chapters. Some large HIs can be observed. For example, in Peru indigenous people
have seven years of schooling on average, half that of whites, while the proportion with
secondary schooling is only one-fifth that of whites (Chapter 10). The rate of extreme poverty
among the indigenous population in 2002 was more than four times the rate among the
nonindigenous (Chapter 11). In Bolivia, too, the sharpest HIs are between the indigenous and
nonindigenous populations. For instance, in 2001 the indigenous population had on average
almost four years less schooling than the nonindigenous population, and the infant mortality rate
among the indigenous population was about 50 per cent higher than among the nonindigenous
population (Molina, 2007). In Guatemala, a similar picture emerges. Even though literacy rates
among the indigenous population have improved considerably since the end of the 1980s, there
is still a significant gap between the indigenous and nonindigenous populations: in 1998, while
about 58 per cent of the indigenous population was literate, among the nonindigenous
population that figure was about 79 per cent (Caumartin, 2005). Twenty per cent of the
indigenous population in Guatemala were in extreme poverty in 2000 while the rate among the
411
nonindigenous population was just 5 per cent (Chapter 11). In Ghana, the Northern Region’s
child mortality rate is nearly two and a half times that of Ghana as a whole, and three times that
of the Greater Accra Region. Access to health services in the Northern Region of Ghana is less
than a quarter of that in the Greater Accra Region (Chapter 4). In Nigeria, maternal mortality
rates in the northeast are nine times those in the southwest. In Côte d’Ivoire, the literacy rate
among the Northern Mande is just 23 per cent, half the rate among the Akan (Langer et al.,
2007). In Malaysia, despite considerable improvement, Chinese incomes were, on average, over
1.6 times those of Malays in 2005, while the percentage of professionals of Chinese origin was
almost twice the percentage of Malays, even though they account for a much smaller proportion
of the population than the Malays (Volpi, 2007). This summary focuses on groups where there
are large socioeconomic HIs. There are, of course, situations where HIs are quite low – for
example, as between the Ewes and the Akans in Ghana, or the Akan and Krou in Côte d’Ivoire,
and some situations where HIs were large at one point, but have diminished sharply, as among
some immigrant groups in the U.S. and between Catholics and Protestants in Northern Ireland.
This section reviews some general findings from CRISE research. These are based on our eight
country studies from three regions of the world and some intercountry analysis of a broader
412
Each country is multicultural and one country in each region has avoided serious national
conflict, while the other(s) has/have experienced considerable violent conflict in the recent past.
Bolivia, Malaysia and Ghana are the countries that have succeeded in remaining broadly
peaceful, while Nigeria has not had national-level conflict for several decades. Our intention was
Here we elucidate the ten major findings that emerge from the analysis in this book, with
1. The probability of conflict occurring rises where socioeconomic HIs are higher
This is supported by intercountry analysis by Østby (Chapter 7) who shows a significant rise in
the probability of the onset of conflict across countries, for countries with severe social and
economic HIs, for 1986-2003. In her models she defines groups alternatively by ethnicity, then
by religion, then by region and finds a significant relationship between HIs and the onset of
violent conflict for each definition. Social HIs are measured by average years of education and
economic HIs by average household assets. The effect of HIs is quite high: the probability of
conflict increases threefold when comparing the expected conflict onset at mean values of all the
explanatory variables to a situation where the extent of horizontal inequality of assets among
ethnic groups is at the 95th percentile. In the case of the interregional HIs, the probability of
conflict increases two and a half times, as HIs rise from mean value to the 95th percentile value
Other statistical cross work supporting this relationship includes Gurr’s successive studies of
relative deprivation and conflict (Gurr, 1968, 1993; Gurr and Moore, 1997), and Barrows’
investigation of sub-Saharan African countries in of the 1960s. Gurr finds a positive relationship
across countries between minority rebellion and protests and relative deprivation, defined in
economic, political and cultural terms, with the precise definitions and relationship varying over
413
the investigations. Barrows (1976) finds that horizontal inequalities showed consistently positive
correlation with political instability across 32 sub-Saharan African countries in the 1960s, with
There are also intracountry studies showing a positive relationship between the level of HIs and
the incidence (or intensity) of conflict. In Chapter 6, Mancini uses district-level data to examine
the relationship between HIs and the incidence of conflict in districts in Indonesia. After
controlling for a number of intervening factors including population size, ethnic diversity and
economic development, he finds that horizontal inequality in child mortality rates and its change
over time are positively (and significantly) associated with the occurrence of deadly
unemployment, landless agricultural labourers, and civil service employment were also related to
the incidence of conflict, but the effects were less than that of child mortality, and were not
significant when the child mortality was included. Results also suggest that violent conflict is
more likely to occur in areas with relatively low levels of economic development and greater
religious polarization. Standard measures of (vertical) income inequality as well as other purely
demographic indicators of ethnic diversity are found to have no significant impact on the
Studies in other conflict-ridden countries have found that the intensity of conflict is related to HIs.
In an early study of the Moro rebellion in southern Philippines, Magdalena (1977) measures
relative deprivation in terms of differential asset returns to education among Muslims and
Christians at the municipality level, and finds this strongly related to the intensity of the conflict.
In Nepal, Gates and Murshed (2005) analyze the relationship between ‘spatial horizontal
inequalities’ and the intensity of the Maoist insurgency. Their analysis does not consider ethnic
and caste groups per se, but rather focuses on regional differences. Using a ‘gap’ measure of
414
human development they find strong econometric support for a relationship between regional
deprivation and the intensity of the rebellion, as measured by the number of conflict fatalities by
district. A subsequent econometric analysis by Do and Iyer (2007) replicates this finding of
regional deprivation – in this case measured by the regional poverty rate and the literacy rate –
but adds additional variables relating to demographic characteristics. They find that caste
polarization – the extent to which the provincial population is divided into two large caste groups
– has an additional impact on conflict intensity, even when holding for the level of regional
deprivation.
While higher HIs are correlated with a higher risk of conflict, not all violent mobilization in high-HI
countries is primarily identity-driven, at least in terms of discourse. This was the case in the
Peruvian and Guatemalan examples explored in Chapter 11, where the rebellions were
presented in ideological terms and some of the leaders of the movements came from outside the
deprived indigenous groups. In these societies ‘race/ethnicity’ and ‘class’ were virtually
coterminous, that is, the societies approached complete ethnic hierarchical stratification, which
Donald Horowitz has termed ethnically ‘ranked systems’ (Horowitz, 1985: 22). In such societies,
mobilization by class may alternate or substitute for moblization by ethnicity. In both Peru and
Guatemala some of the prominent leaders of the rebellions were nonindigenous, and were
motivated by ideology, not by ethnicity1. However, the willingness among indigenous people to
be mobilized against the state and the victimization – indeed the almost genocidal targeting – of
as an underlying element in these conflicts. The commission which investigated the historical
origins of the Guatemalan conflict found that its roots lay in the ‘exclusionary, racist, authoritarian
and centralist’ characteristics of the Guatemalan state, economy and society (Comisión para el
Esclarecimiento Histórico (CEH), 1999: 81). Francisco Bianchi, a government adviser in the
415
early 1980s, openly declared that ‘for the most part the Indians are subversives; and how can
one counter this? Obviously by killing the Indians2’ (cited in CEH, 1999: 1823).
We should emphasize that what we (and others) have found are increased probabilities of
greater incidence of conflict as HIs increase. Not all countries with high HIs experience conflict.
Indeed in our own country studies, both Ghana and Bolivia have high socioeconomic HIs yet
have avoided substantial conflict. It is therefore important to investigate when high HIs lead to
conflict and when they do not. While a few of the studies cited above include political HIs, most
do not. The nature of political HIs is one factor that determines whether high socioeconomic HIs
lead to conflict.
2. Conflict is more likely where political and socioeconomic HIs are high and run in the same
direction, or are consistent. Where they run in different directions, conflict is less likely
Where political and socioeconomic HIs are severe and consistent, both leadership and the mass
of the population in the deprived group(s) have a motive for mobilization – the leadership
because they are politically excluded (that is, they suffer from political HIs) and the population
because they suffer from socioeconomic HIs, and these inequalities can be used by leaders to
mobilize people. The Côte d’Ivoire story illustrates this. During Houphouët-Boigny’s time, there
was political inclusion, and the country was peaceful, despite some severe HIs on a north-south
socioeconomic deprivation and the lack of inequality in cultural status, particularly the lack of
recognition of the Muslim religion, was articulated in the Chartre du Nord of 1992, which
explicitly pointed to the unfairness of the system. But after the Houphouët-Boigny regime came
to an end, much more explicit political exclusion occurred, as Alassane Ouattara, a presidential
candidate from the north, was barred from standing in both the 1995 and 2000 elections. No
416
The Nigerian Civil War (1967-70), which resulted in the deaths of hundreds of thousands of
people, also illustrates the importance of the coincidence of political and economic incentives
and interests in provoking violent conflict. Politics in Nigeria’s First Republic (1963-66) had a
winner-takes-all character: ‘the winners appropriated all the fruits of office to themselves, and
excluded their opponents from them’ (Williams, 1982: 38). In particular, the two relatively well-
educated ethnic groups, the Yorubas and Igbos, were in competition with one another over high-
level positions within the federation (Nafziger, 1973). The coup d’état of 29 July 1966, led by Lt
Col Murtala Mohammed, a northerner, resulted in the exclusion of the Igbos from power and
initiated an increasingly anti-Igbo climate (particularly in the northern regions). Due to the
widespread anti-Igbo feelings and attacks in the period 1966-67, many Igbos migrated to their
home regions in the eastern part of Nigeria. ’Civil servants and university lecturers migrating to
the East, who had severed their previous employment ties, became a powerful lobbying group
for an independent Biafra, in which they now had a vested economic interest’ (ibid: 529). Both
economic and political exclusion was compounded by the fear that, for those without political
Socioeconomic deprivation tends to affect the mass of the people. They thus have grievances
which make mobilization for opposition or even violence possible. However, for effective
mobilization both elite and mass participation is required, especially since most serious conflicts
are organized, not spontaneous, and thus require strong leadership. The motives of the elite of a
group, or its potential leaders, are particularly important, because the elite controls resources
(including sometimes military ones), and they can rouse support by accentuating common
identities, and that of the ‘other’, while increasing perceptions of intergroup inequalities. In Brass’
terms, they play the role of ‘conflict entrepreneurs’. However, where the group’s elite enjoys
417
power, they are not likely to encourage or lead rebellion. This also holds true when they are
politically included without being dominant, as they can still enjoy the ‘perks’ of office, including
Consequently, even in the presence of quite sharp socioeconomic HIs, people are unlikely to
take to violent conflict if their own group leaders are politically included, and even less so if they
This is exemplified by the experience of Malaysia and Nigeria. In both countries, the group that
the population. In Malaysia, the bumiputera account for roughly two-thirds of the population
(depending on the precise categorization), while in Nigeria northern peoples are estimated to
account for over 50 per cent of the population.4 In each country, this numerical advantage has
translated into dominance of political power (continuously in the case of Malaysia and for most of
the time in the case of Nigeria). Having political power – and the ‘pork-barrel’ gains this confers –
obviously greatly reduces the motive of a group’s elite to lead a rebellion. It also permits action
to be taken to correct other inequalities. In the case of Nigeria, such action has been mainly
confined to the political sphere (including the bureaucracy and army through the Federal
Character Principle), but in Malaysia systematic action has also been taken in the
At a local level, contrasting the experiences of two Nigerian cities – conflict-ridden Warri and
more peaceful Calabar – in Chapter 9, Ukiwo also shows the importance of there being
consistent HIs, in socioeconomic and political dimensions, if they are to lead to conflict. In Warri
there were both political and socioeconomic HIs; but in Calabar, although the leaders of certain
groups did feel excluded and tried to mobilize support, their potential group followers believed
themselves to be well treated and consequently were not ready to be mobilized in protest. This
418
shows that consistent economic and political HIs are more likely to lead to conflict – but in the
opposite way to the country examples just cited, because in this case there was socioeconomic
Østby (Chapter 7) provides econometric support for the importance of consistency between
socioeconomic and political HIs. She shows that while political exclusion on its own as an
independent variable does not affect the probability of conflict, it has a strong interactive affect
with interregional asset inequality. That is to say, asset inequality has a stronger effect in
increasing the probability of conflict in the presence of political HIs. She finds a similar effect with
This is really a development of the previous finding, since where power is shared, political HIs
should be lower, hence making peace likely even where there are severe socioeconomic HIs.
When there is genuine power-sharing, no single group dominates political power, but all (major)
groups have some real sense of participation in government. Econometric evidence has shown
that formal power-sharing arrangements do reduce the potential for conflict as argued by Lijphart
(for example, Binningsbø (2005), explores the impact of proportional representation and
territorial autonomy within countries; and Reynal-Querol (2002a) finds a positive impact of PR on
the reduction of conflict-propensity). In the federal context, Bakke and Wibbels (2006) find that
‘co-partisanship’ between central and subnational governments, which implies shared political
power (at least regionally) and consequently lower political HIs, significantly reduces the chance
of conflict. In our studies, both Bolivia and Ghana have included deprived groups in government.
In the case of Ghana, there is an informal tradition in the Fourth Republic that whenever a
involved the political participation of indigenous representatives for much of recent history. Côte
419
d’Ivoire had a similar arrangement under Houphouët-Boigny, but northerners were subsequently
excluded and this was a major cause of the recent conflict (see Chapter 8). Guatemala, Peru
and Indonesia, each conflict-ridden at certain times, practised exclusionary government prior to
The implication of this finding, it is important to note, is that the political cooption of the
conflict without the necessity of undertaking policies to improve the socioeconomic position of
these groups. This has arguably been the case with respect to the Indian population of Malaysia,
which is represented in the governing coalition through the Malaysian Indian Congress, but
which has received little in the way of targeted developmental aid, despite severe pockets of
socioeconomic deprivation (Loh, 2003). Also in Nigeria, as noted above, while northern political
power has helped avoid major north-south confrontations, the northern part of the country has
remained seriously deprived in socioeconomic terms. It does not follow that this is a satisfactory
situation, given that severe horizontal inequalities are undesirable in themselves, in addition to
their instrumental role in fomenting violent conflict. Nonetheless, political inclusion does appear
to play an important role in preventing violence, and may constitute an important step towards
more inclusive development as ethnic leaders who do not ‘deliver’ development to their
constituency are likely to be challenged in the long run by new leadership contenders more
If political inclusion can reduce conflict, then the converse is also true; exclusionary political
practices can provoke conflict. Historical examples of rebellion and genocide show the powerful
role of the elite in this context. In the countries on which we focus here, this was exemplified by
the role of ethnic leaders in forming ethnic militias in Nigeria (Guichaoua, 2006), by the elite’s
behaviour and its use of the media in Côte d’Ivoire (Akindès, 2007), and by the importance of
420
(sometimes nonindigenous) leaders in Peru and Guatemala, and of local leaders in the
Indonesian separatist and communal conflicts. While political exclusion, as well as a shared
sense of identity with the deprivations of their group, can constitute a powerful motive for elites, it
is particularly likely to act as a trigger for conflict if there is a change in the political situation, as
in Côte d’Ivoire, where the government moved from an inclusionary to an exclusionary stance,
and in Indonesia, where the transition from autocracy gave rise to new opportunities for
mobilization.
Citizenship brings a variety of economic and political entitlements, as Gibney showed in Chapter
2. Not only political participation but also entitlements to a range of social and economic benefits
frequently depend on citizenship. Exclusion from citizenship is a form of HI in itself and also
may deny people the right to work, to join a union, or to receive government assistance. Denial
of citizenship is often a deliberate political act, taken for a variety of reasons, as elucidated by
Gibney. Historically, indigenous groups in the Latin American countries were denied citizenship
rights of both a political and economic nature. Moreover in some countries, there are, informally
Sources of loss of citizenship rights vary. Migration (legal and illegal) is a common source, and
in some cases the loss of citizenship is handed down across the generations. Less common are
instances when states explicitly revoke citizenship rights, as happened to the Jews in Nazi
Germany, and to the Asians in Uganda. A third way in which citizenship can be lost occurs when
the state itself changes form. Gibney cites the case of the Roma population, who became
stateless when the Czech Republic separated from Slovakia in the 1990s.
421
At a national level, denial of citizenship has been critical in inciting rebellion in the case of Côte
d’Ivoire (Chapter 8). Similarly, in Nigeria, the settler/indigene distinction has been the source of
many local-level conflicts (Bach, 2006; Human Rights Watch, 2006) while it has also been a
major source of local conflicts in Ghana (see, for example, Tsikata and Seini, 2004; Jönsson,
2006).
Gibney suggests three principles upon which citizenship might be based: that everyone should
be a citizen somewhere, and those without citizenship should be given it in the country where
they are located; first, that de facto membership of a state should confer the right to citizenship,
where de facto membership is defined by contributions and ties to the society; and that an
It is where any or all of these three principles are breached for significant numbers of people,
particularly if they belong to a common ethnic or religious group, that denial of citizenship can
provoke conflict. Moreover, given the close connection between citizenship and other economic
and social benefits (such as the right to work, or access to state services or land), exclusion from
5. Inequality of cultural recognition among groups is an additional motivation for conflict and
Cultural status inequalities (explored by Langer and Brown in Chapter 3) can be extremely
important. In the first place, culture (ethnicity or religion) itself is often the factor binding people
together as a group. Hence the more important it becomes in the way people see themselves
and others, the more likely it is that they mobilize along group lines. Cultural status inequalities
can increase the salience of one’s own and others’ cultural identities. As argued in Chapter 3,
there are three important elements involved in cultural recognition: differential treatment (formal
and informal) with respect to religion and religious observation, language recognition and use,
422
and ethnocultural practices. In some countries, conscious efforts are made to give equal
recognition with regard to each element. This is notable in Ghana, for example. In others, there
have been periods of explicit cultural discrimination (for instance, against the use of indigenous
events are also frequently a trigger for riots and even conflict, as exemplified by the Orange
marches in Northern Ireland, which set off the ‘Troubles’ in the 1970s, as well as by language
policy in Sri Lanka, and the desecration of religious buildings and sites in India and Palestine,
Much of this book has been concerned with reporting on observable HIs. Yet people take action
because of perceived injustices rather than on the basis of data of which they might not be
aware. Normally, one would expect perceived and observed inequalities to be related, so these
‘objective’ HIs are clearly relevant to political action. Yet it is also important to investigate
perceptions of inequality, even when the underlying reality remains unchanged. Chapter 10
reported on such an investigation in Ghana and Nigeria. The results were particularly interesting
in four respects. First, of those questioned in both countries (608 people in Ghana and 597 in
Nigeria, representative of three locations in each case but not of the country as a whole) the
majority perceived very little difference in educational access according to group despite the fact
that records of school attendance show large differences (Chapter 10). This may be because
perceptions of difference are based on opportunities at the local level or within the region in
which they live and much of the recorded difference is between regions. Secondly, in both
countries, respondents considered their religion to be much more important than ethnicity in the
423
private sphere (for example, in relation to social interactions, including marriage) but ethnicity
was much more important in the public sphere, in terms of their views on government job and
contract allocation. This points to the importance of maintaining ethnic balance in the political-
administrative sphere. A third significant finding was that, in general, Nigerians felt ethnicity to be
more important to them than did Ghanaians, both in terms of their view of their own individual
identities and also, they believed, as determinants of the allocation of government jobs and
contracts. Fourthly, significantly greater numbers of respondents in Ghana stated that their
national identity was important to them than in Nigeria. These last two findings may in part, be a
result of Nigeria’s more turbulent political history, but it may also be accounted for by national
leadership and the way the political and educational systems have developed historically in the
two countries.
The importance of the role played by perceptions in provoking action means that leaders,
institutions and policies that influence perceptions can affect the likeilhood of political
leaders to ‘market’ identities and differences via the media (Akindès, 2007). In Ghana, Nkrumah
himself, the first postcolonial leader, put a huge emphasis on national unity, in contrast to
leaders in Nigeria who had a much more regional perspective. Educational institutions are
relevant here too. In Ghana, boarding schools dating back to colonial times have brought the
future elite together from across the country and have contributed to a national project. These
factors may partly account for the findings of the perceptions surveys showing that Ghanaians
different action towards them. Perceptions can also be influenced by a variety of actions
(including symbolic ones). For example, both Nkrumah and Houphouët-Boigny initiated
424
programmes of investment in the deprived Northern regions with the intention of reducing
inequalities. Although they were insufficient to close the gaps, these measures led people to
believe that that there was some attempt to produce a fairer distribution of resources. Moreover,
the elites in the North benefited particularly from some of the programmes through, for example,
the allocation of contracts and this reduced their incentive to mobilize their members for group
conflict. The postconflict support for indigenous social and economic programmes in Guatemala,
while also too small to make a major difference, has changed perceptions of inequality, with
7. The presence of natural resources can be a significant cause of separatist conflict, as well as
of local conflict, often working through the impact this has on HIs
There is a well-established econometric link between the presence of natural resources, such as
gas and oil, and the incidence of conflict, but the precise causal mechanism is disputed (Le
Billon, 2001; Ross, 2004; Humphreys, 2005). In addition to encouraging increased competition
for power among the elite (because of the ‘spoils’ of controlling the state) our research suggests
that the conflict-inducing potential of natural resources is often mediated through their impact on
horizontal inequalities, and that this can translate into both separatist struggles and local-level
conflict (Brown, Chapter 12; Tadjoeddin, 2007). The discovery of natural resources can generate
sharp increases in regional inequality, and where these resources are located in ethnically or
religiously distinct regions of the country, separatist conflict may emerge, particularly if the
groups are relatively underdeveloped or feel they are not benefiting from the exploitation of the
resources.
In Brown’s analysis of separatist movements in Southeast Asia (Chapter 12), he finds that the
discovery of natural resources in the Indonesian province of Aceh was a vital step in the
425
Indonesian state turned into rebellion aimed at seceding from Indonesia altogether. This finding
is backed up by Tadjoeddin (Tadjoeddin, 2007; Tadjoeddin et al., 2001), who argues that natural
resources in Indonesia have generated an ‘aspiration to inequality’ in provinces where they are
located. Similarly, Treisman (1997) argues that natural resources played an important role in
stoking ethnic separatist claims in postcommunist Russia. In Sudan, the discovery of oil in the
southern regions came after the first outbreak of civil war in the country, but its discovery
nonetheless drastically affected the dynamics of the conflict. Indeed, the second outbreak of the
civil war in 1983 came shortly after Nimeiri completed a ‘redivision’ of the south which created a
This is the source of a major dilemma, and the cause of much debate in places such as
Indonesia and Nigeria. Do the people in resource-rich regions have some special rights over
resources found in the region? If these are conceded, then HIs will arise as the resource-rich
regions become much richer than others. Or, should the state redistribute the revenues (as for
example with the INPRES programme in Indonesia under Suharto and the redistributive formula
in Nigeria), which will moderate HIs, but may also lead to unrest (via Tadjoeddin’s ‘aspiration to
Natural resources are also often associated with local-level conflicts, as in the Niger Delta region
in Nigeria. Here too, the distribution of the resources among local groups, or between local
groups and companies, is often unequal and can consequently feed local-level conflict. Local
conflict in the Indonesian province of North Maluku was also instigated by ethnic disputes over
426
8. The nature of the state is of enormous importance to whether serious conflict erupts and
persists
Although highly repressive regimes can prevent conflict (for example, the New Order regime in
Indonesia was effective in preventing communal conflict in much of the country), an aggressive
state can also fuel and sustain a conflict. In both Guatemala and Indonesia (with respect to
separatist conflicts) the harsh and aggressive state reaction to rebellion sustained conflict for
many years, causing deaths on a massive scale and provoking further rebellion. In Guatemala,
state reaction to rebellion has been described as ‘a campaign of state terror’ (Caumartin 2005:
22) with massive killings, particularly focused on the indigenous population. In Indonesia, the
viciousness of the Indonesian armed forces’ response to the original, small-scale Acehnese
rebellion boosted support for the movement when it reemerged (Brown, Chapter 12; Kell, 1995).
In contrast, the respective state’s handling of disputes in Ghana and of some local-level conflicts
in Indonesia has dampened some conflicts and avoided others. But in Nigeria, it seems, state
action is often late and one-sided, making the conflicts more severe than they need have been.
This is exemplified by recent events in the Middle Belt. The government’s passive and late
reaction to the emergence of the violence between Muslims and Christians on 7 September
2001 in the Jos area is illustrative in this regard. It took the Nigerian military and police forces
more than 12 hours to arrive at the scene of the violent conflict and many areas were left without
security for the first 24 hours of the crisis (Higazi, 2007). Moreover, the intervention of the
security forces was perceived to be biased against Muslims and they alleged that police forces
Local institutions are also important in determining the trajectory of violence. This has been
shown by research in Indonesia, Ghana and Nigeria. For example, Asante (2007) has shown
how in Ghana a conflict between adherents of the Ga traditional religion and some Christian
427
Churches over drumming (which forms an important aspect of the celebration of the ‘Homowo’
festival of the Ga people) was prevented from escalating by the way it was handled by local
institutions, particularly through the adoption of several bylaws regulating noise-making in Accra
by the Accra Metropolitan Authority. A similar pattern is found in many other incidents in Ghana
(Tsikata and Seini, 2004). In the conflict in and around Jos in 2001, by contrast, local authorities
did little to stop the conflict, and indeed may have contributed to it, leaving any solution to
We showed how persistent HIs can be in Chapter 4. Many HIs have colonial origins, caused by
colonial powers privileging some groups or regions (or both). This was the case in all our
studies. In Latin American inequalities were caused by the privileged settlers taking the best
resources for themselves and sustaining their privilege through discrimination and unequal
access to every type of capital. Figueroa has shown not only that indigenous people in Peru
have much less access to education than the mestizo population, who in turn have less than the
whites, but also that for any particular level of education the returns, in terms of additional
incomes earned as a result of such education, are significantly lower for the indigenous
population than for the mestizos, which are again lower than the returns the whites secure. This
is due to a combination of poorer quality education, less good social networks and discrimination
In West Africa, the inequalities were partly due to geographical and climatic differences, but
these were made worse by colonial economic policies that favoured the south of each country in
terms of economic and social infrastructure. In Malaysia, the colonial ‘ethnic division of labour’
(Brown, 1997) ensured that the Malays remained in subsistence agriculture, while migrant
Chinese came to dominate the domestic economy. The persistence of these inequalities is due
428
to the cumulative inequalities arising from unequal access to different types of capital – including
financial, land, education and social capital. Asymmetries of social capital, in particular, arising
because group members have stronger contacts within their group than across groups, have
made it almost impossible for some groups to escape from these inequalities.
However, there are cases of ‘catch up’, where group differences are reduced over time, although
differences are rarely completely eradicated. Mostly these are policy-related, where a conscious
and systematic effort has been made to correct the inequalities, as in Malaysia after 1970. In a
few cases, groups have succeeded in catching up without government support – Chinese and
Vietnamese immigrants in the US are an example of this. Their success seems to have been
partly due to the selective immigration policy – which only allowed the more educated to enter
the US – and partly to the culture of work, education and achievement which they brought with
them, cultural capital of a type which is typically absent in long-deprived groups. The nature of
the interlocking forces which typically perpetuate HIs over generations implies that very
10. International policies and statistics are too often blind to the issue of HIs, though national
The international policy community is primarily concerned with poverty reduction and with
promoting economic growth, and almost all policies are directed to these ends. Vertical
inequality is beginning to be recognized as a problem (Kanbur and Lustig, 1999; Cornia, 2004;
UNDP, 2005; World Bank, 2006) because inequality has risen in the majority of countries in
recent years and because high and growing inequality makes poverty reduction more difficult.
Even so, vertical inequality has not received serious policy attention, and there is still less
attention paid to horizontal inequality. The growth and poverty reduction agendas are both blind
to HIs.
429
Internationally advocated policies to support economic growth mainly involve macropolicies
designed to secure economic stability and openness, and mesopolicies intended to support
economic infrastructure and enhance the role of the market in order to improve efficiency.
Poverty reduction policies are mainly derived from the Poverty Reduction Strategy Papers
(PRSPs). These focus particularly on social sector expenditures and on some special schemes
for the poor such as microcredit. According to a review, ‘the participation of minorities or
indigenous peoples is either often overlooked or simply regarded as impractical due to their
marginalisation’ (Booth and Curran, 2005:12). Analyzing the content of PRSPs shows universal
inclusion of the ‘normal’ macro conditions, and policies to promote the social sectors. Gender
mentioned in only a quarter of the cases. The cases where ethnic minorities were not mentioned
include countries which are evidently heterogeneous, such as Azerbaijan, Benin, Burkina Faso,
Chad, Guyana, Malawi, Mali, Mauritania, Mozambique, Niger, Rwanda, Tanzania and Uganda
(for example, World Bank, 2005; DFID, 2007), but it does not form a systematic part of reporting,
and rarely enters policy discussions, apart from the regional dimension. Indeed, national policies
which are explicit in addressing HIs, such as the Federal Character Principle in Nigeria and the
International policies towards political systems also tend to ignore the question of HIs. Thus
Indeed this has been used to defend one-party systems in many multicultural societies. In some
postconflict societies, the issue is so obvious that power-sharing arrangements have been
430
supported by the international community, as in Bosnia Herzegovina and Lebanon. But wider
A dearth of international statistics on the issue reflects this lack of focus. For example, neither
the World Bank nor the UNDP includes statistics on ethnic, religious or regional HIs in their well-
known datasets, although some national-level Human Development Reports, such as those on
Kosovo and Nepal, have provided ethnic or religiously disaggregated data. A notable exception
is the Demographic and Health Surveys (to date covering 77 countries) funded by USAID with
contributions from other donors. These contain ethnic and religious variables in quite a number
of cases, permitting the investigation of relationships across countries, as has been done by
Østby. But these surveys do not cover all countries, and are not carried out at regular intervals.
Nor do they include political variables or other variables of obvious interest, such as household
income.
As far as national policies are concerned, there is a much higher consciousness of the
importance of HIs in many heterogeneous countries, and a vast array of policy approaches have
been adopted, as will be indicated in the next chapter. Nonetheless, by no means all culturally
diverse countries acknowledge the importance of HIs, or take policy action towards them, some
because, like the international community policies they often adopt, they are blind to these
In the countries studied in this book, among the Latin American cases deliberately exclusionary
policies were practiced in colonial times and carried over into independence. More recently,
formally policies have been inclusionary, but informally there is a great deal of inequality and no
conscious effort to correct this, especially in the cases of Peru and Guatemala. Bolivia has been
more inclusionary politically, but not in terms of economic approaches. After the conflict in
431
Guatemala ended, some of the country’s high inequalities, notably with regard to education and
culture, have begun to be tackled. In West Africa, policies have varied over time. As in the case
of Latin American, the colonial governments created HIs. Since independence there have been
weak attempts to correct socioeconomic HIs, but these have been largely offset by
macroeconomic policies which have pulled in the opposite direction. On the political front,
Ghana has generally adopted informal inclusionary policies, and Côte d’Ivoire had a similar
stance for several decades until exclusionary policies were adopted after Houphouët-Boigny.
Nigeria, in contrast, has made a conscious attempt to correct political HIs via the Federal
Character Principle (Mustapha, 2007) (but has done little to correct socioeconomic HIs). The
Federal Character Principle was partly a response to the devastation of the 1967-70 civil war
and the determination that this should not be repeated. In Southeast Asia, Malaysia has been
most explicit, systematic and effective in correcting socioeconomic HIs via the NEP. A national
coalition government involved all groups in society and thus, although Malays predominated,
political HIs were kept largely in check. In Indonesia, there was a conscious but limited attempt
to secure regional equity via the INPRES programmes, but political HIs were sharp under
Suharto. Since democratization, extensive political and fiscal decentralization has effectively
reduced these problems by making the district the main object of political power. An equalization
formula is also in place to ensure that poorer districts receive a greater proportion of central
funds.
4. Conclusions
In summary, this book has elucidated the meaning of HIs, pointing to the multidimensionality of
the concept. We have shown that severe HIs can be an important source of conflict, especially
where they are consistent across dimensions. While socioeconomic HIs generate generally
fertile ground for conflict to emerge and cultural status inequalities act to bind groups together,
political HIs provide incentives for leaders to mobilize people for rebellion. In conditions of
432
severe HIs, abrupt changes in political HIs, or cultural events in which important cultural or
In the introductory chapter we put forward four hypotheses about the relationship of HIs to
violent conflict. Evidence in this book has supported the first three:
• that conflict is more likely where there are significant political or economic HIs, or both;
We were unable to explore the fourth hypothesis systematically – that political mobilization and
possibly conflict is more likely to occur when HIs are widening – because of a dearth of accurate
evidence on HIs over time, although evidence for districts in Indonesia (in Chapter 6) supported
the hypothesis. Thus this hypothesis remains a presumption, supported at present by piecemeal
evidence.
Other factors are, of course, also important in determining whether a conflict emerges. One is
the nature of the state and its reactions, another is the role of local institutions in pacifying or
dynamizing conflict once it has started, a third factor is the presence of natural resources, often
working through the impact this has on HIs. For the most part, especially among the international
community, little attention is paid to the issue of horizontal inequalities, and the policies
advocated can sometimes accentuate them. This is true both of economic policies, such as
structural adjustment, and political polices, such as multiparty democracy. The next chapter
discusses what policies would be appropriate once the importance of the issue has been
acknowledged.
433
.
1
Ideological motivation was also much more common in the Cold War era, with leaders often
adopting Marxist positions. Following the demise of most communist regimes, ethnic
matanda a los indios’ (Francisco Bianchi, spokesman for the governor of Rioss Montt).
3
Caumartin (2005: 28).
4
Political resistance to censuses in Nigeria casts doubt on all population estimates.
5
P. Verney, ‘Raising the stakes: Oil and conflict in Sudan’. Sudan Update, December, 1999.
434