Mapping Poverty - Combat Poverty Agency
Mapping Poverty - Combat Poverty Agency
Mapping Poverty - Combat Poverty Agency
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
RESEARCH<br />
MAPPING POVERTY: NATIONAL, REGIONAL<br />
AND COUNTY PATTERNS<br />
DOROTHY WATSON, CHRISTOPHER T. WHELAN,<br />
JAMES WILLIAMS, SYLVIA BLACKWELL
AA.Prelims 7/7/05 6:53 am Page iii<br />
<strong>Mapping</strong> <strong>Poverty</strong>:<br />
National, Regional and<br />
County Patterns<br />
Dorothy Watson<br />
Christopher T. Whelan<br />
James Williams<br />
Sylvia Blackwell
AA.Prelims 7/7/05 6:53 am Page iv<br />
First published 2005<br />
by the<br />
Institute of Public Administration<br />
57–61 Lansdowne Road<br />
Dublin 4<br />
and<br />
<strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong><br />
Bridgewater Centre<br />
Conyngham Road<br />
Islandbridge<br />
Dublin 8<br />
© 2005 <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong><br />
British Library Cataloguing-in-Publication Data<br />
A catalogue record for this book is available from the British Library.<br />
ISBN 1 904541 25 9<br />
This study forms part of the <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong> Research Series, in which it<br />
is No. 37.<br />
The views expressed in this text are the authors’ own and not necessarily those of<br />
<strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong>.<br />
Publications and printed matter will be made available, on request, in a range of<br />
formats, including audio tape, large print, Braille and computer disc.<br />
Cover design by Red Dog Design Consultants, Dublin<br />
Typeset by Computertype Limited, Dublin<br />
Printed in Ireland by Betaprint, Dublin
AA.Prelims 7/7/05 6:53 am Page v<br />
Table of Contents<br />
List of Tables<br />
List of Maps<br />
List of Figures<br />
About the Authors<br />
Acknowledgements<br />
Foreword<br />
Executive Summary<br />
ix<br />
xiv<br />
xvi<br />
xvii<br />
xx<br />
xxi<br />
xxxi<br />
Chapter 1: Introduction 1<br />
1.1 Background to the Study 1<br />
1.2 Rationales for Spatial Intervention 5<br />
1.3 Data Sources 9<br />
1.4 Outline of the Study 10<br />
Chapter 2: Spatial Trends in Deprivation Surrogates 11<br />
2.1 Introduction 11<br />
2.2 Age Structures 12<br />
2.3 Economic Status and Activity 16<br />
2.4 Variations in Levels of<br />
Educational Attainment 30<br />
2.5 Variations in Levels of Social Class 37<br />
2.6 Conclusions 45<br />
v
AA.Prelims 7/7/05 6:53 am Page vi<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Chapter 3: Measurement of <strong>Poverty</strong> at the<br />
Household Level 46<br />
3.1 Introduction 46<br />
3.2 Measuring <strong>Poverty</strong> and<br />
Deprivation in the LII Surveys 47<br />
3.3 Measuring <strong>Poverty</strong> and<br />
Deprivation in the NSHQ 49<br />
3.4 The National Context 55<br />
3.5 Conclusions 57<br />
Chapter 4: <strong>Poverty</strong> and Deprivation by<br />
Region and Local Authority Area 59<br />
4.1 Introduction 59<br />
4.2 <strong>Poverty</strong> Risk and Incidence by Regional<br />
Authority in the LII Survey 60<br />
4.3 Disparities in Income <strong>Poverty</strong> Risk<br />
by Local Authority Area from the NSHQ 64<br />
4.4 Modified Consistent <strong>Poverty</strong> (MCP) 69<br />
4.5 MCP by Local Authority Area 70<br />
4.6 Direct and Indirect Measures of<br />
<strong>Poverty</strong> and Deprivation 73<br />
4.7 Conclusions 76<br />
Chapter 5: <strong>Poverty</strong> by Area and Tenure Type 78<br />
5.1 Introduction 78<br />
5.2 Measuring Area Type and Tenure 79<br />
5.3 <strong>Poverty</strong> and Deprivation by Area Type 80<br />
5.4 <strong>Poverty</strong> and Deprivation by Tenure Type 85<br />
5.5 Disparities in <strong>Poverty</strong> Risk by Size<br />
of Place and Tenure 89<br />
5.6 Modified Consistent <strong>Poverty</strong><br />
by Area Type and Tenure 93<br />
5.7 Conclusions 95<br />
vi
AA.Prelims 7/7/05 6:53 am Page vii<br />
Contents<br />
Chapter 6: Disparities in Deprivation 97<br />
6.1 Introduction 97<br />
6.2 Methodology 97<br />
6.3 Disparities in Deprivation by<br />
Region and Local Authority Area 98<br />
6.4 Disparities in Deprivation by<br />
Size of Place and Tenure 113<br />
6.5 Conclusions 116<br />
Chapter 7: Understanding Variation in <strong>Poverty</strong><br />
by Area and Tenure Type 119<br />
7.1 Introduction 119<br />
7.2 Assessing the Net Effect of Location<br />
and Tenure 119<br />
7.3 Methodology 120<br />
7.4 Gross and Net Effects of Region<br />
on <strong>Poverty</strong> in the NSHQ 122<br />
7.5 Disparities in <strong>Poverty</strong> by<br />
Local Authority Areas (LAA) 125<br />
7.6 Gross and Net Effects of Size of<br />
Area and Tenure on Risk of<br />
<strong>Poverty</strong> in the NSHQ 127<br />
7.7 Conclusions 136<br />
Chapter 8: Conclusions 139<br />
8.1 Introduction 139<br />
8.2 Surrogate Measures of Deprivation 140<br />
8.3 Overview of CSO Results 141<br />
8.4 Geographical Variation: Evidence from<br />
Analysis of Indirect Measures of<br />
Deprivation 141<br />
8.5 Variation by Size of Area and Tenure 143<br />
8.6 Policy Implications 143<br />
vii
AA.Prelims 7/7/05 6:53 am Page viii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Appendix 1: Additional Tables 151<br />
Appendix 2: Methodology 167<br />
Glossary 193<br />
References 195<br />
viii
AA.Prelims 7/7/05 6:53 am Page ix<br />
List of Tables<br />
Table 2.1:<br />
Table 2.2:<br />
Table 2.3:<br />
Table 2.4:<br />
Table 2.5:<br />
Table 2.6:<br />
Table 2.7:<br />
Percentage of persons in each region<br />
according to age cohort 13<br />
Regional labour force participation and<br />
unemployment rates 19<br />
Distribution of the unemployed<br />
across regions 21<br />
Percentage of all persons engaged in<br />
farming by number of acres farmed 22<br />
Index of economic dependency<br />
in each region 28<br />
Distribution of persons 15 years and over<br />
whose education has ceased, classified by<br />
highest level of educational attainment 31<br />
Distribution of persons whose education<br />
has ceased in highest and lowest levels of<br />
educational attainment across regions 36<br />
Table 2.8: Percentage of persons in each social class 38<br />
Table 2.9:<br />
Table 3.1:<br />
Regional distribution of Professional and<br />
Unskilled Manual class categories 44<br />
Items used in the deprivation scales in<br />
the NSHQ 54<br />
ix
AA.Prelims 7/7/05 6:53 am Page x<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 3.2:<br />
Table 4.1:<br />
Table 4.2:<br />
Table 4.3:<br />
Table 4.4:<br />
Table 4.5:<br />
Trends in percentage of households below<br />
poverty lines 57<br />
<strong>Poverty</strong> risk in 2000 by regional authority<br />
and regional assembly 62<br />
<strong>Poverty</strong> incidence in 2000: Per cent of<br />
poor in each regional authority and<br />
regional assembly area (relative income<br />
poverty and consistent poverty) 64<br />
Disparities in income poverty risk by local<br />
authority area 65<br />
Disparities in risk of modified consistent<br />
poverty by local authority area 71<br />
Correlation of socio-demographic<br />
characteristics of county with risk of poverty 75<br />
Table 5.1: Risk of poverty by type of area for 1987,<br />
1994 and 2000 82<br />
Table 5.2:<br />
Incidence of poverty by type of area for<br />
1987, 1994 and 2000 83<br />
Table 5.3: Risk of poverty by tenure type for 1987,<br />
1994 and 2000 88<br />
Table 5.4:<br />
Table 5.5:<br />
Incidence of poverty by tenure type for<br />
1987, 1994 and 2000 90<br />
Disparities in income poverty risk by<br />
urban/rural location 91<br />
Table 5.6: Disparities in income poverty risk by tenure 92<br />
Table 5.7:<br />
Disparities in risk of MCP by urban/rural<br />
location 94<br />
x
AA.Prelims 7/7/05 6:53 am Page xi<br />
List of Tables<br />
Table 5.8: Disparities in risk of MCP by tenure 95<br />
Table 6.1:<br />
Table 6.2:<br />
Table 6.3:<br />
Table 6.4:<br />
Table 6.5:<br />
Table 7.1:<br />
Table 7.2:<br />
Table A1.1:<br />
Table A1.2:<br />
Table A1.3:<br />
Table A1.4:<br />
Dimensions of deprivation (cannot afford<br />
items) 98<br />
Disparities in risk of basic, secondary,<br />
housing and environmental deprivation<br />
by regional authority area 100<br />
Disparities in risk of basic deprivation,<br />
secondary deprivation, housing<br />
deprivation and environmental deprivation<br />
by local authority 107<br />
Disparities in risk of basic, secondary,<br />
housing and environmental deprivation<br />
by urban/rural area 114<br />
Disparities in risk of basic, secondary,<br />
housing and environmental deprivation<br />
by tenure 115<br />
Gross and net odds ratios of being poor<br />
versus non-poor by regional authority 125<br />
<strong>Poverty</strong> variation between and within<br />
local authority areas 126<br />
Percentage of persons in each county<br />
according to age cohort 152<br />
County-level labour force participation<br />
and unemployment rates 154<br />
Distribution of the unemployed across<br />
regions 155<br />
Percentage of all persons engaged in<br />
agriculture classified by number of acres<br />
farmed 156<br />
xi
AA.Prelims 7/7/05 6:53 am Page xii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.5:<br />
Table A1.6:<br />
Table A1.7:<br />
Index of economic dependency in<br />
each county 157<br />
Distributions of persons 15 years and over<br />
classified by highest level of educational<br />
attainment 158<br />
Distribution of persons whose education<br />
has ceased in highest and lowest levels of<br />
educational attainment across counties 159<br />
Table A1.8: Percentage of persons in each social class 160<br />
Table A1.9:<br />
Distributions of persons in Professional<br />
and Unskilled Manual class categories<br />
across region 161<br />
Table A1.10: Risk of poverty by planning region, 1987,<br />
1994 and 2000 163<br />
Table A1.11: Incidence of poverty by planning region,<br />
1987, 1994 and 2000 164<br />
Table A2.1:<br />
External population characteristics used<br />
in the construction of household weights<br />
for the Living in Ireland Surveys 174<br />
Table A2.2: Population checks for sample weighting 179<br />
Table A2.3:<br />
Level of missing information on key<br />
variables and imputation procedure 181<br />
Table A2.4: Mean weekly household income (in £)<br />
using full measure and single-item<br />
measure by number of adults and number<br />
of persons at work 183<br />
xii
AA.Prelims 7/7/05 6:53 am Page xiii<br />
List of Tables<br />
Table A2.5:<br />
Table A2.6:<br />
Table A2.7:<br />
Table A2.8:<br />
Table A2.9:<br />
Model based on LII to correct for<br />
understatement of income when a<br />
single-item measure is used 184<br />
Income category midpoints and coefficients<br />
applied to the survey of house quality 185<br />
Mean ‘corrected’ income for each original<br />
income category in the NSHQ 187<br />
Average income before and after correction<br />
by number of adults and number at work<br />
in the NSHQ 188<br />
Disparities in risk of income poverty by<br />
region, tenure, household size, economic<br />
status and social class from the LII Survey<br />
and the NSHQ 189<br />
Table A2.10: Disparities in risk of (modified) consistent<br />
poverty (60% mean and basic) by region,<br />
tenure, household size, economic status<br />
and social class from the LII Survey and<br />
the NSHQ 191<br />
xiii
AA.Prelims 7/7/05 6:53 am Page xiv<br />
List of Maps<br />
Map 2.1:<br />
Map 2.2:<br />
Map 2.3:<br />
Per cent age-dependent by local authority<br />
area 15<br />
Per cent of persons age 65 and over and<br />
living alone 16<br />
Labour force participation rate by local<br />
authority area 17<br />
Map 2.4: Unemployment rate by local authority area 18<br />
Map 2.5: Per cent of persons in farming 23<br />
Map 2.6:<br />
Per cent of farmers farming less than<br />
30 acres 24<br />
Map 2.7: Per cent of farmers farming 30–49 acres 25<br />
Map 2.8: Per cent of farmers farming 50+ acres 26<br />
Map 2.9:<br />
Economic dependency rates by local<br />
authority area 29<br />
Map 2.10: Per cent of persons age 15+ who have left<br />
school with no education or primary<br />
education 32<br />
Map 2.11: Per cent of persons age 15+ who have<br />
left school with lower secondary education 33<br />
xiv
AA.Prelims 7/7/05 6:53 am Page xv<br />
List of Maps<br />
Map 2.12: Per cent of persons age 15+ who have left<br />
school with upper secondary education 34<br />
Map 2.13: Per cent of persons age 15+ who have left<br />
school with third-level education 35<br />
Map 2.14: Per cent of persons classified as<br />
Professional/ Managerial/Technical 39<br />
Map 2.15: Per cent of persons classified as ‘Other<br />
Non-Manual’ 40<br />
Map 2.16: Per cent of persons classified as Skilled<br />
Manual 41<br />
Map 2.17: Per cent of persons classified as<br />
Semi-skilled Manual 42<br />
Map 2.18: Per cent of persons classified as<br />
Unskilled Manual 43<br />
Map 4.1: Consistent poverty risk by region 63<br />
Map 4.2:<br />
Map 4.3:<br />
Map 4.4:<br />
Disparities in income poverty risk by local<br />
authority area (50 per cent poverty line) 67<br />
Disparities in income poverty risk by local<br />
authority area (60 per cent poverty line) 68<br />
Disparities in risk of modified consistent<br />
poverty by local authority area 72<br />
Maps 6.1a to 6.1e: Disparities in risk of deprivation by<br />
regional authority area 101<br />
Maps 6.2a to 6.2e: Disparities in risk of deprivation by<br />
local authority area 108<br />
xv
AA.Prelims 7/7/05 6:53 am Page xvi<br />
List of Figures<br />
Figure 7.1: Gross and net ratios of being poor versus nonpoor<br />
at the 50 per cent income line by tenure<br />
and urban-rural location 129<br />
Figure 7.2: Gross and net ratios of being poor versus<br />
non-poor at the 50 per cent income line by<br />
tenure and Dublin versus elsewhere 131<br />
Figure 7.3: Gross and net ratios of being poor versus<br />
non-poor at the 60 per cent income line by<br />
tenure and urban-rural location 132<br />
Figure 7.4: Gross and net ratios of being poor versus<br />
non-poor at the 60 per cent income line<br />
by tenure and Dublin versus elsewhere 133<br />
Figure 7.5: Gross and net ratios of being poor versus<br />
non-poor at the MCP line by tenure and<br />
urban-rural location 134<br />
Figure 7.6: Gross and net ratios of being poor versus<br />
non-poor at the 50 per cent MCP line by<br />
tenure and Dublin versus elsewhere 136<br />
xvi
AA.Prelims 7/7/05 6:53 am Page xvii<br />
About the Authors<br />
Dorothy Watson is a Senior Research Officer at the<br />
Economic and Social Research Institute. She managed the<br />
Living in Ireland Survey and, with James Williams, the<br />
2001/2002 Survey of Housing Quality and the 2003 National<br />
Survey of Domestic Abuse. Together with Bernadette Ryan<br />
and Bertrand Maitre of the ESRI, she is presently working on<br />
an international consortium to establish the Euro-Panel User<br />
Network to provide support and avenues of communication<br />
to users of the European Community Household Panel<br />
(ECHP). She has been a member of the EU-SILC Task Force<br />
and a consultant to Eurostat on the EU-SILC Pilot. Her<br />
current research includes a study of the changes in income<br />
associated with retirement (with Gerry Hughes), the<br />
development of a questionnaire for a national study of<br />
disability in Ireland (with Brian Nolan and James Williams) and<br />
an analysis of the spatial distribution of poverty and<br />
disadvantage in Ireland (with James Williams and Chris<br />
Whelan).<br />
Christopher T. Whelan is a Research Professor at the<br />
Economic and Social Research Institute. His current research<br />
interests include the causes and consequences of poverty<br />
and inequality, the measurement and monitoring of poverty<br />
and social exclusion, social mobility and inequality of<br />
opportunity and quality of life. He is involved in a range of<br />
xvii
AA.Prelims 7/7/05 6:53 am Page xviii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Irish and European projects involving analysis of the Living in<br />
Ireland Panel Survey and the European Community<br />
Household Panel Study. These activities include acting as a<br />
co-ordinator of an EU-funded network of European<br />
institutions focusing on the inter-disciplinary study of<br />
Economic Change, Unequal Life-Chances and Quality of Life<br />
(CHANGEQUAL). He has contributed on these topics to a<br />
range of academic journals. He is currently an associate<br />
editor of the European Sociological Review and is<br />
Chairperson of the Standing Committee for the Social<br />
Sciences of the European Science Foundation.<br />
James Williams is a Research Professor at the Economic<br />
and Social Research Institute and is also Head of the<br />
Institute’s Survey Unit, where he is responsible for all aspects<br />
of survey-based data collection. From 1993 to 1996 he<br />
managed the Irish component of the European Community’s<br />
Household Panel Survey (ECHP). In that role he had a major<br />
involvement with Eurostat in the planning and implementation<br />
of the survey. He has undertaken an extensive range of<br />
commissioned research and consultancy projects and has<br />
both contributed numerous chapters to books and coauthored<br />
others. He has also been involved in a wide range<br />
of commissioned reports and consultancy projects - both<br />
national and international - spanning a diverse range of<br />
topics. He also carries out both ad hoc and ongoing surveys<br />
for a range of organisations and bodies, mostly Irish public<br />
sector, European Commission and Eurostat.<br />
Sylvia Blackwell has been the Surveys Executive in the<br />
ESRI Survey Unit since 2000, where she has worked on<br />
business projects such as the Annual Business Surveys coordinated<br />
by Forfás among client bases of the IDA; Enterprise<br />
xviii
AA.Prelims 7/7/05 6:53 am Page xix<br />
About the Authors<br />
Ireland; Údarás na Gaeltachta and Shannon Development<br />
and the Survey of Business Expenditure on R & D within<br />
Ireland. More recently she has worked on such projects as<br />
the Dublin Docklands Authority on their Educational<br />
Database, a Survey of Employers’ and Employees’ Views and<br />
Experiences and also on the National Survey of Vacancies in<br />
the Private Non-Agriculture Sector and Public Sector.<br />
xix
AA.Prelims 7/7/05 6:53 am Page xx<br />
Acknowledgements<br />
This study forms part of on going research programme being<br />
carried out by the Economic and Social Research Institute<br />
(ESRI) that is sponsored by the Department of Social and<br />
Family Affairs. The study draws on a number of major data<br />
sources, including the National Survey of Housing Quality. The<br />
<strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong> has provided specific funding for this<br />
project and we would like to express our appreciation of their<br />
support.<br />
The Department of the Environment, Heritage and Local<br />
Government provided funding for the National Survey of<br />
Housing Quality 2001/2002, which is an important source of<br />
data for some of the detailed regional analysis presented here.<br />
In carrying out this study we have benefited from the<br />
comments of Jim Walsh and Vanessa Coffey of the <strong>Combat</strong><br />
<strong>Poverty</strong> <strong>Agency</strong> and an anonymous referee.<br />
We also gratefully acknowledge the assistance of Mary<br />
Dowling, Daphne McNamara Lancha and Pat Hopkins.<br />
xx
AA.Prelims 7/7/05 6:53 am Page xxi<br />
Foreword<br />
Introduction<br />
<strong>Combat</strong> <strong>Poverty</strong> is a state advisory body which develops and<br />
promotes evidence-based proposals and measures to<br />
combat poverty. The spatial aspects of poverty are a longstanding<br />
concern of <strong>Combat</strong> <strong>Poverty</strong>, in terms of both<br />
research and area-based policies. 1 This study updates and<br />
extends previous research on the spatial distribution of<br />
poverty using recent national data sources: the Census of<br />
Population (2002), the Living in Ireland Survey (2000) and, for<br />
the first time, the National Survey of Housing Quality<br />
(2001/2002). It applies direct and indirect measures of poverty<br />
(household income, deprivation, socio-demographic<br />
1 Brian Nolan, et al. (1998), Where are Poor Households The Spatial<br />
Distribution of <strong>Poverty</strong> and Deprivation in Ireland, Dublin: Oak Tree<br />
Press and <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong>; Brian Nolan and Christopher<br />
Whelan (1999), Loading the Dice A Study of Cumulative<br />
Disadvantage, Dublin: Oak Tree Press and <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong>;<br />
Sarah Craig and Kieran McKeown (1994), Progress through<br />
Partnership: Final Evaluation Report on the PESP Pilot Initiative on<br />
Long-Term Unemployment, Dublin: <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong>; Brian<br />
Harvey (1994), <strong>Combat</strong>ing Exclusion: Lessons from the Third EU<br />
<strong>Poverty</strong> Programme in Ireland 1989–1994, Dublin: <strong>Combat</strong> <strong>Poverty</strong><br />
<strong>Agency</strong>; Barry Cullen (1994), A Programme in the Making: A Review of<br />
the Community Development Programme, Dublin: <strong>Combat</strong> <strong>Poverty</strong><br />
<strong>Agency</strong>.<br />
xxi
AA.Prelims 7/7/05 6:53 am Page xxii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
indicators) to a range of spatial categories, including<br />
regional and local administrative units, area type and<br />
housing tenure. 2<br />
The study addresses three key aspects of the spatial<br />
distribution of poverty:<br />
• identifies patterns with regard to the concentration of<br />
poverty and how these have evolved over time<br />
• assesses if these patters are significant in terms of the<br />
overall incidence of poverty<br />
• considers the processes underlying poverty clustering,<br />
distinguishing between factors which influence the<br />
location and the causes of poverty.<br />
These questions are also relevant to the government policy<br />
priority attached to tackling ‘poverty blackspots’ in urban<br />
and rural areas. The National Anti-<strong>Poverty</strong> Strategy<br />
highlights the ‘concentrated and cumulative nature’ of<br />
poverty in urban and rural areas, which has prompted<br />
numerous area programmes and related initiatives. 3 The<br />
uneven distribution of poverty is also linked to a wider<br />
policy concern with balanced regional development, as is<br />
reflected in the regional and Peace II programmes under the<br />
National Development Plan. From this policy perspective,<br />
2 A fivefold area classification is used of open countryside, village, town,<br />
provincial city and Dublin. While tenure is not a spatial category per se,<br />
it does have a spatial expression through the segmented nature of the<br />
housing market. The categories include owned outright, owned with<br />
mortgage, tenant purchase, public rented and private rented.<br />
3 These include the Local Development Social Inclusion Programme,<br />
Community Development Programme, RAPID Programme and Clár<br />
Programme. Spatial initiatives also exist to address educational<br />
disadvantage, drug misuse, unemployment, child care and other forms<br />
of disadvantage.<br />
xxii
AA.Prelims 7/7/05 6:53 am Page xxiii<br />
Foreword<br />
research on the spatial dimension of poverty is needed to<br />
accurately identify poor areas in the first instance, to<br />
understand why these areas exist and to assess the impact<br />
of area policies.<br />
Key Research Findings<br />
The study finds a pattern of regional and county variation in<br />
poverty-related indicators such as age dependency,<br />
economic activity, educational attainment and social class.<br />
This provides an interpretative context for subsequent<br />
analyses of direct measures of poverty.<br />
The study identifies an uneven spatial distribution of poor<br />
households at regional and local levels, with a more<br />
differentiated pattern of poverty apparent for each<br />
descending unit of analysis. Thus, the slightly higher poverty<br />
rate in the NUTS 2 Border, Midlands and West region is<br />
accentuated in NUTS 3 planning regions such as the Border,<br />
Mid-West and West. Among local authorities, Donegal,<br />
Cavan, Leitrim, Longford, Mayo and the provincial cities<br />
record above-average rates of poverty.<br />
Turning to area differences, this study finds that the<br />
countryside and villages have a slighter higher poverty risk,<br />
especially compared to Dublin and the larger cities. The<br />
differential for income poverty has worsened in recent years,<br />
though the variation in the consistent poverty measure has<br />
narrowed.<br />
Tenure differences are very marked, with local authority<br />
tenants at particular risk of poverty and deprivation (33 per<br />
cent compared with an average of 6 per cent). Furthermore,<br />
their relative position has worsened over time though the<br />
xxiii
AA.Prelims 7/7/05 6:53 am Page xxiv<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
overall level of poverty has fallen and their share of the<br />
housing market has declined. Another high-risk category is<br />
tenants in private housing, who have twice the average<br />
poverty rate.<br />
The ranking of regional and local authorities varies a lot<br />
depending on the specific deprivation measures. Thus,<br />
Dublin, which otherwise has a low poverty risk, scores high in<br />
housing and environmental items. Deprivation disparities by<br />
tenure are striking, with public tenants experiencing the<br />
highest levels. Furthermore, these disparities are consistent,<br />
indicating a pattern of multiple deprivation among this<br />
housing tenure.<br />
In summary, poverty remains a spatially diffuse phenomenon,<br />
with variations in the risk of poverty being counterbalanced<br />
by a more even incidence of poverty as higher-risk areas tend<br />
to have a lower share of the total population. The greatest<br />
poverty concentration is by housing tenure, with local<br />
authority and private rented tenant households accounting for<br />
60 per cent of all households experiencing poverty and<br />
deprivation, though only representing 17 per cent of the total<br />
population.<br />
The study is constrained by data limitations in exploring the<br />
micro distribution of poverty within counties. However, it is<br />
likely that the more acute poverty clusters at this spatial level<br />
represent a relatively small proportion of the total numbers in<br />
poverty.<br />
In reviewing the factors which underlie the observed spatial<br />
patterns in poverty, the study finds little evidence of a causal<br />
relationship with particular locations or tenures. Where<br />
variation exists, the determining factors relate to the socioeconomic<br />
composition of poor households rather than<br />
xxiv
AA.Prelims 7/7/05 6:53 am Page xxv<br />
Foreword<br />
location per se. The significance of tenure over other spatial<br />
units is its selection effect: the housing market, aided by<br />
public policy, reserves public and social rented housing for<br />
low-income households in the main. However, the study does<br />
indicate an additional tenure effect in urban areas which<br />
warrants further investigation.<br />
Policy Implications<br />
The main policy implications arising from this study identified<br />
by <strong>Combat</strong> <strong>Poverty</strong> are as follows.<br />
• Policies to tackle poverty must continue to prioritise the<br />
structural causes of the problem over a focus on the<br />
spatial outcomes.<br />
• Area programmes cannot be justified on a targeting basis<br />
alone given the diffuse nature of poverty, though other<br />
rationales have merit.<br />
• Policies should distinguish between ‘people poverty’,<br />
which is linked to structural factors, and ‘place poverty’,<br />
which reflects more local issues.<br />
• Greater policy emphasis must be placed on<br />
neighbourhood issues, such as access to services and<br />
estate management in area regeneration policies.<br />
• Action must be taken to address the segregated nature of<br />
the housing market, which goes beyond a concern with<br />
planning issues to address core inequalities.<br />
• Regional and local poverty trends should be monitored<br />
and targets set for poverty reduction in line with national<br />
targets.<br />
xxv
AA.Prelims 7/7/05 6:53 am Page xxvi<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
This study reaffirms that poverty is primarily a structural,<br />
rather than a spatial, phenomenon. It arises from sociodemographic<br />
variables associated with an increased risk of<br />
poverty, e.g. unemployment and low-paid work, low<br />
educational attainment, old age, child dependency and lone<br />
parenthood. The spatial distribution of poverty largely reflects<br />
variations in the existence of these variables among the<br />
population. Policies to tackle poverty must prioritise these<br />
structural issues. At the same time, there is a positive<br />
message in the report’s rejection of the idea of a spatially<br />
concentrated ‘underclass’ that is cut off socially and<br />
economically from the rest of society.<br />
From a targeting perspective, spatial categories are of limited<br />
benefit, especially as compared with a traditional focus on<br />
socio-demographic factors. While it may be possible to<br />
distinguish micro clusters of poor households, often<br />
associated with public housing estates, these are not<br />
significant in terms of the total number of households in<br />
poverty. Spatial programmes have other policy rationales<br />
besides targeting. They can improve the delivery of existing<br />
services through promoting greater integration, community<br />
access and user involvement and develop new outreach<br />
mechanisms to meet the needs of vulnerable groups. They<br />
also achieve a multiplier effect by combining investment in<br />
physical regeneration with economic and social initiatives.<br />
More attention should be focused on the added value of<br />
area-based interventions in relation to these objectives as<br />
distinct from a crude targeting mechanism.<br />
Targeting is relevant when applied to tenants in local authority<br />
and private accommodation. Even then, this must be<br />
contextualised in wider policy issues affecting this group,<br />
especially in urban areas. The main challenge is the selection<br />
xxvi
AA.Prelims 7/7/05 6:53 am Page xxvii<br />
Foreword<br />
effect associated with social housing. While there has been<br />
a policy shift in regard to the scale of social housing in order<br />
to avoid large concentrations, there has been no change in<br />
the restrictive allocation of social housing. Furthermore, the<br />
dynamic nature of social housing ensures a constant<br />
recruitment of poor tenants to replace those who, having<br />
improved their situation, move on. In addition, there is an<br />
internal process within social housing whereby tenants<br />
move from different estates, usually from low-demand to<br />
high-demand areas. These often reflect differences in how<br />
estates are managed as much as the quality of housing.<br />
Previous research indicates that within local authority<br />
estates, there can be subtle differences between individual<br />
streets and blocks. 4 Urgent measures are required to<br />
counter this internal process of ghettoising vulnerable<br />
households.<br />
A stronger focus on ‘place’ or neighbourhood poverty is<br />
warranted. The residualisation of public housing has resulted<br />
in tenants experiencing inferior housing quality and a range<br />
of other neighbourhood drawbacks. The discontinuity<br />
between regeneration initiatives and social programmes is<br />
disconcerting, with the exception of some localised<br />
initiatives, e.g. Ballymun. One example of how the physical<br />
and social dimensions can be combined is in regard to fuel<br />
poverty. If combined with social programmes, energy<br />
efficiency measures can have a multiplier effect for poor<br />
tenants. For this synergy to develop, local authorities, along<br />
with other supporters and providers of social housing (health<br />
4 Tony Fahey (ed.) (1999), Social Housing in Ireland: A Study of Success,<br />
Failure and Lessons Learned, Dublin: Oak Tree Press in association<br />
with <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong> and the Katharine Howard Foundation.<br />
xxvii
AA.Prelims 7/7/05 6:53 am Page xxviii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
boards and voluntary housing associations) should have a<br />
role in designing and delivering area programmes which<br />
address this high-risk poverty sector.<br />
A neighbourhood effect is also likely to arise from the<br />
limited access to services and opportunities for residents<br />
of poor areas. For example, employer recruitment<br />
practices may discriminate against people from certain<br />
localities, either through crude selection policies or<br />
because of limited access to informal recruitment<br />
networks. Also, the quantity and quality of services<br />
provided can be less than are available in other, more<br />
affluent areas, e.g. health care, schools, financial and retail<br />
services, sports and recreation facilities, transport. Antidiscrimination<br />
monitoring of public and private services is<br />
required to address these area effects. Another factor is<br />
the level of social capital, as reflected in voluntary activity<br />
and social networking, in poor neighbourhoods.<br />
Finally, there is a need to monitor poverty trends at<br />
regional and local levels on an ongoing basis and to relate<br />
these to national trends. The provision of sub-national data<br />
on poverty provides a baseline to measure progress and to<br />
inform regional and local poverty strategies. Regional and<br />
local poverty reduction targets should also be set in line<br />
with the national target to reduce the combined poverty<br />
and deprivation measure to 2 per cent by 2007. It is<br />
possible to adapt this target based on reducing the<br />
poverty differentials that exist between regional and local<br />
authorities. These targets should be set in the context of<br />
administrative authorities drawing up plans to reduce<br />
poverty in their areas as part of an overall national<br />
strategy.<br />
xxviii
AA.Prelims 7/7/05 6:53 am Page xxix<br />
Foreword<br />
Data Issues<br />
Geographical analysis of poverty remains underdeveloped.<br />
A key constraint is the lack of poverty-specific data on a<br />
spatially disaggregated basis. The limitations of survey data<br />
as used in this study have already been highlighted. The<br />
Census of Population, though inclusive of all households, has<br />
acute drawbacks in regard to the measurement of poverty<br />
and the spatial units of analysis. The Department of the<br />
Environment, Heritage and Local Government recently issued<br />
a consultation paper on spatial data infrastructure. Various<br />
avenues should be explored to gather specific data on<br />
poverty at the local level, including the Census of Population.<br />
A related task is to develop a system for the gathering and<br />
presentation of existing poverty-related data at the local level,<br />
e.g. welfare payments, medical cards, unemployment<br />
records. In Northern Ireland and the United Kingdom, quite<br />
advanced methodologies have been developed to analyse<br />
poverty-related indicators. In Ireland, Hasse has developed<br />
earlier work by <strong>Combat</strong> <strong>Poverty</strong> to produce a composite<br />
index of deprivation based on the Small Areas Statistics<br />
Population from the Census of Population. 5 Further work is<br />
required to refine this methodology and to incorporate other<br />
data sources so as to create a comprehensive spatial<br />
database on poverty.<br />
Separately, there is a requirement for better research and<br />
monitoring of poor neighbourhoods. A dedicated study is<br />
required which would address two key issues: what is the<br />
5 Trutz Haase (1999), ‘Affluence and Deprivation: A Spatial Analysis<br />
Based on the 1991 Census of Population’, in Dennis Pringle et al.<br />
(eds.), Poor People, Poor Places, Dublin: Oak Tree Press and the<br />
Geographical Society of Ireland.<br />
xxix
AA.Prelims 7/7/05 6:53 am Page xxx<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
neighbourhood effect on clusters of poor households, and<br />
what is the impact of government policies on the living<br />
conditions in poverty clusters An earlier study by <strong>Combat</strong><br />
<strong>Poverty</strong> and the Katharine Howard Foundation demonstrates<br />
the potential of micro studies in understanding the factors<br />
shaping local conditions. 6 Ideally, this type of study should be<br />
established on a longitudinal and a comparative basis so that<br />
change over time can be tracked and comparisons made<br />
between the evolution of different localities. Such a study will<br />
be considered in <strong>Combat</strong> <strong>Poverty</strong>’s strategic plan 2005–2007.<br />
6 Tony Fahey (ed.), op cit.<br />
xxx
AA.Prelims 7/7/05 6:53 am Page xxxi<br />
Executive Summary<br />
Is poverty concentrated in certain ‘black spots’ If so, what is<br />
the extent of this concentration, what causes such<br />
concentration and which areas are particularly at risk These<br />
are important questions for policy analysts because of their<br />
implications for strategies to combat poverty and promote<br />
social inclusion.<br />
These are the questions addressed by this report on the<br />
spatial distribution of poverty in Ireland. The report was<br />
commissioned by the <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong> to look at the<br />
following questions:<br />
• Is poverty concentrated in particular areas If so, what is<br />
the extent of that concentration<br />
• Where are the concentrations of poverty<br />
• Why are there concentrations of poverty<br />
• What are the characteristics of such concentrations of<br />
poverty<br />
• What can be done about reducing such concentrations of<br />
poverty<br />
The study brings together data from three national sources:<br />
the 2002 Census, the 2000 Living in Ireland Survey, with very<br />
xxxi
AA.Prelims 7/7/05 6:53 am Page xxxii<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
detailed information on income and living standards for a<br />
sample of 3,400 households, and the 2001/2002 National<br />
Survey of Housing Quality, with a sample of over 40,000<br />
households.<br />
A key feature of the report is the concern with the causal<br />
processes underlying any association between area and<br />
poverty. Targeting programmes to combat poverty need<br />
to be fundamentally concerned with the causes of poverty<br />
rather than with patterns that are accidental. This is<br />
particularly important from the point of view of area-based<br />
programmes. Without an understanding of why people in<br />
certain areas experience a greater risk of poverty,<br />
programmes to improve their situation cannot be properly<br />
targeted to those in need.<br />
Key Findings<br />
There are clear regional and local differences in both the risk<br />
of poverty and in the levels of key related social indicators<br />
such as unemployment, education and social class. Further,<br />
the differences in risk become larger as we move to lower<br />
levels of aggregation, e.g. from region to county level. The<br />
highest poverty risk is found in counties Donegal, Leitrim and<br />
Mayo and the lowest in the counties around Dublin. However,<br />
there is considerable diversity within administrative planning<br />
regions. Louth and Sligo, for instance, show much lower<br />
relative poverty risk than the other counties in the Border<br />
region. This points to the importance of avoiding broad<br />
generalisations based on administrative units which are<br />
based on historical and political rather than socio-economic<br />
categories.<br />
xxxii
AA.Prelims 7/7/05 6:53 am Page xxxiii<br />
Executive Summary<br />
However, the scale of differences based on geographic<br />
location is rather modest compared to the differences<br />
between socio-economic groups or people with different<br />
housing tenures. Furthermore, the pattern varies depending<br />
on the dimension of deprivation being considered. For<br />
instance, Dublin, which has a low income poverty risk, scores<br />
high in terms of housing and environmental deprivation. On<br />
the other hand, Donegal has a much higher income poverty<br />
rate than the national average, but it is less likely than<br />
average to suffer deprivation in terms of housing deterioration<br />
or environmental deprivation.<br />
There are modest differences between urban and rural areas,<br />
but large differences by housing tenure. Local authority<br />
renters fare worst in terms of almost all measures of<br />
deprivation. This is largely due to a selection effect, whereby<br />
the dynamics of the housing market and public housing<br />
policy result in public and social rented housing being<br />
reserved for low-income households.<br />
Evidence of multiple deprivations structured along spatial<br />
lines is extremely weak. Over time, income poverty rates for<br />
local authority tenants have increased sharply, and while their<br />
consistent poverty rate (which takes account of living<br />
standard as well as income) has shown some improvement,<br />
they have fallen farther behind homeowners. The decline in<br />
their numbers, however, means that they constitute a smaller<br />
proportion of the poor.<br />
What accounts for the differences between areas at risk of<br />
poverty and deprivation The key factors accounting for<br />
poverty and deprivation are socio-economic: unemployment,<br />
non-participation in the labour force due to old age or illness,<br />
lone parenthood, low levels of education and social class.<br />
xxxiii
AA.Prelims 7/7/05 6:53 am Page xxxiv<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
However, the study does indicate an additional effect of<br />
tenure type. Differences between geographic areas in terms<br />
of poverty and deprivation are largely due to differences<br />
between these areas in the socio-economic composition of<br />
their populations.<br />
Implications<br />
The study clearly demonstrates that poverty is a structural<br />
rather than a spatial phenomenon. It arises from socioeconomic<br />
processes such as unemployment and low-paid<br />
work, low educational attainment, old age, child dependency<br />
and lone parenthood. The spatial distribution of poverty<br />
largely reflects spatial variations in these variables. Policies to<br />
tackle poverty must prioritise these structural issues.<br />
Area-based policies have little role in targeting poor<br />
households – most poor households do not live in clearly<br />
identifiable geographically concentrated areas. However,<br />
there may be a role for area-based initiatives in local authority<br />
estates, where they may enhance service delivery through<br />
promoting greater integration, community access and user<br />
involvement. The emphasis of such programmes should be<br />
on the efficient delivery of services and mobilisation of<br />
community resources rather than targeting poor households.<br />
Such locally based programmes may also be used to provide<br />
an outreach mechanism to meet the multiple needs of<br />
vulnerable populations, such as long-term unemployed, exprisoners,<br />
former drug users and early school-leavers. More<br />
attention should be focused on the added value of areabased<br />
interventions in relation to these objectives rather than<br />
as a simple targeting mechanism.<br />
xxxiv
Chapter 01 7/7/05 6:53 am Page 1<br />
Chapter 1<br />
Introduction<br />
1.1 Background to the Study<br />
In 1998 the <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong> published Where are<br />
Poor Households The Spatial Distribution of <strong>Poverty</strong> in<br />
Ireland by Nolan, Whelan and Williams. The study employed<br />
data for the early to mid-1990s. It was conducted in the<br />
context of a growing concern about the extent of<br />
geographical concentration of poverty alongside increasing<br />
fears that such concentration was associated with processes<br />
of cumulative disadvantage of a vicious cycle nature. It<br />
addressed the extent to which poverty in Ireland is<br />
concentrated in particular areas or types of areas, the nature<br />
of the causal processes underlying the spatial distribution of<br />
poverty and the implications for policy.<br />
The key questions of the study were:<br />
• Is poverty concentrated in particular areas If so, what is<br />
the extent of that concentration<br />
• Where are the concentrations of poverty<br />
• Why are there concentrations of poverty<br />
• What are the characteristics of such concentrations of<br />
poverty<br />
1
Chapter 01 7/7/05 6:53 am Page 2<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
• What can be done about reducing such concentrations of<br />
poverty<br />
Ireland changed greatly over the second half of the 1990s,<br />
during which an unprecedented economic boom was<br />
experienced. Data are now coming on stream to allow these<br />
issues, which still remain highly relevant, to be addressed<br />
once again in light of these changes. These data comprise<br />
not only updated census data but a national survey that<br />
enables us to address these issues on a sample of<br />
households whose number goes well beyond anything that<br />
has been available to us in the past.<br />
Much of the debate about the impact of economic<br />
restructuring, industrial modernisation and infrastructural<br />
development in Ireland has a strong spatial dimension.<br />
Concern is expressed that the effects of socio-economic<br />
change vary not only by socio-economic group but also by<br />
location. In popular discourse in Ireland, certain places (often<br />
quite vaguely defined) are often said to have fared badly,<br />
such as ‘the West’, ‘remote rural areas’, ‘new suburbs’ and<br />
‘poverty black spots in the city’. More recently, concern in a<br />
number of European countries with vicious cycle processes<br />
of cumulative disadvantage and the emergence of an urban<br />
‘underclass’ has led to issues relating to spatial segregation<br />
figuring prominently in popular and academic debates.<br />
Since the early 1990s there has been a growing emphasis on<br />
policy options involving spatial programmes aimed at tackling<br />
unemployment, poverty and social exclusion. Following a<br />
number of pilot schemes, area-based programmes have<br />
become a significant part of government policy aimed at<br />
tackling poverty. As Walsh (1999: 279) notes, this is clearly<br />
evidenced in the National Anti-<strong>Poverty</strong> Strategy (1997), which<br />
2
Chapter 01 7/7/05 6:53 am Page 3<br />
Introduction<br />
exhibits an explicit spatial dimension in two of its five priority<br />
themes: disadvantaged urban areas and marginalised rural<br />
communities. In 1992 the Irish government, in collaboration<br />
with the European Commission, established Area<br />
Development Management (ADM). Its primary objective was<br />
to promote social inclusion, reconciliation and equality and to<br />
counter disadvantage through local, social and economic<br />
development. As Haase, McKeown and Rourke (1996)<br />
document, the initiative was representative of a concern to<br />
develop area-based initiatives in response to emerging<br />
evidence relating to unemployment black spots.<br />
The number and range of projects managed by ADM has<br />
evolved considerably since 1992. Among its current<br />
programmes are the Local Development Social Inclusion<br />
(LDSIP) and Revitalising Areas by Planning Investment and<br />
Development (RAPID) programmes. The former is a National<br />
Development Programme aimed specifically at addressing<br />
social inclusion issues at local level. The LDSIP provides<br />
funding to Partnership Community Groups and Employment<br />
Pacts that adopt a partnership approach to tackling local<br />
issues on the basis of comprehensive, integrated local action<br />
plans. The RAPID programme is intended as a response to<br />
the need for more and better-targeted investment in<br />
disadvantaged areas. The focus in terms of service delivery is<br />
on integrating services more efficiently and tailoring them to<br />
community needs. It is also intended to encourage<br />
investment in new facilities and services. Among the<br />
objectives identified as fundamental to the RAPID programme<br />
are the development of an integrated focus on social groups<br />
experiencing cumulative disadvantage, reduction in spatial<br />
concentration of poverty, unemployment and social exclusion<br />
and the mobilisation of social capital and capacity for<br />
3
Chapter 01 7/7/05 6:53 am Page 4<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
economic and social development. Other programmes<br />
include the Equal Opportunities Childcare Programme (EOCP)<br />
and the Millennium Partnership Fund for Disadvantage.<br />
One of the important considerations, but by no means the<br />
only consideration, in the allocation of funding under these<br />
programmes is the socio-demographic profile of the<br />
geographical areas. This can involve the use of indices of<br />
deprivation, which rank district electoral divisions (DEDs)<br />
according to indicators such as unemployment, education,<br />
class composition and housing quality.<br />
This discussion of ADM activities illustrates that while area<br />
interventions all appear to share the objective of targeting<br />
scarce resources, the rationales or justifications associated<br />
with such interventions are variable and can take on a good<br />
deal of complexity. In our earlier report, Where Are Poor<br />
Households (Nolan, Whelan and Williams 1998), we pointed<br />
to the need to consider a number of related, but still relatively<br />
distinct, arguments for focusing on the spatial distribution of<br />
poverty. Some of these involve specific assumptions<br />
regarding the causal basis of poverty, while others take a<br />
more pragmatic form. The most straightforward justification is<br />
based on the assumption that if poor households are highly<br />
concentrated in specific areas, then it is possible to target<br />
resources on these areas in order to maximise the number of<br />
households reached. Thus, the argument outlined in the<br />
NESC (1990) document Strategies for the Nineties was based<br />
on the assumption that general measures to improve<br />
employment creation would not be sufficient to have an<br />
impact on those experiencing long-term unemployment.<br />
Special employment measures were required which would be<br />
‘targeted in an integrated fashion in the context of local areabased’<br />
strategies. Increased polarisation, and its spatial<br />
4
Chapter 01 7/7/05 6:53 am Page 5<br />
Introduction<br />
manifestations, provides an important component of the<br />
rationale for spatial interventions.<br />
1.2 Rationales for Spatial Intervention<br />
A recent review of area-based targeting by Tunstall and<br />
Lupton (2003) distinguishes between five different rationales.<br />
• The first focuses on ‘efficiency’ and completeness in<br />
reaching poor individuals and derives its logic from the<br />
concentration of deprivation and disadvantage. Thus, as<br />
Walsh (1999: 283) notes, developments within mainstream<br />
welfare policy have encouraged a greater local focus on<br />
the design and delivery of services. In the context of a<br />
growing desire among policy makers for better targeting<br />
of resources, spatial programmes seem to offer an<br />
attractive means of responding to social needs. Here,<br />
evidence relating to the distribution of unemployment,<br />
unskilled manual work, lack of educational qualifications<br />
and low income plays a crucial role.<br />
• The second rationale is based on the argument that<br />
concentrated poverty may have cumulative and<br />
qualitatively different effects on individuals, organisations<br />
and infrastructure than less concentrated poverty. <strong>Poverty</strong><br />
‘black spots’ could result in a qualitatively different<br />
experience of poverty in terms of factors such as physical<br />
and mental health, degree of economic strain and<br />
alienation from social and political participation, ranging<br />
from social isolation to declining church attendance and<br />
participation and confidence in the political process. Thus,<br />
NESC (1993) referred to a localised process of interaction<br />
between labour market, education, housing and<br />
5
Chapter 01 7/7/05 6:53 am Page 6<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
environmental factors, which is most likely to be<br />
addressed effectively by an area-based strategy. This<br />
rationale could provide the justification for interventions<br />
offering support targeted not just at individuals, but<br />
organisations and infrastructure. A crucial objective<br />
from this perspective is to provide a focus for<br />
enhancing service provision in responses to multiple<br />
deprivation. Local co-ordination is a potentially effective<br />
means of improving the quality of services. Initiatives of<br />
this kind have been directed at service co-ordination<br />
relating to areas such as labour market and training<br />
access and indebtedness and access to credit.<br />
• The third justification relates to the choice of areas as a<br />
form of rationing by taking advantage of the fact that<br />
targeting areas may be a good deal simpler than<br />
targeting individuals. As Tunstall and Lupton (2003: 4–5)<br />
observe, in some cases governments make new money<br />
available for a specific purpose which will only be<br />
targeted towards areas with specific needs, though<br />
usually fewer than the total number of areas with such<br />
needs are reached on account of limited resources.<br />
• The fourth relates to the use of area-based initiatives as<br />
a form of piloting. In assessing such schemes’<br />
potential, capacity to innovate or deliver may also be<br />
taken into account. Thus, Walsh (1999: 288) notes that<br />
initiatives such as the Area-Based Response to Long-<br />
Term Unemployment (ABR) and its linked programme,<br />
the Global Grant for Local Development (GGLD),<br />
operated on a pilot basis between 1991 and 1995<br />
before being subsumed into the Local Development<br />
Programme.<br />
6
Chapter 01 7/7/05 6:53 am Page 7<br />
Introduction<br />
• The fifth rationale focuses on additional benefits deriving<br />
from area initiatives such as community involvement and<br />
the development of partnerships. Among other things, this<br />
justification draws attention to the particular importance of<br />
co-ordination and more accurate identification of needs in<br />
deprived areas. Thus, the integration and co-ordination of<br />
existing services on the basis of partnership arrangements<br />
may be seen as of particular value in areas where<br />
deprivation takes a multidimensional form. The NESC<br />
document, referred to earlier, argues that intensive coordinated<br />
spatial interventions containing elements of<br />
housing and environment improvements, as well as<br />
retraining and employment schemes, could have an<br />
impact over and above their separate effects. In addition,<br />
it was argued that participation of local communities in<br />
the planning and delivery of area-based projects would<br />
help ensure that they more accurately reflected local<br />
needs and priorities. The community development<br />
rationales for spatial intervention attempt to develop the<br />
argument that involvement of local communities in the<br />
process of economic and social change has an intrinsic<br />
value. Thus, rather than such involvement being peripheral<br />
to a process of economic change, it is seen as a crucial<br />
means by which marginalised groups can come to share<br />
in the benefits of economic progress. This thesis relies not<br />
just on arguments relating to effectiveness and efficiency,<br />
but rather bases its arguments on an understanding of<br />
social exclusion which identifies loss of skills, selfconfidence<br />
and motivation as crucial elements in the<br />
process.<br />
A particular justification for targeted intervention does not<br />
necessarily involve a commitment to a specific understanding<br />
7
Chapter 01 7/7/05 6:53 am Page 8<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
of the causal processes. However, almost inevitably, some<br />
such set of assumptions underlies spatial interventions. It is<br />
possible, though, to see spatial concentration of deprivation<br />
as having no causal significance or distinctive consequences.<br />
Instead, concentration could be seen to arise simply as a<br />
consequence of variations across other genuinely causal<br />
variables such as human capital. Alternatively, location could<br />
be thought to play a potentially independent role in a number<br />
of ways. Thus, employers’ hiring behaviour could mean that<br />
merely residing at a particular address could increase one’s<br />
risk of unemployment and poverty at a given level of human<br />
capital. More broadly, those coming from areas where the<br />
resource stock, in terms of access to training, education,<br />
financial institutions, etc., is poorer could be seen as being<br />
additionally disadvantaged. Much more controversially, at the<br />
centre of recent debates concerning the creation of a<br />
spatially concentrated underclass is the highly contested idea<br />
that persistent poverty is transmitted through a fundamental<br />
altering of norms and ‘tastes’ in relation to welfare<br />
dependency, employment commitment and non-marital<br />
fertility.<br />
It is useful to keep in mind that area-based responses to<br />
poverty are not an entirely new phenomenon. Pringle (1999)<br />
notes that many of the initiatives pursued in Ireland in recent<br />
years, while innovative in the Irish context, are not entirely<br />
dissimilar to those introduced in others countries in the 1960s<br />
and 1970s. We should also acknowledge that there is a<br />
significant literature that is sceptical regarding the extent of<br />
the contribution that spatial strategies can make to reducing<br />
or ameliorating poverty. However, it seems more sensible to<br />
defer consideration of that debate until we have set the<br />
evidence for the nature and significance of spatial variation in<br />
poverty and deprivation.<br />
8
Chapter 01 7/7/05 6:53 am Page 9<br />
Introduction<br />
1.3 Data Sources<br />
In the analysis that follows we shall seek to provide a<br />
relatively up-to-date account of the spatial distribution of<br />
poverty in Ireland and critically assess its implications for<br />
the range of issues that we have outlined. In so doing we<br />
draw on a number of different data sources. The 1998 study<br />
used several distinct data sources. The first was the 1994<br />
Living in Ireland Survey, where results could be compared<br />
with those from a comparable household survey conducted<br />
by the ESRI in 1987. The second source was the Small Area<br />
Population Statistics (SAPS) derived from the Census of the<br />
Population and available for 1986 and 1991, constituting the<br />
most geographically disaggregated and comprehensive<br />
information on the socio-demographic structure of the<br />
country. In seeking to look once again at the spatial<br />
concentration of poverty, we can now make use of the<br />
SAPS data from the 2002 Census, the Living in Ireland<br />
Survey conducted in 2000 and a major new source of<br />
information that has become available, namely the<br />
2001/2002 National Survey of Housing Quality (NSHQ). The<br />
NSHQ was conducted by the ESRI on behalf of the<br />
Department of the Environment and Local Government in<br />
over 40,000 households. By combining these sources we<br />
hope to provide a reasonably comprehensive account of the<br />
spatial distribution of poverty and deprivation, making use of<br />
both direct and indirect measures of deprivation.<br />
Furthermore, the fact that the NSHQ combines a very large<br />
sample with detailed household information will allow us to<br />
go beyond the analysis reported in the previous study in<br />
some important respects. While the NSHQ does not allow<br />
for the same level of disaggregation as the SAPS, crucially it<br />
contains measures of income and deprivation for the<br />
9
Chapter 01 7/7/05 6:53 am Page 10<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
participating households, which can be analysed at<br />
household level. It is precisely the absence of such<br />
information in the data made available from SAPS that leads<br />
to the use of composite indicators for small areas as<br />
indicators of socio-economic deprivation, about which<br />
researchers have expressed serious reservations. This issue<br />
is taken up in detail in the analysis of the 2002 SAPS that<br />
follows in Chapter 2.<br />
1.4 Outline of the Study<br />
In Chapter 2 we consider geographical variations in some of<br />
the underlying socio-demographic correlates of poverty and<br />
deprivation risk as contained in the 2002 Census of<br />
Population. Findings are reported at the level of the 34<br />
counties and county boroughs in the Republic of Ireland. In<br />
Chapter 3 we focus on measurement of poverty and<br />
deprivation at the household level. In so doing we draw<br />
primarily on the series of Living in Ireland Surveys but also<br />
make use of the National Survey of Housing Quality. Chapter<br />
4 documents variation in poverty and deprivation by regional<br />
authority and local authority areas. In Chapter 5 we deal with<br />
variations in poverty by area type in the sense of population<br />
density and type of housing tenure. In Chapter 6 we focus on<br />
non-monetary deprivation. In particular we address the issue<br />
of whether different types of deprivation tend to vary in a<br />
similar manner across spatial categories.<br />
A further question that is explored is the extent to which there<br />
is evidence of concentrations of multiple disadvantage within<br />
particular areas.<br />
10
Chapter 02 7/7/05 6:54 am Page 11<br />
Chapter 2<br />
Spatial Trends in Deprivation<br />
Surrogates<br />
2.1 Introduction<br />
In this chapter we consider geographical variations in some<br />
of the underlying socio-demographic correlates of poverty<br />
and deprivation risk as contained in the 2002 Census of<br />
Population. The variables considered are largely consistent<br />
with those identified in other studies, e.g. Nolan, Whelan and<br />
Williams 1998; Nolan et al. 2002, as the main drivers of<br />
poverty and deprivation. By presenting this type of analysis<br />
we can provide an understanding of spatial variations in<br />
socio-demographic structures throughout the country and,<br />
accordingly, an interpretational framework for much of the<br />
subsequent analysis.<br />
Four main areas are considered below, as follows:<br />
• age structures<br />
• economic status and activity (including levels of<br />
farming)<br />
• levels of educational attainment<br />
• social class structures.<br />
11
Chapter 02 7/7/05 6:54 am Page 12<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
In the analysis that follows, we present figures and maps at<br />
the level of the 34 counties and County Boroughs in the<br />
Republic of Ireland. The maps presented are based on<br />
quintile distributions at county level, which explains why each<br />
map has five categories in its legend. Each of the five<br />
categories contains approximately seven counties. Therefore,<br />
the numeric range for each category is different from one<br />
map to the other. The legends distribute counties in five<br />
groups from the lowest to highest incidence of risk of the<br />
specific indicators under consideration, but they do not<br />
purport to have an equal distribution in terms of average rates<br />
or levels. Hence the maps provide a visual, graphic<br />
representation of a ranking of the county-level data from<br />
highest to lowest across the country in terms of the five<br />
categories. Each distribution must, of course, have a high<br />
and low point. To properly interpret the maps, the reader<br />
must take account of the difference between the highest<br />
figure in the top quintile and the lowest in the bottom quintile.<br />
The authors would also point out that we are acutely aware<br />
that the county distributions may mask very substantial subcounty<br />
variations at, for example, a rural/urban district<br />
(RD/UD) or district electoral division (DED) level. There are<br />
approximately 220 RD/UDs in the country and 3,400 DEDs.<br />
While it would clearly be preferable to undertake the analysis<br />
contained in this chapter at that level of spatial<br />
disaggregation, it would be well beyond the scope of the<br />
current report to do so.<br />
2.2 Age Structures<br />
Regional variations in age structures are outlined in Table 2.1.<br />
(Further county variations are shown in Appendix 1, Table<br />
A1.1.) Details are presented on the percentage of persons in<br />
12
Chapter 02 7/7/05 6:54 am Page 13<br />
Spatial Trends in Deprivation Surrogates<br />
Table 2.1: Percentage of persons in each region according to age cohort<br />
Persons<br />
65+<br />
Age living<br />
0–14 15–29 30–44 45–64 65+ dependent alone<br />
Region % Rank % Rank % Rank % Rank % Rank % Rank % Rank<br />
Border 22.5 3 22.3 8 21.0 7 21.7 5 12.5 2 35.0 1 3.5 1<br />
Midlands 23.0 2 22.5 6 21.8 4 21.0 6 11.6 5 34.6 2 3.1 3<br />
West 21.1 6 23.0 5 20.9 8 22.0 1 12.9 1 34.0 3 3.4 2<br />
Dublin 19.2 8 27.7 1 22.7 2 20.2 8 10.2 7 29.3 8 2.6 7<br />
Mid-East 23.4 1 23.6 3 24.2 1 20.6 7 8.2 8 31.6 7 1.9 8<br />
Mid-West 21.2 5 23.8 2 21.4 6 21.9 3 11.6 5 32.8 5 3.0 5<br />
South-East 22.1 4 22.4 7 21.8 4 21.9 3 11.8 4 33.9 4 3.0 5<br />
South-West 20.8 7 23.3 4 21.9 3 22.0 1 12.0 3 32.7 6 3.0 3<br />
Total 21.1 24.4 22.1 21.2 11.1 32.3 2.9<br />
Source: Census 2002<br />
13
Chapter 02 7/7/05 6:54 am Page 14<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
each of five age cohorts, i.e. 0–14 years, 15–29 years, 30–44<br />
years, 45–64 years and 65 years and over. At a broad regional<br />
level one can see that the West, Border and South-West<br />
regions have the oldest age profiles, with 12–13 per cent of<br />
their populations aged over 65 years. At the other extreme,<br />
the Mid-East region has the lowest proportion (8.2 per cent)<br />
aged over 65, followed by the Dublin region with 10.2 per<br />
cent.<br />
From the details of Appendix 1, Table A1.1 one can see that<br />
the counties with the oldest population structures include<br />
Leitrim (16.1 per cent aged 65 and over), Roscommon (15.5<br />
per cent), Mayo (14.7 per cent) and Cavan (13.8 per cent).<br />
An alternative, and more succinct, way of considering<br />
imbalances in age structures is with the concept of age<br />
dependency. This is expressed as the ratio of the combined<br />
population aged 0–14 years and 65+ years to the total<br />
population. As such, it gives a summary measure of the ratio<br />
of the dependent to the non-dependent population in a<br />
region. The summary age dependency figures are shown in<br />
Map 2.1 as well as Tables 2.1 and Appendix 1, Table A1.1.<br />
Map 2.1 graphically illustrates that there is a much higher<br />
level of age dependency in the Border and Western regions,<br />
especially in Donegal, Mayo, Roscommon, Leitrim, Longford<br />
and Cavan. It is equally obvious that the lowest levels are<br />
found in the Dublin and Mid-East regions.<br />
Details on the percentage of persons aged 65 years or more<br />
and living alone are outlined in Map 2.2. The concentrations<br />
of elderly persons in parts of the Border, Midland and<br />
Western (BMW) Region is very clear from the map,<br />
particularly in Leitrim, Roscommon, Mayo, Cavan and<br />
Longford. Rates of 4.0 to 5.1 per cent are obvious in each of<br />
14
Chapter 02 7/7/05 6:54 am Page 15<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.1: Per cent age-dependent by local authority area<br />
these counties, compared with a national total of 2.9 per<br />
cent. This implies, for example, that Leitrim, with 5.1 per cent<br />
of its population aged 65 years and living alone, is ‘overrepresented’<br />
to the order of 76 per cent in terms of this<br />
elderly segment of its population.<br />
15
Chapter 02 7/7/05 6:54 am Page 16<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 2.2: Per cent of persons age 65 and over and living alone<br />
<br />
In summary, therefore, Maps 2.1 and 2.2 indicate a<br />
substantial over-representation of older persons in the Border<br />
and West regions.<br />
2.3 Economic Status and Activity<br />
Commonly used measures of disadvantage, deprivation and<br />
socio-economic imbalance are labour force participation<br />
rates, unemployment rates and levels of economic<br />
dependency.<br />
16
Chapter 02 7/7/05 6:54 am Page 17<br />
Spatial Trends in Deprivation Surrogates<br />
Table 2.2 and Appendix 1, Table A1.2 along with Maps 2.3 and<br />
2.4 present details on labour force participation and<br />
unemployment rates. One can see from these that participation<br />
rates are highest in the Mid-East, Mid-West and Dublin. For<br />
example, counties Fingal and South Dublin have the highest<br />
levels of labour force participation with rates of 64.8 and 64.6<br />
per cent, respectively.<br />
Map 2.3: Labour force participation rate by local authority area<br />
17
Chapter 02 7/7/05 6:54 am Page 18<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
The lowest rates are in Cork County Borough, Leitrim,<br />
Roscommon, Mayo, Donegal and Longford, all in the range of<br />
52–54 per cent – nearly six percentage points below the<br />
national average. Generally, these are the areas identified in<br />
the previous section as having an above average percentage<br />
of their population in the older cohorts, most of whom are<br />
retired rather than economically active.<br />
Map 2.4: Unemployment rate by local authority area<br />
18
Chapter 02 7/7/05 6:54 am Page 19<br />
Spatial Trends in Deprivation Surrogates<br />
Unemployment rates among labour force participants are also<br />
outlined in Table 2.2 and are plotted in Map 2.4. These show<br />
that the national rate was 8.8 per cent. Rates are the highest<br />
in the Border region (12 per cent) followed by the South-East<br />
(9.6 per cent), with Dublin and the West having the lowest per<br />
cent (5.2 per cent).<br />
Table 2.2: Regional labour force participation and<br />
unemployment rates<br />
Labour force Unemployment<br />
participation rate<br />
rate<br />
Rate Rank Rate Rank<br />
Border 56.1 6 12.0 1<br />
Midlands 57.4 4 9.1 3<br />
West 55.6 8 5.2 7<br />
Dublin 61.2 3 5.2 7<br />
Mid-East 61.5 1 6.8 6<br />
Mid-West 61.5 1 8.2 4<br />
South-East 56.6 5 9.6 2<br />
South-West 56.0 7 8.2 4<br />
Total 58.3 8.8<br />
Source: Census 2002<br />
Some of these regional trends are strongly influenced by<br />
individual counties. For example, the unemployment rates in<br />
Donegal and Louth are significantly high compared to the<br />
national average of 8.8 percent, being 15.6 per cent and 13.2<br />
per cent, respectively. The levels in the county boroughs of<br />
Limerick and Cork are also particularly high.<br />
An alternative way of considering the spatial distribution of<br />
the unemployed is presented in Table 2.3 (the regional level)<br />
19
Chapter 02 7/7/05 6:54 am Page 20<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
and Appendix 1, Table A1.3 (the county level). The tables<br />
show the percentage of the total number of unemployed<br />
persons according to their distribution across regions and<br />
counties. For comparative purposes the table also shows the<br />
distribution of all persons aged 15 years and over who are in<br />
the labour force. If the unemployed were distributed on a pro<br />
rata basis across the counties in line with the total labour<br />
force, the percentage figures in Columns A and B of the table<br />
would be identical. The disparity between the figure in<br />
Columns A and B for any county is a measure of the over- or<br />
under-concentration of the unemployed in a given county. In<br />
Column C we provide details of the ratio of the percentage of<br />
the unemployed to the percentage of the labour force<br />
contained in each county. A figure which is greater than 1 in<br />
Column C indicates an over-representation of the<br />
unemployed in a given county. A figure of less than 1 in<br />
Column C indicates an under-representation of the<br />
unemployed in a county.<br />
If one considers the figures for Donegal as shown in<br />
Appendix 1, Table A1.3, for example, one can see that it<br />
contained 5.6 per cent of the unemployed, as recorded in the<br />
2002 Census of Population. The county contained 3.2 per<br />
cent of the labour force as a whole. This implies an overconcentration<br />
of the unemployed of the order of 76 per cent<br />
in the county. In order words, the proportion of unemployed<br />
in the county is 1.76 times the level it would be if the<br />
unemployed were distributed on a pro rata basis with the<br />
total labour force. The counties and regions are ranked in<br />
Column D in terms of their level of over- or underconcentration<br />
of the unemployed, using the total labour force<br />
as a comparator. This shows that Donegal has the highest<br />
level of over-concentration (1.76 times the proportion if<br />
distributed on a pro rata basis). This is followed by Limerick<br />
20
Chapter 02 7/7/05 6:54 am Page 21<br />
Spatial Trends in Deprivation Surrogates<br />
Table 2.3: Distribution of the unemployed across regions<br />
% of the % 15+ in Ratio of<br />
Region unemployment labour force unemployment<br />
to labour force<br />
A B C D<br />
Ratio Rank<br />
Border 14.1 10.4 1.35 1<br />
Midlands 5.7 5.5 1.03 4<br />
West 9.8 9.3 1.05 3<br />
Dublin 29.7 30.8 0.96 5<br />
Mid-East 8.3 10.8 0.77 8<br />
Mid-West 7.9 8.5 0.93 6<br />
South-East 11.2 10.4 1.08 2<br />
South-West 13.3 14.3 0.93 6<br />
Total 100.0 100.0<br />
Source: Census 2002<br />
County Borough (1.57 times), Louth (1.50 times), Cork and<br />
Waterford County Boroughs (each approximately 1.37–1.40<br />
times) and so on. At the other extreme, Dún Laoghaire-<br />
Rathdown and Kildare are as low as 0.65 and 0.70,<br />
respectively, followed by Cork county, Meath and Limerick<br />
county (each in the region of 0.72–0.74).<br />
Given the importance of unemployment as a determinant of<br />
poverty and deprivation, it would be tempting to focus on it in<br />
isolation as an all-encompassing proxy for disadvantage.<br />
Accordingly, given the very low levels of unemployment in the<br />
Border counties mentioned above, one may feel that the risk<br />
of poverty and disadvantage in these areas may be<br />
correspondingly below the national average.<br />
21
Chapter 02 7/7/05 6:54 am Page 22<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
In interpreting the figures, however, one must take account<br />
of the importance of farming in determining regional<br />
unemployment rates. Tables 2.4 and Appendix 1,<br />
Table A1.4 along with Maps 2.5 to 2.8 provide details<br />
on regional variations in the extent of farming in Ireland.<br />
Table 2.4: Percentage of all persons engaged in farming by<br />
number of acres farmed<br />
Region % of 15+ Farmers Farmers Farmers<br />
persons
Chapter 02 7/7/05 6:54 am Page 23<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.5: Per cent of persons in farming<br />
From Map 2.5 one can see that a high percentage of persons<br />
aged 15 years and over engaged in farming is found in the<br />
counties of Cavan, Roscommon and Leitrim – each with rates<br />
of the order of 10 to 12 per cent compared with the national<br />
total of 4.4 per cent.<br />
23
Chapter 02 7/7/05 6:54 am Page 24<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
In Maps 2.6 to 2.8 we present details on the scale of farming<br />
in question. These clearly indicate that counties with the<br />
greatest extent of small-scale farming activity are<br />
substantially located in the Border and West regions.<br />
Appendix 1, Table A1.4 indicates that at a national level 12.3<br />
per cent of all farmers work less than 30 acres. Apart from<br />
Dublin, where less than 1 per cent of the population is<br />
Map 2.6: Per cent of farmers farming less than 30 acres<br />
24
Chapter 02 7/7/05 6:54 am Page 25<br />
Spatial Trends in Deprivation Surrogates<br />
engaged in farming, the regional rates of farming less than 30<br />
acres are highest in the Border and the West regions (18.4<br />
and 18.2 per cent, respectively). At a county level, excluding<br />
the county boroughs, we find that the figure for Mayo is 23.4<br />
per cent, 21.8 per cent in Donegal and 20.4 per cent in<br />
Monaghan. Similarly, from Map 2.7 one can see a high<br />
proportion of farmers in the range of 30–49 acres in these<br />
Map 2.7: Per cent of farmers farming 30–49 acres<br />
25
Chapter 02 7/7/05 6:54 am Page 26<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
same counties. In contrast, Map 2.8 shows the much higher<br />
percentages of larger commercial farming enterprises in the<br />
South-East region. A total of 74 per cent of all farmers in the<br />
country were recorded as farming 50 or more acres. In<br />
Kilkenny the figure was 86.9 per cent, with high rates also<br />
found in Waterford County (86.8 per cent) and North<br />
Tipperary (86.3 per cent).<br />
Map 2.8: Per cent of farmers farming 50+ acres<br />
26
Chapter 02 7/7/05 6:54 am Page 27<br />
Spatial Trends in Deprivation Surrogates<br />
The pattern of agriculture in Ireland, therefore, can be<br />
characterised as being dominated by small-scale activity in<br />
the Western and Border counties with larger-scale<br />
enterprises being much more generally found in the South-<br />
East regions. Although no accurate figures are available,<br />
relatively higher levels of under-employment are often<br />
associated with small-scale farming. Accordingly, when<br />
considering regional trends in unemployment rates as<br />
shown in Map 2.4, one must take account of the role played<br />
by small-scale farming in determining measured<br />
unemployment rates. In general, therefore, although it would<br />
appear that unemployment rates are relatively low in many<br />
Border and Western counties, one must bear in mind that<br />
many of these same counties contain the highest levels of<br />
small-scale farming activity in the country. This latter may,<br />
at least to some degree, be masking recorded<br />
unemployment in the form of under-employment.<br />
In Table 2.5, Appendix 1, Table A1.5 and Map 2.9 we<br />
present details on a summary measure of economic<br />
dependency. This can be defined in various ways. For the<br />
purposes of the current report we have derived a<br />
dependency measure based on the ratio of those who are<br />
economically inactive to those who are economically active.<br />
The former includes the following categories: persons aged<br />
0–14 years, students, unemployed, first job seekers, those<br />
on home duties, retired, unable to work and ‘other’ labour<br />
force status categories. The economically active are simply<br />
those classified as being ‘at work’. The figures in Table 2.5<br />
and Map 2.9, therefore, provide details on the extent to<br />
which those who are economically active in each region or<br />
county are economically ‘supporting’ those who are<br />
inactive.<br />
27
Chapter 02 7/7/05 6:54 am Page 28<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 2.5: Index of economic dependency in each region<br />
Region<br />
Economic Dependency<br />
Index<br />
Rank<br />
Border 1.61 1<br />
Midlands 1.49 4<br />
West 1.51 2<br />
Dublin 1.21 8<br />
Mid-East 1.28 7<br />
Mid-West 1.43 6<br />
South-East 1.51 2<br />
South-West 1.46 5<br />
Total 1.39<br />
From Table 2.5 one can see that at a national level we had<br />
1.39 inactive persons for every person who was ‘at work’ in<br />
the 2002 Census. In general terms the figures suggest that<br />
levels of economic dependency are quite mixed across the<br />
country.<br />
Highest levels are found in the Border region (1.61), but these<br />
are largely driven by Donegal (1.84 per cent). This reflects the<br />
relatively elderly age structure of the county, combined with<br />
its very high level of recorded unemployment. The South-East<br />
and West also have high levels of economic dependency<br />
(1.51 per cent each). This is followed by the Midlands and the<br />
South-West having levels of economic dependency higher<br />
than the national average (1.49 and 1.46 per cent,<br />
respectively). This variability across the country and lack of a<br />
well-defined regional pattern reflects the fact that economic<br />
dependency is substantially affected by a number of factors,<br />
including regional age structure, unemployment rates and the<br />
related issues of agricultural activity, especially small-scale<br />
farming, as discussed above.<br />
28
Chapter 02 7/7/05 6:54 am Page 29<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.9: Economic dependency rates by local authority area<br />
In summary, therefore, in this section we saw that the lowest<br />
levels of labour force participation were in counties such as<br />
Leitrim, Cork County Borough, Roscommon, Mayo, Donegal<br />
and Longford. Unemployment rates were highest in parts of<br />
the Border region (largely driven by the very high rate<br />
recorded in Donegal) as well as the South-East. The level of<br />
29
Chapter 02 7/7/05 6:54 am Page 30<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
unemployment in some Border and Western counties seemed<br />
to be lower than national levels. For example, Cavan and<br />
Roscommon both had recorded unemployment rates below<br />
8 per cent compared with a national figure of 8.8 per cent.<br />
However, we found that the Border and Western counties<br />
which were associated with low levels of recorded<br />
unemployment were equally characterised with high levels of<br />
small-scale farming. One can surmise that this latter may<br />
reflect elderly age structures and may, at least to some<br />
degree, mask recorded unemployment by under-employment.<br />
2.4 Variations in Levels of Educational Attainment<br />
Given the demonstrated relationship between poverty and<br />
disadvantage on the one hand and level of educational<br />
attainment on the other, it is clearly important to consider<br />
regional variations in the level of education as an aid to<br />
understanding the spatial patterning of disadvantage. The<br />
relevant figures are outlined in Table 2.6, Appendix 1, Table<br />
A1.6 and Maps 2.10 to 2.13.<br />
It is clear that the Border and West regions stand out as<br />
having an above average percentage of persons with lower<br />
levels of attainment. The national figure of 22.2 per cent of<br />
persons who are recorded as having left education with no<br />
qualifications or primary level only compares with a figure of<br />
29 per cent in the Border counties and 26 per cent for the<br />
Western region. One can see that 34 per cent of persons in<br />
Donegal, 31 per cent in Cavan, 30 per cent in Mayo and 29<br />
per cent each in Leitrim and Monaghan have left full-time<br />
education with, at most, primary-level education. Outside the<br />
counties in the Border and West regions only Longford,<br />
Offaly, Laois and Wexford (each with 26–30 per cent) have a<br />
30
Chapter 02 7/7/05 6:54 am Page 31<br />
Spatial Trends in Deprivation Surrogates<br />
Table 2.6: Distribution of persons 15 years and over whose<br />
education has ceased, classified by highest level of<br />
educational attainment<br />
Region None/ Lower Upper Third<br />
primary secondary secondary level<br />
% Rank % Rank % Rank % Rank<br />
Border 29.3 1 25.2 3 25.7 8 19.9 6<br />
Midlands 25.5 3 25.4 2 29.9 3 19.3 8<br />
West 25.9 2 21.2 7 29.1 6 23.7 4<br />
Dublin 18.7 7 19.2 8 28.7 7 33.5 1<br />
Mid-East 18.5 8 23.1 6 31.3 1 27.2 2<br />
Mid-West 22.3 5 23.8 4 30.6 2 23.3 5<br />
South-East 24.0 4 26.7 1 29.5 4 19.8 7<br />
South-West 21.3 6.0 23.8 4.0 29.4 5.0 25.5 3.0<br />
Total 22.2 22.7 29.1 26.0<br />
Source: Census 2002<br />
particularly high percentage of persons with at most<br />
primary-level education (see Map 2.10). In contrast, the<br />
counties with the highest percentages of third-level<br />
graduates are generally in Dublin and the Mid-East regions<br />
(see Map 2.13).<br />
In interpreting these figures it is important to remember that<br />
educational structures are to a large extent influenced by age<br />
structures of a region. The areas identified above as having<br />
lower than average levels of attainment are generally regions<br />
of the country which have an over-concentration of older<br />
persons. Overall, therefore, Maps 2.10 to 2.13 would seem to<br />
confirm a relatively higher level of educational disadvantage<br />
(much of which is related to age structures) in the counties of<br />
the Border and Western regions.<br />
31
Chapter 02 7/7/05 6:54 am Page 32<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 2.10: Per cent of persons age 15+ who have left school<br />
with no education or primary education<br />
In Table 2.7 we provide information on the levels of over- or<br />
under-concentration of those with the lowest and highest<br />
levels of educational attainment at a regional level. In Column<br />
A of the table we show the percentage of persons aged 15<br />
years and over in each county who have left full-time<br />
32
Chapter 02 7/7/05 6:54 am Page 33<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.11: Per cent of persons age 15+ who have left school<br />
with lower secondary education<br />
education with no formal education or primary level only.<br />
Column B presents a comparable figure in respect of those<br />
who leave with third level. Column C provides details on the<br />
percentage of all persons who have left education in the<br />
region. This is the most appropriate comparator of the<br />
33
Chapter 02 7/7/05 6:54 am Page 34<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 2.12: Per cent of persons age 15+ who have left school<br />
with upper secondary education<br />
distributions in Columns A and B. The ratio of the figures in<br />
Columns A and B to those in Column C provides a measure<br />
of the level of over- or under-concentration of persons into<br />
low and high levels of education. From Column D of the table<br />
34
Chapter 02 7/7/05 6:54 am Page 35<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.13: Per cent of persons age 15+ who have left school<br />
with third-level education<br />
one can see that over-concentration of persons with lower<br />
levels of educational attainment is characteristic of Border,<br />
West and Midland regions to the extent of 33 per cent, 18 per<br />
cent and 15 per cent, respectively. In contrast, one can see<br />
35
Chapter 02 7/7/05 6:54 am Page 36<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
that only in the Mid-East and Dublin regions do we have an<br />
over-representation of those with third-level qualifications by<br />
5 per cent and 27 per cent, respectively.<br />
Table 2.7: Distribution of persons whose education has ceased<br />
in highest and lowest levels of educational attainment across<br />
regions<br />
A B C D E<br />
Region % None/ % Third % Pop. Ratio Ratio<br />
primary level who left none/ third level<br />
education primary to to total<br />
total left left<br />
education education<br />
Ratio Rank Ratio Rank<br />
Border 14.6 8.4 11.0 1.33 1 0.77 6<br />
Midlands 6.5 4.2 5.7 1.15 3 0.74 8<br />
West 11.3 8.8 9.6 1.18 2 0.92 4<br />
Dublin 24.0 36.8 29.0 0.83 8 1.27 1<br />
Mid-East 8.7 10.9 10.3 0.84 7 1.05 2<br />
Mid-West 8.7 7.8 8.6 1.01 5 0.91 5<br />
South-East 11.8 8.3 10.9 1.09 4 0.77 6<br />
South-West 14.3 14.7 14.9 0.96 6 0.99 3<br />
Total 100.0 100.0 100.0<br />
Source: Census 2002<br />
At a county level one can see from Appendix 1, Table A1.7<br />
that counties such as Donegal, Cavan, Mayo and Monaghan<br />
have over-concentrations of persons with low levels of<br />
education to the extent of 32–53 per cent. At the other<br />
extreme, Dún Laoghaire-Rathdown is over-represented by<br />
persons with higher level of attainment by 76 per cent. Also in<br />
36
Chapter 02 7/7/05 6:54 am Page 37<br />
Spatial Trends in Deprivation Surrogates<br />
the Dublin region, Fingal is over-represented by 28 per cent.<br />
Galway County Borough also has a high over-representation<br />
of persons with higher levels of educational attainment by 55<br />
per cent.<br />
2.5 Variations in Levels of Social Class<br />
One of the more direct proxies of advantage and<br />
disadvantage available from census data is social class. The<br />
importance of class in determining poverty and disadvantage<br />
is well established. The census classifies all persons into five<br />
broad class categories, as follows:<br />
• higher professional workers/managerial/technical<br />
• non-manual<br />
• skilled manual<br />
• semi-skilled manual<br />
• unskilled manual.<br />
The classification is based on current or former occupation.<br />
Persons engaged in home duties, those classified as<br />
‘students’ and those who are otherwise economically inactive<br />
are assigned the class status of their family head. 1<br />
The percentage of persons in each social class is shown in<br />
Table 2.8, Appendix 1, Table A1.8 and Maps 2.14 to 2.18. If<br />
one concentrates initially on the percentage of persons in<br />
1 There is a residual ‘unknown’ category. This is used for persons who<br />
have no occupations in their own right and who live in households<br />
headed by someone who has never worked.<br />
37
Chapter 02 7/7/05 6:54 am Page 38<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Professional/Managerial and Technical classes, one can see<br />
that the regions with the highest rates are in Dublin (44.4 per<br />
cent) and the Mid-East region (41 per cent). Counties such as<br />
Dún Laoghaire-Rathdown have an extremely high proportion<br />
of their population classified as Higher/Lower Professional –<br />
60 per cent compared with a national total of 39 per cent.<br />
This is followed by Fingal and Galway County Borough (both<br />
47 per cent). At the other extreme, counties in the Border and<br />
West regions have a much lower percentage of population<br />
classified in the highest class category. In Donegal only 30<br />
per cent are classified as Professional/Managerial or<br />
Technical. The rates are also low in Monaghan and Cavan<br />
(31 per cent each) and Offaly and Mayo (32 per cent each)<br />
and in some of the main urban areas such as Limerick<br />
County Borough, with only 31 per cent of the population<br />
classified as being Professional/Managerial/Technical.<br />
Table 2.8: Percentage of persons in each social class<br />
Professional/<br />
Managerial/ Non- Skilled Semi-<br />
Region Technical manual manual skilled Unskilled<br />
% Rank % Rank % Rank % Rank % Rank<br />
Border 32.2 8 19.7 6 23.9 1 15.8 1 8.4 2<br />
Midlands 33.9 7 20.2 2 23.1 2 14.5 3 8.3 3<br />
West 36.5 5 19.9 4 22.0 4 14.4 4 7.1 5<br />
Dublin 44.4 1 21.3 1 18.1 8 11.0 8 5.2 8<br />
Mid-East 41.0 2 19.5 7 21.1 6 11.8 7 6.6 7<br />
Mid-West 37.4 4 20.2 2 21.0 7 14.2 5 7.2 4<br />
South-East 34.5 6 19.0 8 23.1 2 14.7 2 8.8 1<br />
South-West 37.8 3 19.8 5 21.2 5 14.2 5 6.9 6<br />
Total 38.6 20.2 21.0 13.3 6.9<br />
Source: Census 2002<br />
38
Chapter 02 7/7/05 6:54 am Page 39<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.14: Per cent of persons classified as Professional/<br />
Managerial/Technical<br />
In broad terms one can see from Map 2.14 that the overall<br />
pattern across the country clearly indicates that the areas<br />
with the lowest percentage of higher/lower professionals are<br />
the Border and Western counties while Dublin, the Mid-East<br />
and Cork County have the highest proportions in the country.<br />
39
Chapter 02 7/7/05 6:54 am Page 40<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 2.15: Per cent of persons classified as ‘Other Non-Manual’<br />
In contrast, one can see from Map 2.18 that at the other end<br />
of the social class spectrum, Donegal (10.3 per cent) and<br />
several counties in the South-East (South Tipperary 9.7 per<br />
cent, Wexford 9.3 per cent and Carlow 9.2 per cent) have the<br />
highest percentages of persons in the Unskilled Manual<br />
category.<br />
40
Chapter 02 7/7/05 6:54 am Page 41<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.16: Per cent of persons classified as Skilled Manual<br />
In interpreting regional distributions of social class, the reader<br />
should note that farmers are assigned to social class<br />
categories according to acreage farmed. Small farmers (less<br />
than 30 acres) are assigned to the Semi-skilled Manual<br />
category. Large farmers (with 200 or more acres) are assigned<br />
to the Higher Professional category. No farmers are assigned<br />
to the unskilled manual category.<br />
41
Chapter 02 7/7/05 6:54 am Page 42<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 2.17: Per cent of persons classified as Semi-skilled<br />
Manual<br />
Table 2.9 provides details on the over- or under-concentration<br />
of persons from the top and bottom social class categories<br />
across the regions. The details in the table are comparable to<br />
those presented in Table 2.7 in respect of level of educational<br />
attainment. From Columns D and E in the table one can see<br />
42
Chapter 02 7/7/05 6:54 am Page 43<br />
Spatial Trends in Deprivation Surrogates<br />
Map 2.18: Per cent of persons classified as Unskilled Manual<br />
that the largest over-concentration of the Professional social<br />
class category is in the Dublin region (13 per cent), followed<br />
by the Mid-East (10 per cent). In contrast, we find the<br />
greatest over-concentration of the Unskilled Manual category<br />
in the South-East region (29 per cent). At the county level<br />
43
Chapter 02 7/7/05 6:54 am Page 44<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
(see Appendix 1, Table A1.9), the highest relative<br />
concentration of this class group is found in Donegal,<br />
Wexford and South Tipperary (each in the range of 41–49<br />
per cent).<br />
Table 2.9: Regional distribution of Professional and Unskilled<br />
Manual class categories<br />
Region<br />
A B C D E<br />
Professional/ Un- % Ratio Ratio<br />
Managerial/ skilled Total profes- Unskilled<br />
Technical Manual popula- sional- Manual<br />
tion to total to total<br />
ratio<br />
Ratio Rank Ratio Rank<br />
Border 9.2 13.5 11.0 0.84 8 1.23 2<br />
Midlands 5.0 6.9 5.8 0.87 7 1.19 3<br />
West 9.0 9.9 9.7 0.93 5 1.02 5<br />
Dublin 32.4 21.2 28.7 1.13 1 0.74 8<br />
Mid-East 11.6 10.4 10.5 1.10 2 0.99 7<br />
Mid-West 8.4 9.1 8.7 0.97 4 1.04 4<br />
South-East 9.8 14.0 10.8 0.90 6 1.29 1<br />
South-West 14.6 15.0 14.8 0.99 3 1.02 5<br />
Total 100.0 100.0 100.0<br />
In general terms, therefore, it would appear that Dublin, the<br />
Mid-East and parts of the South-West region (specifically<br />
Cork) have the highest levels of their populations in the<br />
Professional/Managerial/Technical category. In terms of the<br />
Unskilled Manual group, although some counties in the<br />
South-East of the country have among the highest rates,<br />
there would generally appear to be a less clearly defined and<br />
more mixed pattern than is apparent with some of the other<br />
distributions considered.<br />
44
Chapter 02 7/7/05 6:54 am Page 45<br />
Spatial Trends in Deprivation Surrogates<br />
2.6 Conclusions<br />
In this chapter we considered spatial variations in sociodemographic<br />
variables related to poverty and deprivation<br />
from the 2002 Census of Population. The somewhat<br />
aggregated county level of the analysis undoubtedly masked<br />
a substantial degree of sub-county variation. Nonetheless,<br />
the chapter provides the reader with a clear picture of<br />
regional variations in the relevant indicators and in so doing<br />
adds considerably to the interpretative context for much of<br />
our discussion in later chapters of regional trends in direct<br />
measures of poverty and deprivation.<br />
Throughout the chapter we saw that the Border and West<br />
regions contained counties with the highest percentages of<br />
their population who were elderly, with the lowest levels of<br />
educational attainment, highest incidence of small farming<br />
activity and high levels of economic dependency.<br />
Throughout the chapter we emphasised that what we were<br />
examining were spatial patterns and outcomes. The<br />
underlying causal processes and drivers will be explored in<br />
later chapters, particularly in Chapter 7. These drivers may<br />
well be individual- or household-level characteristics such as<br />
age and housing tenure.<br />
45
Chapter 03 7/7/05 6:55 am Page 46<br />
Chapter 3<br />
Measurement of <strong>Poverty</strong> at the<br />
Household Level<br />
3.1 Introduction<br />
In this chapter we focus on the measurement of poverty and<br />
deprivation at the household level. In doing so we draw<br />
primarily on the series of Living in Ireland Surveys (LII)<br />
conducted in 1987, 1994 and 2000 but also on the National<br />
Survey of Housing Quality (NSHQ), which was conducted in<br />
2001/2002. All four surveys were designed to provide a<br />
nationally representative sample of the population resident in<br />
private households. The technical properties of these surveys<br />
have been described in detail elsewhere 1 and a summary of<br />
their major features is contained in Appendix 2.<br />
The major advantage of the LII surveys is that they contain<br />
the key indicators on which national reports monitoring<br />
poverty and deprivation are based and they allow us to track<br />
change over time using the same set of indicators. The major<br />
limitation of the LII surveys for our present purposes is that<br />
despite the fact that they constitute very large samples by<br />
1 Full descriptions are provided in Callan et al. (1989 and 1996) and<br />
Nolan et al. (2002), respectively, and Watson and Williams (2003).<br />
46
Chapter 03 7/7/05 6:55 am Page 47<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
social science standards, the extent to which we can<br />
disaggregate by spatial units is strictly limited.<br />
The NSHQ, which was conducted by the ESRI on behalf of<br />
the Department of the Environment and Local Government,<br />
has a substantially larger sample with 40,000 households.<br />
The NSHQ contains measures of income and deprivation for<br />
the participating households, which can be analysed at<br />
household level. However, as will become clear, these<br />
measures do not correspond precisely to those employed in<br />
the LII surveys. Fortunately, as we explain below, steps can<br />
be taken to ensure that while statements concerning absolute<br />
levels of deprivation and poverty are not possible using the<br />
NSHQ data, conclusions concerning relativities are robust.<br />
This allows us to take advantage of the much larger sample<br />
size in the NSHQ to achieve a much greater degree of<br />
geographical disaggregation when considering such<br />
relativities.<br />
3.2 Measuring <strong>Poverty</strong> and Deprivation in the LII<br />
Surveys<br />
We have extensively discussed elsewhere the many issues<br />
involved in the definition and measurement of poverty<br />
(most recently in Layte et al. 2001 and Nolan et al. 2002).<br />
Without duplicating that discussion, for the present study<br />
we take as our starting point that the concept of poverty is<br />
an explicitly relative one, relating to exclusion from the<br />
ordinary life of society due to lack of resources. We<br />
continue to stress the importance of acknowledging, where<br />
relevant, any uncertainty and absence of robustness in<br />
results. We therefore employ several different relative<br />
income poverty lines and a measure that takes into<br />
47
Chapter 03 7/7/05 6:55 am Page 48<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
account both income and non-monetary indicators of<br />
deprivation.<br />
Relative income poverty lines are calculated as a particular<br />
percentage of average household income, taking differences<br />
in household size and composition into account. We take into<br />
account the fact that adults have greater needs than children<br />
and that there are economies of scale in consumption by<br />
using equivalence scales to adjust household income. Here<br />
we report results employing an equivalence scale of 0.66 for<br />
additional adults and 0.33 for children. The two relative<br />
income lines employed in this study are 50 and 60 per cent of<br />
mean equivalent income. For an adult living alone this<br />
produced poverty lines in 2000 of about £113/€144 and<br />
£136/€173 per week. The corresponding lines for 1994 were<br />
£65/€83 and £78/€99, respectively, and for 1987 they were<br />
£43/€55 and £51/€65, respectively.<br />
Reliance on income alone is open to the criticism that it may<br />
not be a good measure of low consumption and therefore<br />
deprivation. If poverty is defined as exclusion from the<br />
ordinary life of society due to lack of resources – understood<br />
as a state of generalised deprivation – it should be<br />
characterised by both a low standard of consumption and a<br />
low level of income. We therefore also make use of the<br />
consistent measure of poverty that has been employed in the<br />
National <strong>Poverty</strong> Strategy (Callan, Nolan and Whelan 1993;<br />
Nolan and Whelan 1996). This combines information on<br />
relative income with that relating to basic deprivation. This<br />
comprises a set of eight indicators (see Table 3.1) that relate<br />
to the enforced absence of such items as food, clothing and<br />
heat and going into debt to meet ordinary living expenses.<br />
Previous analysis has shown that these indicators reflect<br />
rather basic aspects of current material deprivation of items<br />
48
Chapter 03 7/7/05 6:55 am Page 49<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
that are possessed by most people, are considered as<br />
necessities by most people and cluster together, providing<br />
support for the view that they serve as valid indicators of<br />
underlying generalised deprivation. The combined income<br />
and deprivation standard that we employ here requires that a<br />
household is counted as poor where it reports enforced lack<br />
of at least one of these basic items and also falls below 60<br />
per cent of mean income. This combined measure is one that<br />
we felt best captures our underlying conceptualisation of<br />
poverty as involving enforced absence of socially defined<br />
necessities. In the current context it is important to stress that<br />
unlike items such as cars, telephones and holidays, there is<br />
no evidence for significant urban-rural difference in the extent<br />
to which the items included in the basic deprivation index are<br />
considered to be necessities.<br />
3.3 Measuring <strong>Poverty</strong> and Deprivation in the NSHQ<br />
The discussion above outlined the issues involved in<br />
measuring poverty and deprivation using the LII surveys,<br />
which had been specifically designed for this purpose. In the<br />
NSHQ we are more limited in terms of what is possible.<br />
Income in the National Survey of Housing Quality is measured<br />
by a single item, which asks for the approximate level of net<br />
household income, and records the answer into one of 16<br />
categories. The wording is as follows:<br />
Finally, a few questions about how you are able to<br />
manage financially. Could I ask about the<br />
approximate level of net household income This<br />
means the total income, after tax and PRSI, of ALL<br />
MEMBERS of the household. It includes ALL TYPES<br />
49
Chapter 03 7/7/05 6:55 am Page 50<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
of income: income from employment, social welfare<br />
payments, Child Benefit, rents, interest, pensions etc.<br />
We would just like to know into which broad group<br />
the total income of your household falls. I’d like to<br />
assure you once again that all information you give<br />
me is entirely confidential.<br />
Respondents were first of all presented with a card showing<br />
four broad income categories. Then they were presented with<br />
a second card that broke down each of these four broad<br />
categories into four more detailed categories. The result was<br />
a 16-category variable for total household income. This item<br />
had a reasonably good response rate, with 87.7 per cent of<br />
respondents providing information on the initial four-category<br />
breakdown and 85.3 per cent providing information on the<br />
more detailed 16-category breakdown. Income category was<br />
imputed for the 12.3 per cent of households for whom the<br />
information was missing using information on household size,<br />
number of persons at work, social class, local authority area<br />
and sample cluster. Further details on the imputation of<br />
missing information in the NSHQ are given in Appendix 2.<br />
The NSHQ single-item measure of income will tend to<br />
understate total household income, particularly in larger<br />
households, in comparison with the Living in Ireland Survey.<br />
The understatement arises for a number of reasons:<br />
incomplete information on the part of the householder<br />
regarding earnings and income of other people in the<br />
household and a tendency to forget some components (such<br />
as Child Benefit and irregular payments) when responding to<br />
a single question.<br />
Data from the 2000 Living in Ireland Survey was used to<br />
develop a correction for the NSHQ single-item income<br />
50
Chapter 03 7/7/05 6:55 am Page 51<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
measure. Details of how this was done are provided in<br />
Appendix 2. Nevertheless, because of the differences in the<br />
measurement of income compared to the LII, it is not<br />
possible to reproduce a measure with the same overall level. 2<br />
What we are relying on in this analysis is that the measure of<br />
poverty risk derived from the NSHQ will faithfully reproduce<br />
the same patterns of variation across sub-groups of<br />
households as the LII measure. To the extent that it was<br />
possible to compare the distribution of risk across regions,<br />
tenure, household size categories, economic status and<br />
social class, the NSHQ measure compares very favourably to<br />
the LII measure 3 of income.<br />
The non-monetary indicators of deprivation available on the<br />
NSHQ are also different from those available in the Living in<br />
Ireland Survey. Some of the items likely to capture an<br />
absence of very basic necessities (such as shoes, coat and a<br />
hot meal) are not available. There are also some important<br />
differences in how ‘basic’ deprivation items are measured. In<br />
the NSHQ, we ask whether the household can afford each<br />
item if it wanted it. In the LII, a two-step approach is taken to<br />
most deprivation items. The household is first asked if it<br />
possesses an item. If the answer is no, the household is<br />
asked if this is something it would like to have but cannot<br />
afford. ‘Enforced lack’ in the LII context, then, refers to a<br />
household (a) lacking an item (b) unable to afford it and (c)<br />
would like it. In the NSHQ, the last condition is missing, so<br />
some households who would not want an item would be<br />
counted as experiencing an enforced lack in the sense that<br />
2 In fact, the measure of average income in the NSHQ tends to be higher<br />
than in the LII and the ‘poverty’ levels about 5–6 percentage points<br />
lower.<br />
3 Detailed comparisons are provided in Appendix 2.<br />
51
Chapter 03 7/7/05 6:55 am Page 52<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
they could not afford the item, even if they did want it. For<br />
this reason, the levels on the NSHQ for enforced lack of<br />
comparable items tend to be higher.<br />
Table 3.1 shows the items used in each deprivation<br />
dimension and the percentage of households that lack them<br />
on the NSHQ. Note that the items are measured differently to<br />
those in the Living in Ireland Survey, as noted earlier, so that<br />
the levels are not directly comparable. A factor analysis 4 was<br />
undertaken to check if the same type of underlying factor<br />
structure emerged in the NSHQ data as in the LII. The<br />
factor loadings set out in Table 3.1 show which items figure<br />
prominently on each of the dimensions. The structure<br />
represented by the set of dimensions identified is very similar<br />
to that identified using the LII data, although there are some<br />
differences in terms of the factors on which specific items<br />
load most highly. The five-factor solution resulted in a basic<br />
dimension, a secondary dimension, two housing dimensions<br />
(housing amenities and housing deterioration) and an<br />
environment dimension.<br />
The basic dimension reflects an inability to afford items such<br />
as a meal with meat every second day, new clothes,<br />
adequate heating, an ability to replace worn furniture, to<br />
socialise with family or friends once a month and to go on a<br />
week’s annual holiday. It also includes an item on whether the<br />
household experienced arrears in the past 12 months on<br />
housing or utility bills. Note that this measure does not<br />
contain the same set of items as were used in the analysis of<br />
the LII data – the same set of items is not available on the<br />
NSHQ. However, this set of items emerged in analyses of the<br />
4 The method used was principal component extraction and oblique<br />
rotation.<br />
52
Chapter 03 7/7/05 6:55 am Page 53<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
internationally comparable European Community Household<br />
Panel Survey (ECHP) as forming a coherent dimension of<br />
deprivation across European countries (Whelan et al. 2001).<br />
The secondary dimension includes an item on inability to<br />
afford a car, as well as items on whether the household would<br />
like but cannot afford a range of household appliances<br />
(microwave, dishwasher, colour TV, video recorder and<br />
telephone). The levels on these household appliance items<br />
are lower than on some of the basic items because of the use<br />
of the two-step method in measuring them. The household<br />
was first asked if it had the item. If not, it was asked whether<br />
the item was something it would like to have but could not<br />
afford. Unlike the basic items, then, households who would<br />
not like the item are not counted as deprived.<br />
The housing amenities dimension captures an absence of<br />
basic housing amenities: a flush toilet, bath or shower and<br />
hot running water.<br />
The housing deterioration index captures the presence of<br />
problems in the accommodation that are regarded as<br />
moderate or major problems by the householder. The<br />
problems are leaks (in roof/doors/windows), dampness (rising<br />
damp, condensation dampness or dampness of unknown<br />
origin), rot (in doors, windows or other timbers) and noise<br />
from neighbouring houses. The index also includes an item<br />
on the adequacy of the space available. It also measures the<br />
householder’s judgement as to whether the accommodation<br />
is too small for the household’s needs.<br />
The final dimension, labelled environment, refers more to the<br />
social environment than to the physical environment. It<br />
includes problems of public order such as vandalism, graffiti,<br />
rubbish or litter lying about, homes or gardens in bad<br />
53
Chapter 03 7/7/05 6:55 am Page 54<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 3.1: Items used in the deprivation scales in the NSHQ<br />
Factor loading<br />
(principal component<br />
extraction; oblique rotation)<br />
Basic<br />
Secondary<br />
Housing<br />
amenities<br />
Housing<br />
deterioration<br />
Environment<br />
%<br />
Deprived<br />
Cannot afford:<br />
Meal with meat, etc. 0.55 4.5<br />
New clothes 0.66 8.0<br />
Adequate heating 0.60 7.9<br />
Socialising once a month 0.74 24.2<br />
Replacing worn furniture 0.76 27.3<br />
Week’s holiday per year 0.77 31.2<br />
Arrears on housing/utility bills 0.26 8.9<br />
Cannot afford car or van 0.24 22.0<br />
Would like but cannot afford:<br />
Microwave 0.66 4.1<br />
Dishwasher 0.51 14.7<br />
Colour TV 0.51 0.4<br />
Video recorder 0.70 2.9<br />
Telephone, incl. mobile 0.56 2.3<br />
Accommodation lacks:<br />
Flush toilet 0.65 1.6<br />
Bath/shower 0.83 1.4<br />
Hot running water 0.79 1.6<br />
Moderate or major problem with:<br />
Leaking roof/doors/windows 0.77 5.8<br />
Dampness 0.78 6.5<br />
Rot in doors/windows/floors 0.69 3.4<br />
Noise from neighbours 0.21 2.8<br />
Accommodation too small for needs 0.30 13.5<br />
Very common in area:<br />
Vandalism 0.79 2.8<br />
Graffiti 0.76 2.6<br />
Rubbish/litter 0.73 5.0<br />
Homes/gardens in bad condition 0.74 1.8<br />
Public drunkenness 0.70 3.1<br />
Reliability (alpha) 0.76 0.46 0.66 0.50 0.79 –<br />
Source: Irish National Survey of Housing Quality 2001/2002<br />
54
Chapter 03 7/7/05 6:55 am Page 55<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
condition or people being drunk in public. The items capture<br />
the householder’s judgement that these problems are very<br />
common in the area where the accommodation is located.<br />
The reliabilities of the scales, i.e. the extent to which the items<br />
are tapping a common underlying dimension, are shown in the<br />
last row of Table 3.1. The values of the indices of reliability are<br />
very respectable for the basic and environment dimensions<br />
and satisfactory for the housing amenity dimension. The<br />
reliability of the secondary dimension and of the housing<br />
deterioration dimension is lower than we would like.<br />
In constructing indices, the items were weighted by the<br />
proportion that does not lack the item. For instance, 96 per<br />
cent of households can afford a meal with meat, chicken or fish<br />
every second day while 69 per cent can afford a week’s annual<br />
holiday away from home. Enforced lack of a meal with meat,<br />
etc. is given a weight of 0.96, while enforced lack of a holiday<br />
is given a lower weight of 0.69. Such weighting by the<br />
proportion not lacking an item captures the intuition that<br />
enforced lack of something possessed by most people is more<br />
‘serious’ than enforced lack of something that fewer people<br />
possess.<br />
3.4 The National Context<br />
Before proceeding to a spatial analysis of poverty and<br />
deprivation, we wish to put our later results in context by<br />
presenting results from earlier Living in Ireland Surveys. This<br />
is done in Table 3.2. In 2000 over one in four respondents<br />
were below the 50 per cent line, in comparison with just less<br />
than one in five in 1994 and one in six in 1987. 5 The picture at<br />
5 The most up-to-date national poverty rate trends can be found in<br />
Whelan et al. (2003).<br />
55
Chapter 03 7/7/05 6:55 am Page 56<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
the 60 per cent line is somewhat different. By 2000 the<br />
poverty rate was lower that that observed in 1994 but higher<br />
than that for 1987. The actual number of households below<br />
the 50 and 60 per cent lines in 2000 was 26 per cent and 33<br />
per cent, respectively.<br />
Significant changes in the composition of households below<br />
the relative income lines took place during the period under<br />
study. Among the most striking findings was the increasing<br />
risk for those in single-person households, in households<br />
where the reference person is ill/disabled or retired and for<br />
those who are themselves aged 65 or over. Those in<br />
households where the reference person is unemployed still<br />
faced a relatively high risk of falling below the income<br />
thresholds but continued to decline as a proportion of all<br />
those below the lines. Conversely, those in households where<br />
the reference person is an employee still faced the lowest risk<br />
by far but are becoming more important among those below<br />
the thresholds (as the numbers of unemployed continued to<br />
fall and the number of employees continued to rise from 1998<br />
to 2000). Those aged 65 or over faced a much higher risk of<br />
falling below 60 or 70 per cent of median income than those<br />
aged 18–65, with children then facing an intermediate level of<br />
risk. Women faced a higher risk of falling below those lines<br />
than men, but this gap was most marked among the elderly.<br />
As well as household income, non-monetary indicators of<br />
deprivation have also been employed in attempting to identify<br />
those in each of the LII surveys who were experiencing<br />
exclusion from ordinary living patterns due to lack of<br />
resources. Analysis of available indicators led us to focus on<br />
the eight indicators included in the basic deprivation index.<br />
The combined income and deprivation consistent poverty<br />
standard employed here is that a household is counted as<br />
56
Chapter 03 7/7/05 6:55 am Page 57<br />
Measurement of <strong>Poverty</strong> at the Household Level<br />
poor where it reports enforced lack of at least one of these<br />
basic items and falls below 60 per cent of mean income.<br />
About 16 per cent of households were in that position in<br />
1987. By 1994 the percentage had fallen marginally to 15 per<br />
cent. In 2000 we observed a substantial reduction to 6 per<br />
cent.<br />
Table 3.2: Trends in percentage of households below poverty<br />
lines<br />
(Equivalence Scale A – 1.0/0.66/0.33)<br />
1987 1994 2000<br />
Percentage of households<br />
below line:<br />
50% line 16.3 18.6 25.8<br />
60% line 28.5 34.2 32.9<br />
60% consistent line 16.0 15.1 6.2<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
3.5 Conclusions<br />
In this chapter we have dealt with the manner in which<br />
income and deprivation are measured at the household level<br />
in the major data sources on which we have based our<br />
analysis. As we have pointed out, the manner of measuring<br />
both variables is sufficiently different in the NSHQ survey that<br />
we feel it would be confusing rather than helpful to report<br />
poverty rates from that survey. However, our exploratory<br />
analysis indicates that very similar conclusions emerge from<br />
both sources when we focus not on levels but on differentials<br />
or disparities and that is the strategy we will pursue when<br />
57
Chapter 03 7/7/05 6:55 am Page 58<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
utilising this data in our subsequent analysis. This will allow<br />
us to take advantage of the fact that the much greater<br />
sample size of the NSHQ allows us to make much finer<br />
distinctions in terms of units of analysis.<br />
58
Chapter 04 7/7/05 6:55 am Page 59<br />
Chapter 4<br />
<strong>Poverty</strong> and Deprivation by Region and<br />
Local Authority Area<br />
4.1 Introduction<br />
In this chapter we present results on poverty and deprivation<br />
by regional authority and by local authority area. Two different<br />
sources of data are used for this purpose: the Living in<br />
Ireland Survey (LII) for 2000 and the National Survey of<br />
Housing Quality (NSHQ) for 2001/2002. The LII Surveys were<br />
designed to provide nationally representative samples of the<br />
population resident in private households. While the numbers<br />
of households in the achieved samples 1 are relatively large,<br />
both sample size and the sampling designs limit the extent to<br />
which they can be disaggregated by geographical area. Thus,<br />
while it would be possible to derive poverty rates by county<br />
from the samples, it would not be sensible to do so because<br />
of the wide margin of error associated with the figures.<br />
Analysis at the level of planning region or regional authority,<br />
each comprising a number of counties, is possible. In this<br />
chapter we focus on the latter. 2 59<br />
1 The completed sample was 3,467 in 2000.<br />
2 However, in order to allow trends over time to be considered, we<br />
report figures for planning region by year in Appendix 1.
Chapter 04 7/7/05 6:55 am Page 60<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
On the other hand, at over 40,000 households, the NSHQ<br />
sample is large enough to provide reliable estimates at the<br />
level of local authority. The NSHQ data are used for this<br />
purpose in this chapter. In using the NSHQ data, we focus on<br />
disparities – or variations across local authorities – in poverty<br />
and deprivation. As noted in the previous chapter, it is not<br />
possible to use the NSHQ to replicate the levels of income<br />
and deprivation used in the LII Survey, but the patterns of<br />
variation across regions are very similar.<br />
In the final section of the chapter, the regional distribution of<br />
poverty and deprivation is compared to the distribution of<br />
characteristics of individuals and households from the 2002<br />
Census, as presented in Chapter 2.<br />
4.2 <strong>Poverty</strong> Risk and Incidence by Regional<br />
Authority in the LII Survey<br />
In Table 4.1 we show the distribution of risk of poverty across<br />
regional authority for all three poverty lines. The risk of<br />
poverty, or poverty rate, is the percentage of households with<br />
equivalised income below the relevant poverty line. 3 The<br />
regional authorities are made up as follows.<br />
• Border: Cavan, Donegal, Leitrim, Louth, Monaghan, Sligo.<br />
• Dublin: Dublin City, Fingal, Dún Laoghaire-Rathdown,<br />
Dublin South.<br />
• Mid-East: Kildare, Meath, Wicklow.<br />
3 Equivalised income, or income per adult equivalent, is calculated by<br />
allowing a ‘weight’ of 1 for the first adult in the household, 0.66 for<br />
each subsequent adult and 0.33 for each child under age 15. The<br />
poverty line is expressed in terms of a percentage, e.g. 50 or 60 per<br />
cent, of the average equivalised income.<br />
60
Chapter 04 7/7/05 6:55 am Page 61<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
• Midlands: Laois, Longford, Offaly, Westmeath.<br />
• Mid-West: Clare, Limerick, Tipperary North Riding.<br />
• South-East: Carlow, Kilkenny, Tipperary South Riding,<br />
Waterford, Wexford.<br />
• South-West: Cork, Kerry.<br />
• West: Galway, Mayo, Roscommon.<br />
At the 50 per cent line the highest poverty rate of 36 per cent<br />
was observed for the Border region. However, it was closely<br />
followed by the South-West and West with rates of 32 per<br />
cent. Dublin has the lowest rate by far at 17 per cent. The<br />
remaining regions are found in the range running from 22 to<br />
30 per cent, with the Mid-East at the lower end of the range<br />
and the Midlands and South-East at the upper end. The<br />
highest rate at the 60 per cent line was for the Border region<br />
and the West, with rates of 46 per cent and 44 per cent,<br />
respectively. Once again, the lowest rate of 21 per cent was<br />
found for Dublin, followed by the Mid-East with a rate of 26<br />
per cent. The poverty risk for the remaining regions was<br />
about 35 to 40 per cent.<br />
The highest values for the consistent poverty line, as shown<br />
in Table 4.1 and Map 4.1, are found in the Border and Mid-<br />
West regions with rates of about 10 per cent. The lowest rate<br />
of 3 per cent is observed in Dublin. The values for the other<br />
regions range between 4 per cent and 8 per cent.<br />
The bottom panel of Table 4.1 aggregates the regional<br />
authorities into the two regional assembly areas: the Border,<br />
Midlands and West (BMW) and the South and East (Dublin,<br />
Mid-East, Mid-West, South-East and South-West). The<br />
greater risk of poverty in the BMW region compared to the<br />
South and East is very evident on all three measures.<br />
61
Chapter 04 7/7/05 6:55 am Page 62<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 4.1: <strong>Poverty</strong> risk in 2000 by regional authority and<br />
regional assembly<br />
Region 50% 60% 60% line+<br />
income income deprivation<br />
line risk line risk risk<br />
Border 35.5 46.4 10.1<br />
Dublin 17.1 21.3 3.3<br />
Mid-East 22.2 25.9 5.8<br />
Midlands 29.7 34.5 4.2<br />
Mid-West 25.9 35.2 10.4<br />
South-East 26.9 35.4 7.7<br />
South-West 32.4 40.0 7.5<br />
West 31.6 43.5 5.0<br />
Regional Assembly<br />
BMW 32.9 42.7 7.1<br />
South & East 23.0 29.0 5.8<br />
Total 25.8 32.9 6.2<br />
Source: Living in Ireland Survey, 2000<br />
Table 4.2 sets out the incidence figures broken down by<br />
regional authority and regional assembly. The incidence of<br />
poverty refers to the percentage of all poor households that<br />
are located in each region. At all three lines approximately<br />
one in five to one in six of the poor are located in each of the<br />
Border regions, Dublin and the South-West; one in seven to<br />
one in eight are in the South-East. The figure for the<br />
consistent line is somewhat lower for Dublin and higher for<br />
the South-East. For the other regions the modal figure is 7<br />
per cent. However, the concentration of consistent poverty is<br />
rather higher in the Mid-West, at 13 per cent, and low in the<br />
Midlands, at 4 per cent.<br />
The figures in the bottom panel of Table 4.2 indicate that<br />
close to two-thirds of the poor are located in the South and<br />
62
Chapter 04 7/7/05 6:55 am Page 63<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
Map 4.1: Consistent poverty risk by region<br />
63
Chapter 04 7/7/05 6:55 am Page 64<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
East, in spite of the higher risk of poverty (as seen in Table<br />
4.1) in the BMW region. This reflects the higher population of<br />
the South and East, as can be seen in the final column of the<br />
table: 72 per cent of households are located in the South and<br />
East, compared to 28 per cent in the BMW region.<br />
Table 4.2: <strong>Poverty</strong> incidence in 2000: Per cent of poor in each<br />
regional authority and regional assembly area (relative income<br />
poverty and consistent poverty)<br />
Region 50% relative 60% relative Consistent Per cent<br />
income income poverty of all<br />
poverty poverty 60% line+ households<br />
line line deprivation<br />
Border 17.4 17.8 20.5 12.6<br />
Dublin 20.7 20.3 16.7 31.3<br />
Mid-East 6.9 6.3 7.6 8.0<br />
Midlands 7.6 7.0 4.4 6.6<br />
Mid-West 7.7 8.2 13.1 7.7<br />
South-East 11.7 12.1 14.1 11.2<br />
South-West 16.8 16.3 16.3 13.4<br />
West 11.2 12.1 7.4 9.2<br />
Regional Assembly<br />
BMW 36.2 36.9 32.3 28.4<br />
South & East 63.8 63.1 67.7 71.6<br />
Total 100.0 100.0 100.0 100.0<br />
Source: Living in Ireland Survey, 2000<br />
4.3 Disparities in Income <strong>Poverty</strong> Risk by Local<br />
Authority Area from the NSHQ<br />
At this point we turn to data from the NSHQ to examine<br />
regional variation in poverty risk at a more detailed level.<br />
64
Chapter 04 7/7/05 6:55 am Page 65<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
Table 4.3: Disparities in income poverty risk by local authority<br />
area<br />
Region County Income poverty Income poverty<br />
risk (50%) risk (60%)<br />
Border Cavan 1.3 1.3<br />
Donegal 1.7 1.6<br />
Leitrim 1.3 1.4<br />
Louth 1.1 1.1<br />
Monaghan 1.1 1.0<br />
Sligo 1.2 1.1<br />
Midlands Laois 1.0 1.0<br />
Longford 1.5 1.4<br />
Offaly 1.3 1.2<br />
Westmeath 1.0 1.1<br />
West Galway City 0.7 0.7<br />
Galway County 1.2 1.2<br />
Mayo 1.5 1.4<br />
Roscommon 1.2 1.1<br />
Dublin Dublin City Council 1.0 1.0<br />
Dublin Fingal 0.4 0.5<br />
Dublin South 0.5 0.6<br />
Dún Laoghaire-Rathdown 0.4 0.5<br />
Mid-East Kildare 0.7 0.7<br />
Meath 0.7 0.8<br />
Wicklow 0.9 1.0<br />
Mid-West Clare 1.0 1.1<br />
Limerick City 1.5 1.3<br />
Limerick County 1.1 1.1<br />
Tipperary North Riding 0.9 0.9<br />
South-East Carlow 1.3 1.2<br />
Kilkenny 0.9 0.9<br />
Tipperary South Riding 1.3 1.2<br />
Waterford City 1.2 1.1<br />
Waterford County 1.1 1.1<br />
Wexford 1.2 1.2<br />
South-West Cork City 1.4 1.3<br />
Cork County 1.0 0.9<br />
Kerry 1.3 1.3<br />
Total 1.0 1.0<br />
Source: National Survey of Housing Quality, 2001/2002<br />
65
Chapter 04 7/7/05 6:55 am Page 66<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 4.3 and Maps 4.2 and 4.3 show the disparities in<br />
poverty risk by local authority area using data from the<br />
NSHQ. The figures in the table show the ratio of the poverty<br />
risk in each local authority area to the national figure. Thus,<br />
for instance, the risk of poverty in Cavan – using the 50 per<br />
cent mean income line – is 30 per cent higher than the<br />
national average. On the other hand, it is 60 per cent lower<br />
than the national average in Dublin Fingal.<br />
It is clear from Table 4.3 that there are considerable<br />
differences within regions. For example, in the Border region,<br />
all of the counties (except Monaghan using the 60 per cent<br />
line) tend to have poverty risk levels above average. However,<br />
the risk of poverty is 60–70 per cent greater than the national<br />
average in Donegal, 30–40 per cent above that national figure<br />
in Cavan and Leitrim and drops to 10–20 per cent above the<br />
average in Louth, Monaghan and Sligo. There is a somewhat<br />
smaller level of variation among the Midland counties, with<br />
Longford and Offaly faring 20–50 per cent worse than<br />
average in terms of poverty risk, while Laois and Westmeath<br />
are close to average.<br />
The differences are even more dramatic in the West: Galway<br />
City has a poverty risk level that is well below average (30 per<br />
cent), while the risk in Mayo is 40–50 per cent above average.<br />
Galway County and Roscommon occupy an intermediate<br />
position, with risk levels 10–20 per cent above average. The<br />
risk of poverty is just at the national average in Dublin City,<br />
but is lowest of all the local authority areas nationally in the<br />
surrounding Dublin counties at 40–60 per cent below the<br />
average.<br />
In the Mid-East, Kildare and Meath have risk levels that tend<br />
to be 20–30 per cent below the national level, but the risk<br />
level in Wicklow is closer to the national average.<br />
66
Chapter 04 7/7/05 6:55 am Page 67<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
Map 4.2: Disparities in income poverty risk by local authority<br />
area (50 per cent poverty line)<br />
Among the counties in the Mid-West, the risk is highest in<br />
Limerick City (30–50 per cent above average). In fact, the<br />
poverty risk level at the 50 per cent line in Limerick City is the<br />
highest among the five county boroughs (Dublin, Cork,<br />
Limerick, Galway and Waterford cities). The risk level is<br />
somewhat below average in Tipperary North (10 per cent<br />
67
Chapter 04 7/7/05 6:55 am Page 68<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 4.3: Disparities in income poverty risk by local authority<br />
area (60 per cent poverty line)<br />
below average), while Clare and Limerick County show risk<br />
levels that are slightly above average.<br />
In the South-Eastern region, only Kilkenny fares better than<br />
the nation as a whole (10 per cent below the average) in<br />
terms of poverty risk, while Tipperary South, Wexford and<br />
Carlow (20–30 per cent above) fare worst.<br />
68
Chapter 04 7/7/05 6:55 am Page 69<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
In the South-West, the level of poverty risk is 30–40 per<br />
cent above the national level in Cork City and in Kerry, but<br />
is at or slightly below the overall average in Cork County<br />
(0–10 per cent below).<br />
4.4 Modified Consistent <strong>Poverty</strong> (MCP)<br />
As noted earlier, the NSHQ does not contain the set of<br />
basic deprivation items used in the construction of the<br />
measure of consistent poverty based on the LII data, so<br />
even if we could ignore the differences in how income is<br />
measured, it would not be possible to reproduce the<br />
measure of consistent poverty that is used in monitoring<br />
the National Anti-<strong>Poverty</strong> Strategy. Nevertheless, we can<br />
construct a ‘modified’ consistent poverty index which takes<br />
account of both income and living standard and examines<br />
the way in which it varies by region.<br />
Modified consistent poverty (MCP) is defined as<br />
households below 60 per cent of the mean equivalised<br />
household income (using the NSHQ income measure) and<br />
experiencing basic deprivation, i.e. lacking one or more of<br />
the items in the basic deprivation index. The basic<br />
deprivation items in the NSHQ include an inability to afford<br />
adequate food, clothing and heating for the home, as well<br />
as socialising, an annual holiday and the presence of<br />
arrears on utility bills.<br />
As before, we do not present levels of MCP in the following<br />
tables to avoid confusion with the full measure of<br />
consistent poverty based on the LII data. Instead, the<br />
results are presented in terms of ratios of the national<br />
average figure.<br />
69
Chapter 04 7/7/05 6:55 am Page 70<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
4.5 MCP by Local Authority Area<br />
Table 4.4 and Map 4.4 show the disparities in MCP by local<br />
authority area. Earlier tables revealed considerable variation in<br />
income poverty by local authority area within regions. As can<br />
be seen in Table 4.4, the differences persist when the<br />
combined measure of disadvantage is used, taking account<br />
of both income and basic deprivation. Overall, the highest<br />
level of MCP is found in Donegal (1.9 times the average) and<br />
the lowest level in Dún-Laoghaire-Rathdown (0.60 times the<br />
average). Nevertheless, within the Border region the range of<br />
MCP is from Monaghan’s low rate at 10 per cent below the<br />
average to Donegal’s high figure of 90 per cent above. The<br />
figures in the Midlands are all above average, but range from<br />
a modest 10 per cent above average in Laois, Offaly and<br />
Westmeath to 50 per cent above average in Longford.<br />
Galway City stands apart in the Western region with an MCP<br />
30 per cent below the average, while in the other counties in<br />
this region the figure ranges from 1.0 to 1.5.<br />
Dublin City has an MCP which is on a par with the national<br />
average, but it is well below the national figure in the<br />
surrounding counties of Fingal, Dublin South and Dún<br />
Laoghaire-Rathdown. The Mid-East evinces a similar, though<br />
less marked, pattern of variation: the level in Wicklow is the<br />
same as the national level but it is 20–30 per cent lower in<br />
Kildare and Meath. The MCP in the Mid-West ranges from a<br />
figure equal to the national average in Limerick County to 1.5<br />
in Limerick City.<br />
In the South-East, Kilkenny maintains its relatively favourable<br />
position with an MCP 20 per cent below the national figure,<br />
while high figures (30 per cent above the average) are found<br />
in Waterford City and Carlow. Cork City and Kerry have MCP<br />
70
Chapter 04 7/7/05 6:55 am Page 71<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
Table 4.4: Disparities in risk of modified consistent poverty by<br />
local authority area<br />
Region County Modified consistent poverty<br />
Border Cavan 1.2<br />
Donegal 1.9<br />
Leitrim 1.5<br />
Louth 1.2<br />
Monaghan 0.9<br />
Sligo 1.3<br />
Midlands Laois 1.1<br />
Longford 1.5<br />
Offaly 1.1<br />
Westmeath 1.1<br />
West Galway City 0.7<br />
Galway County 1.0<br />
Mayo 1.5<br />
Roscommon 1.1<br />
Dublin Dublin City Council 1.0<br />
Dublin Fingal 0.5<br />
Dublin South 0.6<br />
Dún Laoghaire-Rathdown 0.4<br />
Mid-East Kildare 0.7<br />
Meath 0.8<br />
Wicklow 1.0<br />
South-West Clare 1.2<br />
Limerick City 1.5<br />
Limerick County 1.0<br />
Tipperary North Riding 1.1<br />
South-East Carlow 1.3<br />
Kilkenny 0.8<br />
Tipperary South Riding 1.2<br />
Waterford City 1.3<br />
Waterford County 1.0<br />
Wexford 1.0<br />
South-West Cork City 1.3<br />
Cork County 0.9<br />
Kerry 1.3<br />
Total 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
71
Chapter 04 7/7/05 6:55 am Page 72<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
levels 30 per cent above the national figure, while the level in<br />
Cork County is about 10 per cent lower.<br />
Map 4.4: Disparities in risk of modified consistent poverty by<br />
local authority area<br />
72
Chapter 04 7/7/05 6:55 am Page 73<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
One pattern that is becoming clearer at this stage is a<br />
tendency for levels of deprivation to be higher in the large<br />
cities themselves (Dublin, Cork, Limerick, Waterford) than in<br />
the surrounding counties. This pattern fits with a model where<br />
better-off households move to newer and more spacious<br />
accommodation in the suburbs, which are often located<br />
outside the city boundaries or even in neighbouring counties.<br />
This leaves poorer households trapped in often older and less<br />
attractive accommodation in the cities. Galway is an<br />
important exception to this pattern: the city itself is in a more<br />
favourable situation than the county with respect to<br />
deprivation indicators.<br />
4.6 Direct and Indirect Measures of <strong>Poverty</strong> and<br />
Deprivation<br />
Having examined the direct measures of poverty and<br />
deprivation from the LII Survey and the NSHQ, we now turn<br />
to the relationship between these direct measures at an<br />
aggregate level and the indirect measures based on census<br />
data presented in Chapter 2.<br />
We saw earlier in this chapter that counties in the Border and<br />
Western regions emerge as having the highest risk in terms of<br />
relative income poverty. As previously shown in Chapter 2,<br />
these were areas with (a) the oldest and most age-dependent<br />
population structures (b) the lowest levels of educational<br />
achievement (c) the lowest levels of professional classes and<br />
(d) the highest levels of small farm activities. In contrast, at<br />
the other extreme, most of the Dublin and Mid-East region<br />
along with commuter counties such as Meath, Kildare and<br />
counties such as Kilkenny, North Tipperary and Cork County<br />
are among the lowest in terms of relative income poverty risk.<br />
73
Chapter 04 7/7/05 6:55 am Page 74<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
These were also identified in Chapter 2 as being among the<br />
counties with (a) the youngest age profiles (b) the highest<br />
levels of large-scale farming (c) the lowest unemployment<br />
rates and (d) the highest social class and educational<br />
attainment structures in the country.<br />
A useful tool in examining the relationship between direct and<br />
indirect measures of poverty is correlation analysis. A<br />
correlation coefficient ranges from –1.0 to +1.0 depending on<br />
the strength of a relationship between two variables. If the<br />
two variables in question move together in perfect unison, i.e.<br />
as one increases/decreases the other increases/decreases at<br />
exactly the same rate, the correlation coefficient will be +1.0.<br />
If one increases as the other decreases (and vice versa), the<br />
coefficient will be negative.<br />
The figures in Table 4.5 present correlation coefficients of the<br />
socio-demographic variables used throughout Chapter 2 with<br />
three key measures of poverty discussed in the present<br />
chapter. These figures provide a quantifiable measure of the<br />
strength of the relationship at the county level of trends in the<br />
poverty measures with the underlying characteristics.<br />
From the table one can see that there is a strong relationship<br />
between county-level poverty risk (using the 50 per cent, 60<br />
per cent and modified consistent lines) and several of the<br />
socio-demographic variables considered. For example, one<br />
can see that the percentage of the county population<br />
classified as being 65 years and over, having no formal<br />
education/primary certificate only and also being in the<br />
professional class categories are strongly correlated (some<br />
negatively) to risk of poverty.<br />
The reader should note that the correlation coefficients<br />
presented here refer to characteristics measured at the level<br />
74
Chapter 04 7/7/05 6:55 am Page 75<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
Table 4.5: Correlation of socio-demographic characteristics of<br />
county with risk of poverty<br />
<strong>Poverty</strong> <strong>Poverty</strong> MCP<br />
50% 60% line<br />
Per cent persons age 65+<br />
and living alone 0.693 0.718 0.653<br />
Age dependency index 0.576 0.628 0.536<br />
Unemployment rate 0.650 0.607 0.686<br />
Per cent all persons at work who<br />
are small farmers (
Chapter 04 7/7/05 6:55 am Page 76<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
4.7 Conclusions<br />
This chapter drew on data from the 2000 LII Survey and the<br />
2001/2002 NSHQ to examine patterns of variation in income<br />
poverty and deprivation by area. At the level of regional<br />
authority, the Border area emerges as the most clearly<br />
disadvantaged area. The West and South-West tend to be<br />
next in line. At the other extreme is Dublin, and to a slightly<br />
lesser extent the Mid-East, with low levels of poverty and<br />
deprivation.<br />
When we differentiate within regions by identifying local<br />
authority areas, Donegal, Cavan, Leitrim, Longford, Mayo and<br />
the major urban centres outside Dublin emerge as<br />
experiencing the highest levels of disadvantage.<br />
One finding to emerge rather clearly is that there are big<br />
differences between the counties comprising a regional<br />
authority area. In other words, spatial contiguity does not<br />
correspond to a common fate in terms of income poverty and<br />
deprivation. The boundaries between regions are drawn on<br />
the basis of political and pragmatic considerations and do not<br />
necessarily reflect a confluence of similar causal processes.<br />
This has implications for area-based strategies to combat<br />
poverty in that any remedies applied at the broad level of<br />
regional authority will target areas that are quite different from<br />
each other in terms of the level and intensity of deprivation.<br />
In this chapter we complemented the direct measures of<br />
poverty risk and deprivation with a consideration of spatial<br />
variations in relevant socio-demographic variables from the<br />
Census of Population. The somewhat aggregated county<br />
level of the analysis undoubtedly masked a substantial<br />
degree of sub-county variation. Nonetheless, the correlation<br />
coefficients indicated that at the county level there is a<br />
76
Chapter 04 7/7/05 6:55 am Page 77<br />
<strong>Poverty</strong> and Deprivation by Region and Local Authority Area<br />
reasonably strong relationship between the direct measures<br />
of poverty on the one hand and indirect measures such as<br />
age dependency, unemployment rate, rate of educational<br />
disadvantage and membership of the unskilled manual social<br />
class on the other.<br />
In the next chapter we turn to two other aspects of spatial<br />
variation: size of place (or urban-rural location) and tenure to<br />
ask to what extent these are associated with significant<br />
differences in income poverty and deprivation.<br />
77
Chapter 05 7/7/05 6:55 am Page 78<br />
Chapter 5<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
5.1 Introduction<br />
In this chapter we draw on data from a number of sources to<br />
examine variations in poverty by area type, tenure and a<br />
combination of these two variables. As it is used here, ‘area<br />
type’ refers to the population density of the area where the<br />
household is located.<br />
The National Survey of Housing Quality (NSHQ) results<br />
consistently pointed to the importance of housing tenure as a<br />
marker of disadvantage. Local authority renters in particular<br />
tended to live in accommodation that was in poorer<br />
condition, to have lower incomes and to report experiencing<br />
more financial strain than those in other tenures (Watson and<br />
Williams 2003). The disadvantage experienced by local<br />
authority renters is largely a selection effect: households have<br />
to be low-income in order to qualify for local authority<br />
housing in the first place and if their income position<br />
improves they have traditionally bought out their rented<br />
dwellings, becoming homeowners.<br />
The data analysed in this chapter comes from the Living in<br />
Ireland (LII) Surveys of 1987, 1994 and 2000 and the National<br />
Survey of Housing Quality 2001/2002 (NSHQ). The inclusion<br />
78
Chapter 05 7/7/05 6:55 am Page 79<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
of the 1987 and 1994 LII Surveys permits us to examine<br />
trends over time. The larger sample size of the NSHQ allows<br />
us to examine the patterns in greater detail. We begin by<br />
describing how area type and tenure are measured. Next we<br />
examine the risk and incidence of poverty by area type and<br />
tenure using data from the LII surveys. Finally, we examine<br />
disparities in income poverty and in modified consistent<br />
poverty (MCP) by area type and tenure using data from the<br />
NSHQ. As in the previous chapter, since the measures of<br />
income poverty and consistent poverty derived from the LII<br />
cannot be replicated using the more limited variables<br />
available on the NSHQ, we focus on disparities in poverty<br />
rather than levels of poverty when analysing data from the<br />
NSHQ.<br />
5.2 Measuring Area Type and Tenure<br />
We distinguish the following five categories for area type:<br />
• Open country<br />
• Village or town with population of less than 3,000<br />
• Town with population of 3,000 or over<br />
• Cork, Limerick, Waterford or Galway cities<br />
• Dublin City or County.<br />
In both the LII Survey and the NSHQ we will make use of a<br />
variable that captures the nature of the household’s tenure<br />
arrangement. It is reasonable to expect that local authority<br />
housing, and in particular rented local authority housing, will<br />
account for a large proportion of concentrated pockets of<br />
poverty. In establishing the extent to which the poor are<br />
79
Chapter 05 7/7/05 6:55 am Page 80<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
located in such households, we will provide one estimate of<br />
the extent to which poverty is spatially concentrated.<br />
For the LII Survey we distinguish between the following<br />
categories in relation to tenure arrangements:<br />
• Property owned outright<br />
• Owned with mortgage<br />
• Local authority purchaser<br />
• Local authority tenant<br />
• Private rented.<br />
For the NSHQ data we further distinguish within the house<br />
owner categories (the first and second points above) between<br />
those who purchase through local authority tenant purchase<br />
schemes or mortgages on the one hand and those who<br />
purchase through a private lending agency on the other.<br />
5.3 <strong>Poverty</strong> and Deprivation by Area Type<br />
In Table 5.1 we examine the manner in which risk varies<br />
across area type and how this is patterned over time. As we<br />
have seen earlier, there was a substantial increase between<br />
1994 and 2000 in the numbers falling below the 50 per cent<br />
poverty line. This trend is evident in Table 5.1 for each of the<br />
types of area we identified. However, it is particularly true for<br />
open country areas where the poverty rate rose from 20 per<br />
cent in 1994 to 32 per cent in 2000 and for villages and<br />
towns with a population of less than 3,000 where the<br />
corresponding figures were 26 per cent and 36 per cent. For<br />
the other categories the rise was a good deal more modest,<br />
80
Chapter 05 7/7/05 6:55 am Page 81<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
being of the order of 6 per cent for towns greater than 3,000,<br />
3 per cent for the major urban centres outside Dublin and 2<br />
per cent for Dublin City and County.<br />
As in earlier years. the highest poverty rate at the 50 per cent<br />
line in 2000 (36 per cent) was observed in villages and towns<br />
of less than 3,000 followed by open country, where the rate<br />
was 32 per cent. In towns greater than 3,000 the rate fell to<br />
one in four; for urban centres outside Dublin it was just above<br />
one in five and for Dublin one in six. Thus the disparities<br />
between areas increased in the second half of the 1990s.<br />
At the 60 per cent line, on the other hand, the poverty rate fell<br />
slightly between 1994 and 2000. This occurred fairly evenly<br />
across areas, apart from an increase in the rate (from 37 to 41<br />
per cent) in open country areas. However, the ordering of<br />
areas remained constant over time and identical to that at the<br />
50 per cent line. The highest risk of 46 per cent was in<br />
villages and towns less than 3,000 followed by open country,<br />
with a rate of 42 per cent. The rate then fell significantly to 32<br />
per cent for towns less than 3,000, to 28 per cent for cities<br />
outside Dublin and finally to 21 per cent for Dublin.<br />
Turning to the consistent poverty line, involving falling below<br />
the 60 per cent income line and the enforced lack of at least<br />
one basic item, we can see that in earlier years variation by<br />
type of area was a good deal less evident than in the case of<br />
income poverty lines. In particular, the situation of those living<br />
in open country was more favourable. This remained true in<br />
2000. Between 1994 and 2000 there was a very sharp decline<br />
in consistent poverty. Substantial reductions were observed<br />
for all types of areas, but the most dramatic decline was for<br />
Dublin, where the rate fell from 19 per cent to 3 per cent. In<br />
the other urban centres it declined from 19 per cent to 7 per<br />
81
Chapter 05 7/7/05 6:55 am Page 82<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 5.1: Risk of poverty by type of area for 1987, 1994 and 2000<br />
50% income line 60% income line 60% income line<br />
and basic deprivation<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
Area: % of households<br />
Open country 21.6 19.6 32.4 34.7 37.4 41.5 15.5 9.6 5.3<br />
Village/town 3,000 14.2 19.2 25.2 31.5 37.2 32.4 18.1 18.7 10.2<br />
Waterford, Cork, Galway and<br />
Limerick cities 18.0 18.4 21.3 28.0 32.0 27.7 21.0 19.2 6.8<br />
Dublin City and County 8.7 15.1 17.1 16.7 27.4 21.3 12.0 19.2 3.3<br />
All 17.0 18.6 25.8 29.1 34.8 32.9 16.4 14.9 6.2<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
82
Chapter 05 7/7/05 6:55 am Page 83<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
Table 5.2: Incidence of poverty by type of area for 1987, 1994 and 2000<br />
50% 60% 60% Percentage of<br />
income line income line income line and all households<br />
basic deprivation<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
Area: % of households<br />
Open country 46.5 34.3 40.0 42.7 35.6 40.2 33.9 21.3 27.3 35.8 33.1 31.9<br />
Village/town 3,000 15.1 19.0 19.1 19.5 19.4 19.3 19.8 22.8 32.4 18.0 18.1 19.6<br />
Waterford, Cork, Galway,<br />
Limerick cities 9.1 8.5 5.7 8.3 8.0 5.8 11.0 11.3 7.7 8.6 8.7 6.9<br />
Dublin City and County 13.5 23.5 20.7 15.1 22.9 20.3 19.2 29.5 16.7 26.3 29.5 31.3<br />
All 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
83
Chapter 05 7/7/05 6:55 am Page 84<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
cent and in towns of more than 3,000 from 19 per cent to 10<br />
per cent. In towns less than 3,000 the rate goes from 22 per<br />
cent to 10 per cent and in open country from 10 per cent to 5<br />
per cent.<br />
The highest level of consistent poverty in 2000 (10 per cent)<br />
is observed for villages and towns outside the country’s major<br />
urban centres. They are followed by urban centres outside<br />
Dublin with a rate of 7 per cent, open country (5 per cent) and<br />
Dublin (4 per cent).<br />
In Table 5.2 we turn from the risk of poverty to the incidence<br />
of poverty, i.e. the percentage of all poor people who are<br />
located in each type of area. At the 50 per cent line we<br />
observe a partial reversal between 1994 and 2000 of a trend<br />
observed between 1987 and 1994. During the earlier period<br />
there had been a sharp increase in the proportion of the poor<br />
located in Dublin (from 14 to 24 per cent), while there was a<br />
corresponding decline (47 to 34 per cent) in open country<br />
areas. The figures for the other areas remained relatively<br />
constant with 15 per cent located in towns and villages less<br />
than 3,000, 19 per cent in towns greater than 3,000 and 6 per<br />
cent in the county boroughs outside Dublin (Cork, Galway,<br />
Limerick and Waterford cities).<br />
A similar but less pronounced pattern was observed at the 60<br />
per cent line, once again giving a distribution by 2000 that<br />
was very close to that observed in 1987 and one just about<br />
identical to that just described for the 50 per cent line. For<br />
the consistent poverty line the change over time was uneven.<br />
In 1987 one-third were located in open country areas. This fell<br />
to one-fifth in 1994 but increased to over one-quarter in 2000.<br />
For Dublin, on the other hand, there was a sharp increase<br />
between 1987 and 1994 and a corresponding decrease by<br />
84
Chapter 05 7/7/05 6:55 am Page 85<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
2000. By that time one in six of the consistent poor were found<br />
in Dublin. For other urban areas a modest decrease was<br />
observed between 1994 and 2000, while for villages and towns<br />
there was no significant change. The clearest trend involved a<br />
steady increase over time for villages and towns with<br />
populations less than 3,000 from one in five to one in three.<br />
Overall, villages and towns of less than 3,000 emerge as<br />
having a disproportionate number of the poor for each of the<br />
income poverty lines but not for the consistent poverty<br />
measure. Towns greater than 3,000, on the other hand, are<br />
over-represented in terms of consistent poverty. Urban centres<br />
outside Dublin are slightly under-represented at the income<br />
lines but not for the consistent poverty line. Dublin, on the<br />
other hand, is significantly under-represented for all three lines.<br />
This pattern, whereby rural areas appear less disadvantaged<br />
according to the consistent poverty measure that takes<br />
account of standard of living as well as income, is largely due<br />
to the situation of farm households (see Whelan et al. 2003,<br />
Tables 4.11 and 5.10). On average, these households emerge<br />
as having a higher standard of living than their current income<br />
level would lead us to predict. This may reflect a number of<br />
factors, including the accumulation of wealth over time in the<br />
form of savings, the capacity to sell off assets and a lower cost<br />
of housing than in urban areas.<br />
5.4 <strong>Poverty</strong> and Deprivation by Tenure Type<br />
One of the limitations of the analysis conducted to this point is<br />
that it is perfectly possible that a more detailed disaggregation<br />
would reveal pockets of deprivation within the areas we have<br />
been able to distinguish. We will deal with this in two ways.<br />
85
Chapter 05 7/7/05 6:55 am Page 86<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Later in the chapter we will take advantage of the<br />
opportunities provided by the NSHQ. In this section we focus<br />
on the relationship between poverty and tenure type and in the<br />
section that follows we combine information about type of<br />
area and type of tenure.<br />
In Table 5.3 we look at variation in poverty risk level by tenure<br />
across all three poverty standards and for each of the three<br />
years in our analysis. Between 1987 and 1994 there was a<br />
sharp increase in poverty risk at the 50 per cent line for local<br />
authority tenants from 37 per cent to 50 per cent, even though<br />
the overall increase in the poverty rate was rather modest.<br />
Between 1994 and 2000, in the context of a much sharper<br />
increase in the overall poverty rate, the poverty rate for local<br />
authority tenants rose to 66 per cent. At the same time, the<br />
rate for outright owners increased from 18 per cent to 33 per<br />
cent. There was also a more modest increase for private<br />
renters from 15 to 22 per cent. The other categories remained<br />
relatively stable.<br />
At the 60 per cent line the picture remained a good deal more<br />
stable between 1994 and 2000. A slight rise was noted for<br />
outright owners and local authority tenants and a slight decline<br />
for all other categories. The risk rate reaches a remarkably<br />
high level of 78 per cent for local authority tenants. The nexthighest<br />
rate was observed for outright owners, with a figure of<br />
42 per cent. This rather high figure is undoubtedly related to<br />
the relatively high proportion of elderly in this category. Local<br />
authority tenant purchasers display a risk rate of 33 per cent<br />
and private renters one of 28 per cent. By far the lowest rate<br />
is observed for mortgage holders, at 12 per cent.<br />
Turning our attention to the consistent poverty lines, we find<br />
that the rate for local authority tenants has fallen from 52 per<br />
86
Chapter 05 7/7/05 6:55 am Page 87<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
cent in 1994 to 32 per cent in 2000. However, given the<br />
marked decline in risk levels for all of the other categories,<br />
the relative position of local authority tenants actually<br />
deteriorated. Thus, for outright owners the rate went from<br />
11 to 4 per cent, for mortgage holders from 6 to 2 per<br />
cent and for local authority tenant purchasers from 25 to 5<br />
per cent. The decline was least for private tenants, where it<br />
went from 15 to 13 per cent. It is clear that the distribution<br />
of risk levels is rather different for the consistent poverty<br />
measure than for the income poverty lines. The greatest<br />
discrepancy is observed for private renters. Their relative<br />
position is much worse for the consistent poverty measures.<br />
Thus, a relatively clear hierarchy of disadvantage emerges<br />
for consistent poverty, with local authority tenants worst<br />
off by a significant margin, followed at a distance by private<br />
renters and then, at a further distance, by the remaining<br />
groups.<br />
In Table 5.4 we document the incidence of poverty by tenure<br />
type. This outcome depends on both the risk levels for each<br />
type of tenure and the distribution of households across<br />
tenure types.<br />
From the last two columns of Table 5.4 we can see that the<br />
main change in the distribution of households between 1994<br />
and 2000 involved the continuation of two trends already<br />
evident in the earlier period. The first involved a further<br />
decline in the number of local authority tenants from 12 to 7<br />
per cent. The second involved a continued increase in the<br />
number of private renters from 8 to 10 per cent. One<br />
consequence of these patterns, taken together with the<br />
overall increase in the poverty rate at the 50 per cent line, is<br />
that despite the sharp increase in the poverty risk for local<br />
authority tenants, by 2000 they constituted a substantially<br />
lower proportion of the poor than was previously the case. In<br />
87
Chapter 05 7/7/05 6:55 am Page 88<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 5.3: Risk of poverty by tenure type for 1987, 1994 and 2000<br />
50% income line 60% income line 60% income line<br />
and basic deprivation<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
% % %<br />
Owned outright 16.8 18.1 32.9 30.0 37.8 42.3 12.6 10.5 3.9<br />
Owned with mortgage 6.7 8.7 8.6 2.5 14.6 11.5 6.3 5.5 1.8<br />
Local Authority Tenant<br />
Purchase (LATP) 17.8 21.8 24.7 2.5 41.6 33.9 15.1 24.5 4.5<br />
Local authority rented 37.4 49.8 66.1 59.1 74.6 77.7 46.8 52.0 32.4<br />
Other rented* 14.4 15.1 21.9 27.7 34.0 27.5 21.3 15.3 12.6<br />
All 17.0 18.8 25.9 29.1 34.6 33.0 16.4 14.9 6.2<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
*Most ‘Other rented’ dwellings are privately rented, but a small number are rented from voluntary housing<br />
associations.<br />
88
Chapter 05 7/7/05 6:55 am Page 89<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
1987 and 1994 they made up almost one-third of those below<br />
the 50 per cent line, but by 2000 this figure had fallen to less<br />
than one in five. At the same time, the corresponding figure for<br />
outright owners rose from 41 per cent in 1994 to 56 per cent in<br />
2000, while the corresponding figures for private renters went<br />
from 7 to 9 per cent. At the 60 per cent line a similar, although<br />
less pronounced, pattern was observed. Local authority<br />
tenants constituted 30 per cent of those below the 60 per cent<br />
line in 1987. By 1994 this had fallen to 25 per cent, and by<br />
2000 to 17 per cent.<br />
There was a corresponding increase between 1994 and 2000<br />
in the figure for outright owners from 46 to 57 per cent and for<br />
private renters from 8 to 9 per cent.<br />
For the consistent poverty line the changes were somewhat<br />
less dramatic, except in the case of private renters. In 1994<br />
this group made up 8 per cent of the consistent poor, but by<br />
2000 this had risen to 21 per cent. The most substantial<br />
decline was observed for local authority purchasers, with the<br />
rate going from 10 to 5 per cent. In all other cases the trend<br />
was downward but of very modest proportions.<br />
5.5 Disparities in <strong>Poverty</strong> Risk by Size of Place and<br />
Tenure<br />
In Tables 5.5 and 5.6 we present comparable figures from the<br />
NSHQ. In a later section we will move to a further level of<br />
disaggregation in this regard. Table 5.5 presents disparities in<br />
the risk of income poverty by size of place using a more<br />
detailed breakdown. The figures in the table show the relative<br />
risk in each category compared to the national average. Thus,<br />
at the 50 per cent line, the poverty risk in open country areas<br />
is 20 per cent higher than (or 1.2 times) the national average.<br />
89
Chapter 05 7/7/05 6:55 am Page 90<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 5.4: Incidence of poverty by tenure type for 1987, 1994 and 2000<br />
50% 60% 60% Percentage of<br />
income line income line income line and all households<br />
basic deprivation<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
% % % %<br />
Owned outright 44.2 40.5 56.2 45.9 45.9 56.7 34.0 29.4 28.1 44.4 42.0 44.2<br />
Owned with mortgage 11.1 14.5 10.6 12.2 13.5 11.1 10.8 11.8 9.4 28.1 32.0 31.9<br />
Local Authority Tenant<br />
Purchase (LATP) 8.1 7.3 6.2 7.3 7.6 6.6 7.1 10.2 4.7 7.7 6.3 6.4<br />
Local Authority rented 32.1 30.9 18.5 29.6 25.1 17.1 41.3 40.4 37.1 14.5 11.7 7.3<br />
Other rented* 4.5 6.5 8.6 5.0 7.9 8.5 6.8 8.2 20.8 5.2 8.0 10.2<br />
All 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
*Most ‘Other rented’ dwellings are privately rented, but a small number are rented from voluntary housing<br />
associations.<br />
90
Chapter 05 7/7/05 6:55 am Page 91<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
Table 5.5: Disparities in income poverty risk by urban/rural<br />
location<br />
Urban/rural Income Income<br />
location poverty poverty<br />
risk<br />
risk<br />
(50%) (60%)<br />
Open country 1.2 1.2<br />
Town 10,000 1.0 1.0<br />
Cork, Galway,<br />
Limerick, Waterford 1.1 1.0<br />
Dublin City/County 0.7 0.8<br />
Total 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
The figures indicate that although the risk tends to be highest<br />
in rural areas and small towns (population under 3,000) and<br />
lowest in Dublin City and County, there is not a clear trend by<br />
size of place at intermediate population densities. The rate is<br />
below average in towns with a population between 3,000 and<br />
5,000, about 20 per cent above average in towns with a<br />
population between 5,000 and 10,000, at the average in<br />
towns over 10,000 and about 10 per cent above the average<br />
(but only at the 50 per cent line) in the group comprising the<br />
cities of Cork, Galway, Limerick and Waterford.<br />
Table 5.6 shows the disparities in income poverty risk by<br />
tenure. In this table, the group of householders who own their<br />
accommodation outright is split in two: those who purchased<br />
91
Chapter 05 7/7/05 6:55 am Page 92<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
through a Local Authority Tenant Purchase scheme (LATP)<br />
and others. The risk is by far the highest for local authority<br />
renters, at over twice the average. It is also considerably<br />
above average (70–90 per cent) for those who now own the<br />
home outright, but who purchased through a LATP scheme.<br />
Other households who own the home outright are above<br />
average in terms of poverty risk, but by a smaller amount (30<br />
per cent). Private sector renters and local authority tenant<br />
purchasers have income poverty risk levels that are 10–20 per<br />
cent below the average across households. Those purchasing<br />
their accommodation through a private mortgage with a<br />
lending institution fare best, with risk levels that are 20–30 per<br />
cent of the overall level.<br />
To a large extent, the variations across tenure types are due<br />
to selection issues. In order to obtain a mortgage, a<br />
household must demonstrate that it has the resources to<br />
make repayments. Similarly, with increasing rent levels, the<br />
Table 5.6: Disparities in income poverty risk by tenure<br />
Tenure Income Income<br />
poverty risk poverty risk<br />
(50%) (60%)<br />
Own outright, LATP 1.9 1.7<br />
Own outright, other 1.3 1.3<br />
Purchasing, LATP 0.8 0.9<br />
Purchasing, other 0.2 0.3<br />
Local authority renter 2.4 2.2<br />
Other renter 0.8 0.8<br />
Rent free 1.2 1.1<br />
Total 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
92
Chapter 05 7/7/05 6:55 am Page 93<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
private rental market has become increasingly unaffordable<br />
for households on low incomes. The local authority rented<br />
sector is a select group by definition: households are eligible<br />
for local authority housing by virtue of being unable to afford<br />
to provide adequate accommodation from their own<br />
resources. Traditionally, as the circumstances of a local<br />
authority tenant improve, they have become tenant<br />
purchasers.<br />
We will turn to a more complete examination of the factors<br />
accounting for these variations by size of place and tenure in<br />
the multivariate analysis of Chapter 7.<br />
5.6 Modified Consistent <strong>Poverty</strong> by Area Type<br />
and Tenure<br />
In the following section, we turn once more to the measure of<br />
modified consistent poverty (MCP) in the NSHQ data. As<br />
noted in the previous chapter, MCP involves having an<br />
equivalised household income below 60 per cent of the<br />
average and an enforced lack of at least one of the ‘basic’<br />
items.<br />
Table 5.7 presents the disparities in MCP by size of place.<br />
As we saw before, there is no clear trend by population<br />
density, and when both income and deprivation are taken into<br />
account the range of variation by type of area tends to be<br />
narrower. It is still the case that people living in the open<br />
countryside are more likely to experience MCP than those<br />
living in Dublin, but the pattern at intermediate population<br />
densities is mixed. The smallest towns (population under<br />
3,000) emerge as most disadvantaged (20 per cent above the<br />
national level).<br />
93
Chapter 05 7/7/05 6:55 am Page 94<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 5.7: Disparities in risk of MCP by urban/rural location<br />
MCP<br />
Open country 1.1<br />
Town 10,000 1.0<br />
Cork, Galway, Limerick, Waterford 1.1<br />
Dublin City/County 0.7<br />
Total 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
Table 5.8 examines disparities in MCP by tenure. Not<br />
surprisingly, given their disadvantaged position on virtually all<br />
the measures examined so far, local authority renters emerge<br />
as most likely to experience MCP, with a rate 2.8 times the<br />
national average.<br />
It is also not surprising that those purchasing on a mortgage<br />
are least likely to experience MCP, with a rate only 30 per<br />
cent of the national figure. Among owners, there is a clear<br />
distinction between those who purchased through the local<br />
authority (80 per cent above average) and other owners<br />
(10 per cent above average). Those currently purchasing from<br />
the local authority are at the average level and private sector<br />
renters are slightly below the average in terms of MCP levels.<br />
The small group comprising people occupying the<br />
accommodation rent free have a MCP rate that is 10 per cent<br />
above the national figure.<br />
In order to interpret these patterns it is important to see if<br />
they can be accounted for by patterns of employment,<br />
unemployment and social class in the areas or tenure groups.<br />
94
Chapter 05 7/7/05 6:55 am Page 95<br />
<strong>Poverty</strong> by Area and Tenure Type<br />
Table 5.8: Disparities in risk of MCP by tenure<br />
MCP<br />
Own outright, LATP 1.8<br />
Own outright, other 1.1<br />
Purchasing, LATP 1.0<br />
Purchasing, other 0.3<br />
Local authority renter 2.8<br />
Other renter 0.9<br />
Rent free 1.1<br />
Total 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
This is the type of analysis that we will turn to in Chapter 7.<br />
Before that, in the next chapter we consider the extent to<br />
which the different dimensions of non-monetary deprivation<br />
tend to occur together.<br />
5.7 Conclusions<br />
Over time the different measures of poverty behave quite<br />
differently. There was a sharp increase at the 50 per cent line,<br />
a modest decrease at the 60 per cent line and a sharp<br />
decrease for the consistent poverty measure. The increase at<br />
the 50 per cent line was much sharper in the less populated<br />
areas. As a consequence, urban-rural differentials, which had<br />
narrowed between 1987 and 1994, increased substantially<br />
between 1994 and 2000. At the 60 per cent line, differentials<br />
remained relatively stable. For the consistent poverty<br />
measure, differentials, which were already modest, narrowed<br />
even further.<br />
Overall, less-populated areas have a disproportionate risk of<br />
poverty for each type of measure. Analysis using the NSHQ<br />
95
Chapter 05 7/7/05 6:55 am Page 96<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
also indicates that poverty risk tends to be highest in rural<br />
areas and lowest in Dublin, although the pattern at<br />
intermediate levels tends to be mixed. However, differences<br />
connected with size of location are modest in comparison<br />
with those associated with type of tenure. As before, local<br />
authority tenants emerge as a particularly disadvantaged<br />
group. Over time, as they have declined as a proportion of<br />
the total number of households, they account for a smaller<br />
proportion of poor households. Their poverty rates for both<br />
income lines have increased sharply, such that two-thirds are<br />
below the 50 per cent line and three-quarters are below the<br />
60 per cent line. While their consistent poverty rate has<br />
shown some recent decline, the disparities with other groups<br />
have increased as the level has reached five times the<br />
average rate. However, the decline in their numbers means<br />
that they constitute a declining proportion of the income poor<br />
at each line and a relatively constant proportion of the<br />
consistent poor. Analysis based on the NSHQ and using a<br />
more differentiated measure of tenure confirms the major<br />
disadvantages experienced by local authority tenants and,<br />
indeed, local authority owners. These are issues to which we<br />
will return in our multivariate analysis in Chapter 7.<br />
96
Chapter 06 7/7/05 6:56 am Page 97<br />
Chapter 6<br />
Disparities in Deprivation<br />
6.1 Introduction<br />
In this chapter we move the focus from income poverty to nonmonetary<br />
indicators of deprivation, again using data from the<br />
NSHQ. Five dimensions of deprivation are described in the<br />
next section, and the disparities in deprivation by spatial<br />
characteristics of households are analysed in the remainder of<br />
the chapter. The focus is on the extent to which the different<br />
dimensions of deprivation tend to vary across spatial<br />
categories such as region, local authority, area type and<br />
tenure group. A second focus in this chapter is on whether<br />
there is evidence of multiple deprivation, i.e. where relative<br />
disadvantage along one dimension of deprivation is<br />
associated with relative disadvantage along another.<br />
The dimensions of deprivation are as set out in Table 6.1.<br />
6.2 Methodology<br />
In the following tables, we will focus on disparities in<br />
deprivation across sub-groups in the population rather than on<br />
levels of deprivation. Five distinct dimensions of deprivation<br />
are discussed: basic, secondary, housing amenities, housing<br />
97
Chapter 06 7/7/05 6:56 am Page 98<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 6.1: Dimensions of deprivation (cannot afford items)<br />
Basic<br />
Meal with meat, etc.<br />
New clothes<br />
Adequate heating<br />
Socialising once a month<br />
Replacing worn furniture<br />
Week’s holiday per year<br />
Arrears on housing/utility bills<br />
Housing amenities<br />
Flush toilet<br />
Bath/shower<br />
Hot running water<br />
Secondary<br />
Microwave<br />
Dishwasher<br />
Colour TV<br />
Video recorder<br />
Telephone, incl. mobile<br />
Housing deterioration<br />
Leaking roof/doors/windows<br />
Dampness<br />
Rot in doors/windows/floors<br />
Noise from neighbours<br />
Accommodation too small for needs<br />
Environment<br />
Vandalism<br />
Graffiti<br />
Rubbish/litter<br />
Homes/gardens in bad condition<br />
Public drunkenness<br />
deterioration and environment. The items included in each of<br />
the measures of deprivation are shown in Table 3.1 in Chapter<br />
3. As in the previous chapter, the results will be presented in<br />
terms of the ratio of the level in a sub-group, e.g. a local<br />
authority area, to the overall level nationally.<br />
6.3 Disparities in Deprivation by Region and Local<br />
Authority Area<br />
Table 6.2 presents the disparities in deprivation by regional<br />
authority area. The figures are also shown in Maps 6.1a to<br />
98
Chapter 06 7/7/05 6:56 am Page 99<br />
Disparities in Deprivation<br />
6.1e. Basic deprivation involves an inability to afford basic<br />
necessities such as adequate food, clothing and heating for<br />
the home, as well as socialising, an annual holiday and the<br />
presence of arrears on utility bills. The Border and Midland<br />
regions fare worst in terms of basic deprivation, while Dublin<br />
and the South-East fare most favourably. The Dublin region<br />
fares best, with levels of basic deprivation 20 per cent lower<br />
than the average. Earlier we showed that the South-East<br />
and the West had higher levels of income poverty than the<br />
country as a whole, but this is not apparent for basic<br />
deprivation. The Western region is about average in terms of<br />
basic deprivation, while the South-East is somewhat below<br />
average.<br />
The pattern in terms of secondary deprivation is less<br />
variable across regions. Secondary deprivation consists of<br />
an inability to afford a car and an inability to afford various<br />
common household appliances that the householder would<br />
like to have (microwave, dishwasher, colour TV, video<br />
recorder and telephone). Levels tend to be below average in<br />
Dublin and the South-West and are highest in the West.<br />
The housing amenities items (absence of flush toilet, bath or<br />
shower and hot running water) will mainly reflect an older<br />
housing stock. The levels are highest in the Border (40 per<br />
cent above average), the South-East, South-West and Mid-<br />
West (20–30 per cent above average). The levels are lowest<br />
in the Mid-East (40 per cent below average) and Dublin (30<br />
per cent below), and are also below average in the Midlands<br />
and West.<br />
Housing deterioration includes items such as a leaking<br />
roof/doors/windows, dampness, rot in window or<br />
doorframes or other timbers and noise from neighbouring<br />
99
Chapter 06 7/7/05 6:56 am Page 100<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table 6.2: Disparities in risk of basic, secondary, housing and<br />
environmental deprivation by regional authority area<br />
Region Basic Secondary Housing Housing Environamenities<br />
deterioration ment<br />
Border 1.4 1.0 1.4 1.0 0.5<br />
Dublin 0.8 0.9 0.7 1.2 2.1<br />
Mid-East 1.0 1.0 0.6 0.9 0.8<br />
Midlands 1.2 1.1 0.9 1.1 0.4<br />
Mid-West 1.0 1.0 1.2 1.0 0.7<br />
South-East 0.9 1.0 1.3 1.0 0.7<br />
South-West 1.0 0.9 1.3 0.9 0.6<br />
West 1.0 1.2 0.9 0.8 0.2<br />
Total 1.0 1.0 1.0 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
houses. The items are based on the householder identifying<br />
these problems as present in the accommodation and as<br />
constituting a moderate or major problem. It follows a<br />
different pattern by region than the housing amenities index,<br />
with the greatest level of problems in the Dublin area (20 per<br />
cent above average) and the lowest levels in the West (20 per<br />
cent below average).<br />
Finally, the environmental dimension of deprivation consists<br />
of various problems of public order being very common in the<br />
area, such as graffiti, vandalism and litter. The main contrast<br />
in terms of the environment index is between Dublin, with a<br />
level over twice the average, and the other regions. The<br />
lowest level of environmental problems is reported in the<br />
West (20 per cent of the average level).<br />
We saw in the last chapter that regional patterns may mask<br />
considerable levels of variation between local authority areas<br />
100
Chapter 06 7/7/05 6:56 am Page 101<br />
Disparities in Deprivation<br />
Map 6.1a: Disparities in risk of deprivation by regional authority<br />
area – basic deprivation<br />
101
Chapter 06 7/7/05 6:56 am Page 102<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 6.1b: Disparities in risk of deprivation by regional authority<br />
area – secondary deprivation<br />
102
Chapter 06 7/7/05 6:56 am Page 103<br />
Disparities in Deprivation<br />
Map 6.1c: Disparities in risk of deprivation by regional authority<br />
area – housing amenities<br />
103
Chapter 06 7/7/05 6:56 am Page 104<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 6.1d: Disparities in risk of deprivation by regional authority<br />
area – housing deterioration<br />
104
Chapter 06 7/7/05 6:56 am Page 105<br />
Disparities in Deprivation<br />
Map 6.1e: Disparities in risk of deprivation by regional authority<br />
area – environmental deprivation<br />
105
Chapter 06 7/7/05 6:56 am Page 106<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
within a region. Table 6.2 and Maps 6.2a to 6.2e examine the<br />
disparities in deprivation by local authority area to see if this<br />
is also the case for the non-monetary items.<br />
This is clearly the case for basic deprivation within the Border<br />
region. The level is highest in Donegal, at 80 per cent above<br />
the average nationally, and is lowest in Monaghan, at 20 per<br />
cent below the national average. The variation is more muted<br />
for secondary deprivation, but some counties have a level<br />
below average (Monaghan, Leitrim and Louth), while the<br />
others have a level above the national average. Apart from<br />
Louth, all of the Border counties evince above average levels<br />
of deprivation in terms of housing amenities, rising to double<br />
the average in Cavan and Leitrim.<br />
In the Midland region, basic deprivation is highest in Laois<br />
and Longford (30–40 per cent above average) and is at the<br />
national average in Westmeath. Differences between the<br />
counties of the South-East in terms of basic deprivation are<br />
also quite marked: Carlow has a level 30 per cent above<br />
average while Kilkenny has a figure 40 per cent below<br />
average.<br />
Housing deterioration shows a more mixed pattern, but with<br />
levels in the Border counties tending to be close to or below<br />
the average nationally. The final column in Table 6.3 indicates<br />
that the Border counties are all considerably less likely than<br />
average to face problems in terms of public disorder in the<br />
environment.<br />
A similar set of points can be made in the case of the pattern<br />
by county in the other regions.<br />
• The patterns of deprivation do not necessarily correspond<br />
to the patterns in terms of income poverty. Counties<br />
106
Chapter 06 7/7/05 6:56 am Page 107<br />
Disparities in Deprivation<br />
Table 6.3: Disparities in risk of basic deprivation, secondary<br />
deprivation, housing deprivation and environmental deprivation<br />
by local authority<br />
County Basic Secondary Housing Housing Environamenities<br />
deterioration ment<br />
Cavan 1.2 1.0 2.0 1.0 0.4<br />
Donegal 1.8 1.2 1.2 0.8 0.5<br />
Leitrim 1.1 0.9 2.1 0.5 0.2<br />
Louth 1.4 0.9 1.0 1.1 0.7<br />
Monaghan 0.8 0.7 1.1 1.0 0.3<br />
Sligo 1.6 1.3 1.7 1.0 0.4<br />
All Border region 1.4 1.0 1.4 1.0 0.5<br />
Laois 1.4 1.1 0.9 1.0 0.4<br />
Longford 1.3 1.2 1.4 0.9 0.2<br />
Offaly 1.2 0.9 0.6 1.0 0.4<br />
Westmeath 1.0 1.3 1.0 1.3 0.4<br />
All Midlands region 1.2 1.1 0.9 1.1 0.4<br />
Galway City 0.9 1.2 0.4 0.8 0.6<br />
Galway County 0.9 0.9 1.0 0.9 0.1<br />
Mayo 1.2 1.6 0.9 0.8 0.1<br />
Roscommon 1.2 1.1 1.6 0.7 0.1<br />
All West region 1.0 1.2 0.9 0.8 0.2<br />
Dublin City Council 0.9 1.3 1.0 1.3 3.0<br />
Dublin Fingal 0.7 0.6 0.3 0.9 0.9<br />
Dublin South 0.6 0.6 0.3 1.1 1.6<br />
Dún Laoghaire-Rathdown 0.7 0.7 0.4 1.0 1.0<br />
All Dublin region 0.8 0.9 0.7 1.2 2.1<br />
Kildare 0.8 0.9 0.5 1.0 0.9<br />
Meath 1.1 1.1 0.5 0.8 0.4<br />
Wicklow 1.2 1.0 1.0 1.0 1.2<br />
All Mid-East region 1.0 1.0 0.6 0.9 0.8<br />
Clare 1.2 1.2 1.1 0.7 0.6<br />
Limerick City 1.1 1.1 0.9 1.5 2.0<br />
Limerick County 0.7 0.8 1.4 0.9 0.3<br />
Tipperary North Riding 1.3 1.1 1.5 1.0 0.5<br />
All Mid-West region 1.0 1.0 1.2 1.0 0.7<br />
Carlow 1.3 1.6 1.4 1.2 1.2<br />
Kilkenny 0.6 0.9 1.4 0.8 0.4<br />
Tipperary South Riding 1.1 1.2 1.3 1.0 0.5<br />
Waterford City 1.1 1.5 0.8 1.0 1.5<br />
Waterford County 0.9 0.7 1.5 0.7 0.4<br />
Wexford 0.8 0.8 1.3 1.1 0.5<br />
All South-East region 0.9 1.0 1.3 1.0 0.7<br />
Cork City 1.0 1.0 1.4 1.1 1.0<br />
Cork County 1.0 0.8 1.1 0.9 0.5<br />
Kerry 1.1 1.0 1.9 0.8 0.2<br />
All South-West region 1.0 0.9 1.3 0.9 0.6<br />
Total 1.0 1.0 1.0 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
107
Chapter 06 7/7/05 6:56 am Page 108<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 6.2a: Disparities in risk of deprivation by local authority<br />
area – basic deprivation<br />
108
Chapter 06 7/7/05 6:56 am Page 109<br />
Disparities in Deprivation<br />
Map 6.2b: Disparities in risk of deprivation by local authority<br />
area – secondary deprivation<br />
109
Chapter 06 7/7/05 6:56 am Page 110<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 6.2c: Disparities in risk of deprivation by local authority<br />
area – housing amenities<br />
110
Chapter 06 7/7/05 6:56 am Page 111<br />
Disparities in Deprivation<br />
Map 6.2d: Disparities in risk of deprivation by local authority<br />
area – housing deterioration<br />
111
Chapter 06 7/7/05 6:56 am Page 112<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Map 6.2e: Disparities in risk of deprivation by local authority<br />
area – environment<br />
112
Chapter 06 7/7/05 6:56 am Page 113<br />
Disparities in Deprivation<br />
which rank least favourably in terms of income poverty<br />
within a region (see Table 4.3 in Chapter 4) do not<br />
necessarily rank at the bottom, or even as below average,<br />
in terms of deprivation.<br />
• Basic and secondary deprivation also appear to be<br />
patterned differently, particularly in the counties of the<br />
Border, Midlands and West regions.<br />
• Housing amenities and housing deprivation appear to be<br />
capturing phenomena with a very different distribution<br />
across areas. Counties which are below average on one<br />
dimension will often be above average on the other.<br />
• The environment dimension, which captures problems of<br />
public order, is clearly capturing problems which are more<br />
common in urban than in rural areas.<br />
6.4 Disparities in Deprivation by Size of Place<br />
and Tenure<br />
At this point we turn to disparities in the dimensions of<br />
deprivation by size of place and tenure. Table 6.4 presents<br />
disparities in the dimensions of deprivation by size of place.<br />
As with income poverty, basic deprivation tends to be more<br />
prevalent in rural areas and less prevalent in Dublin, but it is<br />
the smaller towns (under 3,000 and 3,000–5,000) rather than<br />
the open countryside that fare worst. The pattern in terms of<br />
secondary deprivation is different: the open countryside and<br />
the Dublin region are most advantaged in this respect, but<br />
differ from the average by only 10 per cent. An absence of<br />
housing amenities is clearly a problem of open country areas,<br />
where the prevalence is 70 per cent higher than nationally. On<br />
the other hand, dwellings in the open countryside fare better<br />
113
Chapter 06 7/7/05 6:56 am Page 114<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
than average in terms of housing deterioration, while the<br />
Dublin region fares worst. The final column again confirms<br />
that problems in the social environment are related to<br />
urbanisation: they are rarest in the open countryside and<br />
increasingly prevalent as population density increases.<br />
Table 6.4: Disparities in risk of basic, secondary, housing and<br />
environmental deprivation by urban/rural area<br />
Urban/ Basic Second- Housing Housing Environrural<br />
area ary amenities deteriora- ment<br />
tion<br />
Open country 1.1 0.9 1.7 0.8 0.2<br />
Town 10,000 1.0 1.1 0.8 1.0 0.9<br />
Cork, Galway, Limerick,<br />
Waterford 1.1 1.2 0.8 0.9 1.1<br />
Dublin City/County 0.8 0.9 0.7 1.2 2.1<br />
Total 1.0 1.0 1.0 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
Table 6.5 shows that the disparities in deprivation by tenure<br />
tend to be greater than by size of place. Apart from housing<br />
amenities, local authority renters fare worst on all dimensions,<br />
although private sector renters are not far behind them in<br />
terms of secondary deprivation. Those who own outright<br />
having purchased through a local authority have levels of<br />
basic and secondary deprivation that are about 30 per cent<br />
above average, but they are closer to average in terms of the<br />
114
Chapter 06 7/7/05 6:56 am Page 115<br />
Disparities in Deprivation<br />
housing and environment items. Other households who own<br />
the accommodation outright are somewhat below average in<br />
terms of basic deprivation and secondary deprivation,<br />
housing deterioration and problems in the environment.<br />
However, it is among this group and the much smaller group<br />
occupying the accommodation rent free that housing<br />
amenities are most likely to be lacking.<br />
Table 6.5: Disparities in risk of basic, secondary, housing and<br />
environmental deprivation by tenure<br />
Basic Second- Housing Housing Environary<br />
amenities deteriora- ment<br />
tion<br />
Own outright, LATP 1.3 1.3 1.1 0.9 0.9<br />
Own outright, other 0.9 0.8 1.6 0.7 0.6<br />
Purchasing, LATP 1.1 1.0 0.4 1.4 1.8<br />
Purchasing, other 0.6 0.5 0.2 0.7 0.7<br />
Local authority renter 2.6 2.8 1.3 2.6 4.0<br />
Other renter 1.3 2.0 0.9 1.5 1.2<br />
Rent free 1.1 1.3 2.4 1.0 0.3<br />
Total 1.0 1.0 1.0 1.0 1.0<br />
Source: Irish National Survey of Housing Quality, 2001/2002<br />
Tenant purchasers and those purchasing on a private<br />
mortgage looked rather similar in terms of income poverty<br />
risk, at least in so far as both groups had a lower risk than<br />
average. However, the deprivation items point to some<br />
important differences between these groups. While private<br />
purchasers are much less likely than average to experience<br />
any of the types of deprivation shown in the table and are<br />
115
Chapter 06 7/7/05 6:56 am Page 116<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
least likely to experience basic or secondary deprivation or<br />
to lack housing amenities, tenant purchasers fare worse<br />
than average on a number of dimensions. They are slightly<br />
above average in terms of basic deprivation and<br />
considerably above average in terms of housing<br />
deterioration (40 per cent) and problems of public disorder<br />
(80 per cent).<br />
Private sector renters rank second to local authority renters<br />
in terms of levels of basic and secondary deprivation and<br />
housing deterioration and are also less favourably placed<br />
than average in terms of problems of public order, but they<br />
fare slightly better than average in terms of housing<br />
amenities.<br />
6.5 Conclusions<br />
Regional disparities differ across the five dimensions of<br />
deprivation. For the basic deprivation items, the by now<br />
familiar pattern whereby the Border region is at one end of<br />
the continuum and Dublin at the other is observed. For the<br />
secondary deprivation index, which incorporates consumer<br />
items that can be accumulated over a period of time, spatial<br />
variation is extremely modest. When we focus on housing<br />
amenities, which cover the presence of rather basic<br />
facilities, Dublin and the Mid-East fare best. In contrast,<br />
housing deterioration is at its worst in Dublin, although<br />
regional variation is rather modest. Environmental problems,<br />
which are primarily of a social character, serve to sharply<br />
differentiate Dublin from all other regions.<br />
When we disaggregate the regions and look at the individual<br />
local authority it is clear that, as with income poverty, there<br />
116
Chapter 06 7/7/05 6:56 am Page 117<br />
Disparities in Deprivation<br />
is considerable variation within regions. Further, the two<br />
housing deprivation sets of items – housing amenities and<br />
housing deterioration – are clearly seen to be independent<br />
phenomenon.<br />
The implications of urban or rural location vary by dimension<br />
of deprivation, as was clear from the analysis by area type.<br />
For the basic and secondary dimensions such variation is<br />
modest, although Dublin occupies the most favourable<br />
position on both dimensions. Variation is similarly modest for<br />
housing deterioration, but on this occasion Dublin occupies<br />
the least favourable position. Problems in the social<br />
environment display a clear urban-rural gradient, being<br />
considerably more prevalent in urban areas.<br />
The pattern by tenure shows both substantially higher<br />
disparities and more striking patterns of overlapping or<br />
multiple deprivation. Thus, local authority tenants are by<br />
some distance the most disadvantaged group on four of the<br />
five dimensions. The exception is housing amenities, where<br />
the least favourable circumstances are associated with being<br />
a private tenant.<br />
Overall, then, the main findings in this chapter were that:<br />
• The disparities in deprivation by tenure tend to be much<br />
greater than the differences between regions, local<br />
authority areas or by size of place.<br />
• As was the case for income poverty, there are substantial<br />
differences in deprivation level between the counties<br />
within a given regional authority area.<br />
• Apart from local authority renters who fare worst on four<br />
of the five dimensions of deprivation, there is only a weak<br />
association between the different types of deprivation.<br />
117
Chapter 06 7/7/05 6:56 am Page 118<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Areas or tenures that fare badly on one dimension do not<br />
necessarily fare badly on the others. Evidence of multiple<br />
deprivation that is structured along spatial lines, then, is<br />
relatively weak.<br />
118
Chapter 07 7/7/05 6:57 am Page 119<br />
Chapter 7<br />
Understanding Variation in <strong>Poverty</strong> by<br />
Area and Tenure Type<br />
7.1 Introduction<br />
In addition to documenting variation in poverty rates by<br />
location and tenure, we also need to get some sense of how<br />
the magnitude of such variation compares with the scale of<br />
variation within areas and tenure type. Following on from this,<br />
we seek to assess the extent to which such variation is<br />
causal in nature. To what extent could the patterns we<br />
observe arise solely because poverty and tenure/location are<br />
jointly associated with other factors that are the true<br />
determinants of poverty<br />
7.2 Assessing the Net Effect of Location and Tenure<br />
The fact that a specific type of area or tenure has a relatively<br />
high poverty rate does not in itself indicate anything about the<br />
impact of location or tenure per se on poverty – such effects<br />
could be entirely attributable to the socio-economic<br />
composition of the households involved. Households renting<br />
local authority housing could be at a high risk of poverty<br />
because applicants for such housing tend to be drawn from<br />
119
Chapter 07 7/7/05 6:57 am Page 120<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
the most vulnerable sectors of the population: the<br />
unemployed, lone parents, the elderly and those unable to<br />
work. Cross-tabulations can only take us so far in<br />
understanding variations in poverty risk by area/tenure type at<br />
a point in time or the changes in this pattern observed over<br />
time. We therefore need to systematically examine the extent<br />
to which the observed variation in risk of poverty by area and<br />
housing tenure may be attributable to differences in the<br />
measured characteristics of those located in different areas or<br />
tenure groups, using logistic regressions. This procedure<br />
allows one to assess the effect of any particular factor on the<br />
odds of being poor while holding the influence of other<br />
factors constant.<br />
7.3 Methodology<br />
The notion of odds is one familiar to all gamblers. Suppose<br />
20 per cent of the population are poor. Instead of saying that<br />
the probability of being poor is 0.2 and of not being poor is<br />
0.8, we can say that the odds of being poor is 0.25, i.e.<br />
0.2/0.8, and the odds on being non-poor is 4:1 (0.8/0.2). An<br />
actual example will help to further illustrate the relationship<br />
between risk of poverty and odds ratios. In 2000, 3.9 per cent<br />
of outright owners fell below the combined 60 per cent mean<br />
income and basic deprivation line, giving them a 0.04<br />
(3.9/96.1) odds of being poor. On the other hand, 32.4 per<br />
cent of local authority tenants were found to be below the<br />
consistent poverty line, giving them a 0.48 (32.4/67.6) odds of<br />
being poor. The disparity in risk of being consistently poor<br />
between these two groups (local authority tenants and<br />
outright owners) can be indexed by the ratio of these odds<br />
(.48/.04), which gives an odds ratio of 12:1. In other words,<br />
120
Chapter 07 7/7/05 6:57 am Page 121<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
the odds on local authority tenants being poor are 12 times<br />
higher than the odds for outright owners. By choosing any<br />
group as a reference point, we can summarise the set of<br />
inequalities between that group and all others. It is more<br />
convenient to work with odds rather than probabilities<br />
because it allows us to make explicit comparisons between<br />
groups.<br />
In the analysis that follows we focus on the NSHQ data in<br />
estimating gross and net effects, although we also report<br />
supplementary analysis relating the 2000 LII Survey in<br />
Appendix 1. We concentrate on the NSHQ because of the<br />
advantages conferred by its substantially larger sample size.<br />
This makes it possible to employ a variable relating to<br />
housing tenure that is a good deal more differentiated than<br />
the version we have used with LII data. Earlier, and in<br />
Appendix 2, we noted the similar patterns of disparities<br />
revealed by the LII Survey and NSHQ data sets. In the<br />
analysis that follows we would expect some differences to<br />
emerge. First, where a more differentiated tenure variable is<br />
employed, we would anticipate that larger disparities would<br />
be observed. In addition, since the range of sociodemographic<br />
variables for which we can control is narrower<br />
in the NSHQ, we would expect the net effects to be larger in<br />
this survey. However, the broad picture painted by both<br />
surveys is extremely consistent.<br />
The analysis proceeds as follows. First we will present the<br />
odds ratios for being poor rather than non-poor by type of<br />
region. We will then consider the impact on such ratios of<br />
controlling for a range of socio-demographic variables,<br />
including employment status, social class, age group of the<br />
householder, number at work in the household and household<br />
type. When implementing such a procedure it is impossible to<br />
121
Chapter 07 7/7/05 6:57 am Page 122<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
be sure that we have controlled for all relevant factors.<br />
However, the survey information available to us allows us<br />
to control for a comprehensive set of variables, and as the<br />
number of variables increases, the impact of such controls<br />
will substantially diminish because of the correlation of<br />
later control variables with earlier ones. We then proceed<br />
to consider the impact of size of location and tenure and,<br />
crucially, the manner in which they interact.<br />
7.4 Gross and Net Effects of Region on <strong>Poverty</strong> in<br />
the NSHQ<br />
In Table 7.1 we show the gross and net odds ratios for<br />
being poor at 50 per cent and 60 per cent of mean income<br />
and for the modified consistent poverty (MCP) measure<br />
broken down by regional authority, with Dublin taken as a<br />
reference point. The gross odds ratios summarise the<br />
differences between groups before taking account of the<br />
socio-demographic controls, whereas the net ratios refer<br />
to the remaining differences once the effects of the sociodemographic<br />
variables have been statistically controlled.<br />
We focus first on the gross effects. At the 50 per cent<br />
income line, Dublin is sharply differentiated from all other<br />
regions. The most disadvantaged regions are the West and<br />
the Border, with odds ratios of 1.78 and 1.75, respectively.<br />
The values for the remaining regions are located in the<br />
range running from 1.38 to 1.46. When we control for the<br />
socio-demographic characteristics of the householder, the<br />
range of differentials is slightly narrowed, with the West<br />
and the Border region now having odds ratios of 1.64 to<br />
1.66.<br />
122
Chapter 07 7/7/05 6:57 am Page 123<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
An explanation of this finding is provided when we examine<br />
the level of variance explained with and without the inclusion<br />
of socio-demographic controls. In order to assess the impact<br />
of variables such as region, we distinguish between the<br />
variance that is unique to each influence and that which is<br />
shared between them. We identify these components by<br />
varying the order of entry of each type of variable. Thus, the<br />
variance that is unique to the variable region is that which is<br />
added after socio-demographic variables have been taken<br />
into account and vice versa. The shared component is that<br />
which cannot be uniquely allocated to either variable. Unlike<br />
ordinary least squares regression, there is no universally<br />
accepted measure of explained variance for logistic<br />
regression. We have reported the Nagelkerke R 2 ; however,<br />
our conclusions relating the relative importance of the types<br />
of effects we are considering would not be substantially<br />
affected by opting for another measure.<br />
From Table 7.1 we can see that identifying the regional<br />
authority does very little to enhance our understanding of the<br />
factors leading to respondents being exposed to poverty at<br />
the 50 per cent income line. In fact, region uniquely accounts<br />
for effectively none of the variance. <strong>Poverty</strong> risk at this level is<br />
being predicted almost entirely by the socio-demographic<br />
characteristics of the householder. This indicates that rather<br />
than it being possible to explain how regional differences vary<br />
by taking socio-demographic differences into account, the<br />
scale of such difference is actually tiny in comparison with<br />
those being produced within regions by differences between<br />
households in socio-demographic background.<br />
When we focus on the 60 per cent line we find a similar but<br />
slightly less extreme position. In terms of gross effects, the<br />
Border region is now clearly the most disadvantaged area,<br />
123
Chapter 07 7/7/05 6:57 am Page 124<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
with a ratio exceeding two to one. At the other end of the<br />
continuum, the South-East is close to Dublin, with a ratio of<br />
1.23. All the remaining regions are found in the narrow range<br />
running from 1.71 to 1.87. Introducing socio-demographic<br />
controls actually increases the coefficients for all regions<br />
relative to Dublin, with the Border, Midlands, South-East,<br />
South-West and West regions also displaying odds ratios<br />
above two. This indicates that rather than it being possible to<br />
explain how regional differences vary by taking sociodemographic<br />
differences into account, the scale of such<br />
difference is actually less than we would have expected on<br />
the basis of socio-demographic variation across region. Once<br />
again, though, the explanatory power of region is modest,<br />
with less than 1 per cent of the explained variance being<br />
uniquely attributable to region, 2 per cent being shared<br />
between it and socio-demographic characteristics and 97 per<br />
cent being uniquely attributable to the latter.<br />
When we turn our attention to the MCP measure, regional<br />
differences are again found to be modest in terms of their net<br />
contribution to explained variation. The Border region is the<br />
only case where the odds ratio exceeds two, and with the<br />
exception of the Mid-East, which is closest to Dublin, all<br />
other regions are located in the range running from 1.77 to<br />
1.84. The pattern of gross coefficients is again very close to<br />
that of their net counterparts. Again, however, only 1 per cent<br />
of the explained variance is uniquely attributable to region.<br />
Thus, a clear picture emerges across poverty lines that while<br />
Dublin and the Mid-East are consistently at one end of the<br />
continuum of disadvantage and the Border region at the<br />
other, exposure to poverty has a great deal to do with sociodemographic<br />
differences between households within a region<br />
and very little to do with other characteristics of the regions.<br />
124
Chapter 07 7/7/05 6:57 am Page 125<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
Table 7.1: Gross and net odds ratios of being poor versus nonpoor<br />
by regional authority<br />
50% income 60% income MCP<br />
line<br />
line<br />
Region Gross Net Gross Net Gross Net<br />
Dublin 1.00 1.00 1.00 1.00 1.00 1.00<br />
Border 1.75 1.66 2.06 2.53 2.23 2.39<br />
Mid-East 1.46 1.23 1.23 1.62 1.34 1.60<br />
Midlands 1.46 1.31 1.72 2.23 1.84 2.11<br />
Mid-West 1.38 1.43 1.71 1.83 1.83 1.85<br />
South-East 1.54 1.59 1.87 2.13 1.78 1.70<br />
South-West 1.47 1.56 1.80 2.02 1.84 1.76<br />
West 1.78 1.64 1.82 2.11 1.77 1.80<br />
% unique to<br />
region 0.00 0.94 1.18<br />
Nagelkerke R 2 1 0.005 0.551 0.014 0.640 0.014 0.509<br />
% common 0.09 2.26 1.57<br />
Source: Living in Ireland Survey, 2000<br />
7.5 Disparities in <strong>Poverty</strong> by Local Authority Areas<br />
(LAA)<br />
It might be argued that the relatively small number of regional<br />
units may lead us to underestimate the importance of spatial<br />
variation since a great deal of such variation may occur within<br />
rather than between regions. In Chapter 4 we observed<br />
1 Nagelkerke R 2 is a modification of Coz and Snell’s R 2 . The latter is an<br />
attempt to imitate the interpretation of the multiple R 2 in OLS.<br />
However, its maximum can be, and usually is, less than one.<br />
Nagelkerke R 2 involves a modification of the Cox and Snell coefficient<br />
to ensure that it can vary from 0 to 1.<br />
125
Chapter 07 7/7/05 6:57 am Page 126<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
that the risk of income poverty and of MCP could vary<br />
substantially between the counties within a region. In Table<br />
7.2 we extend our analysis to take variation across 34 local<br />
authority areas into account. From row one in Table 7.2, it<br />
is clear that spatial variation in poverty risk continues to be<br />
modest, even with a greatly differentiated set of spatial<br />
units. The Nagelkerke R 2 estimate varies from 0.02 at the<br />
50 per cent line to .039 at the 60 per cent line. When our<br />
set of socio-demographic controls is added, the range runs<br />
from .529 for the MCP line to .666 for the 60 per cent line.<br />
The unique contribution of the local authority variable<br />
ranges from 1.7 per cent of the explained variance at the<br />
60 per cent line to 2.6 per cent at the MCP. In contrast, the<br />
socio-demographic variables that we use as controls<br />
explain a low of 92.8 per cent of the explained variance for<br />
the MCP measure and a high of 96.4 per cent of such<br />
variance in the case of the 50 per cent line. Even with the<br />
substantially more differentiated measure, variation across<br />
geographical areas plays a very modest role in explaining<br />
exposure to risk of household poverty in comparison with<br />
variation in the socio-demographic characteristics of<br />
households within such areas.<br />
Table 7.2: <strong>Poverty</strong> variation between and within local authority<br />
areas<br />
Chapter 07 7/7/05 6:57 am Page 127<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
7.6 Gross and Net Effects of Size of Area and Tenure<br />
on Risk of <strong>Poverty</strong> in the NSHQ<br />
In Figure 7.1 we address the issue of the combined impact on<br />
risk of poverty of urban-rural location and tenure at the 50 per<br />
cent level. Urban is defined as being located in Dublin, Cork,<br />
Galway, Limerick or Waterford. For tenure the reference<br />
category is private purchasers and the remaining categories<br />
distinguish between local authority owners, private owners,<br />
local authority purchasers and local authority tenants. It is<br />
immediately evident that variation in risk of poverty by type of<br />
tenure within urban-rural location is a good deal sharper than<br />
by regional authority. There are two countervailing influences<br />
that it is necessary to take into account. In the first place,<br />
rural poverty rates are on average higher than urban rates.<br />
This is clearly the case with private purchasers, where the<br />
odds ratio for the rural group is almost twice that of their<br />
urban counterparts. However, as we shall see, variation in<br />
poverty risk by tenure is substantially weaker in rural areas.<br />
As a consequence, urban-rural differences are considerably<br />
weaker outside the private purchasers group.<br />
If we focus first on gross effects, we find that in urban areas<br />
private purchasers occupy a particularly favoured position.<br />
Private owners have an odds ratio almost five times higher,<br />
local authority (LA) purchasers six times higher, LA owners<br />
seven times higher, private tenants nine times higher and LA<br />
tenants 27 times higher. Thus, in urban areas private<br />
purchasers and LA tenants are found at opposite ends of the<br />
spectrum (1 and 27, respectively), while other tenure groups<br />
are located in a relatively modest range (5 to 9). When we<br />
compare urban and rural residents by type of tenure, we find<br />
fairly similar odds ratios. Thus, urban tenants are not more<br />
vulnerable than rural tenants. However, given the less<br />
127
Chapter 07 7/7/05 6:57 am Page 128<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
favourable position of rural private purchasers, this means<br />
that the impact of type of tenure is considerably stronger in<br />
urban areas. To illustrate this point, we focus on urban and<br />
rural LA tenants. These have almost identical odds ratios of<br />
27:1 and 28:1, in comparison with urban private purchasers.<br />
However, when we compare rural tenants with rural private<br />
purchasers, this reduces to less than 15:1. Similarly, for LA<br />
owners the rural specific ratio is 4:1 compared to 7:1 for their<br />
urban counterparts; for private tenants the respective figures<br />
are 5:1 and 9:1. Therefore, on an overall ranking of<br />
disadvantage, rural residents on average do slightly worse<br />
than their urban counterparts. However, this situation is<br />
brought about by the combined influence of being located in<br />
a rural area and a rather weaker impact of type of tenure than<br />
prevails in urban areas.<br />
When we focus on net odds ratios having taken into account<br />
a range of socio-demographic variables, we observe a<br />
substantial reduction in the magnitude of the observed<br />
coefficients. In particular in the case of LA tenants, we see a<br />
reduction in the size of the relevant odds ratio from 27:1 to<br />
4:1 for the urban group, constituting a reduction of 86 per<br />
cent in the size of the coefficient. For their rural counterparts,<br />
in comparison with rural private purchasers their relative<br />
disadvantage falls from just under 15:1 to just below 3:1,<br />
constituting a reduction of almost 80 per cent in the size of<br />
the effect. The range of effects for other forms of urban<br />
tenure decline from 4.6–8.9 to 1.6–3, involving reductions<br />
running from over two-thirds in the case of private tenants to<br />
just under a half in the case of LA purchasers. The<br />
corresponding rural relativities decline from 5:1–3:1 to<br />
1.8:1–1.4:1. Thus, the vast bulk of the impact of the<br />
combination of urban-rural location and tenure is accounted<br />
128
Chapter 07 7/7/05 6:57 am Page 129<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
for by the socio-demographic control variables. The<br />
remaining effects may be a consequence of the independent<br />
effect of location or tenure or it may, at least in part, be<br />
accounted for by selection effects relating to unmeasured<br />
variables. One further indication of the relative importance of<br />
different factors is provided by an examination of the division<br />
of the explained variances into its unique and common<br />
components. Our analysis shows that that 1.8 per cent of the<br />
variance can be uniquely attributed to the combination of<br />
location and tenure, 16.6 per cent is shared between these<br />
factors and the socio-demographic controls and finally, 81.6<br />
per cent can be uniquely attributed to the latter.<br />
Figure 7.1: Gross and net ratios of being poor versus non-poor<br />
at the 50 per cent income line by tenure and urban-rural<br />
location<br />
It could be argued that the limited importance of location in<br />
accounting for poverty risk arises because the urban category<br />
combines cities with different levels of poverty risk.<br />
129
Chapter 07 7/7/05 6:57 am Page 130<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
In Chapter 4 it was indeed clear that the risk of poverty in<br />
Dublin was quite different from that in the other large cities.<br />
In Figure 7.2 we report a comparable analysis that now<br />
distinguishes between Dublin and all other locations. Again,<br />
the introduction of socio-demographic controls<br />
substantially reduces the size of the coefficients. The<br />
net coefficients showing the impact of tenure within the<br />
Dublin area are not very different for those reported for<br />
urban areas generally in Figure 7.1. However, the net<br />
coefficient for LA tenants rises to almost six (from 4.4 in<br />
Figure 7.1).<br />
However, the comparable rural coefficients all show an<br />
increase indicating a sharper pattern of differentiation by<br />
tenure. The net non-Dublin odds ratios, showing the level of<br />
disadvantage in relation to Dublin private purchasers, are all<br />
(apart from LA purchasers) higher than the corresponding<br />
Dublin figure. Thus, exposure to risk of poverty is higher for<br />
those outside Dublin in each tenure category, including LA<br />
tenants. However, when internal Dublin and non-Dublin<br />
relativities are compared, the former are consistently higher.<br />
Thus, Dublin LA tenants exhibit a net odds ratio that is six<br />
times higher than private purchasers in Dublin, while the<br />
corresponding rural figure is only about half this size<br />
(6.8/2.12 = 3.4). The Dublin/non-Dublin gap is a good deal<br />
less sharp in net terms for other categories of tenure, but in<br />
every case the relative degree of disadvantage is clearly<br />
greater in the case of Dublin. The major factor contributing<br />
to this more substantial disparity in the case of Dublin is<br />
not a distinctive level of disadvantage experienced by<br />
urban LA tenants but the uniquely favourable situation<br />
associated with being a private purchaser in Dublin and the<br />
consequent disparity between the latter and the former.<br />
130
Chapter 07 7/7/05 6:57 am Page 131<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
Figure 7.2: Gross and net ratios of being poor versus non-poor<br />
at the 50 per cent income line by tenure and Dublin versus<br />
elsewhere<br />
Overall we can see that, with the exception of the urban local<br />
authority tenant effect, the net impact of tenure is relatively<br />
modest. Taken together, tenure and location in Dublin or<br />
outside Dublin uniquely account for just over 2 per cent of the<br />
explained variance and jointly account for 16 per cent<br />
together with the socio-demographic controls. The remaining<br />
80 per cent plus can be attributed uniquely to the latter.<br />
In Figure 7.3 we turn our attention to the 60 per cent income<br />
poverty line. Here we observe a pattern not very different<br />
from that observed for the 50 per cent line. Although the<br />
gross and net rural coefficients tend to be higher, relativities<br />
remain larger in the urban area. Once again, the introduction<br />
of socio-demographic controls substantially reduces the size<br />
of the coefficients. However, for urban LA tenants the<br />
reduction is somewhat smaller than was the case for the 50<br />
per cent line and the fall from 27:1 to just over 7:1 constitutes<br />
131
Chapter 07 7/7/05 6:57 am Page 132<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
a reduction of 74 per cent compared to one of 84 per cent in<br />
the case of the 50 per cent line. An almost identical degree of<br />
reduction is observed in the rural case, which once again is<br />
on a somewhat smaller scale than was the case for the 50<br />
per cent line.<br />
Figure 7.3: Gross and net ratios of being poor versus non-poor<br />
at the 60 per cent income line by tenure and urban-rural<br />
location<br />
Figure 7.4 presents a comparable analysis for the 60 per cent<br />
line that now distinguishes between Dublin and elsewhere.<br />
Here we observe higher coefficients, both gross and net, for<br />
those outside Dublin than was the case at the 50 per cent<br />
line. LA owners, both Dublin and other, are more<br />
disadvantaged than was the case at the 50 per cent line. The<br />
odds ratio relating to the comparison of rural local authority<br />
tenants with Dublin private purchasers now comes close to<br />
50:1. However, when the comparison is made to rural private<br />
purchasers, the figure falls to 16:1, a level of inequality that is<br />
just above half that observed when the corresponding Dublin<br />
132
Chapter 07 7/7/05 6:57 am Page 133<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
groups are compared. Once again we see the pattern<br />
whereby rural LA tenants constitute by some distance the<br />
most disadvantaged group, but their distinctive position<br />
comes from a combination of location and tenure effects. If<br />
on the other hand we focus on tenure effects per se within<br />
urban and rural locations, the relative disadvantage enjoyed<br />
by Dublin LA tenants is substantially greater. Similar contrasts<br />
exist between absolute and relative disadvantage for other<br />
tenure groups.<br />
Figure 7.4: Gross and net ratios of being poor versus non-poor<br />
at the 60 per cent income line by tenure and Dublin versus<br />
elsewhere<br />
Once again, the net coefficients are of a considerably smaller<br />
magnitude than their gross counterparts, although in each<br />
case they are substantially larger than those observed at the<br />
50 per cent line. However, despite such increases, the unique<br />
variance accounted for by location and tenure remains a<br />
modest 2.5 per cent compared to the corresponding figure of<br />
almost 80 per cent for socio-demographic factors.<br />
133
Chapter 07 7/7/05 6:57 am Page 134<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Finally, we turn our attention to the modified consistent<br />
poverty (MCP) measure. Figure 7.5 shows the breakdown of<br />
odds ratios by urban-rural location and tenure. While the<br />
pattern here is broadly similar to that observed at the 60 per<br />
cent line, the position of rural local authority tenants appears<br />
to be rather worse than was the case in that instance. In both<br />
the urban and rural cases the gross odds ratio is higher for<br />
the MCP. In the former case it exceeds 30 and in the latter it<br />
exceeds 40. For the 60 per cent line, a focus on internal<br />
urban and rural relativities showed the latter to be<br />
approximately twice the former; for the MCP they come<br />
closer to 60 per cent higher. For both the urban and rural<br />
cases the net coefficients involve a sizeable reduction of 70<br />
per cent on their gross.<br />
Figure 7.5: Gross and net ratios of being poor versus non-poor<br />
at the MCP line by tenure and urban-rural location<br />
The combination of urban-rural location and tenure uniquely<br />
accounts for 4.7 per cent of the explained variance, which is<br />
rather higher than for either of the income lines. The shared<br />
134
Chapter 07 7/7/05 6:57 am Page 135<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
variance of 22 per cent is also slightly higher than in the<br />
earlier cases. The remaining 70 per cent plus is attributable<br />
solely to the socio-demographic controls.<br />
Figure 7.6 provides a comparable analysis for the MCP<br />
measure, distinguishing between Dublin and elsewhere. This<br />
analysis produces the highest odds ratios to date, particularly<br />
in the case of respondents from outside Dublin. However, we<br />
must take care in interpreting the substantive meaning of<br />
these coefficients. We start by focusing on the gross<br />
coefficients. A value of 54:1 is reported for rural local<br />
authority tenants. This implies that the odds of such tenants<br />
being below rather than above the MCP line is 54 times<br />
greater than is the case for private purchasers in Dublin. This<br />
rate is a good deal higher than for the comparison of rural LA<br />
tenants with urban private purchasers. However, the main<br />
reason for this higher effect is not the effect of tenure as such<br />
but the fact that the overall contrast between Dublin and<br />
elsewhere is sharper than in the case of the urban-rural<br />
contrast.<br />
To illustrate we will proceed to calculate the gross and net<br />
disadvantages of LA tenancy within rural areas. The gross<br />
effect has a value of 18.3 (54/2.95) and the corresponding net<br />
figure is 5.4. These figures are actually identical to those<br />
observed for the corresponding urban-rural comparisons.<br />
Thus, the contrast between the two sets of results is entirely<br />
a consequence of the fact that Dublin is more sharply<br />
differentiated from other areas than is the case for the urbanrural<br />
comparison. For both urban and rural local authority<br />
tenure, a reduction of 70 per cent is observed between the<br />
gross and the net coefficients. Taken together, location in<br />
Dublin versus elsewhere and tenure uniquely account for 5.5<br />
per cent of the explained variance and share a further 21.2<br />
135
Chapter 07 7/7/05 6:57 am Page 136<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Figure 7.6: Gross and net ratios of being poor versus non-poor<br />
at the 50 per cent MCP line by tenure and Dublin versus<br />
elsewhere<br />
per cent with socio-demographic controls, leaving almost<br />
three-quarters uniquely accounted for by the latter.<br />
7.7 Conclusions<br />
Our analysis in this chapter shows that the impact of spatial<br />
variation per se, whether defined in terms of regional<br />
authority or local authority area, is extremely modest in<br />
comparison with that of the socio-demographic<br />
characteristics of householders. In this case we must stress<br />
that the most important finding relates not to differences<br />
between gross and net effects but to the modest level of both<br />
compared to the impact of household differences within areas<br />
associated with the socio-demographic characteristics of<br />
households. It is not that spatial effects are being explained<br />
by socio-demographic factors but that they are remarkably<br />
modest in comparison with the latter.<br />
136
Chapter 07 7/7/05 6:57 am Page 137<br />
Understanding Variation in <strong>Poverty</strong> by Area and Tenure Type<br />
In contrast to our spatial findings, we do observe substantial<br />
gross effects that are then very significantly reduced when we<br />
control for socio-demographic factors. This suggests that in<br />
large part, tenure/type of area effects are a consequence of<br />
the socio-demographic profiles of households within each of<br />
the combinations of tenure and location. We observe<br />
significant net effects for the combination of tenure and type<br />
of area. Although we cannot entirely rule out the possibility<br />
that such differences are a consequence of unmeasured<br />
variables for which we cannot control, taken at face value<br />
they point to the operation of additional disadvantages<br />
associated with being a local authority tenant in particular.<br />
The combined effect of size of location and housing tenure is<br />
clearly more substantial than that of location, although still<br />
modest in comparison with factors such as unemployment,<br />
class and household type. The strength of the former arises<br />
from two distinct but interacting influences. The first relates to<br />
the disadvantage experienced by rural respondents and more<br />
generally to those outside Dublin. The second relates to the<br />
impact of tenure, which is generally considerably stronger in<br />
urban rather than rural areas. The consequence of this is that<br />
if we take urban purchasers as the reference category, the<br />
rural groups and, in particular, LA tenants are most exposed<br />
to poverty risk. However, the level of disadvantage is less<br />
than if the impact of location and tenure were operating in a<br />
straightforward cumulative matter. However, if we make our<br />
comparison within the urban and rural sectors and thus focus<br />
on tenure per se, the level of disadvantage associated with<br />
types of tenure, and most particularly with being a local<br />
authority tenant, is significantly greater in urban areas. These<br />
conclusions also hold if the location variable on which we<br />
focus involves Dublin versus elsewhere rather than urban<br />
137
Chapter 07 7/7/05 6:57 am Page 138<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
versus rural. Comparing our findings with those arising from<br />
earlier work, it would appear that there has been a significant<br />
reduction in the additional impact of tenure type in urban<br />
areas and in Dublin, although the differential remains<br />
substantial.<br />
138
Chapter 08 7/7/05 6:58 am Page 139<br />
Chapter 8<br />
Conclusions<br />
8.1 Introduction<br />
In the earlier chapters we have sought to document the<br />
spatial distribution of poverty and deprivation in Ireland. In<br />
doing so we have made use of a variety of sources. Each of<br />
these has its own strengths and weaknesses. Thus, census<br />
data enables us to achieve a much greater degree of<br />
geographical differentiation than the survey data, but it lacks<br />
direct measures of poverty and deprivation. Survey data<br />
contains such measures, but there are strict limits to the<br />
range of geographical variation that we can encompass in our<br />
analysis. By using a variety of sources we hope to provide a<br />
rounded picture of the extent and implication of spatial<br />
variation in poverty and deprivation.<br />
As we have noted, there is no one rationale for spatial<br />
interventions. The justifications for area-based strategies can<br />
range from relatively straightforward efforts at rationing of<br />
scarce resources to much more complex arguments relating<br />
to efficient delivery of services and mobilisation of community<br />
resources. However, in every case the issue of the extent of<br />
geographical concentration of poverty and deprivation is<br />
relevant. Our first effort to address this issue involved analysis<br />
of CSO data relating to administrative counties and county<br />
139
Chapter 08 7/7/05 6:58 am Page 140<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
boroughs and focusing on indirect or surrogate measures of<br />
deprivation.<br />
8.2 Surrogate Measures of Deprivation<br />
In the absence of suitable income data from the census, one<br />
is obliged to use surrogate measures of deprivation such as<br />
age structure, economic status and activity, levels of farming<br />
and social class. In pursuing such analysis, a decision must<br />
be made whether to present results separately for a number<br />
of dimensions or to calculate a composite measure of<br />
multiple deprivation. In line with our earlier work and in<br />
agreement with the arguments made by Pringle et al. (2000),<br />
we have opted for the former strategy. There are a number of<br />
reasons underlying this choice, including the fact that there is<br />
no clear consensus on which indicators to include in such an<br />
index and the fact that crucial differences may be obscured<br />
by such aggregation. The most important factor underlying<br />
our decision, however, is the absence of any objective basis<br />
on which to assign weights to dimensions. While simply<br />
assigning equal weights to each dimension is obviously<br />
arbitrary, it emerges that apparently more sophisticated<br />
statistical approaches can be equally arbitrary.<br />
The most commonly used statistical technique for this<br />
purpose is factor analysis. This procedure can identify<br />
dimensions of socio-demographic characteristics that we<br />
have reason to expect to be associated with poverty and<br />
deprivation. However, as Nolan et al. (1998) observe, in the<br />
absence of direct measures of the latter, the weight derived<br />
for such an analysis can be informative only with regard to<br />
the relationship between the socio-demographic variables but<br />
not concerning the association between such variables and<br />
140
Chapter 08 7/7/05 6:58 am Page 141<br />
Conclusions<br />
deprivation and poverty. Thus, as Pringle et al. (2000: 11)<br />
conclude, it is questionable whether the fact that a variable<br />
happens to be correlated highly with another variable<br />
provides a sound basis for according it more importance.<br />
Pursuing this strategy will give disproportionate weight to<br />
dimensions that have been tapped using multiple but similar<br />
indicators.<br />
8.3 Overview of CSO Results<br />
In Chapter 2 we considered spatial variations in sociodemographic<br />
variables related to poverty and deprivation<br />
from the Census of Population aggregated to the level of<br />
local authority area. Throughout the chapter we saw that the<br />
Border and West regions contained counties with the highest<br />
percentages of population who were elderly, with lowest<br />
levels of educational attainment, highest incidence of small<br />
farming activity and high levels of economic dependency. At<br />
the other end of the spectrum, Dublin and the Mid-East<br />
emerged as the regions with the lowest levels of economic<br />
dependency and the highest levels of labour force<br />
participation, percentage of the population with third-level<br />
education and percentage of the labour force in the higher<br />
social classes.<br />
8.4 Geographical Variation: Evidence from Analysis<br />
of Indirect Measures of Deprivation<br />
Our analysis of geographical variation in direct measures of<br />
poverty and deprivation based on household data from the LII<br />
Surveys and the NSHQ were broadly consistent with the<br />
analysis based on census data. At the level of regional<br />
141
Chapter 08 7/7/05 6:58 am Page 142<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
authority, the Border area clearly exhibits the highest levels of<br />
poverty and deprivation. The sharpest contrast is with Dublin<br />
and the Mid-East area. However, it should be kept clearly in<br />
mind that there is substantial variation between counties<br />
within regions. The boundaries between regions are drawn on<br />
the basis of political and pragmatic considerations and do not<br />
necessarily capture a confluence of similar causal processes.<br />
Further differentiation within regions reveals particularly higher<br />
levels of disadvantage in counties such as Donegal, Cavan,<br />
Leitrim, Longford and Mayo and urban centres other than<br />
Dublin.<br />
When we focus on dimensions of deprivation we find that<br />
regional disparities vary across the five dimensions of<br />
deprivation that we have identified. For the basic deprivation<br />
there is once again a continuum, with the Border region at<br />
one end and Dublin at the other. However, for other<br />
dimensions variation does not follow this familiar pattern. In<br />
the case of secondary deprivation, involving a range of<br />
consumer items, geographical location has little impact.<br />
Absence of basic housing facilities is a problem least<br />
frequently in Dublin, but physical deterioration is worst in<br />
Dublin, as is exposure to social problems. Once again, we<br />
observe very clear differentiation by county within regions. As<br />
one would expect from our summary thus far, the impact of<br />
urban-rural location is highly dependent on the particular<br />
dimension on which one focuses. Indeed, the only evidence<br />
for overlapping patterns of multiple disadvantage comes in<br />
the case of local authority renters, who fare worst on four of<br />
the five dimensions. Otherwise, geographical locations or<br />
types of areas or forms of tenure that fare worst on one<br />
dimension do not necessarily do so on the reminder.<br />
Evidence of multiple deprivation structured along spatial lines<br />
is extremely weak.<br />
142
Chapter 08 7/7/05 6:58 am Page 143<br />
Conclusions<br />
8.5 Variation by Size of Area and Tenure<br />
While there is clear variation in risk of poverty and deprivation<br />
by geographical location, such differentiation is modest in<br />
comparison with that associated with size of area and, more<br />
particularly, tenure. Overall, less populated areas are more<br />
disadvantaged while Dublin is most favoured. The picture at<br />
intermediate levels is somewhat mixed. However, although<br />
variation by size of area is larger than that associated with<br />
geographical location per se, it is extremely modest in<br />
comparison with the impact of housing tenure. Local authority<br />
tenants are a particularly disadvantaged group. Over time<br />
their income poverty rates have increased sharply, and while<br />
their consistent poverty rate has shown some decline, their<br />
relative positions have shown further deterioration. The<br />
decline in their numbers, however, means that they constitute<br />
a smaller proportion of the poor than heretofore.<br />
8.6 Policy Implications<br />
In order to draw sensible policy conclusions from the results<br />
of our analysis it is necessary to go beyond descriptions of<br />
differences by location, type of area or tenure and address<br />
the issue of the underlying processes. In order to do so, in<br />
Chapter 7 we looked not only at the gross or observed<br />
effects of geographical location and the combination of size<br />
of area and location but also at the net effect after controlling<br />
for a range of socio-demographic household characteristics.<br />
In doing so we sought to address issues relating to the<br />
distinction made by such authors as Pringle and Walsh (1999)<br />
between geographical causal processes that increase the<br />
relative risks of people living in one area experiencing poverty<br />
or deprivation compared with individuals living in other areas.<br />
143
Chapter 08 7/7/05 6:58 am Page 144<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
In other words, the risk of poverty is a function of<br />
geographical distribution processes that influence the<br />
likelihood of individuals who experience poverty or<br />
deprivation being located or concentrated in particular areas,<br />
but these geographical processes do not directly influence<br />
the likelihood of an individual experiencing poverty or<br />
deprivation. Such processes influence only the spatial<br />
distribution of poverty – they do not have a major influence<br />
on the total number of poor.<br />
In relation to the impact of geographical location, we found<br />
that such effects were not in any sense statistically spurious.<br />
Even after controlling for a variety of household factors, the<br />
magnitude of such effects remains largely unaffected. The<br />
crucial point that needs to be made in relation to<br />
geographical effects is not that the differences we observe<br />
are in any sense misleading, but that they are extremely<br />
modest when placed in the context of overall variation in risk<br />
between households. Furthermore, the regional and local<br />
authority spatial units that we have employed in our analysis<br />
do not in any way constitute homogeneous blocks in relation<br />
to risk of poverty and deprivation. Finally, as we have<br />
observed, different dimensions of deprivation have rather<br />
different spatial distributions.<br />
It is of course possible to conduct a more fine-grained<br />
analysis of spatial location than it was possible for us to<br />
undertake on this occasion by going to the level of a district<br />
electoral division (DED). However, while such an analysis<br />
would inevitably reveal a more marked pattern of spatial<br />
differentiation, the regional and local authority patterns that<br />
we have documented will also break up, yielding a more<br />
scattered overall picture of spatial differentiation. Thus, using<br />
1996 SAPS data, Nolan and Fahey (2002: 240–41) concluded<br />
144
Chapter 08 7/7/05 6:58 am Page 145<br />
Conclusions<br />
that the national picture is something akin to a partially<br />
patterned mosaic: the DEDs with the highest unemployment<br />
rates are far worse off than those with the lowest<br />
unemployment rates and there is some tendency towards a<br />
clustering of the worst with the worst and the best with the<br />
best. However, the clustering effect is relatively modest when<br />
placed in the context of overall variation, so that the effect is<br />
not of sharply delineated black spots in which the bulk of the<br />
poor or unemployed are concentrated. The most reasonable<br />
conclusion remains that while some geographical<br />
concentration of disadvantage exists, poverty and deprivation<br />
are spatially pervasive and affect almost all parts of the<br />
country at all levels of geographical disaggregation. Our<br />
current analysis confirms earlier findings (Nolan, Whelan and<br />
Williams 1994, 1999; Fahey and Williams 2000) that from a<br />
pure targeting perspective, a focus on geographical location<br />
offers the crudest basis for reaching ‘at-risk populations’. As<br />
Nolan et al. (1999) concluded, area-based strategies cannot<br />
be the panacea for spatially pervasive problems.<br />
When we focus on size of area and type of tenure, a<br />
somewhat different picture emerges. While the effect of the<br />
former per se is modest, in combination they do produce a<br />
level of differentiation in terms of poverty and deprivation that<br />
far exceeds anything arising from geographical location.<br />
Furthermore, although the magnitude of such effects is still<br />
modest in comparison with those associated with the sociodemographic<br />
characteristics of households, unlike those<br />
relating to geographical location they remain highly significant<br />
even after we control for socio-demographic characteristics.<br />
To a very large extent the relative disadvantages suffered by,<br />
for example, local authority tenants are a consequence of the<br />
socio-demographic composition of such households. It is<br />
145
Chapter 08 7/7/05 6:58 am Page 146<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
necessary to caution against assuming that such residual<br />
effects can be entirely attributed to the causal effect of the<br />
combination of type of location and form of tenure. It is<br />
impossible to rule out the possibility that such effects arise as<br />
a consequence of differences relating to unmeasured<br />
household characteristics for which we have been unable to<br />
control. However, the significant net effects that we have<br />
observed do leave open the possibility that such households<br />
face significant additional disadvantages associated with their<br />
tenure that cannot be accounted for by selection on the basis<br />
of socio-demographic characteristics associated with poverty<br />
and deprivation.<br />
As we noted in our introduction, such effects could arise as a<br />
consequence of discriminatory behaviour, a low level of<br />
community resources or as a consequence of cumulative<br />
processes of disadvantage that undermine the coping<br />
capacities of households. However, the methodological<br />
problems that confront one in attempting to establish<br />
neighbourhood effects are formidable. Even in the United<br />
States, where we would expect such effects to be a good<br />
deal more substantial and where data sources and<br />
investment in research on the topic go beyond anything we<br />
can imagine in the Irish situation, it is necessary to adopt a<br />
cautious stance. Thus, Jencks and Mayer (1990), in a review<br />
of the US neighbourhood literature, were ultimately<br />
pessimistic because few studies succeeded in isolating and<br />
measuring the social processes or mechanisms that could<br />
count for such effects. While a great deal of subsequent work<br />
in the US has attempted to rectify these deficiencies and<br />
progress has been made, Sampson et al. (2002: 473–4)<br />
conclude that we continue to know relatively little about key<br />
processes or whether they are responsive to neighbourhood<br />
146
Chapter 08 7/7/05 6:58 am Page 147<br />
Conclusions<br />
policy interventions. Conclusions regarding such effects must<br />
be qualified by a range of caveats relating to intervening<br />
mechanisms, identification of appropriate spatial units and<br />
the role of selection effects (Jargowsky 1996; Brooks-Gunn<br />
et al. 1997).<br />
Efforts by Nolan and Whelan (2000) and Layte, Nolan and<br />
Whelan (2000) to examine the possibility of the emergence of<br />
vicious cycles of deprivation in urban areas exposed to<br />
patterns of multiple disadvantage have also resulted in<br />
cautious conclusions regarding the independent effect of<br />
neighbourhood or community. This stance is in line with that<br />
arising out of a review of the evidence, both US and<br />
European, by Friedrichs (1999). In evaluating the available<br />
European evidence, Friedrichs (1999) cautions that it is<br />
necessary to take into account that many of the most<br />
frequently cited sources are descriptive and small-scale<br />
studies which are highly selective since only neighbourhoods<br />
with high poverty rates are sampled. His overall conclusion is<br />
as follows: ‘The general evidence presented on<br />
neighbourhood effects indicates low or negligible effects;<br />
most context effects can be explained by either individual or<br />
institutional effects’ (Friedrichs 1998: 93). Fahey and Williams<br />
(2000: 241) suggest that if neighbourhoods matter, it is<br />
perhaps in a more complex and fine-grained way than can be<br />
captured by statistical analysis. They suggest an<br />
understanding of neighbourhood that views deprived areas<br />
not as large uniform social environments but as complex<br />
composites of micro areas, each micro area having its own<br />
character and neighbourhood quality.<br />
Effects associated with both geographical location and the<br />
combination of size of area and form of tenure are modest in<br />
comparison with those associated with the<br />
147
Chapter 08 7/7/05 6:58 am Page 148<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
socio-demographic characteristics. <strong>Poverty</strong> and deprivation<br />
are spatially pervasive and it is extremely difficult to justify a<br />
strategy of spatial intervention on the basis of straightforward<br />
targeting arguments. It is difficult to avoid the conclusion that<br />
in discussions of poverty in Ireland, disproportionate attention<br />
has been focused on spatial and community issues in<br />
comparison with that devoted to vulnerable social groups.<br />
However, as we stressed in our introduction, it is possible to<br />
disentangle arguments for area-based intervention from a<br />
focus on rationing or targeting.<br />
Most recent interventions have relied on more complex<br />
justifications that have encompassed factors such as<br />
improved service delivery and mobilisation of community<br />
resources involving a somewhat broader quality of life<br />
perspective. Thus, as Pringle and Walsh (1999: 339) observe,<br />
policies aimed at promoting social inclusion involve more<br />
than simply providing resources and have the objective of<br />
creating mechanisms whereby the disadvantaged can<br />
participate more actively in all aspects of life, including<br />
decision making.<br />
It goes well beyond the brief of this report to offer an<br />
evaluation of the impact of such more complex interventions.<br />
Any assessment of such initiatives would need to be made in<br />
light of evaluations designed specifically to address the<br />
stated objectives of the initiatives. However, the evidence we<br />
have presented relating to the limited role of spatial factors in<br />
comparison with the socio-demographic composition of<br />
households would lead us to warn against developing<br />
unrealistic expectations of area-based initiatives. This is<br />
particularly true in circumstances where, as Pringle and Walsh<br />
(1999: 343) argue, area-based funding has been extremely<br />
modest in comparison to what one would expect to be<br />
148
Chapter 08 7/7/05 6:58 am Page 149<br />
Conclusions<br />
required to have a major impact. It is necessary to strike a<br />
balance between an unduly deterministic perspective that<br />
views area-based initiatives as serving no purpose until<br />
fundamental structural problems are resolved and a<br />
perspective that encourages unrealistic expectations of the<br />
extent to which disadvantaged communities can find<br />
solutions to problems, many of which can be addressed only<br />
by national policies and mobilisation of substantial resources.<br />
149
Chapter 08 7/7/05 6:58 am Page 150
xAppendix 1 7/7/05 6:59 am Page 151<br />
Appendix 1<br />
Additional Tables<br />
151
xAppendix 1 7/7/05 6:59 am Page 152<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
A1.1 Additional Tables from Census 2002<br />
Table A1.1: Percentage of persons in each county according to age cohort<br />
County % 0–14 % 15–29 % 30–44 % 45–64 % 65+ % age % age 65+<br />
dependent live alone<br />
% Rank % Rank % Rank % Rank % Rank % Rank % Rank<br />
Cavan 22.8 7 21.1 26 20.7 25 21.5 19 13.8 4 36.6 3 4.1 3<br />
Donegal 23.4 3 21.6 22 20.5 29 21.8 17 12.6 12 36.0 6 3.5 8<br />
Leitrim 21.0 25 18.7 34 20.1 32 24.1 1 16.1 1 37.1 1 5.1 1<br />
Louth 22.7 9 24.0 12 22.4 8 20.5 27 10.3 28 33.0 21 2.9 21<br />
Monaghan 22.4 16 23.2 14 20.8 24 21.4 20 12.3 15 34.7 10 3.5 8<br />
Sligo 20.8 27 22.5 16 20.7 25 22.8 7 13.2 8 34.0 16 3.7 6<br />
Border 22.5 3 22.3 8 21.0 7 21.7 5 12.5 2 35.0 1 3.5 1<br />
Laois 23.2 4 22.4 17 22.5 7 20.6 23 11.3 22 34.5 13 2.9 21<br />
Longford 22.5 12 20.7 29 20.1 32 22.9 6 13.7 5 36.2 4 4.0 5<br />
Offaly 23.2 4 22.4 17 21.7 15 21.0 21 11.6 21 34.9 9 3.0 19<br />
Westmeath 22.9 6 23.4 13 22.2 12 20.6 23 11.0 24 33.9 18 3.0 19<br />
Midlands 23.0 2 22.5 6 21.8 4 21.0 6 11.6 5 34.6 2 3.1 3<br />
Galway County Borough 16.3 33 37.3 1 20.7 25 17.6 34 8.2 31 24.5 34 1.9 31<br />
Galway County 22.8 7 20.3 31 21.6 18 22.5 10 12.8 10 35.6 7 3.2 15<br />
Mayo 21.5 23 20.2 32 20.4 31 23.2 4 14.7 3 36.2 4 4.1 3<br />
Roscommon 21.2 24 19.2 33 20.6 28 23.5 3 15.5 2 36.7 2 4.4 2<br />
West 21.1 6 23.0 5 20.9 8 22.0 1 12.9 1 34.0 3 3.4 2<br />
Dublin County Borough 16.2 34 29.5 2 22.3 9 19.2 33 12.8 10 29.0 31 3.7 6<br />
Fingal 22.7 9 26.6 7 24.5 2 20.2 29 5.9 34 28.6 33 1.3 33<br />
South Dublin 22.5 12 27.8 5 22.9 6 20.6 23 6.3 33 28.7 32 1.3 33<br />
Dún Laoghaire-Rathdown 19.2 31 24.1 11 21.7 15 22.6 9 12.4 14 31.6 25 3.1 17<br />
Dublin 19.2 8 27.7 1 22.7 2 20.2 8 10.2 7 29.3 8 2.6 7<br />
Kildare 23.7 1 24.9 9 25.0 1 19.7 32 6.7 32 30.4 29 1.5 32<br />
152
xAppendix 1 7/7/05 6:59 am Page 153<br />
Appendix 1<br />
Table A1.1: Percentage of persons in each county according to age cohort (cont’d)<br />
County % 0–14 % 15–29 % 30–44 % 45–64 % 65+ % age % age 65+<br />
dependent live alone<br />
% Rank % Rank % Rank % Rank % Rank % Rank % Rank<br />
Meath 23.6 2 23.0 15 24.1 3 20.5 27 8.7 30 32.3 23 2 30<br />
Wicklow 22.5 12 22.4 17 23.2 4 21.9 16 10.0 29 32.5 22 2.4 29<br />
Mid-East 23.4 1 23.6 3 24.2 1 20.6 7 8.2 8 31.6 7 1.9 8<br />
Clare 22.1 19 21.1 26 22.3 9 22.7 8 11.8 17 34.0 16 3.1 17<br />
Limerick County Borough 19.4 30 28.7 4 20.5 29 19.8 31 11.7 18 31.1 28 3.2 15<br />
Limerick County 20.9 26 25.3 8 21.3 21 22.0 13 10.5 26 31.4 27 2.7 27<br />
Tipperary N.R. 21.8 22 21.2 25 21.1 22 22.5 10 13.4 7 35.2 8 3.4 13<br />
Mid-West 21.2 5 23.8 2 21.4 6 21.9 3 11.6 5 32.8 5 3 5<br />
Carlow 21.9 21 24.9 9 22.3 9 20.6 23 10.4 27 32.2 24 2.6 28<br />
Kilkenny 22.5 12 21.8 21 22.0 13 22.0 13 11.7 18 34.2 15 2.9 21<br />
Tipperary S.R. 22.0 20 21.5 24 21.5 20 22.4 12 12.6 12 34.6 11 3.5 8<br />
Waterford County Borough 20.2 28 26.7 6 21.6 18 20.2 29 11.3 22 31.5 26 2.8 25<br />
Waterford County 22.3 18 20.7 29 21.7 15 23.0 5 12.3 15 34.6 11 3.3 14<br />
Wexford 22.7 11 21.6 22 22.0 13 22.0 13 11.7 18 34.4 14 2.9 21<br />
South-East 22.1 4 22.4 7 21.8 4 21.9 3 11.8 4 33.9 4 3 5<br />
Cork County Borough 17.2 32 29.2 3 19.8 34 21.0 21 12.9 9 30.0 30 3.5 8<br />
Cork County 22.4 16 22.0 20 23.1 5 21.6 18 10.9 25 33.3 20 2.8 25<br />
Kerry 20.2 28 21.1 26 21.1 22 23.9 2 13.7 5 33.9 18 3.5 8<br />
South-West 20.8 7.0 23.3 4.0 21.9 3.0 22.0 1.0 12.0 3.0 32.7 6 3.1 3<br />
Total 21.1 24.4 22.1 21.2 11.1 32.3 2.9<br />
Source: Census 2002<br />
153
xAppendix 1 7/7/05 6:59 am Page 154<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.2: County-level labour force participation and<br />
unemployment rates<br />
County Labour force Unemployment<br />
participation rate<br />
rate<br />
Rate Rank Rate Rank<br />
Cavan 55.8 26 7.9 24<br />
Donegal 54.4 30 15.6 1<br />
Leitrim 53.7 33 8.7 17<br />
Louth 58.2 9 13.2 3<br />
Monaghan 57.4 14 9.9 11<br />
Sligo 56.8 17 8.7 17<br />
Border 56.1 6 12.0 1<br />
Laois 58.3 8 9.8 12<br />
Longford 54.9 29 10.1 9<br />
Offaly 57.4 14 8.8 16<br />
Westmeath 57.6 11 8.5 21<br />
Midlands 57.4 4 9.1 3<br />
Galway County Borough 56.9 16 10.0 10<br />
Galway County 56.5 20 8.7 17<br />
Mayo 54.4 30 10.7 6<br />
Roscommon 54.0 32 7.0 28<br />
West 55.6 8 5.2 7<br />
Dublin County Borough 60.2 5 10.4 8<br />
Fingal 64.8 1 6.9 29<br />
South Dublin 64.6 2 8.0 23<br />
Dún Laoghaire-Rathdown 56.0 25 5.7 34<br />
Dublin 61.2 3 5.2 7<br />
Kildare 63.4 3 6.2 33<br />
Meath 61.7 4 6.6 30<br />
Wicklow 58.5 7 8.1 22<br />
Mid-East 61.5 1 6.8 6<br />
Clare 58.8 6 7.6 25<br />
Limerick County Borough 55.1 28 13.9 2<br />
Limerick County 56.5 20 6.6 30<br />
Tipperary N.R. 56.7 18 7.6 25<br />
Mid-West 61.5 1 8.2 4<br />
Carlow 56.4 22 9.5 13<br />
Kilkenny 57.5 13 7.6 25<br />
Tipperary S.R. 56.1 23 9.5 13<br />
Waterford County Borough 57.6 11 12.1 5<br />
Waterford County 56.6 19 8.7 17<br />
Wexford 56.1 23 10.5 7<br />
South-East 56.6 5 9.6 2<br />
Cork County Borough 52.3 34 12.4 4<br />
Cork County 57.8 10 6.4 32<br />
Kerry 55.3 27 9.0 15<br />
South-West 56.0 7 8.2 4<br />
Total 58.3 8.8<br />
Source: Census 2002<br />
154
xAppendix 1 7/7/05 6:59 am Page 155<br />
Appendix 1<br />
Table A1.3: Distribution of the unemployed across regions<br />
County % of the % of those Ratio of<br />
unemployed 15 + in unemployment to<br />
labour force labour force<br />
Col. % Col. % Ratio Rank<br />
Cavan 1.2 1.4 0.90 23<br />
Donegal 5.6 3.2 1.76 1<br />
Leitrim 0.6 0.6 0.98 19<br />
Louth 3.8 2.5 1.50 3<br />
Monaghan 1.5 1.3 1.12 11<br />
Sligo 1.4 1.5 0.99 16<br />
Border 14.1 10.4 1.35 1<br />
Laois 1.6 1.5 1.11 12<br />
Longford 0.8 0.7 1.14 9<br />
Offaly 1.5 1.6 0.99 16<br />
Westmeath 1.7 1.8 0.96 21<br />
Midlands 5.7 5.5 1.03 4<br />
Galway County Borough 2.0 1.7 1.13 10<br />
Galway County 3.4 3.5 0.99 16<br />
Mayo 3.4 2.8 1.21 6<br />
Roscommon 1.0 1.3 0.79 28<br />
West 9.8 9.3 1.05 3<br />
Dublin County Borough 16.3 13.9 1.18 7<br />
Fingal 4.2 5.5 0.77 29<br />
South Dublin 6.0 6.6 0.90 23<br />
Dún Laoghaire-Rathdown 3.1 4.8 0.65 34<br />
Dublin 29.7 30.8 0.96 5<br />
Kildare 3.1 4.4 0.70 33<br />
Meath 2.6 3.5 0.74 31<br />
Wicklow 2.7 2.9 0.92 22<br />
Mid-East 8.3 10.8 0.77 8<br />
Clare 2.3 2.6 0.86 25<br />
Limerick County Borough 2.1 1.3 1.57 2<br />
Limerick County 2.3 3.0 0.75 30<br />
Tipperary N.R. 1.3 1.5 0.85 25<br />
Mid-West 7.9 8.5 0.93 6<br />
Carlow 1.2 1.1 1.08 13<br />
Kilkenny 1.7 2.0 0.86 25<br />
Tipperary S.R. 2.1 1.9 1.07 14<br />
Waterford County Borough 1.6 1.1 1.37 5<br />
Waterford County 1.4 1.4 0.98 19<br />
Wexford 3.3 2.8 1.18 7<br />
South-East 11.2 10.4 1.08 2<br />
Cork County Borough 4.2 3.0 1.40 4<br />
Cork County 5.9 8.1 0.72 32<br />
Kerry 3.3 3.3 1.01 15<br />
South-West 13.3 14.3 0.93 6<br />
Total 100.0 100.0<br />
Source: Census 2002<br />
155
xAppendix 1 7/7/05 6:59 am Page 156<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.4: Percentage of all persons engaged in agriculture<br />
classified by number of acres farmed<br />
County % of persons Farmers Farmers Farmers 50+<br />
15 years+
xAppendix 1 7/7/05 6:59 am Page 157<br />
Appendix 1<br />
Table A1.5: Index of economic dependency in each county<br />
County<br />
Economic dependency<br />
Index<br />
Rank<br />
Cavan 1.52 10<br />
Donegal 1.84 1<br />
Leitrim 1.58 6<br />
Louth 1.56 8<br />
Monaghan 1.49 14<br />
Sligo 1.43 22<br />
Border 1.61 1<br />
Laois 1.48 18<br />
Longford 1.61 5<br />
Offaly 1.49 14<br />
Westmeath 1.46 20<br />
Midlands 1.49 4<br />
Galway County Borough 1.34 28<br />
Galway County 1.51 12<br />
Mayo 1.62 3<br />
Roscommon 1.53 9<br />
West 1.51 2<br />
Dublin County Borough 1.21 31<br />
Fingal 1.14 34<br />
South Dublin 1.17 33<br />
Dún Laoghaire-Rathdown 1.34 28<br />
Dublin 1.21 8<br />
Kildare 1.21 31<br />
Meath 1.27 30<br />
Wicklow 1.40 24<br />
Mid-East 1.28 7<br />
Clare 1.36 27<br />
Limerick County Borough 1.62 3<br />
Limerick County 1.39 25<br />
Tipperary N.R. 1.44 21<br />
Mid-West 1.43 6<br />
Carlow 1.51 12<br />
Kilkenny 1.43 22<br />
Tipperary S.R. 1.52 10<br />
Waterford County Borough 1.48 18<br />
Waterford County 1.49 14<br />
Wexford 1.58 6<br />
South-East 1.51 2<br />
Cork County Borough 1.64 2<br />
Cork County 1.38 26<br />
Kerry 1.49 14<br />
South-West 1.46 5<br />
Total 1.39<br />
Source: Census 2002<br />
157
xAppendix 1 7/7/05 6:59 am Page 158<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.6: Distributions of persons 15 years and over<br />
classified by highest level of educational attainment<br />
County None/primary Lower sec. Upper sec. Third level<br />
% Rank % Rank % Rank % Rank<br />
Cavan 31.3 2 24.7 13 25.6 31 18.4 26<br />
Donegal 33.7 1 25.4 11 22.5 34 18.3 28<br />
Leitrim 28.7 6 24.3 18 27.6 27 19.3 24<br />
Louth 25.4 12 25.8 9 27.4 29 21.4 21<br />
Monaghan 29.2 5 28.2 1 25.0 33 17.6 33<br />
Sligo 23.9 16 21.9 29 29.6 17 24.5 12<br />
Border 29.3 1 25.2 3 25.7 8 19.9 6<br />
Laois 25.4 12 26.2 8 30.2 13 18.3 28<br />
Longford 29.5 4 24.4 14 28.3 26 17.8 31<br />
Offaly 26.1 10 26.4 6 29.8 16 17.7 32<br />
Westmeath 23.4 19 24.2 20 30.3 11 22.1 16<br />
Midlands 25.5 3 25.4 2 29.9 3 19.3 8<br />
Galway County Borough 13.5 33 15.5 33 30.3 11 40.8 2<br />
Galway County 27.4 8 21.3 30 28.8 21 22.5 14<br />
Mayo 29.6 3 22.5 26 28.7 23 19.1 25<br />
Roscommon 27.7 7 24.4 14 29.6 17 18.3 28<br />
West 25.9 2 21.2 7 29.1 6 23.7 4<br />
Dublin County Borough 23.6 18 19.2 32 25.1 32 32.1 4<br />
Fingal 13.6 32 19.8 31 33.5 1 33.1 3<br />
South Dublin 18.0 30 22.7 25 32.0 3 27.3 7<br />
Dún Laoghaire-Rathdown 11.7 34 14.3 34 29.0 19 45.0 1<br />
Dublin 18.7 7 19.2 8 28.7 7 33.5 1<br />
Kildare 17.1 31 22.2 28 32.3 2 28.4 5<br />
Meath 19.1 29 24.3 18 31.3 5 25.3 10<br />
Wicklow 19.7 27 22.9 24 29.9 15 27.5 6<br />
Mid-East 18.5 8 23.1 6 31.3 1 27.2 2<br />
Clare 21.4 25 22.4 27 31.7 4 24.4 13<br />
Limerick County Borough 26.0 11 24.4 14 27.6 27 22.1 16<br />
Limerick County 20.7 26 23.7 22 30.6 7 25.0 11<br />
Tipperary N.R. 23.3 20 25.8 9 31.3 5 19.5 23<br />
Mid-West 22.3 5 23.8 4 30.6 2 23.3 5<br />
Carlow 24.1 15 26.8 4 28.8 21 20.3 22<br />
Kilkenny 22.0 24 26.3 7 30.2 13 21.5 20<br />
Tipperary S.R. 23.7 17 27.3 3 30.6 7 18.4 26<br />
Waterford County Borough 22.6 22 26.7 5 29.0 19 21.7 19<br />
Waterford County 22.2 23 25.1 12 30.5 10 22.1 16<br />
Wexford 26.7 9 27.4 2 28.4 25 17.5 34<br />
South-East 24.0 4 26.7 1 29.5 4 19.8 7<br />
Cork County Borough 22.8 21 24.1 21 27.2 30 25.9 9<br />
Cork County 19.3 28 23.5 23 30.6 7 26.7 8<br />
Kerry 24.8 14 24.4 14 28.5 24 22.3 15<br />
South-West 21.3 6.0 23.8 4.0 29.4 5.0 25.5 3.0<br />
Total 22.2 22.7 29.1 26.0<br />
Source: Census 2002<br />
158
xAppendix 1 7/7/05 6:59 am Page 159<br />
Appendix 1<br />
Table A1.7: Distribution of persons whose education has<br />
ceased in highest and lowest levels of educational attainment<br />
across counties<br />
County A B C D E<br />
% none/ % third % pop. Ratio Ratio<br />
primary level left none/primary third level<br />
education to total to total<br />
left<br />
left<br />
education education<br />
% % % Ratio Rank Ratio Rank<br />
Cavan 2.0 1.0 1.4 1.42 2 0.71 27<br />
Donegal 5.3 2.5 3.5 1.53 1 0.71 27<br />
Leitrim 0.9 0.5 0.7 1.29 6 0.74 25<br />
Louth 3.0 2.2 2.6 1.16 11 0.84 19<br />
Monaghan 1.8 0.9 1.3 1.32 4 0.68 32<br />
Sligo 1.6 1.4 1.5 1.07 16 0.94 12<br />
Border 14.6 8.4 11.0 1.33 1 0.77 6<br />
Laois 1.7 1.1 1.5 1.16 11 0.71 27<br />
Longford 1.0 0.5 0.8 1.30 5 0.67 33<br />
Offaly 1.9 1.1 1.6 1.16 11 0.67 33<br />
Westmeath 1.9 1.5 1.8 1.06 18 0.86 15<br />
Midlands 6.5 4.2 5.7 1.15 3 0.74 8<br />
Galway County Borough 0.9 2.4 1.6 0.60 33 1.55 2<br />
Galway County 4.5 3.2 3.6 1.24 8 0.87 14<br />
Mayo 4.1 2.3 3.0 1.35 3 0.75 24<br />
Roscommon 1.8 1.0 1.4 1.27 7 0.72 26<br />
West 11.3 8.8 9.6 1.18 2 0.92 4<br />
Dublin County Borough 13.6 15.7 13.3 1.02 20 1.18 4<br />
Fingal 3.0 6.2 4.9 0.62 32 1.28 3<br />
South Dublin 4.8 6.3 6.0 0.81 30 1.05 7<br />
Dún Laoghaire-Rathdown 2.6 8.6 4.9 0.53 34 1.76 1<br />
Dublin 24.0 36.8 29.0 0.83 8 1.27 1<br />
Kildare 3.1 4.5 4.0 0.78 31 1.10 5<br />
Meath 2.9 3.3 3.4 0.87 29 0.99 9<br />
Wicklow 2.6 3.1 2.9 0.89 27 1.06 6<br />
Mid-East 8.7 10.9 10.3 0.84 7 1.05 2<br />
Clare 2.6 2.5 2.7 0.97 25 0.94 12<br />
Limerick County Borough 1.6 1.2 1.4 1.19 10 0.86 15<br />
Limerick County 2.8 2.9 3.0 0.95 26 0.98 11<br />
Tipperary N.R. 1.7 1.2 1.6 1.06 18 0.76 23<br />
Mid-West 8.7 7.8 8.6 1.01 5 0.91 5<br />
Carlow 1.3 0.9 1.2 1.10 14 0.79 22<br />
Kilkenny 2.1 1.7 2.1 1.01 23 0.84 19<br />
Tipperary S.R. 2.2 1.4 2.0 1.07 16 0.71 27<br />
Waterford County Borough 1.1 0.9 1.1 1.00 24 0.82 21<br />
Waterford County 1.5 1.3 1.5 1.02 20 0.86 15<br />
Wexford 3.7 2.1 3.0 1.23 9 0.69 31<br />
South-East 11.8 8.3 10.9 1.09 4 0.77 6<br />
Cork County Borough 3.2 3.1 3.2 1.02 20 0.99 9<br />
Cork County 7.2 8.6 8.2 0.88 28 1.04 8<br />
Kerry 3.9 3.0 3.5 1.10 14 0.85 18<br />
South-West 14.3 14.7 14.9 0.96 6 0.99 3<br />
Total 100.0 100.0 100.0<br />
Source: Census 2002<br />
159
xAppendix 1 7/7/05 6:59 am Page 160<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.8: Percentage of persons in each social class<br />
County Professional/ Non- Skilled Semi- Unskilled<br />
managerial/ manual Manual skilled<br />
technical<br />
% Rank % Rank % Rank % Rank % Rank<br />
Cavan 30.6 32 21.5 5 25.1 3 15.1 8 7.7 17<br />
Donegal 30.1 34 18.4 33 24.3 4 16.9 3 10.3 1<br />
Leitrim 33.0 25 21.1 7 23.4 9 14.9 12 7.5 18<br />
Louth 33.0 25 19.6 20 23.2 10 16.1 4 8.1 13<br />
Monaghan 30.9 31 19.4 22 26.8 1 14.9 12 8.0 15<br />
Sligo 38.0 12 21.1 7 20.2 29 14.5 17 6.2 28<br />
Border 32.2 8 19.7 6 23.9 1 15.8 1 8.4 2<br />
Laois 33.0 25 20.1 15 22.8 13 15.6 7 8.4 10<br />
Longford 33.2 23 20.3 10 24.1 6 13.9 19 8.5 8<br />
Offaly 32.0 30 18.7 31 25.3 2 15.0 10 8.9 5<br />
Westmeath 36.5 17 21.6 4 21.0 25 13.4 21 7.5 18<br />
Midlands 33.9 7 20.2 2 23.1 2 14.5 3 8.3 3<br />
Galway County Borough 46.6 3 20.1 15 15.5 33 13.1 25 4.7 31<br />
Galway County 36.6 15 19.2 25 23.1 11 14.1 18 7.1 22<br />
Mayo 32.4 29 19.4 22 23.6 8 16.1 4 8.5 8<br />
Roscommon 34.5 18 22.6 1 22.8 13 13.0 27 7.0 23<br />
West 36.5 5 19.9 4 22.0 4 14.4 4 7.1 5<br />
Dublin County Borough 39.2 10 21.3 6 19.2 31 13.1 25 7.2 21<br />
Fingal 46.9 2 21.9 3 17.5 32 9.7 33 4.0 33<br />
South Dublin 39.3 9 22.5 2 22.0 19 11.7 31 4.5 32<br />
Dún Laoghaire-Rathdown 60.2 1 18.9 28 11.6 34 6.5 34 2.8 34<br />
Dublin 44.4 1 21.3 1 18.1 8 11.0 8 5.2 8<br />
Kildare 41.3 5 20.3 10 20.3 28 11.7 31 6.3 27<br />
Meath 39.5 7 19.2 25 22.8 13 11.8 30 6.7 25<br />
Wicklow 42.2 4 18.9 28 20.1 30 12.1 29 6.7 25<br />
Mid-East 41.0 2 19.5 7 21.1 6 11.8 7 6.6 7<br />
Clare 38.9 11 20.2 13 21.4 22 13.4 21 6.1 29<br />
Limerick County Borough 30.5 33 20.3 10 21.5 21 18.9 1 8.8 6<br />
Limerick County 39.4 8 19.7 19 20.4 27 13.5 20 7.0 23<br />
Tipperary N.R. 36.6 15 21.0 9 21.3 23 13.2 24 7.9 16<br />
Mid-West 37.4 4 20.2 2 21.0 7 14.2 5 7.2 4<br />
Carlow 33.1 24 18.6 32 24.3 4 14.8 15 9.2 4<br />
Kilkenny 37.7 13 19.2 25 23.1 11 12.5 28 7.5 18<br />
Tipperary S.R. 33.3 21 19.8 18 22.2 17 15.1 8 9.7 2<br />
Waterford County Borough 33.3 21 18.8 30 22.3 16 17.2 2 8.4 10<br />
Waterford County 37.5 14 17.5 34 21.7 20 15.0 10 8.3 12<br />
Wexford 32.5 28 19.3 24 24.1 6 14.7 16 9.3 3<br />
South-East 34.5 6 19.0 8 23.1 2 14.7 2 8.8 1<br />
Cork County Borough 34.1 20 19.9 17 21.3 23 16.0 6 8.7 7<br />
Cork County 40.3 6 19.6 20 20.8 26 13.4 21 5.9 30<br />
Kerry 34.5 18 20.2 13 22.2 17 14.9 12 8.1 13<br />
South-West 37.8 3.0 19.8 5.0 21.2 5.0 14.2 5.0 6.9 6.0<br />
Total 38.6 20.2 21.0 13.3 6.9<br />
Source: Census 2002<br />
160
xAppendix 1 7/7/05 6:59 am Page 161<br />
Appendix 1<br />
Table A1.9: Distributions of persons in Professional and<br />
Unskilled Manual class categories across region<br />
County A B C D E<br />
Professional/ Unskilled % total Ratio Ratio<br />
Managerial/ Manual population Professional Unskilled<br />
Technical to total Manual<br />
to total<br />
Col. % Col. % Col. % Ratio Rank Ratio Rank<br />
Cavan 1.2 1.6 1.4 0.81 28 1.13 17<br />
Donegal 2.7 5.2 3.5 0.78 33 1.49 1<br />
Leitrim 0.6 0.7 0.7 0.84 24 1.07 20<br />
Louth 2.3 3.1 2.6 0.87 21 1.19 10<br />
Monaghan 1.1 1.6 1.3 0.82 27 1.19 10<br />
Sligo 1.4 1.3 1.5 0.96 13 0.88 30<br />
Border 9.2 13.5 11.0 0.84 8 1.23 2<br />
Laois 1.3 1.9 1.5 0.88 20 1.26 6<br />
Longford 0.6 0.9 0.8 0.81 28 1.16 14<br />
Offaly 1.3 2.1 1.6 0.81 28 1.27 5<br />
Westmeath 1.7 2.0 1.8 0.94 16 1.08 19<br />
Midlands 5.0 6.9 5.8 0.87 7 1.19 3<br />
Galway County Borough 1.8 1.0 1.7 1.05 7 0.59 33<br />
Galway County 3.5 3.7 3.7 0.95 15 1.02 24<br />
Mayo 2.5 3.7 3.0 0.84 24 1.24 8<br />
Roscommon 1.2 1.4 1.4 0.91 18 1.03 21<br />
West 9.0 9.9 9.7 0.93 5 1.02 5<br />
Dublin County Borough 11.7 12.0 12.7 0.93 17 0.95 26<br />
Fingal 6.4 3.0 5.0 1.27 2 0.60 32<br />
South Dublin 6.3 4.1 6.1 1.04 8 0.67 31<br />
Dún Laoghaire-Rathdown 7.9 2.1 4.9 1.62 1 0.43 34<br />
Dublin 32.4 21.2 28.7 1.13 1 0.74 8<br />
Kildare 4.6 4.0 4.2 1.11 4 0.95 26<br />
Meath 3.7 3.5 3.4 1.08 6 1.03 21<br />
Wicklow 3.3 2.9 2.9 1.12 3 0.99 25<br />
Mid-East 11.6 10.4 10.5 1.10 2 0.99 7<br />
Clare 2.7 2.4 2.6 1.03 9 0.90 28<br />
Limerick County Borough 1.0 1.7 1.4 0.74 34 1.21 9<br />
Limerick County 3.2 3.2 3.1 1.03 9 1.03 21<br />
Tipperary N.R. 1.5 1.8 1.6 0.96 13 1.17 12<br />
Mid-West 8.4 9.1 8.7 0.97 4 1.04 4<br />
Carlow 1.0 1.5 1.2 0.84 24 1.31 4<br />
Kilkenny 2.1 2.3 2.1 1.00 12 1.13 17<br />
Tipperary S.R. 1.7 2.9 2.0 0.86 23 1.41 2<br />
Waterford County Borough 0.9 1.3 1.1 0.81 28 1.14 16<br />
Waterford County 1.5 1.8 1.5 1.01 11 1.25 7<br />
Wexford 2.6 4.2 3.0 0.87 21 1.40 3<br />
South-East 9.8 14.0 10.8 0.90 6 1.29 1<br />
Cork County Borough 2.6 3.7 3.1 0.81 28 1.16 14<br />
Cork County 9.1 7.4 8.3 1.09 5 0.90 28<br />
Kerry 3.0 4.0 3.4 0.89 19 1.17 12<br />
South-West 14.6 15.0 14.8 0.99 3 1.02 5<br />
Total 100.0 100.0 100.0<br />
Source: Census 2002<br />
161
xAppendix 1 7/7/05 6:59 am Page 162<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
A1.2 Risk of <strong>Poverty</strong> by Planning Region from the LII<br />
Survey<br />
In Chapter 4 we reported poverty and deprivation results for<br />
regional authorities and local authorities. Here, drawing on the<br />
series of Living in Ireland Surveys, we report trends over time<br />
for planning regions, in which counties are grouped as<br />
follows.<br />
• East: Dublin, Kildare, Meath, Wicklow<br />
• South-West: Cork, Kerry<br />
• South-East: Carlow, Kilkenny, South Tipperary, Waterford,<br />
Wexford<br />
• North-East: Cavan, Louth, Monaghan<br />
• Mid-West: Clare, Limerick, North Tipperary<br />
• Midlands: Laois, Longford, Offaly, Roscommon,<br />
Westmeath<br />
• West: Galway, Mayo<br />
• North-West: Leitrim, Sligo<br />
• Donegal: Donegal.<br />
Risk of <strong>Poverty</strong> by Planning Region<br />
In Table A1.10 we show risk of poverty by planning region<br />
for 50 per cent and 60 per cent of mean income and for<br />
the consistent poverty line. At the 50 per cent line, with the<br />
exception of the East and the Mid-West, and to a lesser<br />
extent the South-East, there has been a sharp increase<br />
in the level of poverty between 1994 and 2000. The lowest<br />
rate by far of 18 per cent is observed in the East, while the<br />
highest rate of 37 per cent is observed in Donegal and the<br />
North-West. The range of regional variation is modest, with<br />
162
xAppendix 1 7/7/05 6:59 am Page 163<br />
Appendix 1<br />
Table A1.10: Risk of poverty by planning region, 1987, 1994 and 2000<br />
Planning Region 50% income 60% income 60% line +<br />
line risk line risk deprivation risk<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
% % %<br />
East 9.6 16.0 18.1 19.3 29.1 22.2 13.6 13.6 3.8<br />
South-West 17.7 17.9 32.4 31.0 34.4 40.0 16.3 13.0 7.5<br />
South-East 20.8 23.0 26.9 33.3 40.0 35.4 16.9 18.3 7.7<br />
North-East 24.6 15.9 34.1 35.7 38.2 46.2 20.6 12.8 11.6<br />
Mid-West 20.5 22.0 25.9 32.9 39.2 35.2 19.7 13.4 10.4<br />
Midlands 21.7 21.4 31.4 41.1 39.7 40.4 19.7 13.1 3.8<br />
West 19.6 20.2 30.2 31.0 35.5 39.0 13.8 7.3 5.6<br />
North-West & Donegal 27.3 24.7 36.7 42.6 43.5 46.5 22.9 23.5 8.9<br />
Total 16.9 18.9 25.8 29.0 34.7 32.9 16.4 14.9 6.2<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
163
xAppendix 1 7/7/05 6:59 am Page 164<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A1.11: Incidence of poverty by planning region, 1987, 1994 and 2000<br />
Planning Region 50% income 60% income 60% income Percentage of<br />
line incidence line incidence line + basic households<br />
deprivation incidence<br />
1987 1994 2000 1987 1994 2000 1987 1994 2000 1987 1994 2000<br />
% % % %<br />
East 20.2 32.2 27.6 23.7 31.9 26.6 29.5 39.9 24.3 35.6 38.1 39.8<br />
South-West 16.1 14.8 16.8 16.5 15.4 16.3 15.3 13.7 16.3 15.4 15.5 13.6<br />
South-East 11.0 12.7 11.7 10.2 12.0 12.1 9.2 13.0 14.1 8.9 10.4 9.9<br />
North-East 8.6 4.5 7.5 7.3 5.9 8.0 7.4 4.7 10.8 5.9 5.4 5.8<br />
Mid-West 13.2 10.2 7.7 12.4 9.9 8.2 13.0 7.8 13.1 10.9 8.9 7.7<br />
Midlands 8.6 8.3 10.0 9.5 8.4 10.2 7.9 6.3 5.0 6.7 7.3 8.5<br />
West 11.0 7.9 8.8 10.2 7.6 8.9 8.0 3.6 6.8 9.5 7.4 7.7<br />
North-West & Donegal 11.3 9.2 9.9 10.3 8.8 9.8 9.8 11.0 9.6 9.0 6.9 7.1<br />
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100 100.0 100.0 100.0<br />
Source: Living in Ireland Surveys, 1987, 1994 and 2000<br />
164
xAppendix 1 7/7/05 6:59 am Page 165<br />
Appendix 1<br />
five of the eight regions being found between 29 and 37 per<br />
cent.<br />
At the 60 per cent line, reductions in the level of poverty<br />
between 1994 and 2000 were observed for the East, with<br />
the figure going from 29 to 22 per cent, the Mid-West (39 to<br />
35 per cent) and the South-East (40 to 35 per cent).<br />
Elsewhere slight increases were observed. The lowest rate<br />
of 22 per cent was again observed in the East and the<br />
highest rates of 46 to 47 per cent were found in the North-<br />
East and Donegal and in the North-West. In 2000, seven of<br />
the eight regions were found in the range running from 35 to<br />
47 per cent.<br />
For the combined income and deprivation measure, a sharp<br />
decline in poverty was found in all regions between 1994<br />
and 2000, with the exception of the North-East, where the<br />
decline was more modest. The largest absolute decline was<br />
in Donegal and the North-West, where it went from 24 to 9<br />
per cent, but similar proportionate increases were observed<br />
in a number of areas. The lowest poverty rate in 2000 of<br />
less than 4 per cent was observed in the East and the<br />
Midlands, followed by the West, where the figure was 6 per<br />
cent. It then rises to 8 to 9 per cent for the South-East and<br />
the South-West and the North-West and Donegal. It then<br />
reaches almost 12 per cent for the North-East.<br />
A1.3 Incidence of <strong>Poverty</strong> by Planning Region from<br />
the LII Survey<br />
In Table A1.11 we set out incidence of poverty by planning<br />
region. At the 50 per cent line changes in the distribution of<br />
poverty across planning region over time were extremely<br />
165
xAppendix 1 7/7/05 6:59 am Page 166<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
modest. The most notable change was a decline from 32 to<br />
28 per cent of poor households located in the East.<br />
The picture is very similar for the 60 per cent line, where the<br />
largest change between 1994 and 2000 is a reduction in the<br />
percentage for the East from 32 to 27 per cent. The changes<br />
for the consistent poverty line are somewhat more<br />
substantial. The number located in the East (including Dublin)<br />
increased significantly, from 30 to 40 per cent between 1987<br />
and 1994, and declined very significantly between 1994 and<br />
2000 to 24 per cent. Corresponding increases between 1994<br />
and 2000 are observed in the North-East, where the figure<br />
goes from 5 to 11 per cent, and in the West, where it goes<br />
from 4 to 7 per cent.<br />
166
xAppendix 2 7/7/05 6:59 am Page 167<br />
Appendix 2<br />
Methodology<br />
A2.1 The Surveys<br />
As well as drawing on data from the 2002 Census of<br />
Population, this report makes extensive use of data from<br />
two surveys: the Living in Ireland Surveys (LII) and the<br />
National Survey of Housing Quality (NSHQ). The key<br />
features of each of these surveys are outlined below.<br />
The Living in Ireland Surveys<br />
The Living in Ireland Surveys form the Irish component of<br />
the European Community Household Panel (ECHP), an<br />
EU-wide project co-ordinated by Eurostat to conduct<br />
harmonised longitudinal surveys dealing with household<br />
income and labour situations in the member states. The aim<br />
of the ECHP is to produce a fully harmonised data set<br />
providing information on the social situation, financial<br />
circumstances and living standards of a panel of<br />
households to be followed over several years. The fact that<br />
the same set of households is interviewed each year means<br />
that it is possible to study changes in the characteristics<br />
and circumstances of particular households or individuals<br />
over time.<br />
167
xAppendix 2 7/7/05 6:59 am Page 168<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
The first wave of the LII was conducted in 1994, and the<br />
same individuals and households were followed each year.<br />
The wave conducted in 2001, therefore, was the eighth wave<br />
of the survey. In 2000, the Irish sample of individuals and<br />
households followed from Wave 1 was supplemented by the<br />
addition of 1,500 new households to the total. This was done<br />
in order to increase the overall sample size, which had<br />
declined due to attrition since 1994. A larger sample size<br />
ensures that the precision of estimates of key figures, such as<br />
the poverty rate and average equivalised household income,<br />
remained at a high level. It also allows a greater<br />
disaggregation of the data so that the situation of policyrelevant<br />
sub-groups, such as the unemployed or older adults,<br />
can be examined. These additional households, as well as<br />
the original sample, were followed in 2001.<br />
Survey Structure<br />
The ECHP involves a household questionnaire which is<br />
completed by the ‘reference person’ or person responsible for<br />
the accommodation, and an individual questionnaire which is<br />
completed by each adult (age 16 or over) in the household.<br />
The main items of information collected on the questionnaires<br />
were household size and composition, tenure, standard of<br />
living (things the household has or can afford to do), financial<br />
strain, economic activity, training and education, general<br />
health and very detailed information on income from<br />
employment, self-employment, personal and occupational<br />
pensions, social welfare, education and training-related<br />
allowances and grants, property (interests, dividends, rental<br />
income) and other sources. Income was measured in great<br />
detail and income information was collected from each adult<br />
in the household.<br />
168
xAppendix 2 7/7/05 6:59 am Page 169<br />
Appendix 2<br />
The questionnaires were administered in a face-to-face<br />
interview by the ESRI’s team of interviewers. In farm<br />
households, a farm questionnaire was also completed to<br />
collect information to be used in conjunction with Teagasc’s<br />
National Farm Survey to estimate the income flow (family<br />
farm income) of farm households. This approach was<br />
necessary because the nature of farm income, being a<br />
combination of market profit or loss, grants and subsidies,<br />
makes it difficult for respondents to provide the figure directly.<br />
The Sample<br />
The sample of households was originally selected for the<br />
1994 wave of the survey. The objective of the sample design<br />
was to obtain a representative sample of private households<br />
in Ireland. Those living in institutions such as hospitals,<br />
nursing homes, convents, monasteries and prisons are<br />
excluded from the target population, in line with the<br />
harmonised guidelines set down by Eurostat and standard<br />
practice adopted in surveys of this kind (such as the<br />
Household Budget Survey conducted by the Central<br />
Statistics Office 1 ). Among those effectively excluded from the<br />
target population are a number of small groups that face a<br />
relatively high risk of poverty, such as the homeless and<br />
Travellers. To do justice to the particular circumstances of<br />
groups such as these would require a different research<br />
methodology.<br />
1 Collective households, however, were included. These are private<br />
households containing five or more unrelated persons with a looser<br />
budget-sharing relationship than in the standard private household.<br />
The main examples are boarding or lodging houses and army<br />
barracks. An individual living in a collective household is treated as a<br />
one-person ‘sub-household’.<br />
169
xAppendix 2 7/7/05 6:59 am Page 170<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
The sample was selected using the ESRI’s RANSAM system,<br />
which was developed at the institute and has been<br />
successfully used for selecting random samples from the<br />
electoral register for over two decades. RANSAM allows one<br />
to pre-stratify the sampling frame according to any<br />
combination of census variables. In selecting the sample for<br />
the Living in Ireland Survey, the following strata were used:<br />
• Province: Four categories – Dublin, rest of Leinster,<br />
Munster, Ulster/Connaght.<br />
• Urban/rural: Two categories – district electoral divisions<br />
(DEDs) with more than 50 per cent of their population in<br />
towns with a population of 1,500 or more versus the rest.<br />
• Unemployment: Two categories – DEDs with an<br />
unemployment rate of 16 per cent or more versus the rest.<br />
The target sample selected using the ESRI’s RANSAM<br />
procedure was a sample of persons, not of households. Since<br />
the probability of selection is greater for households with a<br />
larger number of registered voters, this means that the<br />
resulting sample will tend to over-represent larger<br />
households. This was taken into account in reweighting the<br />
sample for analysis.<br />
The total number of households successfully interviewed in<br />
1994 was 4,048, representing 57 per cent of the valid sample.<br />
This response rate is as one would expect in an intensive and<br />
demanding survey of this nature, and is comparable to the<br />
response rates achieved in the Household Budget Surveys. A<br />
total of 14,585 persons were members of the completed<br />
households. Of these, 10,418 were eligible for personal<br />
interview, i.e. born in 1997 or earlier, and 9,904 eligible<br />
respondents completed the full individual questionnaire (964<br />
on a proxy basis).<br />
170
xAppendix 2 7/7/05 6:59 am Page 171<br />
Appendix 2<br />
Reinterviewing Households and Tracing Individuals Who<br />
Move<br />
The sample from the Wave 1 (1994) Living in Ireland Survey<br />
was followed in subsequent years and reinterviewed. The<br />
follow-up rules for the survey meant that new households<br />
might be included in each wave where a sample person from<br />
Wave 1 moved to another household. All individuals in the<br />
Wave 1 sample were to be followed in subsequent waves and<br />
household and individual interviews were to be conducted, as<br />
long as the person still lived in a private or collective<br />
household within the EU.<br />
Sample Supplementation in 2000<br />
Even with a relatively high year-on-year response rate, there<br />
was a substantial loss of respondents over time. Of the<br />
original sample individuals who were still ‘in scope’ in 2000,<br />
5,530 individuals still in scope 2 (40 per cent) were in<br />
completed households.<br />
The main reason for household non-response was refusal<br />
(ranging from 9 per cent of the eligible sample in Wave 2 to 5<br />
per cent in Wave 5). Among the newly generated households,<br />
difficulties in obtaining forwarding addresses for those who<br />
moved also contributed to the non-response rate.<br />
Detailed checks on the pattern of attrition between waves<br />
showed that there was no evidence of a disproportionate loss<br />
of households from the upper or lower ends of the income<br />
distribution of the kind that would tend to bias estimates of<br />
average household incomes or poverty measures.<br />
2 Of the original 14,585 individuals, a total of 400 had died by 2000 and<br />
324 had moved to an institution or outside the EU. This left 13,861<br />
individuals still ‘in scope’ by 2000.<br />
171
xAppendix 2 7/7/05 6:59 am Page 172<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
However, the reduced sample size was addressed by<br />
supplementing the sample in Wave 7. The new sample was<br />
selected using the same procedure as for the first wave of the<br />
survey in 1994, using the ESRI’s RANSAM programme based<br />
on the electoral register. The household response rate<br />
reached 57 per cent for the 2,661 new sample households<br />
contacted by interviewers. This is the same as the rate<br />
achieved in Wave 1 and is in line with the typical response<br />
rate in other surveys of a demanding nature, such as the<br />
Household Budget Survey.<br />
The sample supplementation exercise, together with the<br />
follow-up of continuing households, resulted in a completed<br />
sample in 2000 of 11,450 individuals in 3,467 households.<br />
Individual interviews were conducted with 8,056 respondents,<br />
representing 93 per cent of those who were eligible (born in<br />
1983 or earlier).<br />
Sample Weights for the LII Data<br />
The purpose of sample weighting is to compensate for any<br />
biases in the distribution of characteristics in the completed<br />
survey sample compared to the population of interest,<br />
whether such biases occur because of sampling error, from<br />
the nature of the sampling frame used, differential response<br />
rates or attrition.<br />
Whatever the source of the discrepancy between the sample<br />
and population distributions, we would like to adjust the<br />
distributional characteristics of the sample in terms of factors<br />
such as age, sex, economic status and so on to match that<br />
of the population. In a cross-sectional survey, or in the first<br />
wave of a panel survey, the only way to check the<br />
distributional characteristics of the sample is to compare<br />
172
xAppendix 2 7/7/05 6:59 am Page 173<br />
Appendix 2<br />
sample characteristics to external population figures from<br />
sources such as the census, the Labour Force Survey, official<br />
statistics on the number of social welfare recipients from the<br />
Department of Social Welfare and so on. In waves following<br />
the first wave of a panel survey, we can also compare the<br />
characteristics of the individuals and households successfully<br />
followed to those of the individuals and households in a<br />
previous wave of the survey. In constructing the weights for<br />
the Living in Ireland Survey in Waves 2 and subsequently,<br />
both of these methods were used.<br />
The household weights were developed in a number of steps:<br />
• The first step involved adjusting the continuing sample for<br />
attrition.<br />
• The second step was to calibrate the sample totals<br />
against population totals from external sources.<br />
The external sources of information used were the Quarterly<br />
National Household Survey (QNHS), the Department of<br />
Social, Community and Family Affairs statistics on social<br />
welfare recipiency levels and figures from Teagasc on the<br />
total number of farms by farm size. The result of the<br />
weighting procedure was to ensure as close a match as<br />
possible between the sample and the population in terms of<br />
the distribution of the characteristics shown in Table A2.1.<br />
Apart from incorporating weights to control for attrition from<br />
previous waves and the availability of new technology for<br />
constructing weights, the logic and general strategy in<br />
developing the weights for Waves 2 to 7 was very similar to<br />
that used in Wave 1. The resulting match between the<br />
weighted sample characteristics and the population<br />
173
xAppendix 2 7/7/05 6:59 am Page 174<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
characteristics for the 2001 data was highly satisfactory,<br />
confirming that the weights are effective in adjusting the<br />
achieved sample to population characteristics.<br />
Table A2.1: External population characteristics used in the<br />
construction of household weights for the Living in Ireland<br />
Surveys<br />
Household characteristics:<br />
Household size (total size, number over 18 and number over 65)<br />
Location (Dublin, other county borough, rural)<br />
Number of persons at work (0, 1 and 2 or more)<br />
Head age (under 25, age 25 and over)<br />
Number of farms in each of six size categories<br />
Individual characteristics:<br />
Number of males and females by 10 age categories<br />
Number of males and females age 15+ by 11 age/marital status<br />
categories<br />
Number of recipients of 12 major social welfare payments<br />
Number of males and females by seven economic status categories (at<br />
work (ILO), unemployed (ILO), unemployed (not ILO), student, home<br />
duties, retired, other)<br />
Number of males and females age 20–64 by level of education (four<br />
categories)<br />
The National Survey of Housing Quality (NSHQ)<br />
The Economic and Social Research Institute (ESRI) was<br />
commissioned by the Department of the Environment to carry<br />
out the Irish National Survey of Housing Quality (NSHQ) in<br />
2001/2002. The purpose of the survey was to record very<br />
174
xAppendix 2 7/7/05 6:59 am Page 175<br />
Appendix 2<br />
detailed information on the condition of the national housing<br />
stock. Similar surveys were undertaken in 1981 and 1991, but<br />
were conducted by the local authorities themselves. This is<br />
the first time that the methodology, administration and<br />
protocols for the survey have been completely centralised,<br />
ensuring a harmonised set of data across local authority<br />
areas.<br />
The survey collected detailed information on the condition of<br />
the dwelling; on residents’ satisfaction with aspects of their<br />
dwelling, such as costs, heating system and water supply; on<br />
problems in the area where the dwelling is located; and on<br />
problems with the affordability of the dwelling itself, with<br />
heating the dwelling or with home appliances and furnishings.<br />
The Sample<br />
One of the requirements of the survey was to provide a<br />
database to the Department with a large enough sample to<br />
yield separate breakdowns at local authority level. The NSHQ<br />
completed sample size was over 40,000 households<br />
throughout the country. This is an extremely large sample by<br />
the standards of other sample surveys which have been<br />
previously carried out in Ireland. The sample of addresses<br />
was selected using the ESRI’s RANSAM programme, which<br />
uses a multi-stage randomised design based on the electoral<br />
register.<br />
The Survey<br />
The survey is based on a questionnaire interview of a<br />
household respondent. In each household, the person<br />
responsible for the accommodation (the owner, purchaser or<br />
tenant) was interviewed. A pilot test of the questionnaire was<br />
175
xAppendix 2 7/7/05 6:59 am Page 176<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
conducted in August 2001 and the main survey went into the<br />
field in September. The fieldwork for the main survey<br />
extended from September 2001 to summer 2002. The<br />
questionnaire had an average completion time of 30 minutes.<br />
The overall response rate was 75 per cent. By the standards<br />
of statistical probability surveys currently undertaken in<br />
Ireland, these response levels are extremely high. The higher<br />
than usual response level in the NSHQ can be attributed to<br />
intensive interviewer training and sustained call-backs on the<br />
part of the interviewers.<br />
Sample Weights<br />
As noted above, sample weights are constructed to ensure<br />
that the sample is representative of the population along a<br />
number of key dimensions, such as region, household size,<br />
labour force participation, age of dwelling and so on. These<br />
weights adjust the sample for any lack of overall<br />
representativeness arising from sample design, the sampling<br />
frame available and patterns of non-response. The sample<br />
design for the NSHQ would have over-represented rural areas<br />
because of the requirement, noted above, for a sample of<br />
sufficient size to provide local authority-level tables. This<br />
meant that smaller local authority areas were overrepresented<br />
in the sample compared to their populations. The<br />
sampling frame, based on the electoral register, also tends to<br />
over-represent households with a larger number of persons<br />
over age 18. Differences in response rates are typically found<br />
between urban and rural areas, with higher response rates in<br />
the latter.<br />
The sample weights were constructed by adjusting the<br />
sample proportions to population figures based on the most<br />
176
xAppendix 2 7/7/05 6:59 am Page 177<br />
Appendix 2<br />
up-to-date information available. The population figures drew<br />
on data from reliable external sources, such as the<br />
preliminary figures from the 2002 census, from the Quarterly<br />
National Household Surveys and from the 1996 census with<br />
adjustments for population change.<br />
There were a number of steps involved in constructing the<br />
weights. The first involved constructing a weight to control for<br />
the fact that the sampling frame (based on the electoral<br />
registers) will tend to over-represent households with a larger<br />
number of adults. The weight was Wt1 = 1/A, where A is the<br />
number of adults age 18 or over in the household.<br />
The second weight grossed the number of sample cases in<br />
each local authority area up to the total number of private<br />
households in that local authority area, using preliminary<br />
figures provided by the Central Statistics Office based on the<br />
2002 census. Wt2 = (Wt1*PL )/SL, where PL refers to the total<br />
number of households in the local authority area and SL<br />
refers to the number of sample households in that local<br />
authority area.<br />
The next stage involved what is normally referred to as<br />
calibration (see, for example, Deville and Särndal 1992) – the<br />
second weight (Wt2) was adjusted so as to match the sample<br />
distribution of a given set of characteristics to the population<br />
distribution of these characteristics derived from external<br />
sources. The Gross programme was used to gross this<br />
second weight to local authority- and region-level totals for a<br />
set of control variables. 3 177<br />
3 This programme, developed by Johanna Gomulka, uses a minimum<br />
distance algorithm to adjust an initial weight (in this case Wt2) so that<br />
the distribution of cases in the sample matches a set of control totals.
xAppendix 2 7/7/05 6:59 am Page 178<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
The region-level totals were obtained from the Central<br />
Statistics Office, which provided special tabulations from<br />
the QNHS (second quarter, 2001). The local authority-level<br />
totals were obtained from the 1996 census (household size,<br />
number of persons at work) and the Department of the<br />
Environment Housing Statistics (number of local authority<br />
rented dwellings, new dwellings built after 1991). The countylevel<br />
figures from the 1996 census were updated to 2002<br />
figures using region-level information from the QNHS and<br />
preliminary county-level population and household totals from<br />
the 2002 census.<br />
At the time of constructing the weights, only the total males,<br />
total females and an estimate of the total number of<br />
households was available from the 2002 census. These<br />
figures were used to adjust the total number of males,<br />
females and households for each local authority to the figures<br />
for 2002.<br />
Table A2.2 shows the population checks that were included<br />
and the level (county or region level). For some variables,<br />
recent information was only available at the level of planning<br />
region (from the QNHS).<br />
A2.2 Imputation of Missing Information on NSHQ<br />
Missing information on key background variables in the<br />
NSHQ 2001/2002 was imputed based on other data on the<br />
household. This was done to ensure that all figures in a table<br />
were based on the same set of cases. Imputation was also<br />
conducted for the variables used to construct the weights.<br />
The variables where the level of missing information<br />
exceeded 5 per cent are shown in Table A2.3. The figure also<br />
178
xAppendix 2 7/7/05 6:59 am Page 179<br />
Appendix 2<br />
Table A2.2: Population checks for sample weighting<br />
Population checks at county level:<br />
• Household size (number of persons age 18 or over). From 1996<br />
census adjusted to 2002 figures using QNHS 2001 at region level and<br />
preliminary figures from 2002 census of number of males, females<br />
and households by local authority area.<br />
• Number of persons in household at work (three categories: none, one,<br />
two or more). From 1996 census adjusted to 2002 figures using<br />
QNHS at region level for second quarter 2001 and preliminary figures<br />
from 2002 census.<br />
• Number of local authority rented dwellings (from Department of the<br />
Environment Housing Statistics, September 2001).<br />
• Age of dwelling (from the 1991 census, updated using figures from the<br />
Department of the Environment Housing Statistics 2002 on new<br />
dwellings built since then 4 ).<br />
Population checks at level of planning region:<br />
• Household size (six categories, persons of all ages; from QNHS 2001).<br />
• Household type (five categories; from QNHS 2001).<br />
• Tenure (owner occupied, renter, other tenure; from QNHS 2001).<br />
• Age by sex (10 age groups; from QNHS 2001).<br />
• Occupation of oldest person, if at work (ISCO88, five categories; from<br />
QNHS 2001).<br />
• Education by sex (three categories of education; from QNHS 2001).<br />
• Economic status by sex (at work, unemployed, home duties, retired,<br />
student, other; from QNHS 2001).<br />
4 It was assumed that 0.6 per cent per annum of the 1991 housing stock<br />
was lost through demolition by 2002 – a total of 64,471 dwellings. It<br />
was further assumed that older dwellings would be lost at a greater<br />
rate: 70 per cent from the pre-1919 stock; 20 per cent from the<br />
1919–40 stock and 10 per cent from the 1941–60 stock. Of dwellings<br />
built since 1991, it was assumed that 1 per cent of the total built from<br />
1991–96 were for holiday use, rising to 1.5 per cent of the total built<br />
after 1996.<br />
179
xAppendix 2 7/7/05 6:59 am Page 180<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
shows the percentage of values imputed and the basis on<br />
which imputation was conducted.<br />
The imputation involved matching the household where the<br />
data was missing to a similar household with non-missing<br />
data using a set of related characteristics (typically county,<br />
cluster, tenure, household size and other variables that are<br />
predictive of the variable to be imputed). The imputed value<br />
was taken from the household with the closest match in<br />
terms of these characteristics. This approach is preferable to<br />
imputing an average value since it preserves the variation of<br />
the variable being imputed.<br />
Information on household members (age, sex, education,<br />
economic status and occupation of the oldest person) was<br />
needed for all households to weight the sample.<br />
A2.3 Income Correction Factor<br />
Measures of household income based on a single item, as is<br />
done in the NSHQ, will tend to understate income compared<br />
to the figure obtained if all household members are asked<br />
about their income from different sources. We know this from<br />
the Living in Ireland (LII) Surveys. The LII Surveys make use of<br />
both a single-item measure on the household questionnaire<br />
and a detailed set of questions on each income source posed<br />
to all adults in the household. The single-item measure<br />
understates total household income by 19 per cent (of the full<br />
measure) on average (or 24 per cent of the single-item<br />
measure). Table A2.4 shows that the degree to which the<br />
single-item measure understates total income is greater for<br />
households with a large number of income sources (typically<br />
associated with a larger number of adults) and households<br />
180
xAppendix 2 7/7/05 6:59 am Page 181<br />
Appendix 2<br />
Table A2.3: Level of missing information on key variables and<br />
imputation procedure<br />
Variable % imputed Variables used to impute the value<br />
Sex of household 1.4 Household size, local authority area,<br />
member<br />
cluster, sex of spouse (where applicable)<br />
Age of household 5.8 Age of spouse/parent/child (where<br />
member (all members)<br />
applicable), economic status (retired),<br />
household size, local authority area,<br />
cluster<br />
Highest level of 19.8 Occupation, age, sex, household size,<br />
education achieved<br />
housing tenure, local authority area,<br />
by each household<br />
cluster<br />
member<br />
Economic status 3.7 Age, sex, household size, local<br />
of household<br />
authority area, cluster<br />
member<br />
Occupational group 3.3 Housing tenure, age, education, Local<br />
of oldest person in<br />
authority area, cluster<br />
the household, if<br />
at work<br />
Household income 12.3 Social class of householder, number of<br />
persons at work, number of adults, local<br />
authority area, cluster<br />
Housing tenure 2.2 Household size, local authority area,<br />
cluster<br />
Household type 1.3 Household size, age of reference person,<br />
age of other persons, local authority area,<br />
cluster<br />
Size of place 2.3 Local authority area, cluster<br />
Age of dwelling 1.7 Local authority area, cluster<br />
Floor area 75.0 No imputation<br />
Presence of wall 18.0 No imputation<br />
insulation<br />
Source: National Survey of Housing Quality, 2001/2002<br />
181
xAppendix 2 7/7/05 6:59 am Page 182<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
where the main source of income is from self-employment or<br />
agriculture. The difference between the two measures is<br />
smallest for one-adult or two-adult households relying on<br />
pension or social welfare income.<br />
The understatement is particularly marked where there is<br />
income from work and where there are a large number of<br />
adults in the household. A regression-based model was<br />
developed to correct for this understatement using variables<br />
that are measured on both the NSHQ and the LII Survey from<br />
2000. The model was developed using the LII Survey and<br />
then the coefficients for the model were used to ‘correct’ the<br />
income measure on the NSHQ. The single-item measure of<br />
income in the LII Survey recorded income as a continuous<br />
amount, or into 10 categories if an exact amount could not<br />
be provided. Since the NSHQ used a categorical variable, the<br />
LII incomes were recoded into a categorical format before<br />
running the model. This would enable us to simulate the<br />
relationship between the continuous distribution of income<br />
based on aggregating information collected in detail from all<br />
adults in the household and a categorical measure recorded<br />
by the householder.<br />
The coefficients from the model are shown in Table A2.5. The<br />
r-squared for the model is .644, indicating that the variables<br />
included in the model explain about 64 per cent of the<br />
variance in income. 5<br />
The model used the income category (coded as a<br />
dichotomous variable with a value of 1 for each category), the<br />
number of adults in the household, the number of children in<br />
the household and the number of persons at work. 6 Table<br />
5 The r-square for the model with the income categories alone is .54.<br />
182
xAppendix 2 7/7/05 6:59 am Page 183<br />
Appendix 2<br />
Table A2.4: Mean weekly household income (in £) using full<br />
measure and single-item measure by number of adults and<br />
number of persons at work<br />
Single- Full Difference Difference Difference<br />
item measure as % of as % of<br />
measure single-item full<br />
measure measure<br />
Number over 18<br />
1 184 195 11 6% 5%<br />
2 422 495 73 17% 15%<br />
3 483 633 149 31% 24%<br />
4 or more 621 932 311 50% 33%<br />
Number at work<br />
0 170 169 –2 –1% –1%<br />
1 345 407 62 18% 15%<br />
2 or more 569 748 179 31% 24%<br />
Total 389 482 93 24% 19%<br />
Source: Living in Ireland Survey, 2000<br />
A2.5 shows that incomes clearly bear a strong relationship to<br />
the income category. The category coefficients in Table A2.4<br />
are below the lower bound of the category itself because they<br />
are shown net of the effect of number of adults and number<br />
of persons at work. Each household will have at least one<br />
adult and at higher levels of income are likely to have at least<br />
one person at work. The number of adults and the number of<br />
6 A number of more complex models were tested, including variables<br />
such as tenure, region, education and age of householder and<br />
dichotomous variables for number of adults and number at work, but<br />
no improvement in the predictive power of the model was achieved.<br />
183
xAppendix 2 7/7/05 6:59 am Page 184<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
adults at work also have strong coefficients. The effect of<br />
additional children is much weaker and does not reach<br />
statistical significance. Nevertheless, it was included in the<br />
model because in a household survey such as the NSHQ,<br />
which does not have income as a central focus, it is likely<br />
that many householders did not include Child Benefit in their<br />
estimate of total income.<br />
Table A2.5: Model based on LII to correct for understatement<br />
of income when a single-item measure is used<br />
Variables Coefficient Std. error<br />
Constant –9.18 89.62<br />
Number of adults over 18 55.42 5.03<br />
Number of children under 18 2.73 3.45<br />
Number of adults at work 96.44 5.73<br />
Income £50–£99 50.41 90.43<br />
Income £100–£149 67.34 90.55<br />
Income £150–£199 81.62 90.06<br />
Income £200–£249 110.00 90.48<br />
Income £250–£299 173.06 90.28<br />
Income £300–£399 215.77 89.98<br />
Income £400–£499 275.46 90.16<br />
Income £500–£599 374.17 90.45<br />
Income £600–£999 521.26 90.30<br />
Income over £1,000 888.55 92.44<br />
Source: Living in Ireland Survey, 2000. Note: The omitted category for<br />
income is ‘under £50’.<br />
184
xAppendix 2 7/7/05 6:59 am Page 185<br />
Appendix 2<br />
Table A2.6: Income category midpoints and coefficients applied<br />
to the survey of house quality<br />
Lower Upper Point Lower Upper Point Coeffibound<br />
bound estimate bound bound estimate cients<br />
£ £ £ € € €<br />
Four-category 0 190 132 0 241 167 0.54<br />
measure 191 360 267 243 457 339 0.63<br />
(2.4% of 361 570 454 458 724 576 0.61<br />
households) 571 1,000 787 725 1270 999 0.65<br />
Sixteen- 0 85 75 0 108 54 0.00<br />
category 86 110 98 109 140 124 0.63<br />
measure 111 150 131 141 190 166 0.54<br />
(85.3% of 151 190 171 192 241 216 0.47<br />
households) 191 220 206 243 279 261 0.49<br />
221 270 246 281 343 312 0.63<br />
271 320 296 344 406 375 0.63<br />
321 360 341 408 457 432 0.62<br />
361 400 381 458 508 483 0.62<br />
401 450 426 509 571 540 0.61<br />
451 500 476 573 635 604 0.61<br />
501 570 536 636 724 680 0.68<br />
571 650 611 725 825 775 0.68<br />
651 750 701 827 952 889 0.65<br />
751 950 851 954 1,206 1,080 0.65<br />
951 Open 1,100 1,208 Open 1,333 0.81<br />
In the NSHQ, there were 16 income categories, rather than 10<br />
as in the LII data, and the amounts were presented to the<br />
respondents either in Irish pounds or in euro, depending on<br />
respondent preference, since the survey spanned the period<br />
of the euro changeover. The midpoints of the NSHQ income<br />
185
xAppendix 2 7/7/05 6:59 am Page 186<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
categories were matched to the nearest category from the<br />
LII so that the appropriate correction could be applied to<br />
the income category. The coefficients used were obtained<br />
by dividing the midpoint of each income category (shown<br />
in Table A2.6) by the corresponding coefficient in the<br />
model. The coefficients applied to each category are<br />
shown in Table A2.6.<br />
Table A2.7 shows the mean ‘corrected’ income for each<br />
household income category. Overall, incomes are adjusted<br />
upwards by about 24 per cent (see Table A2.8). In general,<br />
incomes in the lower categories tend to be adjusted<br />
upwards to a greater extent than incomes in the higher<br />
categories.<br />
The final column of Table A2.7 shows the percentage by<br />
which the predicted income would have been understated<br />
if the midpoint of the categories based on the single item<br />
had been used instead of the ‘corrected’ income. The<br />
biggest change is to the lowest category (€0 to €108).<br />
For the lowest income category, taking the midpoint of the<br />
category as a point estimate would not have been a good<br />
choice in any case. The general shape of the income<br />
distribution, rising steeply towards the lower end, would<br />
indicate the choice of a point estimate towards the upper<br />
bound of this category rather in the middle of it.<br />
Table A2.8 shows that the difference between the singleitem<br />
measure and the ‘corrected’ income is minimal where<br />
there is only one adult in the household or where there is<br />
nobody at work in the household. The difference is much<br />
larger where there are several adults in the household (the<br />
average increase is 65 per cent where there are four or<br />
more adults in the household) and where there are adults at<br />
186
xAppendix 2 7/7/05 6:59 am Page 187<br />
Appendix 2<br />
Table A2.7: Mean ‘corrected’ income for each original income<br />
category in the NSHQ<br />
Lower Upper Mean Implied<br />
bound (€) bound (€) ‘corrected’ ‘under-<br />
Income statement’<br />
(€)<br />
Four-category 0 241 292 43<br />
measure 243 457 562 40<br />
(2.4% of households) 458 724 759 24<br />
725 1,270 1,142 13<br />
Sixteen-category 0 108 198 73<br />
measure (85.3% of 109 140 219 43<br />
households) 141 190 260 36<br />
192 241 333 35<br />
243 279 386 32<br />
281 343 491 37<br />
344 406 572 34<br />
408 457 633 32<br />
458 508 678 29<br />
509 571 723 25<br />
573 635 781 23<br />
636 724 899 24<br />
725 825 981 21<br />
827 952 1,050 15<br />
954 1,206 1,193 9<br />
1,208 Open 1,607 17<br />
work. The increase is 12 per cent where there is one person<br />
at work and 37 per cent, on average, where there are two or<br />
more people at work.<br />
187
xAppendix 2 7/7/05 6:59 am Page 188<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Table A2.8: Average income before and after correction by<br />
number of adults and number at work in the NSHQ<br />
A B C D<br />
Household Household Difference Difference<br />
income income (B – A) %<br />
(uncorrected) (corrected) (C ÷ A)<br />
€ per week € per week<br />
Number of adults (18+)<br />
1 316 316 0 0<br />
2 559 643 84 15<br />
3 601 815 214 36<br />
4 or more 726 1195 470 65<br />
Number at work<br />
0 261 260 –2 –1<br />
1 499 558 59 12<br />
2 or more 719 983 263 37<br />
Total 519 641 122 24<br />
A2.4 Comparing Disparities in Relative <strong>Poverty</strong> Risk<br />
in the LII and NSHQ<br />
As noted in Chapter 3, the analysis of poverty risk using the<br />
NSHQ focuses on disparities in risk, i.e. on the ratio of the<br />
risk faced by a particular sub-group to the overall risk at a<br />
national level. The following tables compare the disparities in<br />
risk of income poverty found using the NSHQ to those using<br />
the LII for a comparable set of variables.<br />
As can be seen from Table A2.9, the figures from both<br />
sources are very similar for relative income poverty. However,<br />
there is some tendency for the relative risk associated with<br />
working or not working to be exaggerated in the NSHQ. This<br />
188
xAppendix 2 7/7/05 6:59 am Page 189<br />
Appendix 2<br />
Table A2.9: Disparities in risk of income poverty by region,<br />
tenure, household size, economic status and social class from<br />
the LII Survey and the NSHQ<br />
<strong>Poverty</strong> risk, 50% <strong>Poverty</strong> risk, 50%<br />
mean, Scale A mean, Scale A over<br />
over household household<br />
LII NSHQ LII NSHQ<br />
Region Border 1.4 1.3 1.4 1.3<br />
Dublin 0.7 0.7 0.6 0.8<br />
Mid-East 0.8 0.8 0.8 0.8<br />
Midlands 1.1 1.2 1.0 1.2<br />
Mid-West 1.0 1.1 1.1 1.1<br />
South-East 1.1 1.2 1.1 1.1<br />
South-West 1.2 1.1 1.2 1.1<br />
West 1.3 1.2 1.4 1.1<br />
Tenure Owner 0.8 0.8 0.8 0.8<br />
LATP 1.2 1.4 1.3 1.4<br />
Rent free 1.1 1.2 1.1 1.1<br />
LA tenant 2.6 2.4 2.4 2.2<br />
Private tenant 0.7 0.8 0.7 0.8<br />
Household 1 1.9 1.8 1.6 1.8<br />
size (all ages) 2 1.0 1.3 1.1 1.0<br />
3 0.6 0.6 0.7 0.7<br />
4 0.6 0.4 0.6 0.5<br />
5 or more 0.7 0.7 0.8 0.9<br />
Economic Work 0.4 0.2 0.5 0.3<br />
status of Unemployed 2.5 3.9 2.3 3.3<br />
householder Retired 1.5 2.6 1.6 2.4<br />
Home duties 2.5 3.4 2.1 3.1<br />
Student 2.4 2.2 2.4 2.2<br />
Other 2.7 3.8 2.6 3.2<br />
Social High professional/<br />
class of managerial 0.3 0.2 0.3 0.3<br />
householder Low professional/<br />
managerial 0.5 0.4 0.5 0.5<br />
Other non-manual 0.7 0.7 0.7 0.8<br />
Skilled manual 0.9 0.8 1.0 0.8<br />
Semi-skilled manual 1.5 1.1 1.5 1.1<br />
Unskilled manual 2.0 1.5 1.9 1.5<br />
Unknown 2.4 2.3 2.2 2.1<br />
Number at 0 2.7 3.2 2.4 2.8<br />
work in 1 0.6 0.3 0.8 0.5<br />
household 2 0.2 0.0 0.2 0.1<br />
3 or more 0.1 0.0 0.2 0.0<br />
Total 1.0 1.0 1.0 1.0<br />
189
xAppendix 2 7/7/05 6:59 am Page 190<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
is likely to be due to the use of the work variable in the<br />
income correction equation described above. As we shall<br />
see later, the differences disappear when material<br />
deprivation, as well as income, is taken into consideration.<br />
A2.5 Comparing the Pattern of Disparities in<br />
Deprivation in the LII Survey and NSHQ<br />
Table A2.10 presents corresponding results for the<br />
modified consistent poverty (MCP) measure discussed in<br />
Chapter 3. Note that the measures use different sets of<br />
items for the ‘basic deprivation’ set so that the results from<br />
the NSHQ are for ‘modified’ consistent poverty, rather than<br />
for consistent poverty as measured in the LII Survey.<br />
Consistent poverty in the LII Survey is measured as being<br />
below 60 per cent of the mean equivalised household<br />
income (using Scale A) and lacking at least one item from<br />
the set of basic deprivation items (meat, new clothes,<br />
adequate heat, overcoat, shoes, roast, debt, do without<br />
meal, do without heat). MCP in the NSHQ is measured as<br />
being below 60 per cent of the mean equivalised<br />
household income (Scale A) and lacking at least one of the<br />
items in the basic set from the NSHQ (meat, new clothes,<br />
adequate heat, arrears, socialising, furniture and holiday).<br />
Given the differences between the measures of deprivation,<br />
it should not be surprising that the results diverge to a<br />
greater extent than for income poverty risk. Nevertheless,<br />
the main patterns still hold up well across both measures.<br />
The NSHQ measure shows somewhat less variation by<br />
regional authority, but the Border region remains most<br />
deprived and the Dublin region least deprived. The<br />
190
xAppendix 2 7/7/05 6:59 am Page 191<br />
Appendix 2<br />
Table A2.10: Disparities in risk of (modified) consistent poverty<br />
(60% mean and basic) by region, tenure, household size,<br />
economic status and social class from the LII Survey and the<br />
NSHQ<br />
Consistent<br />
poverty LII<br />
Modified<br />
consistent<br />
poverty NSHQ<br />
Regional authority area<br />
Border 1.8 1.4<br />
Dublin 0.5 0.7<br />
Mid-East 1.0 0.8<br />
Midlands 0.6 1.2<br />
Mid-West 1.5 1.1<br />
South-East 1.1 1.1<br />
South-West 1.1 1.1<br />
West 1.0 1.1<br />
Tenure<br />
Owner 0.5 0.7<br />
LATP 0.8 1.5<br />
Rent free 2.0 1.1<br />
LA tenant 5.2 2.8<br />
Private tenant 1.6 0.9<br />
Household size<br />
1 1.8 1.8<br />
2 0.9 1.0<br />
3 0.9 0.7<br />
4 0.5 0.6<br />
5 0.8 0.9<br />
Economic status of householder<br />
Work 0.3 0.3<br />
Unemployed 5.7 4.1<br />
Retired 1.2 2.2<br />
Home duties 2.4 3.3<br />
Student – 2.6<br />
Other 2.4 4.0<br />
Social class of householder<br />
High professional/managerial 0.1 0.2<br />
Low professional/managerial 0.3 0.4<br />
Other non-manual 1.0 0.7<br />
Skilled manual 0.8 0.8<br />
Semi-skilled manual 2.1 1.1<br />
Unskilled manual 1.7 1.7<br />
Unknown 1.8 2.2<br />
Number at work<br />
0 2.7 2.9<br />
1 0.7 0.5<br />
2 0.1 0.1<br />
3 0.0 0.0<br />
Total 1.0 1.0<br />
191
xAppendix 2 7/7/05 6:59 am Page 192<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Midlands appears more deprived than average on the NSHQ<br />
measure, but less deprived than average on the LII measure.<br />
The Mid-West appears 50 per cent more likely than average<br />
to face consistent poverty on the LII, but is only 10 per cent<br />
above average on the NSHQ.<br />
In terms of tenure, the local authority tenant purchasers show<br />
50 per cent more deprivation than average on the NSHQ<br />
measure, but 20 per cent less on the LII measure. On the<br />
other hand, private sector renters appear 60 per cent more<br />
deprived than average on the LII measure, but 20 per cent<br />
less deprived than average on the NSHQ. The relative<br />
disadvantage of local authority renters is much greater on the<br />
LII measure (5.2 times the average versus 2.8 times).<br />
The patterns using both measures are very similar across<br />
household size categories and for number of persons in the<br />
household at work and are broadly similar for social class of<br />
householder.<br />
Differences are also apparent for economic status of the<br />
householder. The extent of disadvantage experienced by the<br />
unemployed is greater on the LII, while the extent of<br />
disadvantage experienced by those retired, engaged on<br />
home duties or otherwise inactive (including people unable to<br />
work due to illness or disability) is greater on the NSHQ.<br />
192
xxGlossary/Refs 7/7/05 6:59 am Page 193<br />
Glossary<br />
Consistent poverty A measure of poverty that takes account<br />
of living standards as well as income. To be ‘consistently<br />
poor’, a household must fall below 60 per cent of the mean<br />
equivalised household income and lack basic necessities.<br />
Equivalised income An adjustment to total household<br />
income to take account of the household size and<br />
composition. It is the income per ‘adult equivalent’ in the<br />
household. It is calculated by giving a weight to each adult<br />
(age 15 and over) and each child (under age 15) in the<br />
household and dividing total household income by the sum of<br />
the weights for household members.<br />
Incidence of poverty This term refers to the proportion of all<br />
poor households who are found in a specific sub-group or<br />
region, e.g. the percentage of all poor households that are<br />
located in the Border-Midlands-West (BMW) region.<br />
LII The Living in Ireland Surveys – a panel survey of several<br />
thousand households that followed the same individuals over<br />
time between 1994 and 2000. It collected detailed information<br />
on incomes, economic activity, household composition and<br />
living standards. In 2000 the sample was supplemented by<br />
adding 1,500 new households.<br />
193
xxGlossary/Refs 7/7/05 6:59 am Page 194<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Nagelkerke R 2 A measure of how much of the variation in a<br />
dependent variable is explained by a set of independent<br />
variables. Like the R 2 in ordinary least squares regression<br />
(OLS), it can vary between 0 and 1. A value closer to 1<br />
indicates that more variation is ‘explained’ by the<br />
independent variables.<br />
NAPS National Anti-<strong>Poverty</strong> Strategy.<br />
NSHQ National Survey of Housing Quality conducted in<br />
2001/2002, with a sample size of over 40,000 households.<br />
Odds ratio A measure used to compare the risk of different<br />
groups. For instance, in 2000, 3.9 per cent of homeowners<br />
(without a mortgage) fell below the consistent poverty line,<br />
giving them a 0.04 (3.9/96.1) odds of being poor. On the other<br />
hand, 32.4 per cent of local authority tenants were found<br />
below the consistent poverty line, giving them a 0.48<br />
(32.4/67.6) odds of being poor. The difference in risk of being<br />
consistently poor between these two groups (local authority<br />
tenants and outright owners) can be indexed by the ratio of<br />
these odds (.48/.04), which gives an odds ratio of 12:1. Thus,<br />
local authority tenants had a 12 times greater risk of being<br />
poor than homeowners.<br />
Risk of poverty The percentage of persons or households in<br />
a group that fall below the poverty line, e.g. the percentage of<br />
households in the Border-Midlands-West (BMW) region that<br />
are poor.<br />
194
xxGlossary/Refs 7/7/05 6:59 am Page 195<br />
References<br />
Callan, T., Nolan, B. and Whelan, C.T. (1993), ‘Resources, Deprivation and<br />
the Measurement of <strong>Poverty</strong>’, Journal of Social Policy, vol. 22, no. 2,<br />
pp. 141–72.<br />
Callan, T., Nolan, B., Whelan, B.J., Hannan, D.F. and Creighton, S. (1989),<br />
<strong>Poverty</strong>, Income and Welfare, Dublin: Economic and Social Research<br />
Institute.<br />
Callan, T., Nolan, B., Whelan, B.J., Whelan, C.T. and Williams, J. (1996),<br />
<strong>Poverty</strong> in the 90s: Evidence from the 1994 Living in Ireland Survey,<br />
Dublin: Oak Tree Press.<br />
Deville, J.C. and Särndal, C.E. (1992), ‘Calibration Estimators in Survey<br />
Sampling’, Journal of the American Statistical Association, no. 87,<br />
pp. 376–82.<br />
Fahey, T. and Williams, J. (2000), ‘The Spatial Distribution of Disadvantage<br />
in Ireland’, in B. Nolan, P. O’Connell and C.T. Whelan (eds.), Bust to<br />
Boom: The Irish Experience of Growth and Inequality, Dublin:<br />
Institute of Public Administration.<br />
Friedrichs, J. (1998), ‘Do Poor Residents Make their Residents Poorer<br />
Context Effects of Poor Neighbourhoods on Residents’, in H.J.<br />
Andreß (ed.), Empirical <strong>Poverty</strong> Research in Comparative<br />
Perspective, Aldershot: Ashgate.<br />
Gomulka, J. (1992), ‘Grossing-up Revisited’, in R. Hancock and H.<br />
Sutherland (eds.), Microsimulation Models for Public Policy Analysis:<br />
New Frontiers, STICERD Occasional Paper 17, LSE.<br />
Gomulka, J. (1994), ‘Grossing Up: A Note on Calculating Household<br />
Weights from Family Composition Totals’, Microsimulation Unit<br />
Research Note MU/RN/4, University of Cambridge Department of<br />
Economics, March 1994.<br />
195
xxGlossary/Refs 7/7/05 6:59 am Page 196<br />
<strong>Mapping</strong> <strong>Poverty</strong><br />
Haase, T., McKeown, K. and Rourke, S. (1996), Local Development<br />
Strategies for Disadvantaged Areas, 192-195: Evaluation of the<br />
Global Grant in Ireland, Dublin: Area Development Management<br />
(ADM) Ltd.<br />
Jargowski, P.A. (1996), <strong>Poverty</strong> and Place: Ghettos, Barrios and the<br />
American City, New York: Russell Sage Foundation.<br />
Jencks, C. and Mayer, S. (1990), ‘The Social Consequences of Growing<br />
Up in a Poor Neighbourhood’, in L. Lynn, Jr. and M. McGeary,<br />
<strong>Poverty</strong> in the United States, Washington, DC: National Academy<br />
Press.<br />
Layte, R., Nolan, B. and Whelan, C.T. (2000), ‘Cumulative Disadvantage<br />
and Polarisation’, in B. Nolan, P. O’Connell and C.T. Whelan (eds.),<br />
Bust to Boom: The Irish Experience of Growth and Inequality, Dublin:<br />
Institute of Public Administration.<br />
Layte, R., Nolan, B. and Whelan, C.T. (2001), ‘Reassessing Income and<br />
Deprivation Approaches to <strong>Poverty</strong> in Ireland’, Economic and Social<br />
Review, vol. 32, no. 3, pp. 239–61.<br />
Murray, C. (1984), Losing Ground: American Social Policy, 1950–1980,<br />
New York: Basic Books.<br />
National Economic and Social Council (1990), Strategies for the Nineties:<br />
Economic Stability and Structural Change: Report 89, Dublin: NESC.<br />
Nolan, B. and Whelan, C.T. (1996), Resources, Deprivation and <strong>Poverty</strong>,<br />
Oxford: Clarendon Press.<br />
Nolan, B. and Whelan, C.T. (2000), ‘“Urban Housing” and the Role of<br />
“Underclass” Processes: The Case of Ireland’, Journal of European<br />
Social Policy, vol. 10, no. 1, pp. 5–21.<br />
Nolan, B., Gannon, B., Layte, R., Watson, D., Whelan, C.T. and Williams,<br />
J. (2002), Monitoring <strong>Poverty</strong> Trends in Ireland: Results from the<br />
2000 Living in Ireland Survey, Dublin: Economic and Social Research<br />
Institute.<br />
Nolan, B., Whelan, C.T. and Williams, J. (1998), Where are Poor<br />
Households: The Spatial Distribution of <strong>Poverty</strong> in Ireland, Dublin:<br />
Oak Tree Press.<br />
Pringle, D., Cook, S., Poole, M.A. and Moore, A.J. (2000), Cross-border<br />
Deprivation Analysis, Dublin: <strong>Combat</strong> <strong>Poverty</strong> <strong>Agency</strong>.<br />
Pringle, D.G. (1999), ‘Something Old, Something New: Lessons to be<br />
Learnt from Previous Strategies of Positive Territorial Discrimination’,<br />
196
xxGlossary/Refs 7/7/05 6:59 am Page 197<br />
References<br />
in D.G. Pringle, J. Walsh and M. Hennessy (eds.), Poor People: Poor<br />
Places, Dublin: Oak Tree Press.<br />
Pringle, D.G and Walsh, J. (1999), ‘Poor People, Poor Places: Conclusion’<br />
in D.G. Pringle, J. Walsh and M. Hennessy (eds.), Poor People: Poor<br />
Places, Dublin: Oak Tree Press.<br />
Sampson, R.J., Morenoff, J.D. and Gannin-Rowley, T. (2002), ‘Assessing<br />
“Neighbourhood Effects”: Social Processes and New Directions in<br />
Research’, Annual Review of Sociology, no. 28, pp. 443–78.<br />
Tunstall, R. and Lupton, R. (2003), ‘Is Targeting Deprived Areas an<br />
Effective Means to Reach Poor People An Assessment of One<br />
Rationale for Area-Based Gunning Programmes’, CASE paper 70,<br />
Centre for Analysis of Exclusion, London School of Economics.<br />
Walsh, J. (1999), ‘The Role of Area-based Programmes in Tackling<br />
<strong>Poverty</strong>’, in D.G. Pringle, J. Walsh and M. Hennessy, (eds.), Poor<br />
People: Poor Places, Dublin: Oak Tree Press.<br />
Watson, D. and Williams, J. (2003), Irish National Survey of Housing<br />
Quality 2001–2002, Dublin: Economic and Social Research<br />
Institute/Department of the Environment, Local Government and<br />
Heritage, ESRI Books and Monographs Series No. 173.<br />
Wilson (1987), The Truly Disadvantaged: The Inner City, The Underclass<br />
and Public Policy, Chicago, IL: Chicago University Press.<br />
197
<strong>Mapping</strong> poverty is a longstanding concern, both for researchers and<br />
policy makers. This study updates and extends previous research on<br />
the spatial distribution of poverty using recent national data sources:<br />
the Census of Population (2002), the Living in Ireland Survey (2000)<br />
and, for the first time, the National Survey of Housing Quality (2001).<br />
The research uses three poverty indicators (household income,<br />
material deprivation and socio-demographic variables) to measure the<br />
distribution of poverty at various spatial and administrative levels,<br />
including county and city councils for the first time.<br />
The study addresses three key aspects of the spatial distribution<br />
of poverty:<br />
■ It identifies patterns with regard to the concentration of poverty<br />
and how these have evolved over time.<br />
■ It assesses if these patterns are significant in terms of the overall<br />
incidence of poverty.<br />
■ It considers the processes underlying poverty clustering,<br />
distinguishing between factors that influence the location and the<br />
causes of poverty.<br />
The findings of this study increase our understanding of the location<br />
of poverty and highlight numerous implications for policy on<br />
combating poverty and social exclusion, including area programmes,<br />
social housing and local anti-poverty strategies. The study will be of<br />
relevance to anti-poverty organisations, regional and local government<br />
and teachers/students of social policy and geography.<br />
€ 20.00