Inequality: Measurement, Trends, Impacts and Policies

INEQUALITY: MEASUREMENT, TRENDS, IMPACTS AND POLICIES by Tony Addison UNU-WIDER, Helsinki Jukka Pi il€ a* Uni e si y o Tampe e and UNU-WIDER, Helsinki and Finn Ta p UNU-WIDER and Uni e si y o Copenhagen Many low- and middle-income coun ies a e achie ing good a es o economic g ow h, while high inequali y emains a p io i y conce n. Some coun ies meanwhile ha e low g ow h, high inequali y, and pe asi e po e y—o en linked o hei agili y. The e is now ac i e deba e on how coun ies should se hemsel es goals o achie ing bo h absolu e po e y educ ion and lowe inequali y. Bu policy ac ion needs o be be e se ed by analysis and da a. JEL Codes: I3, O1 Keywo ds: de eloping coun ies, inequali y, po e y, ulne abili y These conce ns mo i a ed he UNU-WIDER Con e ence on “Inequali y: Measu emen , ends, impac s, and policies”, held in Helsinki, on 5–6 Sep embe 2014. The con e ence ea u ed keyno e lec u es, in i ed sessions, con ibu ed pa allel sessions and pos e -session p esen a ions. 1 O e se en y pape s we e deli e ed, o which six now appea in his special issue o The Re iew o Income and Weal h. These pape s ha e been subjec o pee e iew, and he edi o o The Re iew o Income and Weal h, P o esso Conchi a DAmb osio, was esponsible o he o e all edi o ial p ocess. The e has been much deba e o e inequali y a he global le el. Acco dingly, his special issue begins wi h he pape by Nino-Za azua, Roope and Ta p, en i- led “Global Inequali y: Rela i ely Lowe , Absolu ely Highe ”. I se s he s age by p o iding upda ed and e ined es ima es o inequali y in he wo ld. The au ho s use da a om he UNU-WIDER Wo ld Income Inequali y Da a Se (WIID) No e: UNU-WIDER acknowledges he inancial con ibu ions o i s wo k p og amme, du ing he pe iod in which he con e ence was held, o he go e nmen s o Denma k, Finland, Sweden and he Uni ed Kingdom. *Co espondence o: P o esso Jukka Pi il€ a, School o Managemen , Uni e si y o Tampe e, 33014 Tampe e, Finland (jukka@wide .unu.edu). 1 A selec ion o pape s, as well as ideos o he sessions a he UNU-WIDER con e ence on inequali y a e a ailable a : h ps://www.wide .unu.edu/e en /inequali y-measu emen - ends-impac s- and-policies. V C2017 UNU-WIDER This is an open access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion IGO License h ps://c ea i ecommons.o g/ licenses/by/3.0/igo/legalcode which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided ha he o iginal wo k is p ope ly ci ed. In any ep oduc ion o his a icle he e should no be any sugges ion ha UNU o he a icle endo se any speci ic o ganiza ion o p oduc s. The use o he UNU logo is no pe mi ed. This no ice should be p ese ed along wi h he a icle's URL. 603 Re iew o Income and Weal h Se ies 63, Numbe 4, Decembe 2017 DOI: 10.1111/ oiw.12352 bs_bs_banne which is a eliable sou ce o coun y-le el income inequali y indices (Jenkins, 2015). The au ho s con ibu e o he li e a u e in h ee main ways. Fi s , hey p o- ide up- o-da e es ima es co e ing he pe iod om 1975 un il 2010, whe eas mos ea lie es ima es use da a ha ends a ound he mid-1990s. Second, hey p o ide indings o bo h adi ional,  ela i e inequali y, measu es—acco ding o which inequali y is una ec ed by changes whe e he incomes o all households inc ease by he same pe cen —and absolu e measu es (which accoun o absolu e income changes) as well as cen is ones, which combine he elemen s o ela i e and abso- lu e measu es. Finally, hey conduc coun e ac ual expe imen s o examine whe he inc easing absolu e income dispa i ies could ha e been a oided i all coun ies we e cha ac e ized by he No dic egions low le els o inequali y. Nino-Za azua e al. ind ha income inequali y in he wo ld has dec eased s eadily o e he pas h ee decades when inequali y is measu ed using con en- ional ela i e measu es. A decomposi ion analysis shows ha he educ ion is due o a lowe ing o be ween-coun y inequali y, whe eas wi hin-coun y inequal- i y shows conside able he e ogenei y. Howe e , absolu e inequali y and e en hei cen is measu e, he K scha index, ha e isen. This la e inding mo i a ed he coun e ac ual analysis whe e coun ies we e assumed o ha e No dic le els o inequali y in 2010. E en in his exe cise, absolu e inequali y indices, measu ed ei he by he s anda d de ia ion o he absolu e Gini, depic an inc ease in global inequali y om 1975 o 2010, whe eas ela i e inequali y declines signi ican ly. The indings a e pa icula ly use ul as hey s a kly illus a e how di e en iews on measu ing inequali y can lead o comple ely opposi e conclusions ega ding inequali y de elopmen s. Pe haps some o he ensions be ween he iews o non- p o essionals and expe s on wo ld inequali y can be explained by hese di e en- ces in app oaches. The second a icle in he collec ion, by Da ies, Llube as and Sho ocks i led “Es ima ing he le el and dis ibu ion o weal h, 2000–2014”, also p o ides global inequali y es ima es, now o weal h. Es ima ing he Gini o weal h is a complex exe cise, in ol ing h ee main s eps. In he i s , he au ho s need o es ima e he mean le el o weal h in each coun y. I household balance-shee da a is no a ail- able, he au ho s use eg ession models o impu e he mean le el o weal h o he coun ies wi h missing di ec in o ma ion on weal h holdings. The second s ep is o es ima e he weal h dis ibu ion o indi iduals wi hin a coun y. Fo some coun ies, di ec dis ibu ional in o ma ion exis s, while o o he s, weal h dis i- bu ion is impu ed based on income inequali y, again using measu es om he WIID. As expec ed, o coun ies whe e da a on bo h income and weal h dis ibu- ion exis , weal h is mo e une enly dis ibu ed. The hi d s ep is o co ec o he unde es ima ion o he weal h holdings o he e y ich ha ypically occu s in su ey-based weal h analysis; his is done by combining in o ma ion om o he sou ces, mos no ably om Fo bes Magazine. The o al o wo ld weal h by household s ood a 250 illion USD in 2014 (measu ed using ma ke exchange a es). In he beginning o he analysis pe iod, he co esponding numbe was abo e 100 illion. So he inc ease has been qui e d ama ic. The sha e o global weal h o people esiding in Eu ope o No h Ame ica is app oxima ely 70 pe cen . The op g oup o weal h holde s (ha ing weal h abo e 1 million USD) comp ises jus 0.7 pe cen o he wo lds adul Re iew o Income and Weal h, Se ies 63, Numbe 4, Decembe 2017 V C2017 UNU-WIDER 604 popula ion whe eas i s household weal h amoun s o 45 pe cen o he global o al. The Gini o he global weal h dis ibu ion is 92.2 pe cen , way abo e he Gini es ima ed o wo ld income inequali y. The au ho s close wi h a s imula ing discussion on whe he global weal h inequali y will ise in he u u e. They con- clude ha while wi hin-coun y weal h inequali y may ise, as p edic ed by Pike y, be ween-coun y weal h inequali y can also decline, i poo e coun ies con inue o ca ch up wi h he cu en ly high-income coun ies; his implies ha o al wo ld weal h inequali y o ecas s emain ambiguous. The emaining pape s in his issue ocus on ad ances in he measu emen o inequali y and applica ions using da a om indi idual coun ies. The hi d a icle, “C oss-Sec ional e sus Panel Income App oaches: Analyzing Income Dis ibu- ion Changes o he Case o Mexico,” by Du al-He nandez, Fields and Jakub- son con as s wo di e en ways o examining changes in inequali y. In he i s , in he mos commonly used app oach, epea ed c oss sec ions a e used o examine inequali y de elopmen s. In such an app oach, he cu en ich a e no he same pe sons as he ich in p e ious da a poin s. The e o e, he app oach can be cha - ac e ized as anonymous. Howe e , one can also use longi udinal, non-anonymous da a, and o en hese da a a e used in a e y di e en way: he e, analys s ha e examined whe he hose who we e ini ially poo ha e expe ienced la ge - han- a e age income gains; i.e. whe he he e has been con e gence in income le els. The au ho s i s show, concep ually, ha i is qui e likely o obse e a si ua- ion whe e con e gence akes place, and ne e heless income inequali y, when measu ed using he anonymous app oach, inc eases. They hen use Mexican labo - o ce su ey da a and apply he wo di e en app oaches. The esul s e eal ha i espec i e o whe he con en ionally measu ed ea nings dispa i y ises o no , ea nings ha e always, on a e age, exhibi ed con e gence. The eason is ha ew o he ini ially poo ha e expe ienced la ge income gains, which a ec s he eg ession pa ame e measu ing con e gence, whe eas mos ha e aced e y small changes in hei incomes. Thei a icle he e o e makes i clea ha i ma e s how inequali y is measu ed, a heme h oughou his special issue. Use o panel da a con inues wi h he ou h a icle “Wel a e Dynamics Measu emen : Two De ini ions o a Vulne abili y Line and Thei Empi ical Applica ion,” by Dang and Lanjouw. Analys s ha e inc easingly become in e es ed in he speci ic shape o he income dis ibu ion, on op o summa y measu es o inequali y (such as he Gini). The mass o people cu en ly jus abo e he po e y line, who a e also ulne able o imminen ly alling in o po e y, is o en o special in e es : a he same ime when absolu e po e y has allen in he wo ld, he e is a h ea ha he numbe o ulne able people has inc eased. Dang and Lanjouw o e wo al e na i e ways o de ining a “ ulne abili y line.” In he i s , hey iden i y indi iduals ha a e no ulne able and hen de ine he ulne - abili y line as a lowe bound o income o his g oup. This gi es ise o hei “Insecu i y Index.” In he second app oach hey conside he popula ion ha is no poo bu who ha e a heigh ened possibili y o expe ience po e y in he u u e. The ulne abili y line is hen se o he uppe bound o he income le el o his g oup. The au ho s illus a e hese app oaches using ue panel da a om Vie nam and he USA, and syn he ic panel da a om India. The esul s sugges ha he Re iew o Income and Weal h, Se ies 63, Numbe 4, Decembe 2017 V C2017 UNU-WIDER 605 educ ions in po e y in Vie nam and India ha e been ma ched wi h an almos simila inc ease in he middle class—people abo e he ulne abili y line—so he sha e o he ulne able pa o he popula ion has emained almos cons an a 35–50 pe cen o he o al numbe o people. I is in e es ing ha in he same pe iod (2004–2008/9), he end in he USA has been ma kedly di e en : an inc ease in he numbe o he poo (measu ed using a much highe po e y line in absolu e e ms), oge he wi h a educ ion in he size o he middle class and an inc ease in he ex en o he ulne able g oup. The e also seems o be less income mobili y in he USA han in India o Vie nam. The middle class is he ocus in he nex a icle, “When he Cen e Canno Hold: Pa e ns o Pola iza ion in Nige ia,” by Clemen i, Dabalen, Molini and Sche ino. The pape add esses a key ea u e o income de elopmen s in many coun ies, he isk o pola iza ion. This happens i he sha e o people ha ing income le els close o he median income is educed and a he same ime he sha e o households a he bo om and he op o income le els inc eases. The phenomenon can ha e undesi able consequences, since he middle class has o en been a sou ce o s abili y o economic and poli ical de elopmen s whe eas he ex emes o he income dis ibu ion can become aliena ed om each o he . Resea che s ha e de eloped measu es o pola iza ion, and hese a e used by he au ho s o his s udy as well. In addi ion, hey demons a e how non-pa ame ic desc ip i e analysis is help ul in con eying in o ma ion abou he de elopmen s o he whole income dis ibu ion. The au ho s o his s udy cons uc wo compa able income dis ibu ion su - eys o Nige ia, o 2003 and 2013. The esul s e eal a sha p inc ease in pola - iza ion in he coun y and some in e es ing egional dispa i ies: he Sou he n pa o he coun y has con ibu ed o pola iza ion in he uppe ail o he dis ibu ion whe eas he No h has expe ienced pola iza ion in he lowe pa o income dis i- bu ion. Clea ly, o unde s anding he pic u e o po e y and inequali y in A ica, Nige ia is a key coun y, and he analysis he au ho s p o ide makes a good case o eplica ion analysis elsewhe e in A ica and o he de eloping coun ies. The inal pape in his issue “Measu ing Inequali y by Asse Indices: A Gen- e al App oach wi h Applica ion o Sou h A ica,” by Wi enbe g and Leibb and , ocuses on he p ope ies o an o en-used measu e o inequali y when di ec income o expendi u e da a a e no a ailable, he so-called asse index. Such an index is de i ed om he owne ship o ce ain asse s (such as a ca , a mobile phone o a TV se ) o gene a e a ough measu e o ma e ial well-being. Some imes hese da a a e mo e egula ly a ailable, om e.g. Demog aphic and Heal h Su eys, han om ully- ledged income o expendi u e su eys. While ea lie s udies ha e assessed he p edic i e p ope ies o asse indices o “ ue” income inequali y, he ocus in he p esen pape is di e en ; i is on he in e nal consis ency o asse indices. The au ho s poin ou ha he ca ego ical na u e o asse owne ship da a (cap u ing whe he he households own a ce ain asse o no ) implies ha he measu e is also ul ima ely o disc e e na u e. Pe haps mo e impo an ly, some imes some asse s ge nega i e weigh s in an o e all asse index, composed o example by he p incipal componen analysis. A p ac ical example is owne - ship o ca le, which ends o be mo e impo an o poo e households han iche Re iew o Income and Weal h, Se ies 63, Numbe 4, Decembe 2017 V C2017 UNU-WIDER 606 ones. Howe e , we know om ex e nal da a ha ca le owne ship is an impo an means o s o ing weal h among u al households in de eloping coun ies, and he e o e ha ing mo e ca le canno eally be a nega i e phenomenon. Wi enbe g and Leibb and also calcula e asse inequali y indices o pos -Apa heid Sou h A ica and highligh a sizable educ ion in asse inequali y. While his may sugges ha con en ional measu es o inequali y (which ha e a ising end o e he same pe iod) need o be complemen ed by addi ional measu es, i may also sugges ha asse inequali y analysis has d awbacks when compa ed wi h, o he , pe haps b oade measu es. The au ho s conclude ha while asse indices emain use ul, hei use s should ensu e ha hey unde s and well how such indices a e compiled. W i ing in he la e 1970s Hen y Aa on amously no ed ha s udying inequal- i y was like “wa ching he g ass g ow,” o he e had been li le change in he le el o US inequali y since he immedia e yea s a e Wo ld Wa Two (Aa on, 1978, p. 17). This was jus be o e inequali y s a ed o change apidly, bo h be ween and wi hin coun ies, om he 1980s onwa ds. The d i e s o hese changes a e now he subjec o in ense deba e, no leas he ela i e impac s o globaliza ion and echnological change. Reade s will no doub ag ee ha wi h new da a, new echni- ques and new policy issues, he e has ha dly been a mo e exci ing ime o s udy inequali y in all i s many dimensions. We hope ha his Special Issue will s imu- la e u he ad ances in he s udy o inequali y, a opic which is now o he highes impo an o de elopmen policy and social policy mo e b oadly. Re e ences Aa on, H. J., Poli ics and he P o esso s: he G ea Socie y in Pe spec i e. Washing on, D.C. The B ookings Ins i u ion, 1978. Jenkins, S., “Wo ld Income Inequali y Da abases: an assessmen o WIID and SWIID,” Jou nal o Economic Inequali y, 13, S629–671, 2015. Re iew o Income and Weal h, Se ies 63, Numbe 4, Decembe 2017 V C2017 UNU-WIDER 607