8/8/2019 47 Growth Development Inequality ion Poverty

1/4

Brieng PaperOverseas Development

Institute

Overseas Development Institute

ODI is the UKs leading independent

think tank on international develop-

ment and humanitarian issues.

ODI Brieng Paperspresent inorma-tion, analysis and key policy recom-

mendations on important develop-

ment and humanitarian topics.

This and other ODI Briefng Papers

are available rom www.odi.org.uk

Key points

Polarisation hindersboth growth and humandevelopment

Measuring incomepolarisation provides adistinct insight into thelink between growth andhuman development

Policy recommendations

based on polarisationmeasures will dierrom those based onconventional inequalitymeasures, and are likely tobe more nuanced

Several countries, particularly in Asia,have seen soaring economic growthover the last decade. But when itcomes to social indicators, such as

health and education, they have made morelimited progress. Inequality is oten blamedor the poor links between economic growthand human development, but data show thatlinks between economic growth, inequality andhuman development are less robust than oten

assumed.Important eatures o income distribution are

not easily captured by conventional measureso inequality such as the Gini coecient, par-ticularly the way in which income is distributedacross society. It was the Wolson polarisationmeasure, or example, that highlighted thephenomenon o the disappearing US middleclass missed by the Gini.

Unlike the measurement o inequality,polarisation measures ocus on the cluster-ing o members o a society at more than oneincome level reerred to as poles capturing

convergence around these income levels.Polarisation measures help to clariy why

economic growth does not always translateinto human development. They provide a dis-tinct and complementary insight into the linkbetween growth and human development,capturing the distributional aspects missedby traditional inequality measures. They lead,thereore, to dierent policy recommendations.

Polarisation: Distinct policy relevanceWhat are the distinct policy implications thatare provided by polarisation measures, and

missed by measures o inequality?The two polarisation indices used most oten

are by Wolson, and by Duclos, Esteban and Ray(DER). The Wolson bipolarisation measure isthe more conventional, assuming two groups inthe income distribution: those above average

income and those below. Developed in 1994, itmarks the rst major attempt to quantiy incomepolarisation and is now used by the World Bankor its household survey analysis (PovcalNet).

A rise in the Wolson index indicates a deep-ening in bipolarisation and a disappearance othe middle class. To mitigate urther hollowing

out o the middle-income groups, two types opolicies are needed. To reduce the poorer group,policy could build human capacity by, or exam-ple, eliminating secondary school ees. It couldsupport income generation through small andmedium enterprises or micro-nance grants.Policy recommendations to spread the wealth inthe high-income group may include redistribu-tive policies, such as progressive taxation.

The DER index, developed in 2004, meas-ures polarisation across an income distribu-tion without assuming a specic number ogroups. It thereore provides insights that

are distinct rom those o the Wolson index,which assumes just two income groups in thepopulation. A rise in the DER does not meanthat the middle class is alling away, ratherthat the income distribution has greater varia-tion, spikiness or multimodality (Duclos et

Growth without development:

Looking beyond inequalityDoes polarisation explain the divergence

between growth and development?

Brieng Paper 47February 2009

A amily takes shelter ater foods in Pakistan

here, poverty means more than lack o income.

8/8/2019 47 Growth Development Inequality ion Poverty

2/42

Brieng Paper

Figure 1: Inequality versus polarisation

Source: adapted rom Wolson (1997).

al. 2005). The DER may, or example, refect risingpolarisation among three income poles extremelypoor, poor and high income earners.

The DER measure can be divided into three com-ponents. The rst is mean alienation: the averagedistance between individuals on an income distri-bution. The theory, oten ound in the literature oninequality, is that a person may eel estranged rom

those ar away in terms o income. The second compo-nent is mean identication, which captures the den-sity o people clustered around a particular incomelevel. Based on theoretical oundations, Duclos et al(2005) assume that an individual at one particularincome level identies with those on the same level an identication that grows as the group expands.Increased identication means that groups that arealready densely populated become more populatedat the expense o smaller groups. The danger is notonly the risk o increasing antagonism between largerpoles, but that these larger groups carry more weightwhen advocating or particular aims, regardless o the

costs or growth and development across the widersociety.

The third component is a correlation component,measuring the relationship between alienation andidentication. Because the DER can be unpacked, itshows whether rising identication or alienation areresponsible or the changes in polarisation. It can,thereore, tell policy makers whether they shouldocus on reducing alienation or identication.

Polarisation vs. inequalityThe terms polarisation and inequality tend to be usedsynonymously. In a New York Times article, Daniel

Altman links the two when he writes ...meaninguldiscussion o the potential widening o inequality the polarization o rich and poor has been sadlylacking (Altman, 2003). A Brookings Institution pub-lication called Globalization and Income Polarizationin Rich Countries discusses inequality throughout.

Distinguishing between the two is important, as theyrequire distinct policy responses.

Inequality measures assume that a transer roma wealthy person to a less wealthy person will reduceinequality (reerred to as the principle o transers).Polarisation measures, however, ocus on where onthe income distribution curve this transer occurs.This determines the impact. Consider a transer o

resources rom a person with a higher income tosomeone with a lower income, both o whom earnabove the average income, such as an educationsubsidy nanced by a progressive tax system. Thetranser results in a convergence o income betweentwo people. One has a little less, the other a littlemore. According to the principle o transers, ine-quality should decrease. In act, polarisation mayincrease, as this transer has contributed to a reduc-tion in the size o the middle income group, and anincrease in the size o the wealthy group.

Figure 1 shows how inequality measures can hidedistributional divergences. The horizontal axes repre-

sent income levels normalised around the average,while the vertical axes represent the population. Thedotted black line on the let denotes an equal popula-tion size or each income (a uniorm density unction).

Consider a series o income transers that reshapethe income distribution, as shown on the right o Figure1. The extremely poor become wealthier (blue arrow),the super-rich become less rich (green arrow), andmiddle income groups become more or less wealthy(orange and purple arrows). These transers mayinclude targeted cash transers, uel subsidies, oodstamps and progressive tax schemes, etc. The incomedistribution, depicted by the red line, now has two

peaks clustered around two income levels thesomewhat poor and the somewhat wealthy andthe middle group alls away. The polarised curve (redline) is considered more equal (based on conventionalinequality measures) than the uniorm density curve(black dotted line) (Wolson, 1997).

population population

uniormdensity

income income

post-transer

density

mean-preservingtransers

.25 .50 .75 1.00 1.25 1.50 1.75 .25 .50 .75 1.00 1.25 1.50 1.75

1.00

.80

.60

.40

.20

1.00

.80

.60

.40

.20

8/8/2019 47 Growth Development Inequality ion Poverty

3/43

Brieng Paper

This shows that conventional inequality measuresail to capture the changes in an income distributionthat are captured by polarisation measures.

Empirical evidenceDuclos, Esteban and Ray (2005) have applied theirmeasure to 21 countries to show the dierencebetween their polarisation measure and a Gini ine-quality measure. The UK and US, or example, havethe same Gini coecient, yet the UK has a higherpolarisation measure. When the DER is examinedmore closely, mean identication outweighs averagealienation in the UK, resulting in a higher polarisa-tion index. So policies to address polarisation in theUK should ocus more on identication issues.

Or take the example o the US and the Czech

Republic. They have similar polarisation measures,but the Czech Republic is more equal by Gini stand-ards. This is captured in the alienation component people eeling estranged rom those with dierentincomes which is higher or the US than the CzechRepublic. The high alienation component in the US,however, is balanced by a lower identication andcorrelation component, resulting in a polarisationmeasure equal to that o the Czech Republic. Thisanalysis suggests that policies in the Czech Republicshould respond more to identication, while thosein the US should ocus more on reducing alienation.Policy measures associated with each o these are

discussed below.Gasparini et al. (2006) have perormed a simi-

lar exercise on household surveys rom 21 LatinAmerican and Caribbean countries. A comparisonbetween Mexico and the Dominican Republicshows that both have the same levels o averagealienation, yet the Dominican Republic has higherpolarisation as a result o greater levels o identica-tion. Policy measures should, thereore, ocus moreon reducing identication than policies in Mexico.

These examples compare two countries, but acomparison can be made across two points in time.Trend analysis will reveal whether changes in polari-

sation are driven by alienation or identication.Gasparini et al. have also looked at the actors

that drive polarisation, in particular, how labourincomes aect income distribution. A strong rela-tionship between polarisation (as measured by theDER index), labour income, and household incomesuggests that individual skill levels hamper engage-ment in an economy, and thus contribute to incomepolarisation. They nd that many Latin Americanand Caribbean countries have experienced a polari-sation between those with the skills to take advan-tage o a new economic environment and those whohave struggled to adapt.

Growth, polarisation and development

How might polarisation have an impact on the linkbetween growth and human development? The roleo inequality in the link between growth and humandevelopment is well documented, but does not ully

explain the phenomenon o growth without develop-ment in many developing countries. The link betweengrowth, human development and polarisation is less

well known. It shows ve major areas o infuence.First, weak social cohesion. Polarisation refects

social cohesion better than inequality, as it capturesalienation between groups in society, and identi-cation within these groups. The little literature thatexists on the relationship between income polarisa-tion and confict shows that, as a society becomesmore polarised, confict is more likely (Horensteinand Olivieri, 2004). A study in Latin America andthe Caribbean (Gasparini et al. 2006) nds that thecombination o weaker institutions with higher polar-ization quickly translated into situations o socialtension and confict. This link is complex. It includes

both economic and non-economic variables. A soci-ety can be equal in economic terms, but have ten-sions across ethnic or religious lines. In contrast, asociety that is polarised by income can exhibit socialcohesion, during times o war or example. But ingeneral, weak social cohesion, social tensions andconfict undermine human development and slowdown economic growth.

Second, political and scal instability. In a cross-country study, Woo (2005) examines the role opolarisation on macro-economic stability, ndingthe greatest scal instability among countries withhighly polarised economic societies. Woo also

examines how polarisation between socio-eco-nomic groups impacts the evolution o scal insta-bilities. When politicians disagree on the composi-tion o government unds, Woo contends that eachhas an incentive to overexploit the common poolo resources, with a negative eect on the whole.This behaviour, he writes, is more likely to occurand be more severe in societies with higher degreeso polarisation. Further, he contends that politicalinstability associated with polarisation shortensthe time horizons o policy-makers, encouragingshort-sighted policies that lead to scal decits atthe expense o macroeconomic stability. This dis-

courages private investments by oreign and localentrepreneurs, penalising economic growth.

Third, bias in the distribution o social spending.Bandyopadhyay (2004) examines how income isdistributed and how this distribution changes acrossIndian States. He nds that unequal distribution insocial spending, refecting socio-ethnic and politi-cal dierences, contributes signicantly to a twinpeak phenomenon where the lower income statesconverge to orm one pole, while other wealthierstates orm another. So, when policy decisionsrefect polarisation, they can actually perpetuate it.This is particularly the case when policies lead to

poor investment in human development or somegroups.

Fourth, disappearing middle income groups andelite domination. The erosion o the middle classhas consequences or growth and human develop-ment, as this class orms the backbone o a healthy

8/8/2019 47 Growth Development Inequality ion Poverty

4/4

Overseas Development

Institute

111 Westminster BridgeRoad, London SE1 7JD

Tel: +44 (0)20 7922 0300

Fax: +44 (0)20 7922 0399

Email:

publications@odi.org.uk

Readers are encouraged

to quote or reproduce

material rom ODI Briefng

Papers or their own pub-

lications, but as copyright

holder, ODI requests dueacknowledgement and a

copy o the publication.

Overseas Development

Institute 2009

ISSN 0140-8682

Brieng Paper

Altman, D. (2003) Eciency And Equity (In the Same

Breath), New York Times, 20 April.

Bandyopadhyay, S. (2004) Twin Peaks: Distribution

Dynamics o Economic Growth across Indian States,

in A. Shorrocks and R. van der Hoeven (eds), Growth,

Inequality and Poverty: Prospects for Pro-Poor Growth.

Oxord: Oxord University Press.

Birdsall, N., Graham, C. & Pettinato, S. (2000) Stuck In The

Tunnel: Is Globalization Muddling The Middle Class?,

Working Paper 14. Center on Social and Economic Dynamics.Duclos, J-Y, Esteban, J. & Ray, D. (2005) Polarization:

Concepts, Measurement, Estimation in C. Barret (ed.)

The Social Economics of Poverty, New York: Routledge.

Easterly, W. (2001) The Middle Class Consensus and

Economic Development,Journal of Economic Growth

6(4): 317-35.

Gasparini, L. Horenstein, M. Molina, E. & Olivieri, S.

(2006) Economic Polarisation in Latin America and the

Caribbean: What do Household Surveys Tell Us? Working

Paper 38. CEDLAS.

Horenstein, M. & Olivieri, S. (2004) Income Polarisation

in Argentina: Pure Income Polarisation, Theory and

Applications, Economica L(1-2):39-66.

Keeer, P. & Knack, S. (2002) Polarization, Politics and

Property Rights: Links Between Inequality and Growth,

Public Choice 111(1-2): 127-54.Wolson, M. (1997) Divergent Inequalities: Theory and Empirical

Results,Review of Income and Wealth 43(4): 401-21.

Woo, J. (2005) Social polarization, scal instability and

growth, European Economic Review49: 1451-77.

economy. Without the middle class, growth cannotbe sustained. Birdsall et al. (2000) stress that ahealthy market economy requires the active involve-

ment o middle groups as stakeholders, entrepre-neurs, skilled workers, and consumers. Supportingthe emergence o the middle class requires invest-ment in human development. However, equitableinvestments are less likely when a society is polar-ised and an elite group dominates politics. Easterly(2001) nds that societies that are dominated byan elite tend to have lower levels o human capitalormation. He concludes that a middle class con-sensus (i.e. a high share o income or the middleclass and a high degree o ethnic homogeneity) isassociated with higher levels o human and inra-structure capital accumulation.

Fith, insecure property rights and legal rights.While the elite may well ensure their own propertyrights, they rarely extend these to the general public.Keeer and Knack (2002) contend that polarisationreduces the enorcement o property and contrac-tual rights, creating an environment o uncertaintythat penalises growth. Insecure property and con-tractual rights, they argue, aect growth directly,by infuencing the choice o production process andthe eciency with which production is carried out,and indirectly by reducing incentives to invest.

Insecure rights can deprive people o their justearnings and, in turn, lead to underinvestment in

human development. I opportunities or incomegeneration are insecure, incentives to invest inhuman capital ormation will be low and the risksassociated with social investments will keep humandevelopment low. It is vital, then, to address polari-sation when translating growth into development.

Policy and polarisation measuresPolicy recommendations based on typical inequalitymeasures imply, in general, a transer o resourcesrom the well o to the less well o, regardless owhere the individuals lie on the income distributioncurve. Transers may, inadvertently, aggravate polari-

sation. Policy responses and recommendations basedon polarisation measures tend to be more nuanced

because they consider disparities and clustering withinsociety and will dier according to whether polarisa-tion stems rom alienation or identication.

The key dierence between policies to reducealienation and those to reduce identication lies intheir aim. Policies to reduce alienation aim to bringgroup averages together, while those to addressidentication aim to ll gaps in an income distribu-tion by increasing the spread o the poles. Both mayentail cash transers, progressive tax schemes, socialprogrammes or subsidies, etc., but they are tailoredand targeted to shape the income distribution in dis-tinct ways. These policies should vary according tothe specic income groups targeted, as policies orwealthy groups dier rom those or poorer groups.

While a detailed discussion on the costs associ-

ated with the use o polarisation measures is beyondthe scope o this brieng, they may include the highadministrative costs o targeting or political tensions,when addressing underlying causes o polarisation.

Polarisation measures, like inequality measures,can be used in analysis to anticipate distributionalimpacts o macro-economic policies or economicshocks. They will, however, provide distinct insights,and can contribute to a better understanding o thesocial and political implications o external shocks.Political or social actors, o course, drive the trendstowards rising polarisation, so any reduction inpolarisation requires an understanding o the politi-

cal context and historical background.A polarisation approach oers distinct, yet

complementary, insights and, thereore, policyresponses. It also leads us to consider a wide arrayo instruments beyond simply spending on thepoor, including property rights, inclusion policiesand policies or the non-poor, and how these can becomplemented to acilitate the link between growthand human development. Many countries ace chal-lenges in orging this link as they try to translate eco-nomic growth into human development. Perhaps apolarisation approach can help them identiy whereinequities arise, and their nature.

Written by Milo Vandemoortele, ODI Research Ofcer(m.vandemoortele@odi.org.uk).

Photo credit: Johanna Vandemoortele

Reerences