The Distribution of Income in Ireland (2000) - Combat Poverty Agency

Combat Poverty Agencyworking for the preventionand elimination of povertyOf l«COtfSBrian NolanBertrand MaitreDonal O'Neill &Olive Sweetman

About the AuthorsBrian Nolan is a Research Professor and Bertrand Maitre is aStatistical Analyst at The Economic and Social Research Institute.Donal O'Neill is a senior lecturer and Olive Sweetman isa lecturer in the Department of Economics, NUI Maynooth.

THE DISTRIBUTION OFINCOME IN IRELANDBrian Nolan, Bertrand Maitre,Donal O'Neill and Olive SweetmanOak Tree PressDublinin association withCombat Poverty Agency

Oak Tree PressMerrion BuildingLower Merrion StreetDublin 2, Irelandwww.oaktreepress.comContents© 2000 Combat Poverty AgencyA catalogue record of this book isavailable from the British Library.List of Tables and FiguresAcknowledgementsForewordExecutive SummaryviixixiiixxxiiiISBN 1 86076 208-5All rights reserved. No part of this publication may be reproduced ortransmitted in any form or by any means, including photocopying andrecording, without written permission of the publisher. Such writtenpermission must also be obtained before any part of this publication isstored in a retrieval system of any nature. Requests for permissionshould be directed to Oak Tree Press, Merrion Building, Lower MerrionStreet, Dublin 2, Ireland and the Combat Poverty Agency, BridgewaterCentre, Conyngham Road, Islandbridge, Dublin 8, Irelandu^Mchft f °r ST- ** COmbat P ° Verty ^ c f sResea «* Series,m which it is No. 32. The views expressed in this report are the authors 1own and not necessarily those of the Combat Poverty Agency.Printed in the Republic of Irelandby Colour Books Ltd.Chapter 1: Introduction 1Brian NolanChapter 2: Measuring the Distribution of Incomein Ireland 3Brian NolanChapter 3: The Distribution of Income in Ireland inthe 1990s 13Brian Nolan and Bertrand MaitreChapter 4: Trends in the Irish Income DistributionSince 1973 35Brian NolanChapter 5: Ireland's Income Distribution inInternational Perspective 51Brian Nolan and Bertrand MaitreChapter 6: The Distribution of Earnings 67Brian Nolan

viThe Distribution of Income in IrelandChapter 7: Female Labour Force Participation andHousehold Income Inequality in Ireland. 79Donal O'Neill and Olive SweetmanChapter 8: Conclusions 89Brian NolanAppendix 1 93Appendix 2 97References ..101List of Tables and FiguresTable 3.1: Decile Shares and Summary InequalityMeasures, Disposable Income among Irish Households,1994 and 1997 Living in Ireland Surveys 15Table 3.2: Decile Shares for Direct and Gross Incomeamong Households, 1994 and 1997 LII Surveys 17Table 3.3: Decile Shares in Equivalised Disposable Incomefor Irish Households in the 1994 and 1997 LII Surveys,Alternative Equivalence Scales 20Table 3.4: Decile Shares in Equivalised Disposable IncomeAmong Households and Among Persons, 1994 and 1997 LIISurveys (1/0.66/0.33 Scale) 21Table 3.5: Decomposition of Inequality in Disposable EquivalisedIncome by Age of Head, 1994 and 1997 LII Survey 22Table 3.6: Decomposition of Inequality in DisposableEquivalised Income by Sex of Head, 1994 and 1997 LII Survey. 23Table 3.7: Decomposition of Inequality in DisposableEquivalised Income by Labour Force Status of Head, 1994and 1997 LII Survey 24Table 3.8: Decomposition of Inequality in DisposableEquivalised Income by Social Welfare Dependency, 1994and 1997 LII Surveys 25Table 3.9: Position in the Income Distribution of PersonsCategorised by Sex of Household Head, 1994 and 1997 LIISurveys 28

VIII The Distribution of Income in IrelandTable 3.10: Position in the Income Distribution of PersonsCategorised by Household Composition Type, 1994 LIISurvey 30Table 3.11: Decile Shares and Summary InequalityMeasures, Disposable Income among Irish Households,1994,1997 and 1998 Living in Ireland Surveys 31Table 3.12: Decile Shares for Direct and Gross Incomeamong Households, 1994,1997 and 1998 LII Survey 32Table 3.13: Decile Shares in Equivalised DisposableIncome for Irish Households, 1994,1997 and 1998 LIISurveys (1/0.66/0.33 Scale) 33Table 4.1: Decile Shares in Disposable Income for IrishHouseholds in 1987 and 1994, ESRI and HBS Surveys 36Table 4.2: Decile Shares in Gross Income for IrishHouseholds in 1987 and 1994, ESRI and HBS Surveys 37Table 4.3: Decile Shares in Equivalised Gross andDisposable Income for Irish Households, 1987 and 1994ESRI Surveys (1/0.66/0.33 Scale) 38Table 4.4: Decile Shares in Equivalised Disposable Incomefor Irish Households, 1987 and 1994 ESRI Surveys,Alternative Equivalence Scales 39Table 4.5: Decile Shares in Disposable Income among IrishHouseholds, 1973, 1980, 1987 and 1994/95 HBS 40Table 4.6: Suits Progressivity Index and Average Tax Ratefor Income Tax and PRSI Contributions, 1973, 1980, 1987and 1994/95 46Table 4.7: Decomposition of Inequality in DisposableEquivalised Income by Labour Force Status of Head,1987 ESRI Survey 48Table 5.1: Decile Shares in Disposable Income among IrishHouseholds, Current versus Annual Income, 1994 Living inIreland Survey/ECHPList of Tables and Figures IXTable 5.2 : Decile Shares in Disposable Income amongHouseholds, Wavel ECHP 54Table 5.3: Summary Inequality Measures for Distribution ofDisposable Income among Households, Wave 1 ECHP,1993 55Table 5.4 : Decile Shares in Equivalised Disposable Incomeamong Persons Wave 1 ECHP, 1993 57Table 5.5: Summary Inequality Measures for Distribution ofEquivalised (Modified OECD Scale) Disposable Incomeamong Persons, Wave 1 ECHP, 1993 58Table 5.6: Gini Coefficient for Equivalised DisposableIncome Among Persons in ECHP Wave 1, 1993, AlternativeEquivalence Scales 59Table 5.7: Gini Coefficient for Equivalised DisposableIncome among Persons in ECHP Wave 1, 1993, and Wave2, 1994, Modified OECD Scale 60Table 5.8: Gini Coefficient, Equivalised Income amongPersons, Square Root Scale, Wave 1 ECHP and Atkinson,Rainwater and Smeeding (ARS) Study (Equivalence scalesquare root of household size) 62 #Table 6.1: Distribution of Earnings, Ireland 1987, 1994 and1997 70Table 6.2: Summary Measures of Earnings Dispersion,Ireland and Other OECD countries, 1994 72Table 6.3: Trends in Earnings Dispersion, Ireland andOther OECD Countries, 1987-1994 73Table 6.4: Education Attainment Level, Ireland 1987 and1994 74Table 6.5: Median Hourly Earnings and Ratio of Top toBottom Decile by Education Category, All Employees, 1987and 1994 (IR£ per hour) 75Table 7.1: Female Participation Rates in the OECD, 1984and 1994 81

XThe Distribution of Income in IrelandTable 7.2: Components of Gross Household Income(households with couples aged 24-55) 1987-1994 82Table 7.3: Correlation between Earnings of MarriedCouples, 1987 and 1994 83Table 7.4: Female Employment Rate by Male Earnings,1987-1994 84Table Al. 1: Decomposition of Changes in the Coefficient ofVariation for Gross Household Income, 1987-1994 94Table A2.1: Decile Shares in Equivalised DisposableIncome for Irish Households, 1994 and 1997 LII Surveys(1/0.66/0.33 Scale) 95Table A2.2: Position in the Income Distribution of PersonsCategorised by Age of Household Head, 1994 and 1997 LIISurveys 96Table A2.3: Position in the Income Distribution of PersonsCategorised by Labour Force Status of Household Head,1994 LII Survey 97Figure 2.1: Lorenz Curve for Income 11Figure 3.1: Decile Shares in Equivalised DisposableIncome for Irish Households, 1994 and 1997 LII Surveys 18Figure 3.2: Position in the Income Distribution of PersonsCategorised by Age of Hcjusehold Head, 1994 and 1997 LIISurveys 27AcknowledgementsThis study draws extensively on the 1994, 1997 and 1998 Wavesof the Living in Ireland Survey, the Irish element of the EuropeanCommunity Household Panel. James Williams and DorothyWatson were responsible for the survey design, data collectionand database creation for 1997 and 1998, while the 1994 design,data collection and database creation were the responsibilityof Brendan Whelan and James Williams. The authors aregreatly in their debt. We are also grateful to colleagues in theESRI and NUI Maynooth for valuable discussions on the issuesand findings in a variety of forums, and to participants at theIrish Economic Association Annual conferences in 1998 and2000 and the International Association for the Review of Incomeand Wealth, 25th Anniversary Conference, Cambridge, UK,1998. Very helpful comments on earlier drafts were also receivedfrom Helen Johnston of the Combat Poverty Agency.Figure 3.3: Position in the Income Distribution of PersonsCategorised by Labour Force Status of Household Head1994 T.II c

ForewordINTRODUCTIONThe Combat Poverty Agency is a statutory body working towardthe elimination and prevention of poverty in Ireland. Oneof the Agency's key functions is to undertake research into thecauses and consequences of poverty with a view to understandinghow poverty can be overcome. The Agency also has astatutory role to provide policy advice to government and inthis regard research findings are extremely important in helpingus identify policy options and directions.This research report examines the very important issue ofincome distribution. It examines how income is spread acrosshouseholds, looking at the gap between rich and poor and thefactors which can operate either to narrow or widen that gap,and it identifies how Ireland fares on these issues relative toother EU and developed OECD countries.The findings show that while Ireland has become increasinglywealthy in recent years, it still has one of the highest levelsof income inequality in the EU. It is clear that our unprecedentedeconomic growth is not lifting all boats in an equitablemanner, and that better-off households are gaining from theboom to a greater extent that those who are less well off. Thesefindings are of great concern to the Agency.The growing gap between rich and poor during our economicturnaround is damaging to society in a number of ways.Firstly, given the broad acceptance that poverty in developedsocieties is a relative concept, it is still likely that there will be alink between the scale of income inequality and the levels of

XIV The Distribution of Income in IrelandForewordxvrelative income poverty, i.e. greater inequality will result ingreater poverty. Secondly, from a social justice or moral pointof view, it is unfair or simply wrong that the benefits of economicgrowth are not shared more equitably, and that thosewho have most, benefit to a greater extent. Thirdly, incomeinequality is bad for social cohesion and inclusion, leading toalienation of marginalised groups, and the high social costs thatthis alienation or marginalisation can impose. Fourthly, severeincome inequality limits choice and diversity, and hinders theability of those on low incomes to participate fully in the political,social, economic and cultural life of society. This is a curtailmentof basic rights and is contrary to the principles andobjectives of the Government's National Anti-Poverty Strategy(NAPS). Fifthly, it also appears from research in epidemiologythat income inequality is bad for a nation's health and that thehealthiest nations are not necessarily the richest ones, but theones where there is the smallest gap between rich and poor. 1Finally, it is now being argued by economists that income inequalityitself can be bad for economic growth. 2Severe income inequality implies a poor redistribution ofresources and opportunities throughout society. If public policydoes not intervene sufficiently to redistribute resources generatedby the market, this in turn this implies a poor rate of publicinvestment in social and human capital, and in public serviceand infrastructural development, all of which are essential tothe long-term viability and sustainability of current economicgood fortune.Whether the motivation is the reduction of poverty, the pursuitof social justice, a concern for greater social inclusion or acommitment to economic growth, reducing income disparitiesis a necessary public policy objective.The Combat Poverty Agency believes that a radical incomeredistribution in favour of those on low incomes is central to reducingpoverty and creating an inclusive society - and later inRo W uu^r R (1 " 6) **"****** ^Miction of meguality. London:this foreword particular policy proposals are identified drawingon the lessons from the report's findings. The Agency's currentstrategic plan (1999-2001) identifies narrowing the gap betweenrich and poor as one of its four key strategic objectives, and suggestsin a series of strategic goals, how this may be achievedthrough a fairer distribution of resources, services and employmentopportunities in favour of people living in poverty. 3POLICY CONTEXTThis interest in a fairer distribution of income is not unique tothe Agency, but is a view reflected throughout the current publicpolicy arena, both at national and international levels.The National Anti-Poverty Strategy (NAPS), the official governmentpolicy on tackling poverty, commits the Government to. . . ensuring that the impact of very rapid economic, socialand demographic change reduces social inequalities andsocial polarisation . . . (and) that the benefits of sound economicmanagement and growth are distributed fairly and inparticular are used to tackle the underlying causes of povertyand social exclusion (Government of Ireland, 1997: 2). 4The Action Programme for the Millennium, the current governmentaction programme, aspires to everyone sharing the benefitsof economic growth. 5 The current national agreement, theProgramme for Prosperity and Fairness (PPF), aims to substantiallyincrease the resources allocated to social exclusion. 6 Therecent UN World Summit on Social Development in Genevaidentifies the need to "encourage Governments to re-evaluate,as appropriate, their national fiscal policies including progressivetax mechanisms, with the aim of reducing income inequali-3 Combat Poverty Agency (1999) Strategic Plan 1999-20001. Dublin: CombatPoverty Agency.4 Government of Ireland (1997) Sharing in Progress: National Anti-PovertyStrategy. Dublin: Stationery Office.Fianna Fail and Progressive Democrats (1997) Action Programme for the NewMillennium. Dublin: Stationery Office.Government of Ireland (2000) Programme for Prosperity and Fairness. Dublin:Stationery Office.

XVIThe Distribution of Income in IrelandForewordxvuties and promoting social equity" as part of the political declarationon proposals for further initiatives on social develop-ment 7A TIMELY REPORTEvidence from the Living in Ireland Survey 8 shows that whileconsistent or severe poverty decreased between 1994 and1997, relative income poverty increased. In other words morehouseholds had less that 50 per cent of average income in 1997than in 1994 and they also fell further below the relative incomepoverty lines than before. This contributed to an increased gapbetween rich and poor.This report is therefore very timely as it examines thisheadline finding on income poverty and tries to establish how agrowing economy could contribute to greater income inequality.It was also important to look more closely at income distributionas Ireland appeared to have a high level of income inequalityrelative to other EU countries. The small cluster of EUcountries with high levels of income inequality also demonstratedhigh levels of child poverty, another matter of seriousconcern to the Agency. It is striking that, whereas in some othercountries rapid economic growth has resulted in a decline inincome inequality, Ireland's performance in this regard hasbeen disappointing.The information and analysis arising from the study are crucialto the Combat Poverty Agency's ultimate objective of devumg^^ocmwMtottoB,for public policy on interventions togenerate and support greater equality and a fairer redistributionof resources in our society. This will require a stronger re-"World * 2 £ S J D S S SeSS1 ° n ° f * e General *—«** «*tod= -opmen, E al ^SSSSST^JSST ******DeVe1 'Survey. Dublin: 5J JJ g j * ^ *f» g »» l*ng in Irelandand the Combat Poverty & ! » ? , T***' Conu ty and Family Affairsfor 1998, which i n S S S S l T 7 ^ ^ " n ° W *• •*"•trend. ^ * ^ ^ "»*»a!ian in the relative income povertydistribution strategy, alongside investment strategies and taxand welfare policies that complement other positive initiativesto prevent poverty. In short, addressing income inequalitiesneeds to be a key public policy priority for the future. 9This report, by a research team in the Economic and SocialResearch Institute and NUI Maynooth who have extensive experiencein the field of poverty and inequality research, is of thehighest quality. It draws on data from the Living in Ireland Surveyundertaken by the Institute for Eurostat, supported by theCombat Poverty Agency and the Department of Social, Communityand Family Affairs, and on Household Budget Surveydata from the Central Statistics Office.The immediate policy context which the recommendationsfrom this report will inform include the Government budgetaryprocess, the review of the National Anti-Poverty Strategy (particularlythe work of the Indexation of Welfare and IncomeAdequacy Working Group), commitments to social inclusionand social investment within both the Programme for Prosperityand Fairness and the National Development Plan and the strategicoverview of tax and welfare policies for the next ten yearscurrently being embarked on by the National Economic andSocial Council.Finally, this report will be complemented by a report on theImpact of Social Spending on Inequality and Poverty, which willbe published by the Agency early in 2001. It will focus on theperiod covered by the last three national agreements, with aview to stimulating further public and political debate on howcurrent fiscal and welfare policies and social spending createor reinforce inequality and poverty.AIMS OF THE STUDYThe study looks primarily at changes in the distribution of incomein Irish society between 1994 and 1997 based on datafrom the Living in Ireland Survey for both of those years and onthe 1994/5 Household Budget Survey. In doing so it specifically:9 Combat Poverty Agency (2000) Annual Report, 1999. Dublin: Combat PovertyAgency.

XVI11The Distribution of Income in Ireland• Places these changes in the longer term context of what hashappened since 1973 through 1987 up to 1998• Assesses how Ireland fared on the issue of income distributionrelative to other EU and industrialised OECD countries• Examines more closely the nature of those changes and thefactors influencing emerging trends, e.g. inequality betweenparticular types of households, the position of thosehouseholds on the income ladder and how earnings, socialwelfare payments, taxation, and women's participation inthe workforce impacts on these changes.How INCOME IS DEFINEDThis study examines household survey income data rather thantax or administrative data. These surveys collect information byway of questionnaire, on income from the following sources:employee earnings, self-employment, farming, secondary jobs,casual employment, state training or work experience schemes,social welfare transfers, child benefit, the renting of land orproperty, mterest or dividends, retirement pensions, pensionsfrom abroad, annuities, covenants or trusts, sick pay, strike pay,maintenance from outside the household and educational grants.The distribution of income is described by way of incomepercentiles, specifically deciles, i.e. the share of total incomegoing to the bottom 10 per cent of households, the next 10 percent and so on up to the top 10 per cent of households; andShaieSPO next 20 S " per le " cent and so ** on.g ° eS t0 *• bottom 2 0 Per cent, thealso^L* 8 l0 ° kln9 St VeraU inCOme° ^tribution,^ " 1 ' , ^contributions.the studymC ° me WhiCh mcludes b °* rnarket *^nunustax andinsuranceof E^ST? T t ** hOUSehold level - •* -uestudy, while ^ZT^ltTT t0 ^ ^ ^g us with a comprehensive analysis of in-Fore wordcome distribution, is not a study of wealth. It does not examinethe holding or accumulation of wealth in the form of profits,savings and other assets, which may be becoming increasinglysignificant and unequal at a time of rapid economic growth.SUMMARY OF KEY FINDINGS• Overall income inequality increased between 1994 and1998. In the mid 1990s the bottom 10 per cent of householdshad about 2 per cent of total income whereas the top 10 percent had about 27 per cent. However, between 1994 and1998 there was a redistribution of over 1 per cent of total incomeaway from the bottom 30 per cent of the income distribution- representing a substantial shift in a short period.The increasing inequality reflects a shift from the bottomhalf of the distribution to the top half, rather than to thoseright at the top.• The growth in earnings inequality continued to increase.Dispersion in the Irish earnings distribution was relativelyhigh by international standards in 1994 and it increasedbetween 1987 and 1994 by more than in almost any otherOECD country for which data are available. The scale of thisincrease in Ireland over the period showed that a rapid increasein the supply of highly educated labour and centralisedwage setting were not enough to limit the growth inearnings inequality. Between 1994 and 1997, when economicgrowth accelerated rapidly, the increase in earningsinequality slowed although the top of the distribution didcontinue to move away from the middle.• Ireland is one of the most unequal countries in the EU.Overall income inequality and inequality in the distributionof earnings have risen sharply during the 1980s and 1990s ina number of industrialised countries. In the mid 1990s, however,Ireland had one of the highest levels of income inequalityin Europe, with the exception of Portugal and alongwith Spain, Portugal and the UK, was one of a group of EUcountries with relatively high levels of income inequality.Based on data from the mid-1980s Ireland also rankedxix

XXThe Distribution of Income in IrelandForewordxxiamong the most unequal in the OECD countries. As notedearlier, this cluster of countries with high levels of incomeinequality is the same cluster that shows relatively high levelsof child poverty.Older households moved down while younger onesmoved up. The period 1994-1997 saw younger householdsmoving up the income distribution towards the top whileolder ones moved down. Households headed by someoneaged between 35 and 65 were fairly evenly spread throughoutthe income ladder. However those headed by someoneunder 35 were more concentrated in the top quintile by1997 while for older households the proportion at the topdeclined and the numbers at the bottom rose markedly.Larger families concentrated at the bottom of the incomeladder. Households with one adult with children werevery heavily concentrated right at the bottom of the distribution,though less so in 1997 than in 1994. While coupleswith one or two children were towards the top of the distributionin both years, it is striking that those with four ormore children on the other hand were very heavily concentratedat the bottom.The equalising effects of social welfare transfers hasdiminished. Social welfare transfers, though having anequalising effect in both 1994 and 1997 had less impact inthe latter year - mainly because the numbers of householdsdependent on such payments declined.^creasing numbers of women at work did not significantlyaffect income inequality. Rapid increases in femalelabour force participation in the 1987-94 period have notbecausf^^ mPaCX ° n inC ° me ^q^ity- This is largelytaPid lnCreaSes inSn wo m ^ 2wome *'* Participaof^otlhoT 'nT" 9 " Stm 0nly accounted fo * 15 Per centde S of Z!, mC ° me * 1994 - e s ? *"* ^ thebecl.e a ^TT " ^ ^ "* mdicates ** they mayZ*Z: g lT * "*"*"faCt ° r * -^standing Z• Overall income inequality has fallen since 1973. Lookingback over a longer period, from 1973-1987 inequality in thedistribution of household income fell markedly, with theshare of the top decile down by 1.4 per cent of total incomeand that of the bottom quintile up by 0.9 per cent. This reflectsthe increasingly redistributive impact of income taxand a substantial increase in the average tax rate. From 1987to 1994 this continued at a much slower rate. However withinthis overall statistic, between 1987 and 1994 wage inequalityin Ireland increased substantially, even though centralisedwage agreements were reintroduced in 1987 and have hada significant impact on improving the economic landscape.UNDERSTANDING THE FINDINGSThe wider economic context for this report is one in which ourwealth is clearly rising, with more people at work, an increasein the real incomes of the poor and a drop in long-term or consistentpoverty and child poverty. However, relative incomepoverty has increased, the depth of poverty is greater, the riskof poverty for some groups is increasing and a quarter of childrenlive in poor households.The key lesson from this research report is that in a fastgrowing economy, a rising tide does not lift all boats equally.While more people are employed now than before, those at thetop of the income ladder have benefited at a faster pace thanthose at the bottom. It is also clear that internationally Irelandfares badly on the income inequality front, and that despite undoubtedadvances in our general prosperity and availability ofemployment, we are still lagging well behind the levels of incomeequality that have been achieved in other moderneconomies in Europe or other industrialised OECD countries.All of this raises questions about how progressive fiscal andwelfare policies can be developed to ensure that this period ofresource buoyancy does not lead to a more divided and moreunequal society. The Combat Poverty Agency believes that it isthe responsibility of government to ensure that the inequalitiesarising from wealth creation are controlled and amelioratedthrough public pohcy interventions. The challenge of good

XX11 The Distribution of Income in Irelandgovernance is to redress imbalances through more redistributivetax and welfare policies and investment in quality publicservices. Until the fair distribution of our new-found wealth isaddressed, the "Celtic Tiger" will be an incomplete achieveinment. 10In addition to the national scenario, the shape and directionof government fiscal policy takes place within a European Unionwhich now has a strengthened focus on tackling povertyand other forms of inequality through the Employment ActionPlan and the Social Action Agenda, and within the context of theUN proposals on reducing income inequality which emergedfrom the recent World Summit and were mentioned earlier.This presents a unique opportunity for Ireland to pursue anew model of development based on equitable growth,whereby the pursuit of economic growth and social cohesionare two sides of the one coin, one reinforcing the other.POST-1998 SCENARIOBefore setting out specific policy recommendations it is worthoutlining briefly the kind of trends that have occurred since1997/98, the point to which this research report brings us. Unemploymenthas fallen dramatically and the economy continuesUs remarkably strong rate of growth — although there havebeen recent concerns about rising inflation and containing demandarising from supply constraints, for instance in the labourand housing markets. 11 In particular there are concerns not tooverheat the economy by further fuelling inflationary pres-M Z!?? 1S a fundamental h i* between tackling poverty anddistobu ing income Simulating a continuation of recenttax/welfare policies 12 it was predicted in 1999 that relative in-^ ' ^ ^ ^ - Family Mairs onForewordxx nicome poverty would increase by almost a half between 1994and 2001 at the 40 per cent income poverty line. The redistributiveimpact of Budget 1999, while better than previousones, primarily benefited those on middle incomes (plus 3 percent), and despite being the most generous budget in years,gave those on the lowest incomes minimal gains. Budget 2000,largely due to tax rate reductions, was even less progressive.The result was an upward distribution of resources, with betteroffhouseholds gaining up to four times more than those on lowincomes. The Programme for Prosperity and Fairness, agreed inearly 2000, (while committing considerable resources in the futureto the issue of social inclusion) continued the approach ofpercentage pay increases along the income distribution. It alsoimplemented the introduction of a minimum wage and contamedgeneral commitments on the future shape of income tax policy.Given this fiscal and welfare approach since 1997/98 itwould seem reasonable to assume that no significant changehas been made in relation the extent of income inequality sincethen, certainly not in the direction of greater equality.POLICY RECOMMENDATIONSThere are now enormous resources at the State's disposal tostrategically develop public policy interventions that will controlincome inequalities arising from a growing economy andprevent the continuation of a more divided society as we moveinto the 21st century. To date, public policy has relied tooheavily on job creation alone to solve the problems of inequality,social exclusion and poverty.A more strategic and planned approach is required. Such astrategic approach needs to focus on five broad public policyinterventions.• Firstly, establishing adequate welfare payments and childbenefit rates and ensuring, in a transparent way, that thevalue of those payments are protected.° * 1 3 £ ^ Commentary.Ma^ a'i *MSoXZ^ir n TL*J-^ **»"- « d *"eny itends. Data from the 1997 Living in IrelandlSurvey. Dublin: ESRI, The Department of Social, Community and Family Affairsand the Combat Poverty Agency.

XXIVThe Distribution of Income in IrelandForewordXxv• Secondly, eliminating welfare and poverty traps therebyeasing the transition to work for those who can take up employment,or are already in low paid employment, and investingin human capital through training and education forthis group.• Thirdly, removing the tax burden on the low paid over timeby taking those on the minimum wage out of the tax net andcreating a more progressive tax system by concentrating oncompleting the move to tax credits and increasing tax allowances(or credits) rather than cutting tax rates.• Fourthly, placing a more explicit emphasis on those on thelowest wages in the centralised pay bargaining process.• Fifthly, strategic investment in public services, particularlyhealth, housing, education, childcare and transport.the UK, for example, shows that only half of poor families havesomeone available to enter employment. 14Budget analyses conducted by the Agency in recent yearsusing the SWITCH model (a micro-simulation model demonstratingthe effects of tax and welfare changes across the incomedistribution) have shown how the distributional benefitshave been in favour of those on middle or higher incomes. Thekey mechanism for sharing our growing wealth is the annualBudget. The challenge for future budgets is to reverse the patternto date and ensure that tax and welfare reforms redistributeresources in favour of the least well-off in society.In this regard the Agency recommends the more rigorousapplication of poverty-proofing procedures to budget proposalsin the future. This commitment is made in the Programme forProsperity and Fairness, particularly around budget taxationproposals.THE ANNUAL BUDGETTo avoid the emergence of a more divided society, a budgetarystrategy which re-balances the tax and welfare package in favourof those on low incomes and on welfare is required. The distributionof resources should not negatively impact on work incentives,given the importance of work as a route out of poverty.The Government, however, should not rely entirely on risingemployment to address growing inequalities. Firstly, becausethere is evidence that one factor contributing to a slowdown in labour force growth in Ireland is that potential laboursupply is reaching a limit, 13 and secondly, because employmentc?meTrf *" ° Ptl0n f ° r 3 P r °P° rti °n °f those on the lowest inchZnLt1 T"** th ° M Wh ° m m or ^led, mothers22whoY ^ h° me WWle "•* Children "e young, orPOverT ° Sr ,T d) W M e'workl^ness * still a cause ofPoverty, particularly for families with children, research from13THE NATIONAL ANTI-POVERTY STRATEGY (NAPS)While consistent poverty (based on income poverty and deprivationmeasures combined) is falling, relative income povertyhas been increasing. The increase in relative income poverty,and the widening of the gap between rich and poor, has implicationsfor NAPS in the longer term. The challenge for NAPS inthe current, benign macro-economic environment is to tackleboth the remaining consistently poor and to address the"broader maldistribution of income". 1SA review of NAPS agreed under the PPF includes the establishmentof an Indexation and Income Adequacy WorkingGroup, which will examine the issue of relative income povertyand the adequacy of welfare payments. In the context of thework of this Working Group, specific consideration should begiven to how our growing income inequality can be curtailedM Aber, L. (2000) The Impact of Child Poverty on Children's Weil-Being. Paperto the Combat Poverty Agency Conference on Child Poverty, July 2000. Dublin:Combat Poverty Agency.15 Johnston, H. and T. O'Brien (2000) Planning for a More Inclusive Society: AnInitial Assessment of the National Anti-Poverty Strategy. Dublin: Combat PovertyAgency.

xxviThe Distribution of Income in IrelandForewordxxviiand the gap between rich and poor narrowed through the NAPSstrategy and its targets.Three proposals for consideration in the light of the reviewof NAPS are as follows:• Introduce a relative income poverty reduction target tocomplement and strengthen existing targets. 16• Introduce a NAPS target for reducing income inequality. Themost commonly used measure of income inequality is theGini Coefficient. This study shows that Ireland has a GiniCoefficient above the EU average, and well above thatfound in countries such as Denmark and the Netherlands.• Integrate an income inequality dimension into the povertyproofing process.SOCIAL WELFAREAdequacy and IndexationIn relation to the role of social welfare this research raises twospecific but related issues: establishing an adequate welfarepayment; and determining in a transparent way how rates canbe indexed so as to maintain the relative value of payments.Adequacy and indexation are critical issues that have beenidentified in the PPF to be examined by a working group underthe terms of that agreement. The research highlights the urgencyof doing this.In recent years the Agency has recommended that paymentsbe raised in line with earnings, as fast rising earningsnave outpaced both inflation and the rate of welfare increases,paving those on welfare behind. The failure to link welfarepayments to increases in earnings means that those outside then7n^L m ltad *•* Situation datively disimproved, thuscontributing to the growing gap between rich and poor.16 ,_toti-tonrtTtoJ^llT* !l° nomic M d Social Forum (2000) The National0pUUOnSooalForuL ^" ^No " 8 - Dublin: National Economic andThe Agency recommendation that welfare increases be indexedto rising earnings was predicated on low rates of inflation,which have been a feature of Irish economic recovery.However we have just experienced a sharp rise in price inflation,which is now not only outstripping welfare increases butalso wage increases agreed in the Programme for Prosperityand Fairness (2000). 17Both price inflation and rising earnings contribute to thepattern of consumer norms and expectations and the resulting"standard of living". There is therefore an argument for indexationto be linked to either of these, depending on which ishigher, in order to offer effective protection to the relativevalue of welfare payments. This is certainly a complex issue,which requires immediate attention in the current economicand policy context.Finally, in the absence of an integrated child income supportpolicy, the indexation of welfare rates should also apply tochild dependent allowances.Transitions to WorkThere is a continued need to ease the transition from welfare towork by addressing remaining disincentives for those takingup work. A strengthened Child Benefit, including a "top-up" tocover childcare costs, would play a significant role in this regard.This was a central policy recommendation from theAgency in its recent submission to the Government on the NationalChildren's Strategy. 18A reformed child income policy should incorporate an enhanceduniversal benefit for all children, set at a given proportionof the total costs of raising a child. The Agency suggeststhat this proportion should be in the region of two-thirds of thetotal costs, which is currently the equivalent of £25 per childper week. In addition, the provision of a universal childcareIt is noted that this impact on wages may not follow through completely tonet pay, depending on the nature of tax improvements that complement wageincreases.18 Combat Poverty Agency (2000) A Better Future for Children: EliminatingPoverty, Promoting Equality. Submission to the National children's Strategy.Dublin: Combat Poverty Agency.

XXVlll The Distribution of Income in Irelandsubsidy (funded by restricting the transferability of tax bandsbetween married couples) would support all families with children,not just those liable for tax. This could play an importantrole in supporting parents making the transition from welfare tofull or part-time work.Reform of the tax system requires action on a number offronts: maintaining tax revenue (as a proportion of GNP/GDP)as a means of supporting a more progressive and inclusive society,payment of taxes, progressivity of personal income tax,and broadening the tax base.Maintaining Tax RevenueWorking towards a more progressive tax system is important toensure fairness, while maintaining tax revenue at a level to fundmuch needed public service expenditure is also important. Recentbudget decisions (e.g. cutting the higher rate of tax) maycompound rather than improve the public service inadequaciesand infrastructural bottlenecks which are being highlighted increasinglyas the two chief obstacles to Irish economicsustainability.Clearly there are limits to the extent to which tax rates canbe cut, if we are to achieve greater social inclusion throughpublic policy. At European level, examining taxation in the 15EU countries, tax revenue as a percentage of GDP increasedfrom 40.7 per cent in 1987 to 41.5 per cent in 1997. In contrast,over the same period the ratio of tax revenue to GDP in Irelandtell from 37.4 per cent to 32.8 per cent. 19 Consideration couldbe gi Ven to setting , target for ^ ^ ^ ^EU average ratio of tax/GDP(or GNP)levels oftend^?T C ° n0mieS *** ***""come equalitya"es^° t r *•* aVera^e ta */GDP ratios. Cuts in the taxicTmter^nreSOUICeS aVaUable to *• st *e for public poltQE^dlcTfror*•*" ****** n citizens "Evidence from opinion surveys suggests public support for theOrganisation for Economic Co rw ^Statistics. Paris: OECD For furth " and Develo Pment (1999) RevenueSeeEconomy", chapter by Colm fyR* f"**" "^equality in the New IrishSpending and Inequality fromthf^ *? fr ° m * e ^coming book on Social-Huauty rrom the Combat Poverty AgencyForewordxxixnotion that taxation is a form of social solidarity. However, theperception that some can get away with tax evasion more easilythan others needs to be dealt with and counteracted. The publicneeds to be confident that the tax system is fair, that revenue iscollected equitably and that resources will be reinvested inpublic services for the common good.Payment of TaxesIt is a core principle of a socially cohesive society that all memberspay their fair share of taxes and claim only their legitimateentitlements. Recent behaviour, as revealed by the investigationsof the Public Accounts Committee, has cast a cloud overthis principle. Every effort should be made to remedy the perceptionthat those with large amounts of wealth can "get awaywith" tax evasion whereas those on low and middle incomespay automatically through the PAYE system.More Progressive Income Tax SystemMoves toward a more progressive income tax system include:• Increasing standardised personal and PAYE allowances (ortax credits). This is a fairer approach to tax reform as itbenefits all taxpayers equally and it improves work incentivesfor low earners. This should work progressively towardsthe objective of taking everyone on the minimumwage out of the tax net.• No further cuts in the top rate of tax. The focus on allowancesmeans that there should be no change in either the taxbands or the tax rates.• Restricting the transferability of tax bands between marriedcouples — on the basis that this is an ineffective subventionfor children — and investing revenue saved in improvedChild Benefit.Broadening the Tax BaseThe Government has indicated its intention to reduce corporationtax to 12.5 per cent while profits in this sector are spiral-

xxxiiThe Distribution of Income in Irelanddifficult to obtain information on wealth it is important to reflecton how a more rigorous knowledge and understanding of itcould contribute to our understanding of how to achievegreater equality and social inclusion.There have been a number of efforts in recent years to examinemore closely how income is distributed within householdsand this report underlines how important it is to revisitthe issue. In this regard, the Agency has now commissioned areport on the intra-household allocation of resources, which willbe available in 2001.Finally the Agency would like to acknowledge and thank theresearch team at the ESRI and NUI Maynooth who undertookthis study: Brian Nolan, Bertrand Maltre, Donal O'Neill andOlive Sweetman. The research is of the highest standard andthe Combat Poverty Agency is very grateful for this excellentreport and the dedicated attention to detail throughout.Combat Poverty AgencyNovember 2000Executive SummaryOverall income inequality and inequality in the distribution ofearnings have risen sharply during the 1980s and 1990s in anumber of industrialised countries, giving rise to widespreadconcern about the factors at work and the societal implications.This makes it particularly important to know how the distributionof income in Ireland has been changing over time, how itcompares with other countries, and what factors contribute toexplaining Ireland's particular experience.This study first uses data from the Living in Ireland surveyscarried out by the Economic and Social Research Institute toprovide a picture of the distribution of household income inIreland in the 1990s. A key finding is that there was a markedshift in the disposable income distribution away from the bottom30 per cent over the period from 1994 to 1998. The sharegoing to the bottom 30 per cent of households declined by almostone and a half per cent of total income (adjusted for differencesin household size and composition). The distributionamong households of income coming directly from the marketdid not become more unequal over the period; instead, an importantfactor was that the equalising effect of social welfaretransfers declined. Increases in income share were seen overthe top half of the distribution, rather than concentrated right atthe top. There was also some change in the composition of bothtop and bottom income groups, with younger householdsmoving up and older ones moving down.Over the period from 1973-87, on the other hand, inequalityin the distribution of disposable household income had fallenmarkedly, with the share of the bottom quintile up by almost 1

XXXIVThe Distribution of Income in Irelandper cent of total income and the share of the top decile falling.This was partly because the redistributive impact of income taxand employees' social insurance contributions increased, reflectingboth increasing progressivity and a very substantialrise in the average tax rate. From 1987 to 1994 this continuedbut at a much slower rate, helping to explain the greater stabilityin the shape of the income distribution over those years.Turning to international comparisons, data from the EuropeanCommunity Household Panel Survey shows Ireland tohave one of the more unequal income distributions in the EU inthe mid-1990s. Ireland is one of a group of EU countries - theothers being the UK, Greece and Spain - with relatively highinequality, though not as high as Portugal. A fairly widespread,though not universal, trend towards increased inequality in thepenod from the mid-1980s to the mid-1990s was found in a recentOECD comparative study. The most notable common underlyingfeature was that the share of earnings going to thelower income groups among the working population decreasedm all the countries covered in the studya n ^ ^ T ** ESW household surveys are also used to examinethe distribution of earnings among Irish employees. ThefonTT?e , arnings is found to be relativel y "igh *>y intema -from !? 8 ff! m 1994 'rrrn^ . ^1994havin 9 inc «-sed -laLely rapidly^1997 . When Irish economicEfd althf Tt raPldlyi *" lnCrease in earni^ EqualityW° men in ** P aid ^our forcehas^mcfef 011 ° f ^ ^ ^bJtSjE aTo 7 ln Ireland ' S ° ^ Wa *eS haVecome. Elsewhere SLI K 9 pro P° rtion of household incomeinequality%^a do'te S6en t0 mCrease household'*>r j*2S3fflfirLKsr^ hr been * e case- female employment ra"s Jve r that ^ T ^ ^ ^among wives marrieri to >, ,! Penod were greatestmarried to husbands with relatively low earnings.Chapter 1IntroductionBrian NolanAn accurate picture of the distribution of income, and an adequategrasp of how it comes about, is crucial for policy formationand for understanding the society in which we live. Recentinternational research has highlighted the fact that both overallincome inequality and inequality in the distribution of earningshave risen sharply during the 1980s and 1990s in a number ofindustrialised countries, notably the UK and the USA. Thismakes it particularly important to know how the distribution ofincome in Ireland has been changing over time, how it compareswith other countries, and what factors contribute to explainingIreland's particular experience. The aim of this study isto address these issues, using household survey data.These data allow us to first provide a picture of the distributionof household income in Ireland in the 1990s, so one canboth see what the shape of that distribution is and how it hasbeen changing recently. This distribution may be comparedwith similar figures for 1987 and earlier years, to assess trendsover a longer period. These results for Ireland can also becompared with estimates for other countries, so that both Ireland'scurrent distribution and trends over time can be placedin comparative perspective. Finally, the survey data for Irelandallow us to explore the factors underlying the distribution, notablythe relationship between the overall household incomedistribution and the distribution of earnings among individualearners.

2 The Distribution of Income in IrelandThe study is structured as follows. Chapter 2 describes thesurvey data on which the study relies and how income and itsdistribution are measured. Chapter 3 examines the distributionof household income in Ireland in the 1990s. Chapter 4 makesuse of data from earlier surveys to analyse how the distributionof income in Ireland changed between 1987 and 1994, and alsoto examine longer-term trends in the income distribution backto 1973. Chapter 5 focuses on how the level of inequality in thedistribution of household income in Ireland compares withother countries. Chapter 6 turns to the distribution of earnings,the most important source of income, and how this evolved inIreland over the decade from 1987. Chapter 7 then looks at therelationship between the earnings distribution and the overalldistribution of income among households, focusing in particularon the way in which the earnings of husbands and their wivesare related, and how that influences the household income distribution.Finally, Chapter 8 brings together the main findings.Chapter 2Measuring the Distribution ofIncome in IrelandBrian Nolan2.1 INTRODUCTIONBefore one can examine empirical evidence about the distributionof income, it is essential to have an understanding of theway incomes, their distribution, and the extent of inequality aremeasured. The aim of this chapter is to describe the data onwhich this study relies, and outline how income and its distributionare to be measured. Section 2.2 focuses on the descriptionof the household surveys to be employed and the informationthey obtain on income. Section 2.3 outlines methodologicalissues which arise in measuring the distribution of income andthe extent of income inequality, and the approaches to beadopted in this study.2.2 MEASURING THE DISTRIBUTION OF INCOME: DATA ANDPREVIOUS RESEARCHStudies of the distribution of income in Ireland rely on householdsurveys rather than administrative tax/social security recordsto provide the database. (See Nolan, 1978 for a discussionof the use of survey versus administrative data on income distributionin the Irish case, and Callan, 1991a, for a discussion ofsurvey and Revenue Commissioners income data). In this studywe rely on data from two sets of large-scale household surveys.The first comprises the survey carried out by the ESRI in 1987,

The Distribution of Income in Irelandand the longitudinal survey initiated in 1994. The 1987 Surveyof Income Distribution, Poverty and Use of State Services is describedin detail in Callan, Nolan et al., 1989, the 1994 Living inIreland Survey is described in Callan et al, 1996, and Callan etal, 1999 describes the 1997 Living in Ireland Survey. Thesesurveys have already provided the foundation on which an extensiveprogramme of research on the extent and nature ofpoverty, and a wide range of related topics, has been based(see Nolan and Callan, 1994; Callan et al., 1996; Callan et al,1999). The 1987 survey has also provided data for Ireland forthe Luxembourg Income Study database, employed in the recentcomprehensive comparative study of income distributionin OECD countries by Atkinson, Rainwater and Smeeding(1995).The other set of surveys from which estimates of the incomedistribution can be derived is the Household Budget Survey(HBS) carried out by the Central Statistics Office. As its nameindicates this is primarily an expenditure survey, but it containsdetailed income data and serves as an invaluable source foranalysis of the income distribution. The HBS has been carriedout in 1973, 1980, 1987 and 1994/95, with results from the mostrecent survey (1999/2000) not yet available. A detailed descnptxonof the HBS is given in Murphy (1984). The income iniSE° n ° btained in the ES W surveys and in the HBS is verysmular^although differences in exact timing between the latest^o HBS surveys and the ESRI1987 and 1994 surveys have to beapeswL'ioor ^ ^ ^com Parisons. The micro-dataand are aTlai„^Y° to re searchers for the first time,fro d mTh e eT 9 ? 3^0 £T 2S" ~ ° f ***** "^trends in incometlrPrCmde a lon 9 erPerspective onlandMhasobeens T ph^ s T e ^ L l T 6 ^ 1 7 197 ° S -a /!, °* distribution of income in Ire-1987 and 1994 ESPT ^ ** Househ °l - d the publishedPrOVlde firstoverall distribution of*• P icture of the6(1984, 1985) exDlor^Ham ° ng Irish h °useholds. Murphy1973 and 1980 HBS h»? T**** **** ** distribution in theHBS, based on analysis of the micro-data fromMeasuring the Distribution of Income in Ireland 5within the CSO. Rottman and Reidy (1988) also used data providedby the CSO from the 1973 and 1980 HBS. Callan andNolan (1997) looked at trends in inequality in the household incomedistribution in the 1973, 1980 and 1987 HBS, based onthese previous studies for 1973 and 1980 and data supplied bythe CSO from the 1987 HBS. Data on household incomes fromthe 1987 ESRI survey was lodged with the Luxembourg IncomeStudy database, and as already noted was used in the recentcomprehensive comparative study of income distribution inOECD countries by Atkinson, Rainwater and Smeeding (1995).Callan and Nolan (1999) analysed trends up to the 1994 Livingin Ireland Survey, while O'Neill and Sweetman (1998) comparedthe 1987 and 1994/95 HBS. The availability of these micro-datasets,together with data from later waves of the ESRI'sLiving in Ireland Survey, opens up new possibilities for exploringthe structure of the income distribution and trends inincome inequality in Ireland.Detailed descriptions of the ESRI and HBS surveys havebeen given elsewhere, but it is important to set out here thenature of the income measures and the information obtained inthe surveys to construct them. The questionnaires collected informationon income from the following sources: employeeearnings, self-employment, farming, secondary jobs, casualemployment, State training or work experience schemes, SocialWelfare transfers by scheme, Child Benefit, the renting of landor property, interest or dividends, retirement pensions, pensionsfrom abroad, annuities, covenants or trusts, sick pay froman employer, trade union strike or sick pay, private or charitablemaintenance from outside the household (including alimonypayments), and educational grants.The time period covered by the income information is important.The ESRI surveys and Household Budget Survey adoptthe same approach, recording details for most sources of income,such as earnings, Social Welfare transfers, and privatepensions, in respect of the amount received in the current payperiod (week, fortnight, month, etc.). A longer reference periodwas used for certain other income sources, because itwould not be very meaningful to collect details on currentweekly or monthly receipts in respect of income from self-

6The Distribution of Income in Irelandemployment, farming, property rental or investment income. Inrespect of these income sources details were recorded on thebasis of the most recently available annual figures, converted toa weekly average for the construction of the income measures.L^TS» PPIOaCh iS f ° UOWed in * e Eur °P ean CommunityHousehold Panel Survey (ECHP), in which an annual accountingpenod is adopted throughout. Since the ESRI Living in Ireland2 H C r p n s e d the Iri *h element of the ECHP, they alsoobtained that annual income data, along with current receipts.ECHP and its annualincomeure unHrrt SCUSS1 ° n ° f ** meas-Z1 H 3P I ' COncentratin 9 ^til then on current incomeas measured in the ESRI surveys and the HBScial oroir 5 inf ° rmatl ° n ° n inC ° me from fa »ng Po^s speot£°^th T th6re ^ S° me **» * meth-X T ? b6tWeen ** ESBI SUrve y s -"* the HBS. Inwas esLard 6^ ^ inC ° me " ** P revious

8The Distribution of Income in IrelandMeasuring the Distribution of Income in Ireland 9While the ultimate source of concern is the welfare of the individual,the income accruing to each individual is not a satisfactorymeasure of their command over resources because incomeis generally shared among individuals in a given familyor broader household. The extent to which income is actuallyshared within the household so as to equalise living standardsis an empirical question which has received some attention (seefor example Lundberg, Pollak and Wales, 1997; Cantillon andNolan, 1998; 2000) but is particularly difficult to address. It isnot pursued here, where we follow the conventional approachof employing the household as the income recipient unit. 2Since a given income will provide a different living standardto the individuals in a large versus a small household, or onecomprising adults rather than mostly children, income has to beadjusted for differences in household size and composition.Equivalence scales are intended to make such an adjustment,with actual household income being divided by the number ofequivalent adults in the household to produce equivalent orequivalised income. Equivalence scales may take only householdsize into account, or they may incorporate both the numberand age of household members. A very wide range ofscales is employed within and across countries, and there is noconsensus as to which set of scales or methodologies for estimatingthem is most satisfactory or appropriate. Studies such asBuhman et al. (1988) and Coulter, Cowell and Jenkins (1992)have shown the extent to which the equivalence scale employedcan affect the measured income distribution (even whenonly size is being taken into account).It is therefore necessary to assess the sensitivity of the resultsto variation in the equivalence scale employed. Here wedo this by using five different equivalence scales. The first isone of the scales which has been employed in analysis of povertyin the ESRI surveys, corresponding to the scales implicit in" The household is defined in the ECHP as comprising "either one personliving alone or a group of persons, not necessarily related, living at the sameaddress with common housekeeping — i.e. sharing a meal on most days orsharing a living or sitting room" (Eurostat 1999, p. 25). The CSO employ asimilar definition in the Household Budget Survey.Irish social welfare rates in the late 1980s. This attributes avalue of 1 to the household head, 0.66 to each other adult, and0.33 to each child in the household. The second scale has beenwidely used in the UK, and is closer to the values for additionaladults towards which Irish social welfare rates have moved inrecent years: this attributes a value of 0.6 to each additionaladult and 0.4 to each child. The three other sets of scales havebeen commonly employed in cross-country income distributionand poverty studies. One is the square root of household size,without distinguishing between adults and children (see for example,Atkinson, Rainwater and Smeeding, 1995). The othertwo are widely known as the OECD and the "modified OECD"scales (see for example Hagenaars, De Vos and Zaidi, 1994).Where the first adult in the household is given a value of 1, underthe OECD scale each other adult is attributed a value of 0.7and each child is attributed a value of 0.5. With the modifiedOECD scale, each adult is attributed a value of 0.5 and eachchild 0.3. As in Hagenaars et al., we take adult here to mean age14 years or over.A further issue is whether one focuses on the distribution ofincome or poverty among households, which attributes eachhousehold equal weight in the analysis, or on the distributionamong individuals. As noted by Atkinson, Rainwater andSmeeding, it makes sense to treat each household as a singleunit (i.e. to apply household weights) if no adjustment is madeto income for household size. When equivalent income is used,though, person weights seem more appropriate. This isachieved by weighting each household in the analysis by thenumber of persons it contains. However, much of the previousresearch on the Irish income distribution refers to the distributionamong households, and it is only on this basis that resultsfrom the 1973 and 1980 Household Budget Surveys are available.For that reason we will be dealing with the distributionamong households at some points, and the distribution amongpersons at others, in the present study.The distribution of income among households and/or personsmay be portrayed and summarised in a number of differentways. Here we generally rank cases by income and thenderive decile shares — the share of total income going to the

10 The Distribution of Income in Irelandbottom 10 per cent, the next per cent, . . . top 10 per cent. Inlooking at the earnings distribution we follow conventionalpractice and the deciles or quartiles of the distribution as percentagesof the median — the earnings at the 10th percentilepoint, 25th percentile . . . 90th percentile point as percentagesof median earnings. In addition, various summary measures ofinequality are employed here. These are the Gini coefficient,Theil's entropy measure, the coefficient of variation, the meanlogarithmic deviation, the Atkinson inequality measure with acoefficient (i.e. inequality aversion parameter) of 0.5 and 1.0,and the ratio of the top to the bottom decile, P 90 /P 10 . Suchmeasures, designed to summarise the degree to which incomesare concentrated, are commonly used in the study of incomeinequality. Since some inequality measures put most weight onincome differences in different parts of the distribution to others,all such measures will not always display the same trends,so it is worth looking at more than one. In addition, some measuresare more suitable than others for specific types of analysis,in particular decomposition into sub-groups or income sources.We will not attempt to review here their derivation and properties,on which there is an extensive literature: a comprehensivedescription is given in e.g. Cowell (1995).Summary measures represent one approach to capturingand comparing the level of inequality in different income distributions.Lorenz curves, on the other hand, show the wholedistribution in graphical form. As illustrated in Figure 2.1, thisshows the share of total income going to the bottom x per centof the distribution, where x goes from 0 to 100 per cent. Wherethe Lorenz curve for one distribution lies above that for anotherdistribution at all points, this means that the bottom x per centof the first distribution has a higher share than in the seconddistribution no matter which value we choose for x. For distributionswith the same mean income, it has been shown that thismeans that the first distribution can be taken to have a higherlevel of what economists term "social welfare", for quite a widevariety of social welfare functions. Where the Lorenz curvesintersect, on the other hand, no such unambiguous ranking ofthe distributions is available - it will depend on the weight weassign to different parts of the distribution (see Cowell, 1995)Measuring the Distribution of Income in Ireland 11Figure 2.1: Lorenz Curve for Income"ot tnCI* 8•a > *-•«i 3O1008060402000 20 40 60 80 100Cumulative Percentage ofIndividuals or HouseholdsWhat about where the two distributions have different levels ofmean income? Generalised Lorenz curves provide a convenientway of incorporating information about average living standardsand inequality into the comparison of the level of socialwelfare yielded by different distributions. This involves plottingcumulative mean incomes (instead of cumulative income sharesin standard Lorenz curves) against cumulative populationshares (see Shorrocks, 1983; Jenkins, 1991). Once again, unambiguousrankings will only be available in certain circumstances,but the value of the approach is precisely in allowingus to identify when that occurs.2.4 CONCLUSIONS•Line of EqualityIncomeThis chapter has outlined the data and methods to be employedin measuring the distribution of income in this study. The studyrelies on data from large-scale household surveys carried outby the ESRI and the CSO. The methodological choices faced insuch a study include the choice of income recipient unit, howbest to adjust income for the size and composition of the household,and how best to present and summarise the shape of theincome distribution. We go on in the next chapter to use thesemethods to analyse the distribution of income in Ireland in the1990s.

Chapter 3The Distribution of Income inIreland in the 1990sBrian Nolan and Bertrand Maitre3.1 INTRODUCTIONIn this chapter we look at the distribution of income among Irishhouseholds and persons in the 1990s. We use for this purposedata from the Living in Ireland Survey, which as we saw in thelast chapter was initiated in 1994. We have been able to analysein some depth income data from that first wave and from thefourth wave of the survey, carried out in 1997. Data from the1998 wave of the survey, coming on stream as this study wascompleted, allowed us to also include some initial results forthat year. This chapter concentrates on the broad pattern of incomedistribution and recent changes as revealed by thesesurveys. In the next chapter we make use of the 1987 ESRIhousehold survey and the CSO's Household Budget Surveys toassess trends from 1987 to 1994 and over a longer period backto 1973. 1We first look in Section 3.2 at the distribution of market,gross and disposable income among households in 1994 and1997, without any adjustment for differences in household sizeand composition. Section 3.3 then looks at the shape of the distributionwhen adjustment for such differences has been madeIn doing so we also compare the results of the Living in Ireland Survey 1994with the 1994/95 HBS; for this chapter we confine our attention here to theLiving in Ireland Surveys.

14 The Distribution of Income in Irelandby equivalisation, using alternative equivalence scales. Thedistributions of equivalised household income among householdsand among persons are also compared. Section 3.4 analyses thecomponents of change in the distribution between 1994 and 1997.Section 3.5 examines where different types of household tend tobe located in the distribution. Section 3.6 presents some initialresults for 1998. Finally, Section 3.7 summarises the main findingsof the chapter.3.2 THE DISTRIBUTION OF DISPOSABLE INCOME AMONGHOUSEHOLDS IN 1994 AND 1997We begin by examining the distribution of disposable income -the income concept which has the most direct relevance forability to spend — among households in the 1994 and 1997Living in Ireland Surveys (LII). Table 3.1 shows the share in totalincome going to each decile. We see that in each year the bottom10 per cent of households had about 2 per cent of total income,while the top 10 per cent had about one-quarter of thetotal. In broad terms a similar shape for the income distributionis seen in other industrialised countries, and the comparisonwith other EU and OECD countries presented in Chapter 5 willhelp to put Ireland's distribution in perspective. For the present,though, we concentrate on the Irish pattern and on how ithas been changing between 1994 and 1997.Two inequality measures designed to summarise the degreeto which incomes are concentrated are also shown in Table 3.1,namely the Gini and Theil measures. As mentioned in the previouschapter, different inequality measures put most weight onincome differences in different parts of the distribution, and allsuch measures will not always display the same trends, so it isworth looking at more than one. We see from the table thatthese summary measures calculated for disposable incomesuggest little difference in the level of inequality between the1994 and 1997 surveys. However, looking at the decile shareswe see that this overall stability masks the fact that there was aslight shift away from both the bottom and the top of the distribution,with those in the middle gaining.The Distribution of Income in the 1990s 15Table 3.1: Decile Shares and Summary Inequality Measures,Disposable Income among Irish Households, 1994and 1997 Living- in Ireland SurveysDecileBottom23456789TopAllGiniTheilShare in Total DisposableIncome (%)1994 LII2.33.34.66.07.59.111.113.516.526.4100.0Inequality Measure0.3770.2371997 LII2.13.34.56.07.79.511.213.416.525.8100.00.3730.236As well as shares, it is worth stating the actual income levels requiredto bring one into for example, the top 10 per cent of thedistribution, and what constitutes an "average" income. In 1997,the median point in the disposable income distribution amonghouseholds — the point which splits the distribution exactly intwo — was about £290 per week or £15,100 per annum. The incomecut-off for the top decile — the lowest income which wouldbring a household into the top 10 per cent — was about £630 perweek or £32,700 per year. The income below which a householdwould be in the bottom 10 per cent, on the other hand, was about£84 per week or £4,400 per year. No account has been taken sofar of differences in household size and composition, which obviouslyaffect the living standard one can reach on these variousincome levels. It is instructive none the less to know what theseabsolute figures are, since there may be widespread misper-

16 The Distribution of Income in Irelandception of, for example, how high an income is required to locateone in the top ranges of the distribution.We now turn to the distribution of income from the market,and market income plus cash transfers, and see how thesecompare with disposable income. Table 3.2 presents decileshares in direct (market), gross and disposable income amonghouseholds in the 1994 and 1997 LII surveys. Unsurprisingly, ineach year the distribution of market income is very much moreunequal than that of gross or disposable income: the bottom 30per cent of the distribution has virtually no income from themarket, while the top 10 per cent has about one-third of the total.State cash transfers bring about a substantial change in theshape of the distribution, with the share of the bottom 30 percent of households rising to about 8 per cent of total income,and the share of the top deciles falling significantly. The differencebetween gross income shares and those for disposableincome is less marked, but the latter does have higher shares atthe bottom and lower ones at the top: income tax and employee'ssocial insurance contributions do move the distributionfurther in the direction of greater equality.This is reflected in the Gini and Theil coefficients for thesedistributions. Going from direct to gross income, State cashtransfers reduced the Gini coefficient by about one-quarter in1994. Going from gross to disposable income, direct tax reducedthe Gini by a further 10 per cent in that year. As is commonlythe case in industrialised countries, both cash transfersand direct tax thus have an equalising impact on the shape ofthe income distribution, with the effect of transfers being substantiallymore pronounced.We can also see from Table 3.2 that for market income, inequalitydid not in fact increase between 1994 and 1997. Both theGini and Theil coefficients suggest that inequality actually fellfor direct income between 1994 and 1997. For gross income, onthe other hand, there was little change in the summary measures.This is illustrated by the fact that in 1997 these transfersreduced the Gini coefficient by 22 per cent, compared with 25per cent in 1994. The relationship between gross and disposableincome was broadly unchanged from 1994 to 1997. Decreasinginequality in market income was thus offset by the di-The Distribution of Income in the 1990s 17minishing impact of State cash transfers in reducing inequality,leaving little change in the disposable income distribution.Table 3.2: Decile Shares for Direct and Gross Incomeamong Households, 1994 and 1997 LII SurveysDecileBottom23456789TopAll19940.00.00.32.86.09.012.115.420.434.0100.0Gini 0.565Theil 0.587Direct19970.00.01.13.86.69.512.015.119.232.8100.00.5360.529Share in Total Income (%)19941.92.73.95.16.88.811.013.717.528.7100.0GrossInequality Measure0.4220.29519971.72.63.95.37.39.111.113.617.228.3100.00.4170.291Disposable19942.33.34.66.07.59.111.113.516.526.4100.00.3770.2373.3 ADJUSTING FOR HOUSEHOLD SIZE AND COMPOSITION19972.13.34.56.07.79.511.213.416.525.8100.00.3730.236We now proceed to the analysis of the distribution of incomeafter adjustment for the size and composition of the household.As made clear in Chapter 2, no consensus exists on the mostappropriate equivalence scale to make this adjustment, so wewill be using the range of scales described there and assessingwhether the choice of scale makes a significant difference to theresults. We begin with the scale giving the value 1 to a singleadult, 0.66 to each additional adult and 0.33 to each child (under15). Figure 3.1 shows the distribution of disposable income,equivalised using this scale, among households in the 1994 and1997 LII surveys.

18 The Distribution of Income in IrelandFigure 3.1: Decile Shares in Equivalised Disposable Incomefor Irish Households, 1994 and 1997 LII Surveys*3 4 5 6 7 8 9 10* See Table A2.1 in Appendix 2 for data.• LII 1994QUI 1997A comparison with Table 3.1 shows that equivalisation producesa more equal distribution in each year. The share of thetop decile is lower than before equivalisation, and the shares ofthe bottom two deciles are higher by what is, in proportion totheir unadjusted share, a very substantial amount.This reflects the fact that larger households have, on average,higher incomes than smaller households. Many of thehouseholds towards the bottom of the unadjusted distribution,for example, comprise single (often elderly) adults or couples.Comparing 1994 and 1997 we observe a similar pattern to thatshown in Table 3.1 for unadjusted income. The bottom and topof the distribution lose share and deciles 5, 6, 7 and 8 all gain,the net result being little change in overall inequality as reflectedin the summary measures.Given the extent of uncertainty about the most appropriateway to adjust income for such differences, we need to test howsensitive these results are to alternative equivalence scales.The other four sets of equivalence scales described in Chapter2 are now employed, to see whether the shape of the disposableincome distribution is affected by the choice of scale. Table3.3 shows decile shares among households in both the 1994and 1997 LII surveys with each of these scales. We see theequalising impact of equivalisation occurs with each of thescales, and the shape of the distribution is very similar acrossThe Distribution of Income in the 1990s 19all five sets of scales. Comparing 1994 and 1997, all the scalesshow once again the marked decline in share for the bottom ofthe distribution and the top decile which we saw in Table 3.3.While there are some differences across the scales in the exactpattern of change from 1994 to 1997, all show a substantial declinein share for the bottom half of the distribution and increasefor the top half.Up to this point, we have been concerned with the distributionof income among households. As discussed in Chapter 2, itmakes sense to treat each household as a single unit and "counthouseholds" if no adjustment is made to income for differencesin household size. When equivalent income is used, however,focusing on persons seems more appropriate since we are primarilyconcerned with the distribution of welfare or living standardsamong persons rather than households. Much of the previousresearch on the Irish income distribution refers to thedistribution among households, and it is only on this basis thatresults from the 1973 and 1980 Household Budget Surveys areavailable. However, it is of interest to now also look at the distributionamong persons, which is achieved by weighting eachhousehold in the analysis by the number of persons it contains.It must be emphasised that, as discussed in Chapter 2, the assumptionis still being made that resources are distributedwithin the household so that each member of a given householdhas the same living standard: we are now in effect simplycounting persons rather than households.

20 The Distribution of Income in IrelandTable 3.3: Decile Shares in EquivalisedDisposable Incomefor Irish Households in the 1994 and 1997 LII Surveys,Alternative Equivalence ScalesDecileBottom23456789TopAll19943.84.75.46.17.18.710.512.815.925.1100.0Share in Total Equivalised Disposable Income (%)7/0.6/0.4Scale19973.54.55.26.27.49.010.613.115.924.7100.01/0.7/0.5Scale19943.64.85.56.17.18.510.412.516.025.4100.019973.44.75.26.27.38.810.612.816.224.9100.07/0.5/0.3Scale19943.94.65.36.17.28.910.712.915.824.8100.079973.54.45.26.17.69.110.913.115.724.5100.0Square RootScale19943.64.45.16.17.49.010.912.916.024.6100.019973.34.15.06.47.79.211.213.115.824.3100.0Table 3.4 compares the distributions of equivalised income(with the 1/0.66/0.33 scale) among households and among personsin the 1994 and 1997 LII surveys. We see that countingpersons rather than households reduces the share of the top 30per cent and increases the share of middle income groups ineach year. Comparing the distribution among persons in 1997with 1994 we do again see a shift in share away from the bottom30 per cent, but this is less pronounced, and there is now littlechange at the top.We go on in subsequent chapters to look at the evolution ofthe equivalised income distribution between 1987 and 1994,and to compare the shape of the equivalised distribution inIreland and other countries. In the rest of this chapter, however,we explore some of the factors underlying the shape of the distributionin Ireland in 1994 and 1997, and why it has changed somarkedly over that short period.The Distribution of Income in the 1990s 21Table 3.4: Decile Shares in Equivalised Disposable IncomeAmong Households and Among Persons, 1994 and 1997 MISurveys (1/0.66/0.33 Scale)DecileBottom23456M89TopAllShare in Total Equivalised (1/0.66/0.33) DisposableIncome (%)AmongHouseholds3.94.85.46.17.18.710.512.715.925.0100.01994 UIAmongPersons3.84.95.66.47.58.910.612.615.324.4100.0AmongHouseholds3.64.65.26.17.59.010.713.015.924.6100.01997 UIAmongPersons3.64.75.56.67.59.210.612.615.424.3100.03.4 SUB-GROUP DECOMPOSITION OF INEQUALITY IN 1994AND 1997We now employ a decomposition technique that allows us toassess the role of inequality between and within particularpopulation sub-groups in overall inequality, and how thischanged between 1994 and 1997. (Decomposition of inequalityby sub-groups in this manner is discussed in, for example,Shorrocks (1980) and (1984) and Cowell (1995)). This is basedon a summary inequality measure particularly suited for thispurpose, called the mean logarithmic deviation (MLD). Withthis measure, overall inequality in the distribution can be convenientlydecomposed into inequality between discrete subgroupsand inequality within each of those sub-groups. (TheGini coefficient, by contrast, cannot be readily decomposed inthis way). Here we look at a range of household characteristicsin this light, categorising households by age, by sex and thenby labour force status of head, by composition type, and finallyby the extent of social welfare recipiency.

22 The Distribution of Income in IrelandWe begin in Table 3.5 with three groups of households distinguishedby whether the head is aged under 35, 35-64, or 65 orover. The table first shows the MLD for each of these sub-groups.Table 3.5: Decomposition of Inequality in Disposable EquivalisedIncome by Age of Head, 1994 and 1997 UI SurveyA:1994GroupUnder 3535 - under 6565 or overAllOf which:Inequalitywithin Group(MLD*1000)162168102160Within group inequality (% of total)Between group inequality (% of total)B: 1997Under 3535 - under 6565 or overAllOf which :175169121167Within group inequality (% of total)Between group inequality (% of total)Group MeanIncome(£ per week)132.13124.90107.20123.64158 (98.8)2(1.2)177.65160.26132.17159.35163 (97.8)4 (2.2)Group Share inPopulation186814100196615100We see that in both years the level of within-group inequality ismuch lower among those headed by someone aged 65 or overthan among the two younger groups. Households headed by anolder person also have lower mean income than the other twogroups, even after adjustment for their smaller size. Inequalitybetween the groups accounts for less than 2 per cent of overallinequality, the rest being attributable to inequality within theage groups. Inequality within each of the age-groups and betweenthem all rose between 1994 and 1997.Table 3.6 focuses on households headed by a man or a coupleversus those with a female head. 2 Only 15 per cent ofand Watson (1998) discuss alternatives to this crude categorisation.The Distribution of Income in the 1990s 23households are in the latter group, and the table shows that thelevel of inequality is less than among them than the rest of thesample, but that female-headed households have lower meanincomes than the rest of the population. Inequality within thesetwo groups rather than between them accounts for almost allthe overall total, and this was unchanged between 1994 and1997.Table 3.6: Decomposition of Inequality in DisposableEquivalised Income by Sex of Head, 1994 and 1997 MI SurveyGroupInequalitywithin Group(MLD*1000)A: 1994Male or Couple161Female134All160Of which:Within group inequality (% of total)Between group inequality (% of total)B: 1997Male or CoupleFemaleAll164158167Of which:Within group inequality (% of total)Between group inequality (% of total)Group MeanIncome(£ per week)127.39103.14123.64157 (98.3)3 (1.7)164.61126.83159.35163 (97.7)4 (2.3)Group Share inPopulation(%)85151008614100Table 3.7 categorises households by the labour force status ofthe head. There is now a great deal of variation in inequalitywithin these groups. There is a much higher degree of inequalityamong households headed by a self-employed person (includingfarmers) than among those headed by an employee,and relatively little inequality among households headed bysomeone who is unemployed or ill, or engaged full-time inworking in the home. There are also now substantial differencesacross the groups in mean equivalised income: householdsheaded by an employee or a self-employed person havemuch higher mean incomes than those with an unemployed or

24 The Distribution of Income in Irelandill head or one working full-time in the home. In 1994, thesedifferences in mean income across the groups accounted forabout 27 per cent of the inequality in the overall sample. By1997, the inequality produced by these differences in mean incomehad fallen, and accounted for 24 per cent of overall inequality.In terms of within-group inequality, there was a sharpincrease for households headed by someone unemployed or ill.The size of some of the groups had also changed, with the proportionof employees increasing and unemployed falling.Table 3.7: Decomposition of Inequality in DisposableEquivalised Income by Labour Force Status of Head, 1994and 1997 LLT SurveyA: 1994GroupEmployeeSelf-employedUnemployed/illRetiredHome DutiesAllOf which:Inequalitywithin Group(MLD*1000)1012405510575160Within group inequality (% of total)Between group inequality (% of total)B: 1997EmployeeSelf-employedUnemployed/illRetiredHome DutiesAllOf which:9923510011665167Within group inequality (% of total)Between group inequality (% of total)Group MeanIncome(£ per week)152.50140.3269.59112.3482.92123.64116(72.9)44(27.1)188.65187.3386.16142.9693.73159.35128 (76.4)39 (23.6)Group Share inPopulation(%)4119171210100461914139100Finally, it is interesting to focus on the role of social welfaretransfers by distinguishing two groups: households which re-The Distribution of Income in the 1990s 25ceive more than half their total income from transfers andhouseholds which do not. Table 3.8 shows that, unsurprisingly,the 30 per cent of households who do receive more than halftheir income from transfers have both lower mean equivalisedincome and lower within-group inequality than the rest of thesample. The table also shows that inequality between these twogroups accounted for about 35 per cent of overall inequality inthe 1994 sample.Table 3.8: Decomposition of Inequality in DisposableEquivalised Income by Social Welfare Dependency, 1994and 1997LLT.SurveysGroupInequalitywithin Group(MLD*1000)A: 1994Social Welfare13650% or Less ofTotal IncomeSocial Welfare29More that 50% ofTotal IncomeAll160Of which:Within group inequality (% of total)Between group inequality (% of total)B: 1997Social Welfare50% or Less ofTotal IncomeSocial WelfareMore that 50% ofTotal IncomeAll13736167Of which :Within group inequality (% of total)Between group inequality (% of total)Group MeanIncome(£ per week)147.3368.02123.64104 (65.2)56 (34.8)183.8379.75159.35113(67.9)54(32.1)Group Share inPopulation(%)By 1997, inequality within both groups had increased, and accountedfor a slightly larger share in total inequality than in1994. This reflects the fact that although the mean income of70301007624100

26 The Distribution of Income in Irelandthose relying on social welfare lagged behind the overall meanbetween 1994 and 1997, the size of that group declined significantlyfrom 30 per cent to 24 per cent of the sample, reducingthe between-group component of inequality.These decomposition results clearly provide valuable insightsinto the structure of income inequality in Ireland and howit has been changing. This is approached from another perspectivein the next section, where we look at where differenttypes of household tend to be located in the income distributionand how that has been evolving.3.5 WHO IS WHERE IN THE INCOME DISTRIBUTION?We begin by looking at the location of persons in the incomedistribution categorised by the age of the head of the householdin which they live, and focus on the proportion falling intoeach quintile of the disposable equivalised income distributionamong persons. (We employ quintiles — successive one-fifthsof the distribution — rather than deciles in order to see theoverall pattern more easily.) We see first in Figure 3.2 thatpeople in households headed by someone aged between 35and 64 were spread rather evenly over the quintiles in both1994 and 1997. In 1994 those in households headed by someoneaged under 35 were relatively heavily concentrated in boththe bottom and the top quintile, but by 1997 they were moreconcentrated at the top. In 1994 those in households headed bysomeone aged 65 or over were heavily concentrated in thesecond quintile from the bottom. By 1997, however, the proportionat the top of the distribution had fallen and the numbersin the bottom quintile had risen markedly. The short periodbetween 1994 and 1997 thus saw a considerable change at boththe top and bottom of the distribution, to the advantage ofyounger households and the disadvantage of older ones.The Distribution of Income in the 1990s 27Figure 3.2: Position in the Income Distribution of PersonsCategorised by Age of Household Head, 1994 and 1997 LITSurveys*199445 Y'403530- '25-20H$^H18-1050Head Aged Under Head Aged 35-64 Head Aged 65 or35 Over45403530252015105ni1997i £ a0Head Aged Under Head Aged 35-64 Head Aged 65 or35 Over* See Table A2.2 in Appendix 2 for data.• Bottom02S3• 4DTop• Bottom• 2S3• 4DTopTable 3.9 distinguishes those living in households headed by acouple or a single man and those headed by a woman. We seethat the latter are significantly more heavily concentrated towardsthe bottom of the income distribution, and that their proportionin the bottom quintile increased from one-quarter toover one-third between 1994 and 1997.

28 The Distribution of Income in IrelandTable 3.9: Position in the Income Distribution of PersonsCategorised by Sex of Household Head, 1994 and 1997 LIISurveysA: 1994B:1997QuintileBottom234TopAllBottom234TopAllPosition in Equivalised (1/0.66/0.33) DisposableIncome Distribution (%)Male or Couple Head19.117.320.921.521.2100.017.419.220.621.221.5100.0Female Head24.335.614.911.813.4100.035.724.915.813.010.7100.0Figure 3.3 categorises people by the labour force status of thehead of their household, and we see that those in householdsheaded by an employee are mostly located in the top threequintiles. Those in households with a self-employed head arerelatively heavily concentrated both at the bottom and at thetop of the income distribution. Those in households headed bya farmer are fairly evenly spread over the distribution. Over 60per cent of those in households headed by an unemployed personwere in the bottom quintile in 1994, and by 1997 this hadrisen to 68 per cent. Those in households headed by a retiredperson were heavily concentrated in the second and thirdquintile from the bottom in 1994, but by 1997 the proportion inthe bottom quintile had risen a good deal. Those in householdswhere the head works full-time in the home were mostly in thebottom 2 quintiles in each year.The Distribution of Income in the 1990s 29Figure 3.3: Position in the Income Distribution of PersonsCategorised by Labour Force Status of Household Head,1994 LII Survey*Employee Self- Farmer Unemployed Retired Home Dutiesemployed• Bottom D2 S3 B4 DTopEmployee Self- Farmer Unemployed Retired Home Dutiesemployed• Bottom D2 S3 B4 OTop* See Table A2.3 in Appendix 2 for data.Finally, Table 3.10 focuses on a categorisation by the number ofadults and children in the household. This shows that those in 1-adult households were heavily concentrated in the secondquintile from the bottom in 1994, and in the bottom quintile in1997. Those in households comprising 1 adult with childrenwere very heavily concentrated right at the bottom of the distribution,though less so in 1997 than in 1994. A relatively highproportion of those in households comprising couples with 1 or

30 The Distribution of Income in Ireland2 children were towards the top of the distribution in bothyears. Those with 4 or more children, on the other hand, werevery heavily concentrated in the bottom quintile.Table 3.10: Position in the Income Distribution of PersonsCategorised by Household Composition Type, 1994 HISurveyQuintileA: 1994Bottom234TopAllB:1997Bottom234TopAllPosition in Equivalised (1/0.66/0.33) DisposableIncome Distribution (%)1 2 1 adult +adult \ adults childrenL .21.235.39.510.024.0100.036.721.18.511.122.7100.09.226.916.216.831.0100.013.624.011.414.936.0100 056.417.818.16.21.5100.042.29.838.38.71.0100.02 adults,1 child14.015.712.423.634.4100.03.6 INITIAL RESULTS FOR 199817.011.217.922.731.3100.02 adults, 212 adults, 3children ! children12.48.820.528.629.6100.012.87.124.132.523.6100.022.515.516.627.218.1100.025.012.514.429.318.8100.02 adults,4+ children37.014.222.919.06.9100.041.619.426.36.26.6100.0While we have been able to analyse the distribution of incomein the 1997 Living in Ireland survey in some depth, data fromthe 1998 round of the survey is just now coming on stream. Inconcluding this chapter it is therefore valuable to look briefly atsome initial results on the overall shape of the distribution inthat sample. Once again it is important to be conscious of thefact that some attrition in the sample takes place from year toyear, and that reweighting is employed to seek to maintain theoverall representativeness of the results. A full description ofThe Distribution of Income in the 1990s 31the 1998 sample and a discussion of attrition and reweighting inthat context is given in Layte et al (2000).We look first in Table 3.11 at the distribution of disposableincome among households, without adjustment for differencesin household size and composition, in 1998 compared with 1997and 1994. We see once again a decline in the share of the bottom30 per cent of the distribution between 1997 and 1998,which is in fact more pronounced than that observed from 1994to 1997. With the share of the top decile rising from 1997 to1998, unlike 1994-97, the summary inequality measures alsorise marginally.Table 3.11: Decile Shares and Summary InequalityMeasures, Disposable Income among Irish Households,1994,1997 and 1998 Living in Ireland SurveysHouseholdsDecileBottomca3456789TopAllGiniTheil1994 UI2.33.34.66.07.59.111.113.516.526.4100.00.3770.237Share in Total Disposable Income (%)Inequality1997 UI2.13.34.56.07.79.511.213.416.525.8100.00.3730.236Measure1998 UI1.83.04.46.07.79.511.313.516.726.1100.00.3860.251Table 3.12 shows the distribution of market and gross income inthe three years. Once again, the distribution among householdsof income from the market is seen to become if anything moreequally-distributed from 1997 to 1998, as it did from 1994 to

32 The Distribution of Income in Ireland1997. Turning to the distribution of gross income, there is onceagain some decline in share for the bottom 30 per cent from1997 to 1998, as there had been from 1994 to 1997.Table 3.12: Decile Shares for Direct and Gross Incomeamong Households, 1994,1997 and 1998 LII SurreyDecileBottom23456789TopAllGmiTheil19940.00.00.32.86.09.012.115.420.434.0100.00.S650.587Direct19970.00.01.13.86.69.512.015.119.232.8100.00.5360.529Share in Total Income (%)19980.00.01.54.27.09.512.015.219.331.4100.079941.92.73.95.16.88.811.013.717.528.7100.0Inequality Measure0.5200.4950.4220.295Gross19971.72.63.95.37.39.111.113.617.228.3100.00.4170.29119981.62.53.75.27.59.211.313.717.627.7100.00.4180.291Finally, Table 3.13 shows the distribution of income amonghouseholds after adjusting for differences in size and compositionusing the 1/0.66/0.33 equivalence scale. We now see theshare of the bottom 30 per cent of households down by 0.7 percent of total income between 1997 and 1998. This brings thecumulative fall in the share of the bottom 30 per cent up to 1.4per cent of total income over the 1994-98 period. A decliningshare for the top decile from 1994 to 1997 had offset this declinein terms of overall inequality as reflected in the summarymeasures. From 1997 to 1998, however, the share of the topdecile rose and so did the summary inequality measuresThe Distribution of Income in the 1990s 33Table 3.13: Decile Shares in Equivalised DisposableIncome for Irish Households, 1994,1997 and 1998 LIISurveys (1/0.66/0.33 Scale)DecileBottom23456789TopAllGiniTheilShare in Total Equivalised (1/0.66/0.33) DisposableIncome (%)1994 Ul1997LII1998 LII3.94.85.46.17.18.710.512.715.925.0100.00.3260.1843.64.65.26.17.59.010.713.015.924.6100.0Inequality Measure0.3290.1853.44.35.06.27.69.110.612.815.825.2100.00.3380.1953.7 CONCLUSIONSThis chapter has looked at the distribution of income amongIrish households and persons in the 1990s, using data fromhousehold surveys carried out by the ESRI in 1994 and 1997 andsome initial results from 1998.Focusing on the period from 1994 to 1997, summary measuressuggested little change in the overall level of inequality inthe distribution of disposable income among Irish householdsover this period. There was however a shift away from the bottom30 per cent and the top decile to the middle of the distribution.The distribution of income directly from the market was ifanything more equally distributed among households in 1997than 1994, but the equalising effects of social welfare transfersdeclined." !

34 The Distribution of Income in IrelandThe shape of the disposable income distribution after adjustmentfor differences in household size and composition,using alternative equivalence scales, was seen to have changedin a similar way to the unadjusted distribution between 1994and 1997. Focusing on the distribution among persons ratherthan households showed a decline in shares for the bottom butnow little change at the top.A decomposition analysis of the distribution of equivalisedincome in 1994 and 1997 looked at inequality within and betweendifferent groups of households, distinguished on the basisof age, gender, labour force status and social welfare dependence.For example, this showed that inequality betweenhouseholds categorised by the labour force status of the headaccounted for about 27 per cent of overall inequality in 1994and 24 per cent in 1997. Within these groups, inequality washighest for households headed by a self-employed person andlowest among the retired.There were also some changes in the location of differenttypes of household in the income distribution. Those in householdsheaded by someone aged between 35 and 64 werespread rather evenly over the quintiles in both 1994 and 1997,but there were considerable shifts for both older and youngerhouseholds. In particular, households headed by someoneaged 65 or over were more heavily concentrated towards thebottom by 1997.Initial results from the 1998 Living in Ireland survey showeda continued decline in the share of equivalised income going tothe bottom 30 per cent of households. This brought the cumulativefall in the share of the bottom 30 per cent from 1994 to 1998up to 1.4 per cent of total income. A declining share for the topdecile from 1994 to 1997 had offset this decline in terms ofoverall inequality, but from 1997 to 1998 the share of that decilerose and so did summary inequality measures.In assessing both the overall shape of the distribution of incomein Ireland and these recent trends, a frame of reference isprovided by comparison with corresponding figures for othercountries. That is the topic of Chapter 5, but first we look inChapter 4 at how the Irish distribution has been evolving over alonger period, back to the early 1970s.Chapter 4Trends in the Irish IncomeDistribution Since 1973Brian Nolan4.1 INTRODUCTIONSo far we have focused on data from the household surveys carriedout by the ESRI in the 1990s. In this chapter we use datafrom a household survey carried out by the ESRI in 1987, andfrom the Household Budget Surveys carried out by the CSO in1973, 1980, 1987 and 1994/95, to see how the Irish income distributionhas been evolving over a longer period. Section 4.2looks at 1987 and 1994, for which estimates of the income distributionare available from both ESRI and HBS surveys. Section4.3 documents trends in the distribution over the period from1973 to 1997. Section 4.4 analyses factors underpinning theselonger-term trends, while Section 4.5 presents an in-depth decompositionanalysis of the 1987-1994 changes. Section 4.6summarises the main findings.4.2 THE INCOME DISTRIBUTION IN 1987 AND 1994We look first at the distribution of disposable income amonghouseholds in the ESRI and CSO samples for 1987 and 1994, inTable 4.1. (We concentrate now on the distribution amonghouseholds rather than persons, because it is only on that basisthat we can push the comparisons back to 1973 in the next section.)We see that in each of the years there was some differencebetween the estimates derived from the ESRI surveys and

36 The Distribution of Income in Irelandthose from HBS, with the share going to the top decile ratherhigher in the ESRI surveys. However, both sets of surveys showa similar pattern of broad stability in the distribution betweenthe two years. 1Table 4. Is Decile Shares in Disposable Income for IrishHouseholds in 1987 and 1994, ESRI and HBS SurveysDecileBottom23456789TopAllESRI2.03.44.85.97.38.810.713.216.527.4100.0Share in Total Disposable Income (%)1987HBS2.23.75.06.37.69.211.013.416.525.1100.0ESRI2.33.34.66.07.59.111.113.516.526.4100.01994HBS2.13.54.86.07.69.211.313.616.725.1100.0Turning to gross income, Table 4.2 shows the decile shares forthe two years from the two sets of surveys. These again show ahigher share going to the top decile in the ESRI surveys butonly modest changes in shares between the two years.This stability is reflected in the Gini and Theil coefficients for the ESRI surveys,but comparable summary measures cannot be accurately computedfrom the HBS microdata because what is termed "top-coding" was employedin releasing these data to researchers. This involves setting very high incomesto a ceiling figure, which would distort the inequality measures. Decileshares however can be corrected for top-coding and this has been done forthe figures presented here. Since mean income for the total sample withouttop-coding is published, total income can be calculated, the income of deciles1-9 can be derived from the micro-data, and that for the top decile — affectedby the top-coding — can be derived as a residual.Trends in Income Distribution since 1973 37Table 4.2: Decile Shares in Gross Income for IrishHouseholds in 1987 and 1994, ESRI and HBS SurveysDecileBottom23456789TopAllESRI1.72.84.05.26.78.610.713.517.629.2100.0Share in Total Gross Income (%)1987HBS1.93.14.25.47.08.811.013.617.627.5100.0ESRI1.92.73.95.16.88.811.013.717.528.7100.01994HBS1.92.94.05.36.98.911.213.817.527.8100.0One important difference between the ESRI and HBS surveys isin the timing of the fieldwork, as mentioned in Chapter 2, andthis may have had an impact on farm incomes in particular.However, the reasons for the divergence in the shape of thedistribution are not clear at this point. As far as trends over the1987-94 period are concerned, though, both sets of surveyssupport the same conclusion of little change in the distributionof gross and disposable income.We now proceed to adjust for the size and composition ofthe household, beginning again with the 1/0.66/0.33 equivalencescale. Table 4.3 shows the distribution of gross and disposableincome equivalised using this scale in the 1987 and1994 ESRI surveys: analysis of equivalised incomes in the 1987and 1994-95 HBS indicate very similar patterns over time.

38 The Distribution of Income in IrelandTable 4.3: Decile Shares in Equivalised Gross and DisposableIncome for Irish Households, 1987 and 1994 ESRISurveys (1/0.66/0.33 Scale)DecileBottom23456789TopAllShare in Total Equivalised (1/0.66/0.33) Income (%)19872.74.14.95.66.88.310.112.416.328.8100.0Gross19943.24.04.55.26.48.210.313.017.228.1100.019873.24.85.86.57.48.710.212.315.225.9100.0Disposable19943.94.85.46.17.18.710.512.715.925.0100.0We see that for gross income there is now some increase in theshare of the bottom decile, with a decline in share for those inthe middle of the distribution — deciles 3, 4, 5 and 6 — and forthe top decile. For disposable income the increase in the shareof the bottom decile is greater, at about two-thirds of one percent of total income, while deciles 3, 4 and 5 and the top decilesee a fall.The other four sets of equivalence scales described in earlierchapters are now employed, to see whether the pattern ofchange between 1987 and 1994 is affected by the choice ofscale. Table 4.4 shows decile shares in disposable income forboth 1987 and 1994 with each of these scales. All show amarked increase in share for the bottom decile, and a fall inshare for deciles 3, 4, 5 and the top decile.Trends in Income Distribution since 1973 39Table 4.4: Decile Shares in Equivalised Disposablefor Irish Households, 1987 and 1994 ESRI Surveys,Alternative Equivalence ScalesDecileBottom23456789TopAllShare in Total Equivalised Disposable Income (%)1/0.6/0.419873.14.85.76.57.48.610.212.315.326.1100.019943.84.75.46.17.18.710.512.815.925.1100.0Equivalence Scale1/0.7/O.S19873.04.75.86.67.48.610.012.215.326.4100.019943.64.85.56.17.18.510.412.516.025.4100.01/0.5/0.319873.24.85.66.47.48.710.312.415.325.9100.019943.94.65.36.17.28.910.712.915.824.8100.0IncomeSquare Root19873.14.55.36.37.48.810.412.615.626.0100.019943.64.45.16.17.49.010.912.916.024.6100.0The increase between 1987 and 1994 in the share of total incomegoing to the very bottom of the income distribution isconsistent with the pattern of declining income poverty gapsover the period reported in Callan et al. (1996). An importantfactor at work there was the extent to which the lowest socialwelfare payments were brought up relatively rapidly, as therecommendations of the Commission on Social Welfare in thisregard were implemented. This meant that Unemployment Assistance,for example, rose a good deal more rapidly than bothsocial welfare pensions and average earnings (see Callan et at,1996 for a discussion). We now move on to overall trends in thedistribution over the longer period back to 1973.4.3 THE IRISH INCOME DISTRIBUTION FROM THE EARLY1970S TO THE MID-1990SWe now look at the distribution of income among householdsover the longer period back to the early 1970s, using the HBS

40 The Distribution of Income in Irelandsamples for each of the years for which these are available. TheHBS was carried out in 1973 and 1980, and although the microdatafrom these surveys are not available for analysis some figureshave been produced by the CSO, published in Murphy(1984, 1985) and Rottman and Reidy (1988). Table 4.5 shows thedecile shares in disposable income among households from theHBS for 1973, 1980, 1987 and 1994/95. It also shows Gini andTheil summary measures computed from these decile shares:while these will understate the extent of inequality at each pointin time (because inequality within each decile is ignored) theyshould capture trends over time adequately.Table 4.5: Decile Shares in Disposable Income among IrishHouseholds, 1973, 1980, 1987 and 1994/95 HBSDecileBottom234S6789TopAllGiniTheil79731.73.35.06.57.89.210.913.016.226.4100.00.3670.221Share in Total Disposable Income (%)19801.73.55.16.67.99.311.013.016.225.7100.00.3600.21119872.23.75.06.37.69.211.013.416.625.0100.00.3520.2001994-952.13.54.86.07.69.211.313.616.725.1100.00.3620.210Source: Decile shares for 1973 and 1980 from Rottman and Reidy (1988), Table7.4, for 1987 and 1994/95 from microdata tapes (with correction for topcoding).Looking first at 1973 to 1980, we see that the share of the bottomdecile was unchanged but those of deciles 2-7 all rose slightly,with the top decile seeing a substantial decline in share. This isreflected in the fall in the Gini and Theil coefficients, also shownTrends in Income Distribution since 1973 41in the table. Between 1980 and 1987, the share of the bottomquintile rose substantially and that of the top decile again fell.Deciles 3-6 now experience a decline in share, however, withdeciles 8 and 9 seeing an increase. The Gini and Theil coefficientsshow a further decline in inequality between 1980 and1987. Over the whole period 1973-87, then, the HBS shows inequalityin the distribution of disposable household income falling,and the Gini coefficient falling from 0.37 to 0.35. This reflectsthe fact that the share of the top decile was down by 1.4per cent of total income and that of the bottom quintile up by0.9 per cent. It is worth noting that the other gainers were not inthe bottom half of the distribution, but rather deciles 8 and 9. 2This pattern contrasts with the increase in inequality observedfor gross income, over both sub-periods and 1973-87 asa whole. The redistributive impact of income tax and employees'social insurance contributions on the gross income distributiontherefore increased substantially over the period, asseen in the growing gap between the summary inequalitymeasures for gross and disposable income. In 1973 the Ginicoefficient for disposable income was 97 per cent of that forgross income, in 1980 it had fallen to 94 per cent, and by 1987 itwas only 88 per cent, while the corresponding percentages forthe Theil measure were 93 per cent in 1973, 87 per cent in 1980and 77 per cent in 1987. 3We saw in the previous section that the decline in inequalityin the distribution of disposable income between 1973 and 1987was not sustained after 1987. It would be desirable to look attrends during the earlier period having adjusted household incomesfor differences in size and composition, in other words toanalyse the distribution of equivalent household incomes.Without micro-data for 1973 and 1980, however, we have todraw on published results for those years. Roche (1984) presentedequivalent gross and disposable income distributions2 In terms of Lorenz curves, 1980 and 1987 both Lorenz-dominate 1973. TheLorertz curve for 1987 lies inside that for 1980 at most points, but those curvescross and there is no unambiguous dominance.3 Murphy (1984) and Rottman and Reidy (1988) note that between 1973 and1980 inequality rose even more for market income — that is, income beforecash transfers — but transfers offset some of this increase.

42 The Distribution of Income in Irelandfor 1973 and 1980 using a set of equivalence scales based onthe social security rates paid at the start of the period. On thisbasis the distribution of gross equivalent income became lessequal between 1973 and 1980, with the shares of each of thebottom five deciles falling. For disposable income the picturewas less clear-cut: the share of the bottom decile fell betweenthe two years, that of the bottom half was unchanged, the topdecile lost share, and the remainder of the top half gained,while the Gini coefficient indicates a fall in inequality.The overall trends between the 1980 and 1987 HBS withequivalised income were broadly similar to those in unadjustedincome — that is, the gap between gross and disposable incomedistributions widened, with the former becoming if anything lessequal and the latter if anything more equal. The ESRI survey for1987 points in the same direction: comparing unadjusted andequivalent income distributions in the ESRI survey does suggestthat the gap between the two was wider than in 1980.It is important to place these trends in the shape of the incomedistribution in the context of the evolution of average realincomes. This is particularly important in comparing the 1970s,1980s and 1990s because the pattern of real income growth wasvery different in the three sub-periods analysed here, that is1973-1980, 1980-1987, and 1987-1994. Between 1973 and 1980,real incomes grew rapidly. Mean household income in the HBSin 1980 was 193 per cent higher than in 1973, while the ConsumerPrice Index rose by 155 per cent, representing a 25 percent rise in average income in real terms. Between 1980 and1987, on the other hand, average incomes rose by 87 per centin nominal terms but the CPI was 91 per cent higher by the endof the period, so real incomes actually fell. Between 1987 and1994, real incomes increased very substantially: averagehousehold income rose by about 40 per cent while consumerprices rose by only 22 per cent. The period from 1994 to 1997also saw real income growth accelerate, with average householdincome rising by 23 per cent compared with only 6 percent for consumer prices. 4J**?"^* real ome growth underlies the relationships betweenGenerahsed Lorenz curves - discussed in Chapter 2 - for tne differentTrends in Income Distribution since 1973 434.4 EXPLAINING TRENDS IN INEQUALITY 1973-1994Why then did inequality in the Irish income distribution evolvein this way over the years 1973-1994? Without micro-data forthe earlier years it is not possible to address this question directlyfor the whole period using the type of decompositionanalysis presented in the previous chapter for 1994-97. Wehave carried out such analyses for the 1987-94 period and discussthe results below, but it is worthwhile to first point to somelikely links over the entire period between the income distributionand developments in the macroeconomy and in tax andsocial security policy. The central features of the Irish economiclandscape in the 1970s and 1980s from this point of view werethe striking increases in both unemployment and direct personaltaxation.Rising unemployment is likely to have been a major influenceon the gross income distribution. The rate of unemploymentdoubled to 7-8 per cent of the labour force between 1973and 1980. 5 A detailed analysis of within-group and betweengroupinequality in the 1973 and 1980 HBS was carried out byMurphy (1984) from within the CSO. This suggested that the increasingnumber of households with an unemployed or retiredhead or with no earner was primarily responsible for the increasinginequality in market incomes, with little increase inwithin-group inequality. 6 The impact on the distribution of grossincomes was ameliorated though not fully offset by the fact thatrates of social security support grew relatively rapidly. This wasparticularly marked for the elderly, which would have been animportant influence on the share of the bottom quintile in theunadjusted distribution. The unemployed, on the other hand,years. For disposable income without any adjustment for household size,generahsed Lorenz curves for both 1980 and 1987 he above 1973 at all points.The curves for 1980 and 1987 themselves cross — because mean income isslightly lower (in real terms) but the income share of the bottom deciles isslightly higher in 1987 than 1980 — so no unambiguous ranking in welfareterms is produced. However 1994 lies above 1987 at all points, and the sameis true for 1997 versus 1994.5 The rate of unemployment in the Labour Force Survey carried out in April1980 was 7.3 per cent, while registered unemployment over the year averagedover 8 per cent.6 See Murphy (1984) p. 84.

44 The Distribution of Income in Irelandtend to be most heavily concentrated in the bottom half but notthe bottom quintile of the unadjusted distribution, which experienceddeclining shares in total gross income.Between 1980 and 1987 unemployment increased verymuch more rapidly, rising to over 18 per cent of the labourforce. This is likely to have played a major part in the fallinggross income shares of the second and third quintiles of the unadjusteddistribution. In assessing the impact on household incomes,it is worth noting that in the Irish case an unusually highproportion of the unemployed are married men living in families,who would conventionally be termed "household heads",rather than secondary earners. 7 Despite the fact that once againsocial security support rates were increased in real terms overthe period, unemployment did in general result in a decline inincome for those affected. 8 In the Irish case a particularly highproportion of the unemployed have also been away from worklong-term: by 1987 about half of all men on the Live Register ofthe unemployed had been in that position for more than oneyear. A substantial majority of the unemployed have thereforeexhausted unemployment insurance entitlements and have torely on (what were in 1987 lower) means-tested social assistancepayments.Between 1973 and 1980, total tax revenue (including socialsecurity contributions) rose from 31 per cent to 34 per cent ofGDP. 9 Within that increasing total, taxes on personal incomeplus employees' social security contributions became more important,rising from 29 per cent of total taxation in 1973 to 37per cent in 1980. 10 This sharp increase in taxation and particularlypersonal taxation reflected not only the growing burden ofunemployment on the public purse, but also the expansion inother areas of state expenditure associated with the ill-fated fiscalpolicy adopted in 1977-1979. This was intended to be acounter-cyclical employment-creating strategy, but its lastingTrends in Income Distribution since 1973 45result was an extremely high level of public debt and associateddebt service payments. After 1980, as the public sectordeficit was brought under control, total tax revenue rose byeven more, reaching 40 per cent of GDP in 1987. Taxes on personalincome and employees' social security contributions continuedto play a more and more important role, and by 1987were accounting for 40 per cent of all tax revenue. (The increasein direct tax revenue was achieved not by raising top taxrates — the top rate was in fact considerably lower in 1987 thanin the early 1970s — but by a higher standard rate and by failureto index allowances and tax bands to take inflation into account.)11The growing redistributive impact of income tax and socialsecurity contributions over the 1973-87 period was attributableto both their growing importance and to the increasing degreeof progressivity with which these taxes operated. Table 4.6shows the average tax rate for income tax and for social insurancecontributions in the HBS, and a summary measure of progressivityknown as the Suits index. This measure is based onthe share of total tax paid by different income groups versustheir share in total income; it has a positive value for a progressivetax and is negative for a regressive one.We see first that in 1973 income tax and employees' socialsecurity contributions came to just under 10 per cent of grossincome of sample households. By 1980 this had risen to 15 percent, as the general level of taxation rose and income taxationaccounted for a steadily rising share of the total. Between 1980and 1987 the percentage of gross income going in income taxagain increased, though less dramatically, but now social securitycontributions rose substantially.7 See OECD Employment Outlook (1990); NESC (1990).For example, Callan and Nolan (1999) show that in most cases the replacementrate is below 80 per cent in the 1994 LII sample.9 OECD Revenue Statistics (1990) Table 3.10 OECD Revenue Statistics (1990) Tables 11 and 15.11 In the early/mid-1970s Irish income tax rates started at about 25 perand rose as high as 80 per cent, whereas in 1987 they went from 35 per c€58 per cent.

46 The Distribution of Income in IrelandTable 4.6: Suits Progressivity Index and Average Tax Ratefor Income Tax and PRSI Contributions, 1973, 1980, 1987and 1994/95Average Tax Rate (%)Income taxPRSI contributionsTotalSuits Progressivity IndexIncome taxPRSI contributionsTotal1973 1980 1987 1994/957.82.09.80.194-0.0740.13812.92.215.10.2070.0560.18515.43.518.90.2750.1330.24914.23.717.80.2820.1470.254Source: Figures for 1973 from Nolan (1981), for 1980, 1987 and 1994-95 calculatedfrom HBS Reports.Focusing on the degree of progressivity, Nolan (1981) used theSuits progressivity index to show that in 1973 income tax wasprogressive but social security contributions were mildly regressive(as indicated by the negative value of the index). 12From Table 4.6 we see that by 1980 income tax was slightlymore progressive but social insurance contributions were nowalso progressive. 13 Between 1980 and 1987, income tax becamea good deal more progressive and so did social insurance contributions.However, contributions became a more importantelement in the total over this period and remained a good dealless progressive than income tax. The net result was that theSuits index for income tax plus social insurance contributionstogether rose by almost as much between 1973-1980 as between1980-1987, registering an increase of 34 per cent in theearner period and 35 per cent in the later one. Between 1987Trends in Income Distribution since 1973 47and 1994, the Suits index rose marginally for both income taxand social insurance contributions, but they both now declinedas a proportion of gross income.We can analyse in more depth the way the income distributionevolved between 1987 and 1994, by exploiting the fact thatmicrodata is available for those years. We do this by carryingout for 1987 the same sub-group decomposition analysis thatwe presented for 1994 and 1997 in Chapter 3, using the meanlog deviation inequality measure. Decomposition by labourforce status was the one to reveal the most substantial changesbetween 1994 and 1997, and this also turned out to be the casefor 1987-1994 so we concentrate on those results here. Table4.7 shows that in 1987 inequality between the groups accountedfor 13 per cent of overall inequality, whereas by 1994 this hadrisen to 27 per cent. This reflects the fact that there was lessvariation in mean incomes across the groups in 1987, as well asthe higher inequality within certain groups — notably the selfemployed,the unemployed or ill, and households headed bysomeone working full-time in the home — in that year. Thus by1994 inequality within these groups had declined while thedifferences between the groups in mean incomes had widened,with households headed by someone who was unemployed orill, or working full-time in the home, falling further behind theaverage.12 The index is calculated on the basis of share of tax paid by gross incomegroups and the share of gross income they receive.Murphy (1984) presents the Suits (and Kakwani) progressivity index calculatedfor equivalent (market) income for 1973 and 1980. Like Table 4.6 (basedon unadjusted gross income), this showed the degree of progressivity of in-" ^ ^Sfed * that of employees' social security contributionssignificantly increased between 1973 and 1980.

48The Distribution of Income in IrelandTable 4.7: Decomposition of Inequality in DisposableEquivalised Income by Labour Force Status of Head,1987 ESBI SurveyA: 1987GroupEmployeeSelf-employedUnemployed/illRetiredHome dutiesAllOf which:Inequalitywithin group(MLD*1000)10538297114148194Within group inequality (% of total)Between group inequality (% of total)B:1994EmployeeSelf-employedUnemployed/illRetiredHome dutiesAllOf which:1012405510575160Within group inequ ility(% of total)Between group ineq uality(% of total)Group meanIncome (£ perweek)94.0292.6250.6077.6867.6482.33168 (86.6)26 (13.4)152.50140.3269.59112.3482.92123.64116(72.9)44(27.1)Group share inpopulation (%)442219971004119171210100Trends in Income Distribution since 1973the ESRI surveys. However, both sets of surveys show a similarpattern of broad stability in the distribution of disposable incomeamong households between the two years. When equivalencescales are used to adjust disposable income for differences inhousehold size and composition, the share of the bottom decile isseen to have risen significantly, at the expense of those in themiddle of the distribution and the top decile.Over the period 1973-87 inequality in the distribution ofdisposable household income fell markedly, with the share ofthe top decile down by 1.4 per cent of total income and that ofthe bottom quintile up by 0.9 per cent. The share of the topdecile fell both from 1973 to 1980 and from 1980 to 1987, but theincrease for the bottom decile was in the latter period. An importantfactor at work was the increasingly redistributive impactof income tax and employees' social insurance contributions,reflecting both increasing progressivity and a very substantialincrease in the average tax rate. From 1987 to 1994 thiscontinued but at a much slower rate, helping to explain thegreater stability in the shape of the distribution over thoseyears.494.6 CONCLUSIONSThis: chapter has used data from household surveys to see howthe M* xncome distribution has changed over the period backFor f 9 87"Idf,o S / S !" ^^ ^ *— SUrVe * S aUow ° ne to *°"oubv tn4 994 '^"nation is available from surveys earned^uTces liST*,****"* 011 GStimateS derived fr thesesources, wxth the share going to the top decile rather higher in

Chapter 5Ireland's Income Distribution inInternational PerspectiveBrian Nolan and Bertrand Maitre5.1 INTRODUCTIONWe have looked at the shape of the income distribution in Irelandand how it has been changing, but a comparative perspectiveis also essential if we are to know what to make ofthese results. The aim of this chapter is thus to compare thedistribution of income in Ireland with other countries. Greatcare is needed in making such comparisons: without careful attentionto comparability in terms of income concept, income unit,time period, nature and coverage of data source, equivalencescale (where relevant) and so on, misleading conclusions can bereached. Much of the chapter is taken up with a comparisonbased on data for the mid-1990s from the European CommunityHousehold Panel (ECHP), which has the major advantage ofbeing a harmonised survey across all the participating countries,covering most of the EU. We then look at available comparisonsfor the 1980s, relying mostly on the income distributiondatabase assembled by the Luxembourg Income Study (LIS)which for Ireland includes data from the 1987 ESRI survey. Wethen draw on the results of some recent comparative analyses ofincome inequality carried out by the OECD, which once again forIreland relies on data provided from the ESRI surveys.The chapter is structured as follows. Section 5.2 looks at thedistribution of income in Ireland versus other EU member

52 The Distribution of Income in Irelandcountries in the first two waves of the European CommunityHousehold Panel survey. Section 5.3 compares these with resultsfrom other cross-country studies for earlier periods. Section5.4 discusses the results of the OECD comparative study ofinequality trends, and Section 5.5 brings together the chapter'sconclusions.5.2 THE DISTRIBUTION OF INCOME IN THE ECHPA full description of the ECHP dataset in terms of sampling, responserates, weighting procedures etc. is in Eurostat (1999).Here we present some results we have derived from Wave 1,carried out in 1994, then look at any major changes by Wave 2,carried out in 1995.The ECHP survey contains detailed information on incomeaccruing to household members from different sources, such asincome from employment, income from self-employment, occupationalpensions, rental income, interest and dividends,cash transfers from the state by type, and regular cash transfersfrom other households. For some income sources, notably themain income from employment, information is sought on bothgross receipt and on net receipt after deductions of tax and socialinsurance contributions. For a number of other sources,notably self-employment, rental and investment income, onlygross or net receipts were obtained. The nature of the data obtainedfor certain countries raises particular problems in thisrespect. For France, for example, self-employment income isonly available on a gross basis and no imputation of net incomewas included on the ECHP file. Some households in the datasetcould not be included in our analysis because they had missingvalues for total household income — this was a particularly seriousproblem for Italy.Unlike the data used in this study so far, the reference periodcovered by the income questions is annual for all sources,in the case of Wave 1 relating to the calendar year 1993 and forWave 2 to 1994. The Irish element of the ECHP, the Living inIreland Survey, collected information on both current and annualincome. Table 5.1 shows the distribution of annual disposableincome in Ireland in 1993 from wave 1 of the ECHP, com-Ireland's Income Distribution in International Perspective 53pared with the estimates based on current income in 1994 fromthe same survey presented in Chapter 3 above. The bottom halfof the income distribution looks very similar, but there are differencesin the top half. The share of the top decile in annualincome is a good deal higher than the share of that decile in thecurrent income distribution, balanced by lower shares for therest of the top half. The shorter accounting period would in itselflead one to expect the current distribution to be less equalthan the annual one, though empirical studies with UK datasuggest not markedly so (see Nolan, 1986). The reasons for thisdifference merit in-depth investigation, though this is complicatedby the fact that the ECHP data are processed internallyby Eurostat, including imputation of missing values.Table 5.1: Decile Shares in Disposable Income among IrishHouseholds, Current versus Annual Income, 1994 Living inIreland Survey/ECHPDecileBottom23456789TopAllShare in Total DisposableIncome (%)Current2.33.34.66.07.59.111.113.516.526.4100.0Annual2.23.24.55.87.38.810.713.016.228.3100.0Table 5.2 shows decile shares in disposable income amonghouseholds in each country in Wave 1 of the ECHP, without anyadjustment for differences in household size and composition.(As noted earlier, it seems appropriate to employ householdweights when dealing with unequivalised incomes.)

54 The Distribution of Income in Ireland8si3rs(%sed Incoa5nequlahare*1orta.IeceJ*Ita•wJ1g•*coco*?oe—1«>o1—1»o1— oI-Hex2COoe-Hca1—4f-xl**>—ieo«—4•-4_ •-•—*eoex00cai-H00cai—ieococacoi-H©eo»-Hco00caCDcai-H«*COp-H•-HCOi-Heoeo*HpH00COex•-HcoCOi-HcoCO•-H•*CO•-HCMcococo•—4COto•-Hx"CO•-Htocococo—COCOi-HCOcoi-HCOCOoeo•-HexcaCDexcaCOx«caeo00CMO00ca00tocaeoCOcaCOx>COcacaCOcacaeo•*cao•-H100.0100.0100.0o8*—1e100qoOo

56 The Distribution of Income in Irelandbroadly similar picture to the Gini, though with some variationin the individual rankings. Since we have not yet adjusted incomefor household size these rankings cannot be taken as reflectingthe distribution of command over resources, livingstandards or welfare, but they provide a point of departure incomparing income distributions across the countries.We now adjust household incomes to take differences insize and composition into account. Following recent Eurostatpractice in such comparisons we employ the modified OECDscale as our central scale — that is, the first adult in the householdis given a value of 1, each additional adult a value of 0.5,and each child a value of 0.3. To assess the overall sensitivity ofthese results we also use the so-called OECD scale (1/0.7/0.5)and the square root of household size scale employed in anumber of comparative income distribution studies. We focuson the distribution of equivalised income among persons, eachindividual being attributed the equivalised disposable incomeof their household.Table 5.4 shows decile shares in equivalised disposable incomeamong persons in the ECHP, using the modified OECDequivalence scale. As was the case for Ireland, equivalisationproduces a more equal distribution, the share of the bottomdecile is now higher in all countries than it was with unadjustedincome, and the share going to the top decile is consistentlylower. The former ranges from about 2 per cent in Greece andPortugal up to 4 l /i per cent in Denmark, while the latter rangesfrom 20 per cent in Denmark up to 29 per cent in Portugal. Thebottom decile in Ireland now has a relatively high share, at over3 per cent, but at 26 per cent the top decile also has a ratherhigh share, similar to the UK.Summary inequality measures for Wave 1 incomeequivalised using the modified OECD scale are shown in Table5.5. The rankings in terms of the Gini measure are generallysimilar to those for unadjusted income, although Denmarkrather than the Netherlands now has the most equal distributionand Portugal and Greece now have the least equal distributions,followed by the UK and then IrelandIreland's Income Distribution in International Perspective 575o>"1spJ)U£is§S20)a.R (Q1•—13osable1\$tl%•aI•5\1fcA1\5"O60o0)CO«—1t>11.5niXICO_.. i•a1o a..5«aCO3o1•o3"3Hga>K3&"3 COd1M£Q>,io>OacsuaQa>i—iCDsiedCOegCOt-«egcotocooCOriCO• **COcdr~*t"HCOcoTTo• *CO• *inCOoCO00txCOocriegoSqCT5COoS• *o>t-«oSCOCDCDod• *dCOdqr—i©>-*codcodCOdCDd•—4CDdCOoCOdt-««edCOesii—it-

58 The Distribution of Income in IrelandTable 5.5: Suxrunary Inequality Measures for DistributionEquivalised (Modified OECD Scale) Disposable Incomeamong Persons, Wave 1ECHP, 1993DenmarkNetherlandsBelgiumGermanyFranceLuxembourgItalySpainIrelandUKGreecePortugalGini0.2220.2700.2820.2930.3090.3110.3210.3380.3410.3460.3530.388Rank1234S6789101112AtkO.S0.0450.0620.0680.0730.0810.0800.0880.0950.0950.0990.1060.125Rank123465789101112Atkl0.0790.1220.1320.1460.1540.1500.1740.1840.1800.1860.2030.238Rank123463798101112Theil0.1020.1270.1390.1490.1740.1710.1770.1980.2100.2120.2240.264RankTo allow the sensitivity of the results to the choice of equivalencescale to be seen, we recalculated the decile shares andsummary inequality measures for both the OECD and squareroot equivalence scales. Table 5.6 summarises the overall patternby simply reporting the Gini inequality measure for eachcountry with each of the three equivalence scales. We see thatthe equivalence scale employed does make a difference to thelevel of the Gini coefficient in some countries. However, whilesome pairwise rankings of countries by inequality level aredifferent, the overall pattern in terms of country groupings isnot affected by the choice across these three scales. In particular,from an Irish perspective it does not affect the placing ofIreland in a group with the UK, Greece and Spain, having thehighest level of inequality other than Portugal.123465789101112ofIreland's Income Distribution in International Perspective 59Table 5.6: Gini Coefficient for Equivalised DisposableIncome Among Persons in ECHP Wave 1, 1993, AlternativeEquivalence ScalesDenmarkNetherlandsBelgiumGermanyFranceLuxembourgItalySpainIrelandUKGreecePortugalGini ModifiedOECDScale0.220.270.280.290.310.310.320.340.340.350.350.39Ranking123455788101012OECDScale0.230.280.290.300.320.320.330.340.350.350.360.39Ranking12345578991112SquareRootscale0.230.270.290.290.310.310.320.340.340.350.350.39RankingIncome distribution results can also be derived from Wave 2 ofthe ECHP. Gini coefficients for Wave 2 (using the modifiedOECD scale) are compared with the estimates we derived fromWave 1 in Table 5.7. We see that the estimates for Wave 2 aregenerally similar to those for Wave 1, though the Wave 2 Gini israther higher in Denmark and The Netherlands, and lower inFrance, the UK, Spain and Portugal. 1123355788101012Income distribution results based on Wave 2 of the ECHP have been publishedby Eurostat (Eurostat 1999). The Gini coefficients shown differ at most°nly marginally from the ones presented here except for Ireland, where O.Jorather than 0.33 is given: this reflects on-going revisions to the data, our resultsbeing based on a later data release.

60 The Distribution of Income in IrelandTable 5.7: Gini Coefficient for Equivalised DisposableIncome among Persons in ECHP Wave 1, 1993, and Wave 2,1994, Modified OECD ScaleDenmarkNetherlandsBelgiumGermanyLuxembourgFranceItalySpainIrelandUKGreecePortugalWavel(1993)0.220.270.280.290.310.310.320.340.340.350.350.39Ranking123455788101012GiniWave 2(1994)0.240.290.280.290.280.290.310.320.330.320.340.375.3 EARLIER COMPARATIVE INCOME DISTRIBUTIONRESULTSRankingHaving presented figures for Ireland's income distributioncompared with other EU countries in the mid-1990s in thefc^HP, it is useful to compare these with results from earliercross-country comparative exercises. The most useful point ofcomparison is the comprehensive study of income inequalitybyESS TAtkinson ' Rainwater and Smeeding (1995) for theOECD. This was based primarily on data from the LuxembourgIncome Study (US) database and focused on surveys carriedout between 1984-1988. The income concept, recipient unit andaccounting period employed was similar to that used in thetCHP, the mam focus being on disposable household incomer^m e ?iTl r 3 y6ar WlthOUt mcludm 9 for ^ p l e imputedent. (Unlike the ECHP, the datasets in LIS come from nationalmharmonisedfcTn?^ ^^ source: areas where datafor particular countries departed from the desired measure of142424781081112Ireland's Income Distribution in International Perspective 61income or the household unit are discussed in Atkinson, Rainwaterand Smeeding, Chapter 3.) They present results on thedistribution of income for nine out of the eleven countries coveredin this study — they did not have data for Denmark orGreece — using the square root equivalence scale and personweighting.These can be compared with the figures presented above,and for convenience we focus in Table 5.8 on the Gini coefficient.There are major differences for some countries betweenthe two sets of results. As summarised in the Gini coefficient,the level of inequality in the ECHP was a good deal higher thanin the Atkinson, Rainwater and Smeeding (1995) results forGermany, Belgium, Luxembourg, the UK, Spain and Portugal.For the Netherlands, France, Ireland and Italy the ECHP estimatesare similar to those in Atkinson et al. It is known from nationalstudies that the level of income inequality did indeed risebetween the mid-1980s and the mid-1990s in some EU countries,notably the UK. On this basis the increase in the UK Ginicoefficient appears broadly consistent with external evidence(see, for example, Goodman, Johnson and Webb, 1997). 2 Someincrease in inequality in Belgium up to 1992 is suggested bynational sources (see Cantillon et al. 1994), but not as great asthe gap between the Atkinson, Rainwater and Smeeding andECHP figures. 3 It would, however, be surprising if the level ofinequality had increased in Luxembourg and Portugal by asmuch as this comparison suggests, particularly when the Atkinson,Smeeding and Rainwater results for Portugal refer to1989/90.2 The latter cannot be compared directly with either Atkinson et al. or theECHP results because a different equivalence scale is used.Cantillon et al. (1994) show the Gini coefficient increasing from 0.225 in 1985to 0.237 in 1992, using a different equivalence scale.

62 The Distribution of Income in IrelandTable 5.8: Gini Coefficient, Equivalised Income amongPersons, Square Root Scale, Wave 1ECHPand Atkinson,Rainwater and Smeeding (ARS) Study (Equivalence scalesquare root of household size)GermanyNetherlandsBelgiumLuxembourgFranceUKIrelandItalySpainPortugalECHP (1993)0.290.270.290.310.310.350.340.320.340.39Gini CoefficientARS0.25 (1984)0.27 (1987)0.24 (1988)0.24 (1985)0.30 (1984)0.30 (1986)0.33 (1987)0.31 (1986)0.31(1990/91)0.31 (1989/90)Source: Atkinson, Rainwater and Smeeding (1995), Tables 4.3 and 4.4 exceptfor Spam from Table 5.21 and Portugal Table 5.20.5.4 A COMPARATIVE PERSPECTIVE ON RECENT TRENDS ININCOME INEQUALITYA recent study on trends in income distribution and poverty bythe OECD (Forster, 2000) provides another basis of comparisonfor the level of inequality in Ireland versus other countries, andis also particularly valuable in terms of a comparative pictureon recent trends. It does not include some of the EU countrieswith relatively high levels of inequality - notably Spain andPortugal - but it does include Greece, Italy and the UK. It relieson figures supplied to the OECD by national experts, includingfor Ireland results from the 1987 ESRI survey and the 1994 Livmgin Ireland survey. On this basis it shows Ireland as havingl6Vel^ T ? ? ° f *»****, as the UK in the mid-1990s,H ° WeVer>H I 1 J'*• ° ECD stud y «*» *ows that Ire-Umd has a higher level of inequality than non-EU countries suchOFrn f, 3 Canad3 'thou tl lower than* «* ^A. While theOECD study tried to harmonise the measurement proceduresIreland's Income Distribution in International Perspective 63adopted across countries, differences inevitably remain; one isthat while annual disposable income was the main focus, for anumber of countries — including Ireland — current income hadto be used. It therefore focused primarily on comparing inequalitytrends rather than levels.As far as income inequality is concerned, for the ten countriesfor which data were available from the mid-1970s to themid-1980s no general trends emerged, with inequality fallingor stable for more countries than it was increasing. In the periodfrom the mid-1980s to the mid-1990s, though, for whichdata on 20 countries was available, more of a common trend isapparent. Inequality increased in twelve countries — in half ofthese by considerable amounts — while it remained stable infour and decreased only slightly in another four. Ireland isshown in the study as belonging to this last group, registering aslight decrease in inequality. Several other aspects of thetrends shown are worth highlighting. The UK was the onlycountry displaying marked increases in inequality both fromthe mid-1970s to the mid-1980s and from the mid-1980s to themid-1990s. The USA saw a substantial increase in inequalityduring the earlier period, but the OECD's figures suggest littleor no further increase there by the mid-1990s. The notion thatthe UK's experience in particular represents a pattern whichother countries will necessarily experience in time thus seemsa highly partial reading of the evidence. However, the fact thatthe majority of countries experienced increases in inequality inthe later period does clearly suggest an underlying dynamic interms of economic forces, policy or both.The OECD study is particularly valuable in that it goes beyondshowing the extent to which trends in inequality wereshared across industrialised countries, to also explore the extentto which common factors were at work. It finds some veryimportant common features, but also many intriguing differencesover the mid-1980s to mid-1990s period. Perhaps themost notable common feature is that the share of earnings goingto the lower income groups among the working populationdecreased in all the countries covered in the study, and theshare going to middle income groups generally declined aswell. The same was true of market incomes generally, including

64The Distribution of Income in Irelandincome from self-employment and capital, but earnings dominatethat total. This was not, or not entirely, translated intohigher inequality of disposable incomes because both transfersand taxes off-set its effects, and indeed in most countries theredistributive impact of taxes and transfers increased over theperiod.The declining share of earnings or market income going tolower income groups here refers to total household earnings,with households ranked by equivalised disposable income. Itcould reflect both changes in the distribution of earningsamong earners, and in the distribution of earners across households.The OECD study does not look at the first of these in detail,being focused on households, but does have some interestinginsights on the second element. It shows that the proportionof households of working age with no earner increased inmost countries, and that such households had much lower levelsof income than those with one or two or more earners. Theproportion with two or more earners also increased in abouthalf the countries, so there was a quite widespread tendencytowards polarisation into work-rich and work-poor households.However, decomposition showed that the more important contributorto increasing inequality in household incomes was increasinginequality within fully-employed versus workless versus"mixed" households.Another important pattern common to many of the countriescovered in the study was an increase in the average incomes ofthe elderly towards the overall average. The main gains herewere for those aged between 66 and 74 rather than those aged75 and over, so it was recent retirees who were doing better.Also, inequality within the retirement-age population decreasedin a considerable number of countries, though publicold-age pensions tended to become less rather than moreequally-distributed among that group reflecting the importanceof earnings-related components in many countries. Nonpensiontransfers, on the other hand, tended to become moreequally distributed among the working-age population, generallyreflecting the impact of family cash benefits.IOQ^16 °f^D StUdy C ° VerS *" P eriod onl y U P to the mid-1990s, and thus does not include the period after 1994 when, asIreland's Income Distribution in International Perspective 65we have seen, inequality appears to have risen quite markedlyin Ireland. None the less, it provides an enormously valuablecomparative framework within which the Irish experience canbe set, and helps to bring out some of the factors which meritfurther investigation in the Irish case.5.5 CONCLUSIONSThis chapter has examined in some detail the distribution ofdisposable income in Ireland compared with other EU memberstates in the mid-1990s, using data from the EuropeanCommunity Household Panel survey. Ireland ranked as one ofthe more unequal in the European Union, along with the UKand Greece, though less so than Portugal. In the earlier studyby Atkinson, Rainwater and Smeeding based on data from themid-late 1980s, Ireland ranked among the most unequal in theOECD. The more recent data suggest rather that Ireland is oneof a group of EU countries with relatively high inequality,much higher than for example Denmark. This conclusion holdswhen one adjusts income for differences in household size andcomposition using equivalence scales. The equivalence scaleemployed was seen to make a difference to the level of theGini coefficient in some countries, but not the overall patternin terms of country groupings. In particular, from an Irish perspectiveit does not affect the placing of Ireland in a groupwith the UK, Greece and Spain, having the highest level ofinequality in the EU except for Portugal.As far as international trends in income inequality are concerned,a fairly widespread though not universal trend towardsincreased inequality in the period from the mid-1980sto the mid-1990s is found in a recent OECD comparative study.The most notable common underlying feature noted was thatthe share of earnings going to the lower income groupsamong the working population decreased in all the countriescovered in the study. This was not, or not entirely, translatedmto higher inequality of disposable incomes because bothtransfers and taxes off-set its effects, and indeed in manycountries the redistributive effects of taxes and transfers increasedover the period. The Irish data included in that study

66 The Distribution of Income in Irelandgoes up only to 1994, but it provides a valuable comparativecontext in which to see the Irish experience up to that pointand beyond.Chapter 6The Distribution of EarningsBrian Nolan6.1 INTRODUCTIONIn this chapter we move from the distribution of income amonghouseholds to the distribution of the largest element in income,namely employee's earnings, among those employees. Sharplywidening inequality in earnings in the United Kingdom and theUnited States in the 1980s and 1990s has been a major preoccupationin recent research on inequality in those countries, andhas been attributed much of the responsibility for rising inequalityin the distribution of total income among households.Attempts to explain this increasing earnings dispersion havehighlighted rising returns to education and skill. 1 This in turnhas been attributed to a shift in demand towards more skilledlabour due to factors such as skill-biased technical change (Kateand Murphy, 1992) and globalisation and competition from developingcountries (Wood, 1994). 2 Some industrialised countrieshave experienced much smaller increases in inequality,however, while others again have maintained stability in theirearnings distributions (OECD, 1993 and 1996b), which has fo-See for example Gosling, Machin and Meghir (1994) and Schmitt (1995) forfoe UK, Levy and Mumane (1992) and Juhn, Murphy and Pierce (1993) for theUSA.2US studies assessing such explanations include Bound and Johnson (1992),B °rjas andRamey (1994), andBurtless (1995).

68The Distributionof Income in IrelandThe Distribution of Earnings 69cused attention on the role of institutional factors such as centralisedwage bargaining.This international context brings to the fore the need to explorethe earnings distribution in Ireland and how it relates tothe structure of the overall household income distribution andtrends m that distribution. That is the aim of this chapter and thefollowing one. In this chapter the distribution of earnings andhow it evolved between 1987 and 1994, and again between 1994Mid 1997, is analysed. Chapter 7 then looks at the relationshipbetween the earnings distribution and the household incomedistribution, focusing on the way earners are grouped together inhouseholds and how that has been changing as the participationof married women in the paid labour force increases. This chapteris structured as follows. Section 6.2 describes the earningsdata to be used. Section 6.3 examines trends in the overall earningsdistribution in Ireland between 1987 and 1994. Section 6.4puts the earnings distribution in Ireland in comparative perspecfcveSection 6.5 looks at explanations for the observed trends inthe Irish earnings distribution between 1987 and 1994, notablychanging rates of return to different levels of education andsuT- m ^? eduCational ^oule of the labour force. Section 6.6summarises the conclusions.6.2 DATAThe datato be employed come once again from the householdw^es ofTe 6 ! ^by ESRI* " 1987 'and ** 1994 *** 1997hold L°l* t 9 m hBland Surve y- Now > r ^ r *an houseriesIn r ° m •" SOUrCeS OUr focus is6'on wages and salariesin T T £ ? * mformation was obtained for em-6y tms2 700 e„mLinf °nnation covered aboutdata^from i JT.L 11 h o u s e; ^ ^h °^- The correspondingS S r S 1 ^ ^ ^ Ireland Survev c °vers over 3,000"pplafto ^Pl ° yees - «* * W about 2,600. These samplesSi.ST"" f employees weU when

70The Distribution of Income in IrelandTable 6.1: Distribution of Earnings, Ireland 1987, 1994 and1997As Proportion of MedianAll employees, hourly earnings:Bottom decileBottom quartileTop quartileTop decileFull-time employees, weekly earnings:Bottom decileBottom quartileTop quartileTop decile19870.470.731.371.960.500.751.351.8219940.470.681.502.240.480.721.431.9719970.480.691.532.320.510.71This shows that from 1987 to 1994 there was a consistent wideningin dispersion for both weekly and hourly earnings, particularlyat the top of the distribution. The ratio of the top decileto the median rises from 1.96 to 2.24 for hourly earnings, andfrom 1.82 to 1.97 for weekly earnings among full-time employees.For hourly earnings the bottom decile is the same proportionof the median in 1987 and in 1994, but for weekly earningsamong full-time employees the bottom decile falls from 0.50 to0.48 of the median. The ratio of the top to the bottom decile,commonly used as a single summary inequality measure in thiscontext, rose from 4.2 to 4.8 for hourly earnings and for weeklyearnings of full-time employees the increase was from 3.6 toBetween 1994 and 1997, the top decile continued to movefurmer away from the median, reaching 2.32 for hourly earndeSlIn t 0tt ° m ^ ° f the **•*•** however, the bottomma^rX, T T* "** ** median if' ***** increasingZSZZL T * a reSUlt the'rati of the° *°P to the bottomlv dol atTn r 9 ^ ^ 48 f r hOUlly°eami^s' and m ~9in-C^h I T WeeWy 6arnmgS amon 9 **-*«» worker!SU^ZTJ,^Peri ° d fr ° m 1987 t0 i997 '1.422.02then - there was awages among all employees. This was more pronounced in theThe Distribution of Earnings 711987-94 period than from 1994 on, so rapid economic growthdid not lead to an acceleration in the trend. It was primarilydriven by relatively rapid increases for those towards the top ofthe distribution, with no indication - unlike for example the UKor the USA - that the bottom was falling behind the median.6.4 IRELAND'S EARNINGS DISTRIBUTION IN COMPARATIVEPERSPECTIVEA comparative perspective on the Irish earnings distribution,and on the way it has been changing, can be obtained usingmeasures of earnings dispersion for a range of developedcountries brought together by the OECD (1996b). There arepotentially important differences in definition and coverageacross countries (including whether earnings are weekly or annual),so these comparisons should be treated with extremecare, but they can serve to highlight some key features of theIrish results. Since they cover only up to the mid-1990s, we usethe Irish figures for 1994 and for trends between 1987-1994 forcomparative purposes.First, Table 6.2 shows measures of the level of earnings dispersionin 1994 for Ireland and other OECD countries, forweekly pay among full-time employees (since the figuresbrought together by the OECD generally refer to full-time employees,and to weekly, monthly or annual rather than hourlygross earnings). Ireland is seen to have a particularly high levelof earnings dispersion. Both the ratio of the top decile to themedian and of the median to the bottom decile are among thehighest of the countries covered. With the top decile/bottomdecile summary measure, only Canada and the USA are seen tohave greater earnings inequality than Ireland.As far as trends in earnings dispersion are concerned, Table6.3 shows the ratio of the top to the bottom decile in 1987 and1994 f °r Ireland and the other OECD countries for which thefigures are available for both points in time. We see that oncea gain Ireland is an outlier: the increase in earnings dispersionls the greatest of any of the countries shown. (The US is not includedin this table because OECD 1996 gives only US figures0rmen and women separately, but from these it appears that

72 The Distribution of Income in Irelandthe increase in earnings dispersion in the USA over the periodmay be similar in scale to that seen for Ireland.)Table 6.2: Summary Measures of EarningsIreland and Other OECD countries, 1994SwedenBelgium*Germany*FinlandNetherlandsSwitzerlandItaly*AustraliaJapanNew ZealandFranceUKAustriaCanadaUSInland*= 1993TopDecile/Median1.59*1.571.611.701.661.681.601.751.851.761.991.861.821.842.071.97Median/BottomDecile1.34*1.431.441.401.561.581.751.641.631.731.651.782.012.282.102.06Source: OECD (1996b), Table 3.1, p. 61-62, andTable 6.1 above.Dispersion,Top/BottomDecile2.132.242.322.382.592.652.802.873.023.053.283.313.664.204.354.06The Distribution of Earnings 73Table 6.3: Trends in Earnings Dispersion, Ireland andOther OECD Countries, 1987-1994Canada**Germany*Belgium*FinlandJapanSwedenAustraliaNetherlandsFranceUKNew Zealand**AustriaItaly*Ireland19874.442.542.442.523.152.092.812.533.193.202.923.472.423.67* = 1993 not 1994; ** = 1988 not 1987.Top Decile/Bottom Decile19944.202.322.242.383.022.132.872.593.283.313.053.662.804.06Source: OECD (1996b), Table 3.1, p. 61-62, andTable 6.1 above.Change-0.24-0.22-0.20-0.14-0.130.040.060.060.090.110.130.190.380.396.5 EXPLAINING THE INCREASE IN EARNINGS INEQUALITYWe have seen that the dispersion in earnings among employeesin Ireland is high by international standards, and rose relativelyrapidly between 1987 and 1994. From 1994 to 1997 theearnings distribution was rather more stable, though the topdid continue to draw away from the middle. We now attempt toidentify factors that may have been driving the increase inearnings inequality, concentrating on the 1987-1994 period anddrawing on results presented in Barrett, Callan and Nolan(1999).As mentioned earlier, sharply widening inequality in earningsin the United Kingdom and the United States in the 1980sand 1990s has been attributed to rising returns to education andskill. This in turn has been attributed to factors such as skill-

74 The Distribution of Income in Irelandbiased technical change and globalisation and competitionfrom developing countries. Here we look at returns to educationand changes in the educational profile of employees in theIrish case, and the extent to which this may have contributed toincreasing earnings dispersion. The fact that this increase inearnings inequality has not been experienced to anything likethe same extent in some other industrialised countries has alsofocused attention on the role of institutional factors such as centralisedwage bargaining and minimum wage legislation. Wetherefore also point to some relevant features of the Irish institutionalbackground.During the 1980s and into the 1990s, there has been a rapidrise in the level of educational attainment of those leaving theIrish education system. This reflects substantially higher numberscompleting full second-level education rather than leavingearly, and a marked increase in the proportion going on tothird-level education. As cohorts containing a high proportionwith relatively low levels of education retire and those withrelatively high levels enter the labour force, this has producedthe rather dramatic change in the education profile of employeesbetween 1987 and 1994 shown in Table 6.4. Whereas in1987 28 per cent of employees did not have a formal qualificationbeyond primary level, by 1994 this had fallen to 17 percent. This was accompanied by a rise in the percentage withpost-secondary school attainment levels from 18 per cent to 24per cent. We now go on to explore the relationship betweenthis changing education profile and the distribution of earnings.Table 6.4: Education Attainment Level, Ireland 1987 andHighest Qualification AchievedPrimary onlyJunior cycleLeaving Cert.Certificate/DiplomaJJegreeAllThe Distribution of Earnings 75Table 6.5 looks first at median earnings by education categoryand how this evolved between 1987 and 1994. The greatestpercentage increase over the period was for those with universitydegrees, for whom the median rose by 52 per cent, comparedwith 30-38 per cent for the other attainment levels. Thetable also looks at dispersion within education categories, interms of the ratio of the top to the bottom decile. This measureshows that the degree of dispersion was widest within the postschoolnon-degree category in each year, but rose considerablywithin the bottom and the top education categories between1987 and 1994.Table 6.5: Median Hourly Earnings and Ratio of Top toBottom Decile by Education Category, All Employees, 1987and 1994 (IR£ per hour)Primary onlyJunior cycleLeaving CertificateDiploma/other thirdlevelDegree'AiPMedian19873.753.834.355.488.404.30Median19945.005.285.687.2312.785.98%Increase33.337.930.632.952.139.1Top Decile/Bottom Decile19872.733.443.804.563.234.1619943.163.863.764.733.904.77With the balance between men and women in the workforcechanging, it is also interesting to look at male and female employeesseparately. The educational profile of male and femaleemployees changed in very much the same way as that for allemployees. Although a higher proportion of men than womenhad only primary-level qualifications, this declined very rapidlyfor both, and the percentage with post-school qualificationsalso rose substantially for both men and women. The relativelyrapid increase in median hourly earnings between 1987 and!994 for those with university degrees was also seen bothamong men and women, and the same is true of the consider-

76The Distribution of Income in IrelandThe Distribution of Earnings 77able increase in dispersion within that education category between1987 and 1994.It is also instructive to look within particular age ranges. Focusingon employees aged 25-39, once again there was a substantialshift in educational profile between 1987 and 1994. Thepercentage with only primary-level qualifications fell from 23per cent to only 9 per cent for this sub-set, and the percentagewith a university degree rose to 16 per cent. The higher levelsof educational attainment again show a relatively rapid rate ofincrease in median earnings. The overall dispersion in hourlyearnings has risen for this age range though more modestlyUian for all employees, with the ratio of the top to the bottomdecile going from 3.1 to 3.3. Dispersion has increased withinmost education groups, with the largest increase again forthose with university degrees.Such analyses of trends in earnings of employees crossciassifiedby sex, age and education category are useful, but*e influence of different factors, notably returns to education,can be distmguished more easily using estimated earningsequations. The extent to which wage premia for different education^qualifications changed over the period has been innotH;d - USm ? that a roach inPP Barrett, Callan and Nolan(1999). Using the 1987 and 1994 datasets, standard humana m^h9e < B T ti0nS 'baSSd ° n edu

78The Distribution of Income in Ireland6.6 CONCLUSIONSThis chapter has used data from household surveys to examinethe distribution of earnings among Irish employees in 1987,1994 and 1997. The results show that the dispersion in the Irishearnings distribution was relatively high by international standardsin 1994, and that it increased between 1987 and 1994 bymore than in almost any other OECD country for which data areavailable. This increase in dispersion was pronounced at thetop of the distribution, and was seen for hourly earnings amongall employees and for weekly earnings among full-time employees.Estimated wage equations showed that returns tohigher levels of education, especially university education, increasedover the period. Taken together, the combination ofthe change in the age-education profile of employees andhigher returns to education account for much of the observedincrease in dispersion, though precisely how much depends onttie specification of the wage equation and other aspects of thedecomposition methodology employed. Between 1994 and1997, when economic growth accelerated rapidly, the increasemeqUalit y slowedZ ^ T ^^ough the top of the distributiondid contmue to move away from the middleChapter 7Female Labour Force Participation andHousehold Income Inequality in Ireland1-1 INTRODUCTIONDonal O'Neill and OliveSweetmanOne of the most notable changes in the Irish labour market inrecent years has been the rapid growth in the number ofwomen working outside the home, in the paid labour force. Inthis chapter we examine the relationship between this trendand the distribution of income among households. While thescale and nature of women's labour force participation canhave a substantial impact on the household income distribution,it is not obvious a priori what this impact will be. Depending onow much those women earn, and on the type of householdom which they come, an increase in women's labour forceParticipation could be equalising or disequalising in terms ofthe household income distribution. We therefore explore empiricallywhat the impact is likely to have been in the Irish case,Jjsmg the period 1987-94 for this purpose. Section 7.2 describese data we use to carry out the analysis, Section 7.3 provides ae ailed description of changes in female participation rates ina household setting, and estimates the contribution theseanges have had on household income inequality. Section 7.4Presents conclusions.

80The Distribution of Income in Ireland7.2 WOMEN'S LABOUR FORCE PARTICIPATION IN IRELANDTo see the Irish situation in comparative context, Table 7.1shows female participation rates for various OECD countries inboth 1984 and 1994. In 1984 the female labour force participationrate in Ireland was 37 per cent. In contrast, many of theother countries had female participation rates of over 50 percent. For instance, the participation rate in the US was approximately63 per cent, 59 per cent in the U.K, 53 per cent in Germanyand 77 per cent in Sweden. Of the OECD countries in thetable, only Spain had a lower participation rate than Ireland. By1994 the participation rate in Ireland had increased to 47 percent. Table 7.1 shows that of the OECD countries for which dataare available only four — Luxembourg, Netherlands, New Zealandand Spain — experienced greater percentage increases inthe female participation rate than Ireland. These increases inparticipation and employment have continued since 1994. Indeed,the employment increases among Irish women between1991 and 1997 exceeded the combined employment increasesover the previous 20 years. Many have argued that this hascontributed substantially to Ireland's impressive growth recordsmce the early 1990's (see for example Bradley, FitzGerald,Honohan and Kearney 1997).To understand the effect of increasing female participationon household income inequality, one needs information showingwhere in the income distribution these changes have beensccumng, and also what has been happening to female wagesas women's participation and employment increased. For this?S?° Se iro e , USe *• *° house «°ld surveys carried out by theESRI m 1987 and 1994, described in detail in earlier chapters.at this point isfocus ^*• household. However, tohn,,.^0^r ° le ° f female Participation we concentrate onhouseholds containing a married couple in which both spousesb0th mus J£Ti?"a 9 ed betwe *n 24 and 55. This leavesus with 1,546 households in 1987, and 1,855 households in 1994.Female Labour Force Participation and Inequality 81Table 7.1: Female Participationand 1994AustraliaAustriaBelgiumCanadaDenmarkFinlandFranceGermanyGreeceIcelandIrelandItalyJapanKoreaCountryLuxembourgNetherlandsNew ZealandNorwayPortugalSpainSwedenSwitzerlandUKUSRates in the OECD, J984Female Participation Rate (%)198452.851.548.763.573.872.954.752.340.962.736.940.757.243.842.240.746.066.356.033.277.355.759.162.8199463.462.155.167.873.869.959.661.844.680.047.242.962.152.756.557.465.071.162.044.174.467.566.270.5Change inParticipationRate (% points)II ls useful now todistinguishocusing on these households, between the two surveys therea significant increase in the percentage of married women^orking as employees. In 1987 approximately 22 per cent of^ves were classified as employees, but by 1994 this had infouT8601 tQ 34 Per C6nt " betweensources of household income: husband's employee earn-+20+20+ 13+70-5+8+ 18+9+28+28+6+9+20+34+41+41+7+11+33-4+21+ 12+ 12

82 The Distribution of Income in Irelandings, wife's earnings, other earnings (accruing to other householdmembers) and all other income (income from selfemployedincome (including farming), from capital, occupationalpensions, and state transfers). We see from Table 7.2 thatthe increase in employment rates for women was also reflectedin the share of household income accounted for by femaleearnings, which increased from under 12 per cent of total grosshousehold income in 1987 to 15 per cent in 1994. In the nextsection we examine these trends in more detail and in particulartheir impact on household income inequality, with this categorisationof income allowing us to highlight the role of wives'earnings versus other income sources.Table 7.2: Components of Gross Household Income (householdswith couples aged 24-55) 1987-1994Man's earningsWoman's earningsEarnings of othersOther incomeShare of Total Household Income (%)738750.311.77.630.3199447.815.08.229.07.3 FEMALE LABOUR SUPPLY AND HOUSEHOLD INEQUALITYWe saw in the previous chapter that the distribution of earningsamong all Irish employees became more widely dispersedbetween 1987 and 1994. This was also true for the married menemployees in the sub-sample of households on whom we are nowfocusing. Their ratio of top to bottom decile (weekly earnings)went up from 2.9 to 3.3.' For married women employees, dispersionalso went up. This is in contrast to the trend in total householdincome inequality for the sub-sample, which has fallenFemale Labour Force Participation and Inequality 83slightly over this period, consistent with the pattern described inChapter 4 for the entire sample.The relationship between the participation and wages ofwomen and their husbands' economic status will clearly havean important role to play in determining patterns of householdinequality. Table 7.3 shows that the strength of the associationbetween the husband's and the wife's earnings, as summarisedby the correlation between them, increased substantially overthis period.Table 7.3: Correlation between Earnings of MarriedCouples, 1987 and 1994Correlation between Husband andWife's Weekly EarningsAll HouseholdsHouseholds with Working Wives19870.050.031994To examine what lay behind this change, we can first look atwhich women moved into employment, in terms of their husband'slabour force status and position in the earnings distribution.Table 7.4 shows that in 1987 employment rates (i.e. thepercentage working as an employee) were higher for womenmarried to husbands with earnings in the top half of the earningsdistribution than those on lower earnings. However, it alsoshows that while wives' employment rates increased throughoutthe male earnings distribution, the bulk of the change inemployment rates was concentrated among women married tomen with below average earnings. The only ones not to experiencean increase in employment rates were women married tounemployed men, whose employment rate fell from 20 per centto 17 per cent. 2 As mentioned in Chapter 5, such a concentrationof employment in work-rich versus work-poor householdshas been noted in various countries.0.120.14Although the results are presented here for the ratio of the top decile to thebottom decile of the earnings distribution the same conclusions hold when weuse alternative measures such as the Gini coefficient or Atkinson's measure ofinequality.2For more information on the labour supply behaviour of womenunemployed men see Doris (1998).

84 The Distribution of Income in IrelandTable 7.4: Female Employment Rate by Male Earnings,1987-1994Husband's Position inEarnings DistributionHusband an employee in theFirst DecileSecond DecileThird DecileFourth DecileFifth DecileSixth DecileSeventh DecileEight DecileNinth DecileTenth DecileBelow MedianAbove MedianFemale EmploymentRate1987 199424.616.618.620.817.826.719.028.331.723.419.425.7Husband unemployed 20.130.845.745.043.743.247.542.042.938.128.541.739.816.7% Change1987-1994+25+ 175+ 141+ 110+143+78+121+52+20+22+115This change in the pattern of female employment might in itselfbe expected to reduce household income inequality. To get acomplete picture of the contribution of female earnings tohousehold income inequality, though, we also need to knowwhat was happening to the level of earnings of women in differenthouseholds over the 1987-1994 period. Compared to theway participation rates have evolved, this in fact reveals a verydifferent story. The greatest wage increase was among workingwomen married to men with above average earnings, for whomreal wages increased by 17 per cent. In contrast, for womenmanned to men with below average earnings, real wages increasedby only 2 per cent. Women employees married to unemployedmen actually saw their real weekly wage fall over thisperiod (though this is based on relatively few observations). 33^Lne^ l°, be CarefUl* """Preting *e wage figure for women marriedto unemployed men as the estimates are imprecise. This reflects the small+55-17Female Labour Force Participation and Inequality 85These wage changes contributed substantially to the increase inthe correlation between spouses' earnings over the period.If the growth in part-time work was more prevalent amongworkers at the lower end of the distribution, the fall in realweekly wages for women married to unemployed men couldthen simply reflect an increasing proportion of these workersworking shorter weeks. However, the same pattern emergeswhen we look at hourly wages, with the real hourly wage forwomen married to high-earning husbands going up 30 percent, compared to 9 per cent for women married to low earninghusbands and once again a decline for women married to unemployedmen. The female earnings distribution itself becamemore unequal, with the ratio of the top to bottom decile of thefemale wage distribution increasing from 9.3 to 10.4 between1987 and 1994." The impact of increasing participation meansthat if we look at all wives, by contrast, we see a reduction indispersion because there are significantly fewer with no earningsat all.This analysis identifies two channels whereby recent trendsin female labour supply and wages would have worked to actuallyincrease household income inequality. The first is that thewage growth among females has been most pronouncedamong women married to high earning husbands, so the correlationbetween male and female wages has increased. Thisincreased correlation would tend to increase household incomedispersion, other things fixed. Secondly, the distribution of femalewages tends to be much more dispersed than that of maleworkers. As female earnings becomes a more important part ofhousehold income the greater dispersion exhibited within thedistribution of female earnings will tend to feed into the distributionof household income. The changing pattern of femalesample sizes on which the wage changes for this category of worker are calculated.A possible explanation for the faster increase in earnings among womenmarried to men with above average earnings is the increase in returns toeducation which have occurred over this period in Ireland, as described inChapter 6. Given the tendency for couples of similar education levels tomarry we might expect the rise in return to skill to be reflected in fasterearnings growth for women married to high income men.

86The Distribution of Income in Irelandemployment, on the other hand, with a greater increase in employmentfor women with low-earning rather than high-earninghusbands, might in itself be expected to reduce household incomeinequality.It is interesting to contrast the Irish experience with whathas been happening in the US, where Juhn and Murphy (1997)examined changes in female earnings and employment between1969 and 1989. Their analysis shows that the trend in femalewage behaviour in the US is similar to Ireland, with thelargest gains being experienced by women married to highearning husbands. However, the employment behaviour overthe period is the opposite of what we have just seen for Ireland:the highest employment rates were initially found amongwomen married to low income men, but over time the largestincreases in employment have been among women married tohigh earning husbands. Thus, in the US both the wage changeeffects and the participation effects for females operated insuch a way as to result in an increase in household income inequality.As we have shown for Ireland these forces operated inopposite directions between 1987 and 1994, and we must turnto a more detailed analysis in order to determine which of thetwo had the greater bearing on inequality.Total household income can be disaggregated into its componentsin order to determine their individual impacts on inequality,allowing us to identify the effect of wives' earnings andemployment on inequality and how it has been changing. Thecoefficient of variation is a summary inequality measure whichis particularly suited to this purpose, and we have used it to lookat ttie period from 1987 to 1994: the details of the methodologyT J T T T Sh ° Wn * Appendbc L »•resul * *WSt that,despite the fact that the correlation in spouses' earnings has in-T ^StantiaUy ° Ver •* period ' *» evolution of wives'fn^ 5 f ? equalising effect on the distribution of householddb«Z * ° Ver "* Peri ° d - *• leased correlationbetween the earnings of spouses would in itself have had a disequahsingeffect. However, its impact was outweighed by the retZl2*^ f ° n * eaJnmgS aCTOSS * "^ed women (as~LnTtr m P ° yed **»»*>> «* fa y * reduction in the correlationbetween wives' earnings and non-labour income ThisFemale Labour Force Participation and Inequality87brings out the complexity of the relationships between individualand household incomes, and the need to carefully disaggregatethe different sources of income accruing to household membersand the ways in which they relate to each other.7.4 CONCLUSIONBetween 1987 and 1994 wage inequality in Ireland increasedsubstantially yet household income inequality, as reflected inthe ESRI surveys carried out in those years, was stable or actuallyfell marginally. In this chapter we have examined one possibleexplanation for this, namely the increased contribution offemale wages to total household income. We have documentedthe changes in wives' employment rates and earnings over thisperiod, paying particular attention to how these varied with theeconomic status of the husband.We saw that while increases in female employment rateswere greatest among wives married to low earning husbands,these women have experienced only modest wage gains whencompared to the wives of husbands with earnings above theaverage. These changes have resulted in a reduction in the dispersionof earnings among all wives (the participation effect)but an increase in dispersion among wives at work and an increasein the correlation between spouses' earnings (bothwage effects).We then used a decomposition of the coefficient of variation,a summary measure of inequality, to examine the contributionof wives' earnings to changes in household income inequalityover the period. The results suggest that the effect of the highercorrelation in spouses' earnings (which would in itself work toincrease inequality) was dominated by the other trends associatedwith wives' earnings. Overall, changes in wives' earningswould have had an equalising rather than disequalising impacton household incomes over the period.Although we have shown that trends in female participationrates have reduced income inequality over the 1987-94 period,one must be careful in extrapolating these findings from thatpoint on. Despite the rapid increases in female participation,female earnings still accounted for only 15 per cent of total

88 The Distribution of Income in Irelandhousehold income in 1994. Given the continued increase inwomen's labour force participation, this share is set to grow inimportance and as it does one would expect the relative importanceof female-specific factors for household inequality willalso increase. Chapter 8ConclusionsBrian NolanOverall income inequality and inequality in the distribution ofearnings have risen sharply during the 1980s and 1990s in anumber of industrialised countries, giving rise to widespreadconcern about the factors at work and about the societal implications.This makes it particularly important to know how thedistribution of income in Ireland has been changing over time,how it compares with other countries, and what factors contributeto explaining Ireland's particular experience. This study hasaddressed these questions, using household survey data.These data allowed us to first provide a picture of the distributionof household income in Ireland in 1994, 1997 and 1998,then to compare these with similar figures for 1987 and earlieryears. These figures for Ireland were also compared with estimatesfor other countries. The evolution of the distribution ofearnings among individual earners, a major factor behind increasinginequality elsewhere, was analysed. Finally, the impactof changes in the extent of women's participation in thepaid labour force was assessed.A key finding from the analysis of data for the 1990s from theLiving in Ireland surveys was that there was a marked shift inthe disposable income distribution away from the bottom 30Per cent. Over the period from 1994 to 1998, and adjusting incomefor differences in household size and composition, theshare of the bottom 30 per cent of households declined by 1.4Per cent of total income.

90 The Distribution of Income in IrelandThe distribution among households of income coming directlyfrom the market did not become more unequal over theperiod; instead, an important factor was that social welfaretransfers, though having an equalising effect in each year, hadmore impact in 1994. The period between 1994 and 1997 alsosaw a considerable change in composition both at the top andbottom of the distribution, with younger households moving upand older ones moving down.Over the period from 1973-87, on the other hand, inequalityin the distribution of disposable household income fell markedly,with the share of the top decile down by 1.4 per cent oftotal income and that of the bottom quintile up by 0.9 per cent.The share of the top decile fell both from 1973 to 1980 and from1980 to 1987, but the increase for the bottom decile was in thelatter period. An important factor at work over this period wasthe increasingly redistributive impact of income tax and employees'social insurance contributions, reflecting both an increasein progressivity and a very substantial rise in the averagetax rate. From 1987 to 1994 this continued but at a muchslower rate, helping to explain the greater stability in the shapeof the distribution over those years.Using data from the European Community Household PanelSurvey, Ireland had one of the more unequal income distributionsin the EU in the mid-1990s. In the earlier study by Atkinsonet al, based on data from the mid-late 1980s, Ireland rankedamong the most unequal in the OECD. The more recent datasuggest rather that Ireland is one of a group of EU countries -the others being the UK, Greece and Spain - with relatively highinequality, though not as high as Portugal. This conclusion holdswhen one adjusts income for differences in household size andcomposition using equivalence scales. The equivalence scaleemployed was seen to make a difference to the level of the Ginicoefficient in some countries, but not to the overall pattern interms of country groupings.As far as international trends in income inequality are concerned,a fairly widespread though not universal trend towardsincreased inequality in the period from the mid-1980s to themid-1990s was found in a recent OECD comparative study. Themost notable common underlying feature noted was that theConclusionsshare of earnings going to the lower income groups among theworking population decreased in all the countries covered in thestudy. This was not, or not entirely, translated into higher inequalityof disposable incomes because both transfers and taxesoff-set its effects, and indeed in many countries the redistributiveeffects of taxes and transfers increased over the period.Data from the ESRI household surveys was also used to examinethe distribution of earnings among Irish employees in1987, 1994 and 1997. The dispersion in Irish earnings was relativelyhigh by international standards in 1994, having increasedfrom 1987 by more than most other OECD countries for whichdata are available. The combination of changes in the ageeducationprofile of employees and higher returns to educationaccounted for much of that increase in dispersion. Between 1994and 1997, when Irish economic growth accelerated rapidly, theincrease in earnings inequality slowed although the top of thedistribution did continue to move away from the middle.The participation of married women in the paid labour forcehas been increasing rapidly in Ireland, so their wages havebeen accounting for a growing proportion of household income.Elsewhere, this has been seen to increase household incomeinequality. That was found not to be the case for Irelandbetween 1987 and 1994, indeed the overall impact of increasedparticipation was seen to have an equalising effect. This wasmostly because increases in female employment rates weregreatest among wives married to husbands with relatively lowearnings. There is no guarantee that this has continued to bethe case as married women's labour force participation hascontinued to rise, but this finding serves to demonstrate thecomplexity of the relationship between trends in the labourmarket and the household income distribution.Much remains to be done before we fully understand howme shape of the Irish income distribution has been evolvingand the nature of the forces producing that distribution. Thisstudy should be seen as providing one of the building-blockson which such an understanding can be built.91

Appendix 1Decomposition of Total HouseholdIncome Inequality by FactorComponents, 1987-1994Chapter 7 made reference to results from the disaggregation oftotal household income into its individual components in orderto identify the effect of wives earnings and employment on inequality.This appendix sets out the details of the decompositionmethodology employed and the results. 1The issues associated with decomposing total inequality byincome components are examined in Shorrocks (1982a,b). Heshows that by appropriate choice of a weighting function onecan find alternative decompositions of a given inequality indexwhich yield vastly different conclusions concerning the importanceof a given component. In fact, the contribution of anyfactor expressed as a proportion of total inequality can bemade to take any value between plus and minus infinity. Furthermore,he shows that there are no strong statistical reasonsfor choosing any one of these decompositions over the other.Shorrocks argues that a potential means of choosing betweenthe multiplicity of outcomes is to focus on what is normallymeant by statements of the form "factor X contributes Z percentof total inequality".Canican and Reed (1998) develop this idea further by comparingtwo common inequality indices, the coefficient of variationand the Gini coefficient. They argue that the standard de-1 For other studies using a similar approach to examining income» "J^JjJJsee Layard and Zabalza (1979), Canican, Danzinger and Gottschalk (1993)and Machin and Waldfogel (1994).

94 The Distribution of Income in Irelandcomposition of the Gini coefficient has no implicit referencedistribution and therefore should not be interpreted as a measureof the effect of an income source on inequality, and that decompositionsbased on the Gini coefficient are not suitable toanalysing changes in inequality over time. They analyse suchchanges using the coefficient of variation, developing several"thought experiments". Here this approach is applied to thecontribution of wives' earnings to household income inequalitybetween 1987 and 1994.The first experiment was to compare the observed inequalityin household incomes in 1987 to the inequality that wouldhave been seen if the distribution of wives' earnings hadchanged to the 1994 pattern, but all other income componentshad stayed fixed at their 1987 levels. The second experimentwas to compare the actual distribution in 1994 with what thelevel of inequality would have been if wives' earnings hadstayed at their 1987 levels. The results using these two alternativecounterfactuals may differ because the base year valuesdiffer between the two — a common problem in other areas, forexample using base-year versus end-year weights in constructingindex numbers. As in those contexts, here we deriveboth sets of results and see if they show the same broad pattern.The results for both decompositions are given in TableAl.l.Appendix 1on the distribution of household income. The reduction in dispersionin earnings across all married women (reflecting increasingnumbers employed), as well as a reduction in the correlationbetween wives' earnings and non-labour income, wereequalising in terms of the household income distribution. Thesewere large enough to outweigh the effect of the increased correlationbetween the earnings of spouses, which in itself wouldhave worked in the opposite direction.95Table Al.l: Decomposition of Changes in the Coefficient ofVariation for Gross Household Income, 1987-1994Observed 19871987 with women's earnings at 1994 levels1994 with women's earnings at 1987 levelsObserved 1994Coefficient of Variation0.6780.671/0.663*0.656/0.662*0.644» * * » »*»« «*«*•• * ** conelationsZ^lnT^!!^! between income sources11 Ae-These results suggest that, despite the fact that the correlationin spouses earnings has increased substantially over this period,the evolution of wives' earnings had an equalising effect

Appendix 2Table A2.1: Decile Shares in Equivalised DisposableIncome for Irish Households, 1994 and 1997 LII Surveys(1/0.66/0.33 Scale)DecileBottom23456789TopAllGiniTheilShare in Total Equivalised(1/0.66/0.33) Disposable Income(%)1994 UI3.94.85.46.17.18.710.512.715.925.0100.0Inequality Measure0.3260.1841997 UI3.64.65.26.17.59.010.713.015.924.6100.00.3290.185

98 The Distribution of Income in IrelandTable A2.2: Position in the Income Distribution of PersonsCategorised by Age of Household Head, 1994 and 1997 HISurveysQuintileA: 1994Bottom234TopAllB: 1997Bottom234TopAllPosition in Equivalised (1/0.66/0.33) DisposableIncome Distribution (%)Head Agedunder 3526.511.314.921.026.2100.018.412.218.619.431.5100.0Head Aged3S-6420.418.020.621.419.7100.019.919.221.020.619.3100.0Head Aged 65or Over9.941.322.013.213.6100.022.033.716.918.88.6100.0Appendix 99Table A2.3: Position in the Income Distribution of PersonsCategorised by Labour Force Status of Household Head,1994 Ln SurveyQuintileA: 1994Bottom234TopAllB: 1997Bottom234TopAllEmployee4.510.023.031.131.5100.04.712.525.028.429.3100.0Position in Equivalised (1/0.66/0.33) DisposableIncome Distribution (%)Selfemployed17.411.021.320.829.4100.016.714.415.718.235.0100.0Farmer22.717.223.419.916.8100.017.625.024.217.318.9100.0Unemployed61.624.69.43.31.1100.068.024.84.52.20.5100.0Retired10.233.326.115.315.0100.015.730.918.622.312.S100.0Homeduties32.843.012.57.34.4100.046.436.110.95.01.6100.0

ReferencesAtkinson, A.B., L. Rainwater and T. M. Smeeding (1995). Income Distributionin OECD Countries: Evidence from the Luxembourg IncomeStudy, Paris: OECD.Barrett, A., T. Callan and B. Nolan (1999). "Returns to Education in theIrish Youth Labour Market", Journal of Population Economics Vol. 12,313-326.Borjas, G. and V. Ramey (1994). "Time-Series Evidence on theSources of Trends in Wage Inequality", American Economic Review,Papers and Proceedings, 84, 10-16.Bound, J. and G. Johnson (1992). "Changes in the Structure of WagesDuring the 1980s: An Evaluation of Alternative Explanations",American Economic Review, 82, 371-92.Bradley, J., J. FitzGerald, P. Honohan, and I. Kearney (1997)."Interpreting the Recent Irish Growth Experience", in Duffy, D., J.FitzGerald, I. Kearney and F. Shortall, (edsj, Medium-Term Review:1997-2003, Dublin: The Economic and Social Research Institute.Buhman, B., L. Rainwater, G. Schmaus and T. Smeeding (1988)."Equivalence Scales, Well-being, Inequality and Poverty: SensitivityEstimates Across Ten Countries Using the Luxembourg Income StudyDatabase", Review of Income and Wealth, Series 34: 115-42.Burtless, G. (1995). "International Trade and the Rise in EarningsInequality", Journal of Economic Literature, 33(2), 800-816.

102 The Distribution of Income in IrelandCallan, T. (1991a). Income Tax and Welfare Reform: MicrosimulationModelling and Analysis, General Research Series No. 154, Dublin: TheEconomic and Social Research Institute.Callan, T. (1991b). "Male-Female Wage Differentials in Ireland",Economic and Social Review, 23, 55-72.Callan, T. and C. Harmon (1997). 77ie Economic Return to Schooling inIreland, Department of Economics Working Paper WP97/23, Dublin:University College.Callan, T., R. Layte, B. Nolan, D. Watson, C.T. Whelan, J. Williams andB. Maitre (1999). Monitoring Poverty Trends, Dublin: Stationery Office.Callan, T. and B. Nolan (1994). "Unemployment and Poverty", in B.Nolan and T. Callan (eds), Poverty and Policy in Ireland, Dublin: Gilland Macmillan.Callan, T. and B. Nolan (1997). "Income Inequality and Poverty inIreland in the 1970s and 1980s," in Gottschalk, P., B. Gustafsson and E.Palmer (eds.)77ie Distribution of Economic Welfare in the 1980s: AnInternational Perspective, Cambridge: Cambridge University Press.Callan, T. and B. Nolan (1999). "Income Inequality in Ireland in the1980s and 1990s," in F. Barry (ed.) Understanding Ireland's EconomicGrowth, Basingstoke: Macmillan.Callan, T., B. Nolan, and B.J. Whelan, D.F. Hannan, with S. Creighton(1989). Poverty, Income and Welfare in Ireland, General ResearchSeries No. 146, Dublin: The Economic and Social Research Institute.Callan, T., B. Nolan, B.J. Whelan, C.T. Whelan and J. Williams (1996).Poverty m the 1990s: Evidence from the 1994 living in Ireland Survey,Dublin: Oak Tree Press.Callan^ T. and H. Sutherland (1997). "Income Supports in Ireland andw>^ UK , in T. Callan (ed.), Income Support and Work Incentives:Ireland and the UK, Policy Research Series Paper No. 30, Dublin: TheEconomic and Social Research InstituteReferences 103Callan, T. and A. Wren (1994). Male-Female Wage Differentials:Analysis and Policy Issues, General Research Series Paper No. 163,Dublin: The Economic and Social Research Institute.Cantillon, S. and B. Nolan (1998). "Are Married Women MoreDeprived than their Husbands?", Journal of Social Policy, 27, 2, 151-171.Cantillon, S. and B. Nolan (2000). "Can Gender Differences in PovertyWithin Households Be Measured Using Non-Monetary Indicators?",Feminist Economics, Vol. 6, No. 3.Cantillon, B., I. Marx, D. Prost and R. Van Dam (1994). IndicateursSociaux: 1985-1992, Antwerp: Centre for Social Policy, University ofAntwerp.Canican, M., S. Danzinger and P. Gottschalk (1993). "Working Wivesand Family Income Inequality Among Married Couples", in Danziger,S. and P. Gottschalk (eds.) Uneven Tides: Rising Inequality in America,New York: Russell Sage Foundation.Canican, M. and D. Reed (1998). "Assessing the Effects of Wives'Earnings on Income Inequality", Review of Economics and Statistics,LXXX, 73-89.Central Statistics Office (1980). Household Budget Survey 1973,Detailed Results for All Households, Dublin: Stationery Office.Central Statistics Office (1982). Household Budget Survey 1980,Detailed Results for All Households, Dublin: Stationery Office.Central Statistics Office (1989). Household Budget Survey 1987,Detailed Results for All Households, Dublin: Stationery Office.Central Statistics Office (1995). Household Budget Survey 1994-95,Detailed Results for All Households, Dublin: Stationery Office.Coulter, F., F. Cowell and S.P. Jenkins (1992). "Equivalence ScaleRelativities and the Extent of Inequality and Poverty", EconomicJournal, 102, 1067-1082.

104 The Distribution of Income in IrelandCowell, F. (1995). Measuring Inequality, LSE Handbooks onEconomics, Hemel Hempstead: Prentice Hall/Harvester Wheatsheaf.Doris, A. (1998). An Analysis of the Labour Supply of British Women totheir Husband's Unemployment, Unpublished Ph.D. Thesis, Florence:European University Institute.Eurostat (1999) European Community Household Panel (ECHP):Selected Indicators from the 1995 Wave, Luxembourg: Office forOfficial Publications of the European Communities.Forster, M. (2000). Trends and Driving Factors in Income Distributionand Poverty in the OECD Area, Labour Market and Social PolicyOccasional Papers No. 42, Paris: OECD.Goodman, A., P. Johnson and S. Webb (1997). Inequality in the UK,Oxford: Oxford University Press.Gosling, A., S. Machin and C. Meghir (1994). "What Has Happened toMen's Wages since the mid-1980s?", Fiscal Studies, 15, 63-87.Hagenaars, A., K. de Vos and M.A. Zaidi (1994). Poverty Statistics in theLate 1980s: Research Based on Micro-data, Luxembourg: Office forOfficial Publications of the European Communities.Jenkins, S.P. (1991). "The Measurement of Income Inequality", inOsberg, L. (ed.J, Economic Inequality and Poverty: InternationalPerspectives, New York: Sharpe.Jenkins, S.P. (1995). "Easy Estimation Methods for Discrete-Time DurationModels", Oxford Bulletin of Economics and Statistics, 57 (1), 129-Juhn, C, K. Murphy and B. Pierce (1993). "Wage Inequality and theRise m Returns to Skill", Journal of Political Economy, 101 (3), 410-42.Juhn, C. and K. Murphy (1997). "Wage Inequality and Family LabourSupply", Journal of Labor Economics, 15, 72-97.J S « and K MUIphy'(1992)> " Chan ges in Relative Wages, 1963-3nd Demand107(1) 3 ^Factors ". Quarterly Journal of Economics,References 105Layard, R. and A. Zabalza (1979). "Family Income Distribution:Explanation and Policy Evaluation", Journal of Political Economy, 87, 5,Part 2, 133-61.Layte, R., B. Maitre, B. Nolan, D. Watson, J. Williams and B. Casey(2000). Monitoring Poverty Trends: Results from the 1998 Living inIreland Survey, Working Paper, Dublin: The Economic and SocialResearch Institute.Levy, F. and R. Murnane (1992). "US Earnings Levels and EarningsInequality: A Review of Recent Tremds and Proposed Explanations",Journal of Economic Literature, 30(3), 1333-81.Lundberg, S., R.A. Pollak, and T.J. Wales (1997). "Do Husbands andWives Pool Their Resources?", Journal of Human Resources, 32 (3),463-80.Machin, S. and J. Waldfogel (1994). 77ie Decline of the MaleBreadwinner: Changing Shares of Husbands' and Wives' Earnings inFamily Incomes, STICERD Working Paper No. WSP/103. London:London School of Economics.Murphy, D. (1984). "The Impact of State Taxes and Benefits on IrishHousehold Incomes", Journal of the Statistical and Social Inquiry Societyof Ireland, Vol. XXV: 55-120.Murphy, D. (1985). "Calculation of Girti and Theil InequalityCoefficients for Irish Household Incomes in 1973 and 1980", Economicand Social Review, Vol. 16: 225-249.National Economic and Social Council (1990). A Strategy for theNineties: Economic Stability and Structural Change, Report No. 89,Dublin: National Economic and Social Council.Nolan, B. (1978). "The Personal Distribution of Income in the Republicof Ireland", Journal of the Statistical and Social Inquiry Society ofIreland, vol. XXIII: 91-139.Nolan, B. (1981). "The Redistribution of Household Income in Irelandby Taxes and Benefits", Economic and Social Review, 13 (1), 59-88.

106 The Distribution of Income in IrelandNolan, B. (1986). "The Distribution of Social Security Transfers in theUK"', Journal of Social Policy, 15(2), 185-204.Nolan, B. (1998). Low Pay in Ireland, Volume II of the Report of theNational Minimum Wage Commission, Dublin: Stationery Office, 1998.Nolan, B. and T. Callan (1989). "Measuring Trends in Poverty OverTime: Some Robust Results for Ireland 1980-1987", Economic andSocial Review, 20, 309-328.Nolan, B. and T. Callan (eds.) (1994). Poverty and Policy in Ireland,Dublin: Gill and Macmillan.Nolan, B., R.M. Hauser and J.P. Zoyem (2000). "The Changing Effectsof Social Protection on Poverty", in Gallie, D. and S. Paugam (eds.),European Welfare Regimes and the Experience of Unemployment, Oxford:Oxford University Press.Nolan, B. andD. Watson (1998). Women and Poverty in Ireland, Dublin:Oak Tree Press.Nolan, B., C. Whelan and J. Williams (1998). Where Are Poor Households?The Spatial Distribution of Poverty and Deprivation in Ireland,Dublin: Oak Tree Press.Nolan, B., G. Boyle, T. Callan, A. Dorris, I. Kearney, J. FitzGerald, S.Machin, D. O'Neill, J. Walsh, J. Williams, B. McCormick and D. Smyth(1999). The Impact of the Minimum Wage in Ireland, published in FinalReport of the Inter-Departmental Group on the Implementation of aNational Minimum Wage, Dublin: Stationery Office.O'Donnell, R. and C. O'Reardon (1996). "The Irish Experiment", NewEconomy, 3(1), 33-38.D- and °-Sweetnam°' N 7_ m '(^SS) Poverty and Inequality in Ireland,1987 1994: A Comparison using Measures of Income and Consumption,Department of Economics Working Paper, Maynooth: NUI MaynoothOrganisation for Economic Co-Operation and Development, (1990).Employment Outlook, Paris: OECD.References 107Organisation for Economic Co-Operation and Development (1990).Revenue Statistics, Paris: OECD.Organisation for Economic Co-Operation and Development (1993).Employment Outlook, Paris: OECD.Organisation for Economic Co-Operation and Development (1996a).Education at a Glance: OECD Indicators, Paris: OECD.Organisation for Economic Co-Operation and Development (1996b).Employment Outlook, Paris: OECD.Roche, J. (1984). Poverty and Income Maintenance Policies in Ireland,Dublin: Institute for Public Administration.Rottman, D. and M. Reidy (1988). Redistribution Through State SocialExpenditure in the Republic of Ireland 1973-1980, Report No. 85,Dublin: National Economic and Social Council.Schmitt, J. (1995). "The Changing Structure of Male Earnings inBritain, 1974-88", in Freeman, R. and L. Katz (eds.), Differences andChanges in Wage Structures, Chicago: University of Chicago Press.Shorrocks, A.F. (1980). "The Class of Additively DecomposableInequality Measures", Econometrica, 48, 613-625.Shorrocks, A.F. (1982a). "Inequality Decomposition by FactorComponents", Econometrica, 50, 193-211.Shorrocks, A.F. (1982b). "The Impact of Income Components on theDistribution of Family Income", Quarterly Journal of Economics, 98,311-326.Shorrocks, A.F. (1983). "Ranking Income Distributions", Economics,50, 3-17.Shorrocks, A.F. (1984). "Inequality Decomposition by Population Sub-Groups", Econometrica, 52, 1369-1385.Wood, A. (1994). North-South Trade, Employment and Inequality:Changing Fortunes in a Skill-driven World, Oxford: Clarendon Press.

directions W*OAK TREE PRESS D U B L I NISBN 1-86076-208-5