Horizontal Inequality: Importance and Measurement

1
Horizontal Inequality Importance and Measurement

• Frances Stewart

2
Plan of lecture

• What HIs are
• How group boundaries are formed
• Why HIs matter
• Some examples
• Some findings on HIs and conflict
• Why HIs persist
• Measurement issues
• Policy conclusions

3
WHAT HIS ARE Perspectives on inequality

• Most economists measure and evaluate VERTICAL
  inequality among individuals or households.
• And most attention paid to income inequality.
• Policies (efficiency/poverty) all in terms of
  individuals.
• Welfare (utilitarians) or wellbeing and
  capabilities focus on individuals.

4
WHAT HIS AREHorizontal Inequality (HI)

• HI is inequality between groups .
• Economists often group people by income
  categories.
• Here aim to classify people into groups which
  have meaning to members, viewed by people
  themselves, or others as important aspect of
  identity.
• Group boundaries vary in different societies (and
  over time)
• Examples of salient identities
• Ethnic/tribe African
• Religious most regions notable N.Ireland
  Middle East
• Race e.g. South Africa Malaysia Fiji
• Regional (overlaps with other identities)
• Caste (India)
• Class (Latin America)

5
WHAT HIS ARERole of class

• LA especially class more of identity apparently
  than ethnicity.
• Performs same role as mobiliser.
• Often coincides with other differences
  (indigenous race)
• Where class is group identity, sharp HI
  inevitable.
• Diminishing role of class post-cold war rising
  role of ethnicity
• a virtual cornucopia of these seemingly
  intractable (and previously invisible) social
  identity conflicts exploded onto the world scene
  and captured the public and policy eyes.
  (Center for Systematic Peace)

6
WHAT HIS AREProportion of conflicts classified
as ethnic
Source Calculated from Marshall (2006)
7
WHAT HIS AREHIs are Multidimensional

• Dimensions are those that matter to members
  affect well-being, sense of injustice, actions.
• Salient dimensions vary according to nature of
  society/economy
• Important dimensions include
• Politics (political participation, power, at all
  levels).
• Economic resources and outcomes (access to
  assets, employment, incomes).
• Social, including services (health/education/water
  .. and social networks).
• Cultural status
• Should also be true of vertical, but despite lip
  service rarely included.

8
GROUP BOUNDARIESImportant question what
determines group boundaries?

• Boundaries often arbitrary.
• Multiple identities. Which do we/others
  emphasise? E.g. white black classification,
  Brazil/US
• Fluid. Mestiso in Latin America. Indigenous,
  cholo.
• Salient groups can change over time. (Moslems in
  Sri Lanka Iwerri in Biafra).

9
GROUP BOUNDARIES Anthroplogists views

• Four views of anthropologists
• Primordial deep historic, even biological
  origins boundaries rigid and unchangeable. (e.e.
  Smith). Yet boundaries change over time.
• Instrumentalist leaders create identities for a
  purpose, administrative, economic, political.
  (Cohen Glazer and Moynihan.)
• Constructionist identities made and
  remade.(Anderson).
• Categorisation approach (by others Barth
  Brubaker and Cooper).

10
GROUP BOUNDARIESInstrumentalism Colonial
construction

• Modern Central Africa tribes are not so much
  survivals from a pre-colonial past but rather
  colonial creations by colonial officers and
  African intellectuals.. (Wim van Binsbergen)
• In nineteenth century far from there being a
  single tribal identity, most Africans moved in
  and out of multiple identities, defining
  themselves at one moment as subject to this
  chief, at another moment as a member of that
  cult, at another moment as part of this clan, and
  at yet another moment as an initiate in that
  professional guild. (Ranger).

11
WHY DO HIS MATTER? Construction and use of
identities for mobilising support for conflict

• Increasing since end of Cold War.
• Identities constructed and emphasised by leaders
  to get support.
• Powerful mobilising agent. Ethnicity or religion.
  E.g. Bosnia, Rwanda, Sudan. N.Ireland. France
• But not plucked from air constrained by
  history, language etc. Those with perceptions of
  common identity share some markers (language,
  behaviour, rituals, religious practices).
• Differences seem real to participants. Turton
  the very effectiveness of ethnicity as a
  means of advancing group interests depends upon
  its being seen as primordial by those who
  make claims in its name

12
WHY DO HIS MATTER?Why do HIs matter?

• Instrumental
• Affects economic growth. Group handicap prevents
  efficient solutions. Unequal access to assets,
  markets, social capital.
• Affects poverty. If discrimination major reason
  for poverty, the group-blind policies may not
  work. Plus conflict reduces growth.
• Direct impact on Wellbeing.
• Group inequality affects peoples
  well-being.Being Black and feeling Blue .
  Akerlof welfare function.
• Resentment of others.
• Knowledge that chances of being discriminated
  against are high.
• Political Instability

13
WHY DO HIS MATTER? Affects political stability

• HIs can lead to group mobilisation
• Ethnic or religious boundaries powerful source of
  mobilisation in general, but
• Especially where there are blatent HIs. Used by
  ethnic entrepreneurs. Maybe worse if growing.
  Many examples
• Rwanda
• N.Ireland
• Kosova.
• Sri Lanka
• Also riots
• US cities in 1970s
• Sporadic, cities in UK
• Many examples in developing countries India.
• Also applies internationally.

14
WHY DO HIS MATTER? Group identity, inequality
and conflict major hypothesis

• Vast majority of groups live together peacefully
• Cultural/group differences only become salient
  or potentially a means of political mobilisation,
  especially for violence, when OTHER factors
  present.
• Cohen AMen may and do certainly joke about or
  ridicule the strange and bizarre customs of men
  from other ethnic groups, because these customs
  are different from their own. But they do not
  fight over such differences alone. When men do
  fight across ethnic lines it is nearly always
  the case that they fight over some fundamental
  issues concerning the distribution and exercise
  of power, whether economic, political, or both_at_
• I.e. cultural differences do not lead to
  violent conflict unless there are also major
  economic and/or political causes.

15
EXAMPLES OF HIS
16
EXAMPLES OF HIS Chiapas, Mexico
17
EXAMPLES OF HIS
18
EXAMPLES OF HIS South Africa
19
CRISE RESEARCH SOME FINDINGSSome findings of
CRISE research on HIs and conflict

• Probability of conflict arises as socio-economic
  HIs increase Econometric cross-country evidence
  by Ostby Gurr Barrows Within country evidence,
  Mancini Gates and Murshed. Using a variety of
  group definitions and HI definitions.
• N.b.PROBABILITY NOT CERTAINTY.

20
CRISE RESEARCH SOME FINDINGS

• 2. Conflict more likely where political and
  socio-economic HIs are consistent.
• - political HIs motivate leaders
  socio-economic followers.
• - examples Cote dIvoire, versus Malaysia and
  Nigeria. Warri versus Calbar.

21
CRISE RESEARCH SOME FINDINGS

• 3. Inclusive government tends to prevent
  conflict.
• - Econometric evidence shows that PR, and
  federalism tends to reduce conflict probability.
• - Contrast Bolivia and Peru and Guatemala Ghana
  and Nigeria and Cote dIvoire.

22
CRISE RESEARCH SOME FINDINGS

• 4. Citizenship can be important source of
  exclusion.
• - citizenship confers political, economic and
  social rights. Therefore source of general
  exclusion (HIs)
• - source of exclusion
• i. States decide to remove citizneship
  rights (Nazis Uganda towards Asians)
• ii. Migrants refused citizenship even after
  generations (Cote dIvoire)
• iii. State changes Czechoslvakia division
  left Roma peoples stateless.
• - can be LOCAL as well as national Indonesia,
  Ghana and Nigeria, Indigenes versus settlers.

23
CRISE RESEARCH SOME FINDINGS

• 5. Perceptions important as well as observed
  inequalities.
• E.g. in Nigeria and Ghana, dont perceive big
  differences in access to jobs, but do perceive
  differences in access to govt. jobs and
  contracts.
• In Nigeria, generally perceive themselves as more
  ethnic and less national.

24
CRISE RESEARCH SOME FINDINGS

• 6. Natural resources can be important source of
  HI, creating
• -regional inequalities in incomes
• - regional resentments about rdistribution
• - within region inequalities.
• This is one mechanism linking NR to conflict
  others include finance, and greed.

25
CRISE RESEARCH SOME FINDINGS

• 7. Nature of state hugely important as to whether
  HIs lead to conflict.
• - state accommodating in Ghana conflicts less
  so in rest of West Africa.
• - state actions in Aceh,Indonesia, and Guatemala
  fueled and prolonged conflict.
• - accommodating state in Sabah, Malaysia
  prevented violent separatism in contrast to
  Thailand, Philippines, East Timor.

26
THE PERSISTENCE OF HIS

• Many HIs very persistent.
• Persistent inequalities in Ghana (North-South)
  US black-white indigenous people, Latin America
• Last centuries.
• Wellbeing impact worse because of this.

27
THE PERSISTENCE OF HIS North south inequalities
in Ghana over 60 years
28
THE PERSISTENCE OF HIS Reasons for persistence
in HIs

• Initial inequality (often colonial cause) leads
  to multiple disadvantages
• 1.Inequality of incomes generates inequalities
  in accumulation, including Hk.
• 2. Asymmetry of social capital, especially
  marked across groups, because groups have more
  intra-group contacts and fewer between group
  contacts. Poorer groups have less advantageous
  contacts.
• 3. Interaction among capitals low Hk worsens
  returns to financial capital low financial
  capital reduces returns to Kk. Low social capital
  reduces returns to each.
• 4. Continued discrimination.
• 5. Political HIs which reinforce disadvantage.

29
THE PERSISTENCE OF HIS Returns to education in
Peru.
30
MEASUREMENT Measurement issues

• What are we trying to measure?
• Many approaches measure group inequality as
  component of total inequality,
• i.e. BG WG inequality (vertical).
• HI BG/WG.
• This approach, HI measure depends on intra-group
  inequality. Same group average difference would
  be much greater if WG is low than if IG is high.
• Want an independent measure of HIs.

31
MEASUREMENT Differences between measuring HI and
VI

• Groups have members, size differs.
• Individuals enjoy welfare, not groups. Measures
  of VI may be welfare related. Less clear that HI
  measures should be.
• Within group distribution may differ not
  applicable to VI.

32
MEASUREMENT Some issues

• Not looking for a measure of welfare. (Note
  Sen/Anand measure wellbeing of whole according to
  gender distribution GDI.) A descriptive measure
  that is helpful for researching consequences of
  group inequality and for policy.
• Should group size influence measure?
• Should intra-group distribution influence
  measure?
• As such, i.e. value group performance differently
  according to distribution within group
• Value differences according to whole
  distribution, not just mean.
• What should we do about multidimensional nature
  of HIs?

33
MEASUREMENT Measuring HIs without looking at
intra-group distribution

• Possible measures
• Group Gini, or Gini weighted by population
  (GGini) compares every group with every other
• Group weighted coefficient of variance (GCOV)
  squares differences from mean.
• Group Theil, weighted by population especially
  sensitive to low end of distribution. Good for
  decomposition, but we are not interested in that
  property.

34
Weighted GCOV
GGINI

GTHEIL
35
MEASUREMENTRejectedmeasures

• Estaban Ray polarisation index. Like Gini but
  weights by population according to a factor , 1
  a, where 1lta lt1.6, typically a1.5. a increases
  weight given to large groups, and index rises as
  population is divided among fewer more equal
  sized groups.
• Measures groups by income difference, not by
  other characteristics (race etc.)

36
ER (k,?)
K1/y, a 1.5
37
MEASUREMENT Zhang/Kanbur

• Uses Theil to decompose.
• The ratio of BG/WG, thus shows relative
  contribution of BG and WG, not BG as such.

38
MEASUREMENT Odds ratio (Chakravarty).

• Ratio of odds that an individual of group A will
  be poor to odds that individual of group B will
  be poor (or other characteristic).
• Two group comparison, like simple ratios.

39
Unwted GCOV
Wted GCOV
Wted Gini
ER measure
White-black income p.head ratio
South Africa 1970-2000
40
MEASUREMENT General empirical findings on
differences In measures

• 1. GCov, GTheil and GGini highly correlated with
  each other.
• 2.Poorly correlated with measures of vertical
  inequality, or with ER or ZK.
• 3. Simple group ratios sometimes most informative.

41
MEASUREMENT Whole distribution

• Some advantage in measures which include allow
  one to look at comparisons across whole
  distribution
• Political and policy differences according to
  where inequality is.

42
C1
Group A
B1
D1
43
MEASUREMENT a-means (Foster).

• a -means (Foster), weights each distribution
  according to a value, a.
• The higher a the more high incomes are valued,
  and conversely.
• where a1 (the arithmetic mean) the mean value is
  most sensitive to median incomes.
• For values of a lt 1, more sensitive to low
  values of X and increasingly so as a decreases.
• Comparing a-means for different values of a thus
  indicates how HIs differ in different parts of
  the income distributions of the two groups and
  thus allows a comparison of group means in
  different quantiles of each groups distribution
  of variable X.

44
(No Transcript)
45
MEASUREMENT Multidimensionality?

• Always lose information.
• Does not make sense at all across major
  dimensions politics, economics, culture.
• Variety of approaches within dimensions
• Standardise and average
• Dominance
• Decisions all somewhat arbitrary.

46
POLICIES Policy conclusions

• HIs can be serious policy needs to address them,
  in all economies with marked inequalities.
• Important for general wellbeing and poverty
  reduction as well as political stability
• NOT included in normal economic or political
  policies.
• Policies similar to those towards exclusion

47
POLICIES

• HIs neglected, in international policy
• Aid, often worsens HIs.
• Structural adjustment
• PRSPs
• Also multiparty democracy
• But more included in Human rights approach
  and Social exclusion approach.
• Much more often included in national policies.

48
POLICIES 3 types of policy

• Direct targeted. Cab be effective. But can
  entrench ethnicity.
• Indirect may be less effective, but also arouse
  less hostility.
• Towards greater intergration very long term.

49
POLICIES Indirect Policies can

• Relate to process. V. important where processes
  previously biassed N.Ireland. (I.e. equal
  opportunities) But generally insufficient. Legal
  policies need backing up with support for legal
  access, and leverage (e.g. via contracts)
• Focus on public sector
• Fiscal policies
• Expenditure, including beneficiaries, contracts
  and jobs
• Use regional policy, where groups are regionally
  concentrated or even district or neighbourhood
  policies.

50
POLICIES Categories of direct economic and
social policies

• Assets
• Land (Malaysia Zimbabwe Fiji Namibia)
• Financial capital (Malaysia S.Africa)
• Terms of privatisation often unequalising
• Credit (Fiji Malaysia)
• Education (Malaysia Sri Lanka).
• Skills and training (Brazil, New Zealand)
• Public sector infrastructure (S.Africa).
• Housing (N.Ireland).
• Social capital? neighbourhoods clubs
• Incomes
• Employment policies
• Public sector (Malaysia Sri Lanka)
• Private sector (S.Africa)
• Fiscal policies inc.
• Transfer payments (often for age or gender, not
  for ethnic group)
• Terms of trade

51
POLICIES Main consequences

• Mostly successful in reducing gaps, but rarely in
  eliminating them. (But N.Ireland education. Sri
  Lanka).
• Does not seem to reduce efficiency. In fact may
  increase it.
• Must work on economic as well as social
  (N.Ireland, Malaysia v. Ghana, US).

52
POLICIES Negative consequences

• May reduce inter-group inequality, but increase
  intra-group. (But intra-group decreased in
  Malaysia more evidence needed).
• Claimed to entrench ethnicity as category. But
  with sharp HIs these may be entrenched anyway
  (N.Ireland, US). If changes ethnic division of
  labour may reduce ethnic salience.
• Can provoke political protest, even violence, Sri
  Lanka clearest example. Micro (non-policy) cases
  in Indonesia. But elsewhere reduces political
  violence Malaysia, N. Ireland, US

53
POLICIES Malaysia a successful case of reducing
econ/social HIs.

• 1971, following anti-Chinese riots, 1969. NEP.
  Aim to secure national unity. 
• Characteristics.
• Two prong to reduce and eventually eradicate
  poverty and to accelerate the process of
  restructuring Malaysian society to correct
  economic imbalance so as to reduce and eventually
  eliminate the identification of race with
  economic function (Second Malaysian Plan
  1971-1975)
• a variety of anti-poverty policies (rural
  development social services).
• restructuring
• expand Bumiputera share of capital ownership to
  30.
• 95 of new lands to be settled on Malays
• educational quotas in public institutions laid
  down, in line with population shares
• credit policies favoured Malays, with credit
  allocations and more favourable interest rates.

54
(No Transcript)
55
(No Transcript)
56
POLICIES N.Ireland economic changes underlying
political ones

• His large, persistent and consistent over all
  dimensions over a long time period
• gtBy the end of the nineteenth century Protestants
  controlled the vast bulk of the economic
  resources of east Ulster - the best of its land,
  its industrial and financial capital, commercial
  and business networks, industrial skills.(Ruane
  and Todd 1996)
• no narrowing of the gap between the communities
  from 1901 to 1970s, with Catholics disadvantaged
  at every level.
• u/e gap widened
• New policies to reduce gaps from late 1970s Fair
  Employment Acts, 1976 1989 housing policy.
  Police Acts 1998,2000, 2003 50 recruitment aim.

57
(No Transcript)
58
N. Ireland intervention on HIs and peace
Troubles
Cease fire
Good Fri agreement
59
Political sensitivity Sri Lanka case
60
In conclusion

• HIs are important for wellbeing and
  instrumentally.
• Neglected in much data collection and policy.
• Range of policies available, economic and
  political, which can be effective without
  sacrifice of efficiency.
• Policies needed in ANY society with sharp
  divisions, not only those with recent conflict.
• NB. Focus on HIs does not discount importance of
  other issues growth, environment, poverty and
  VI.