Barriers to social inclusion in Ireland: Change over time and space, 2016-2022

Devlin, Anne; McGuinness, Séamus; Whelan, Adele Research Report Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 Research Series, No. 212 Provided in Cooperation with: The Economic and Social Research Institute (ESRI), Dublin Suggested Citation: Devlin, Anne; McGuinness, Séamus; Whelan, Adele (2025) : Barriers to social inclusion in Ireland: Change over time and space, 2016-2022, Research Series, No. 212, The Economic and Social Research Institute (ESRI), Dublin, https://doi.org/10.26504/rs212 This Version is available at: https://hdl.handle.net/10419/322442 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 ANNE DEVLIN, SEAMUS MCGUINNESS AND ADELE WHELAN ESRI RESEARCH SERIES Number 212, May 2025 BARRIERS TO SOCIAL INCLUSION IN IRELAND: CHANGE OVER TIME AND SPACE, 2016-2022 Anne Devlin Seamus McGuinness Adele Whelan May 2025 RESEARCH SERIES NUMBER 212 Available to download from www.esri.ie https://doi.org/10.26504/rs212  2025 The Economic and Social Research Institute Whitaker Square, Sir John Rogerson’s Quay, Dublin 2 This Open Access work is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. ABOUT THE ESRI The Economic and Social Research Institute (ESRI) advances evidence-based policymaking that supports economic sustainability and social progress in Ireland. ESRI researchers apply the highest standards of academic excellence to challenges facing policymakers, focusing on ten areas of critical importance to 21st Century Ireland. The Institute was founded in 1960 by a group of senior civil servants led by Dr T.K. Whitaker, who identified the need for independent and in-depth research analysis. Since then, the Institute has remained committed to independent research and its work is free of any expressed ideology or political position. The Institute publishes all research reaching the appropriate academic standard, irrespective of its findings or who funds the research. The ESRI is a company limited by guarantee, answerable to its members and governed by a Council, comprising up to 14 representatives drawn from a crosssection of ESRI members from academia, civil services, state agencies, businesses and civil society. Funding for the ESRI comes from research programmes supported by government departments and agencies, public bodies, competitive research programmes, membership fees, and an annual grant-in-aid from the Department of Public Expenditure, NDP Delivery and Reform. Further information is available at www.esri.ie. THE AUTHORS Anne Devlin is a Research Officer at the Economic and Social Research Institute (ESRI) and an Adjunct Assistant Professor at Trinity College Dublin (TCD). Adele Whelan is a Senior Research Officer at the ESRI and Adjunct Associate Professor at TCD. Seamus McGuinness is a Research Professor at the ESRI and an Adjunct Professor at TCD. ACKNOWLEDGEMENTS The work carried out in this report was funded by Pobal as part of the Research Programme on Community Development and Social Inclusion. We would like to thank all the individuals within Pobal who provided assistance during the project, particularly Alana Ryan, Ela Hogan and Martin Quigley. Valuable contributions were made by members of the Research Programme Steering Committee: Delma Byrne, John Curtis, Paul Geraghty, Deirdre Kelly, Helen Russell, and John O’Toole. We are extremely grateful to the members of the Research Programme Steering Committee for their ongoing support and feedback on the research. Finally, we are sincerely grateful to Jonathan Pratschke and Trutz Haase (now deceased) who created the Pobal HP Deprivation Index, without which this research would not be possible. This report has been accepted for publication by the Institute, which does not itself take institutional policy positions. All ESRI Research Series reports are peer reviewed prior to publication. The authors are solely responsible for the content and the views expressed. Table of contents | iii TABLE OF CONTENTS EXECUTIVE SUMMARY ............................................................................................................................ V CHAPTER 1 INTRODUCTION .................................................................................................................. 1 CHAPTER 2 LITERATURE AND POLICY CONTEXT .................................................................................... 4 2.1 Policy context...................................................................................................................... 4 2.2 Social inclusion and COVID-19 in Ireland ............................................................................ 6 2.3 International evidence ........................................................................................................ 8 CHAPTER 3 DATA AND METHODS ........................................................................................................ 10 3.1 Data .................................................................................................................................. 10 3.2 Methodology .................................................................................................................... 16 CHAPTER 4 RESULTS ............................................................................................................................ 18 4.1 Barrers to social inclusion – economic, social and health ................................................ 18 4.2 Econometric modelling ..................................................................................................... 27 4.3 Robustness checks ............................................................................................................ 40 CHAPTER 5 CONCLUSIONS AND IMPLICATIONS FOR POLICY .............................................................. 43 REFERENCES .......................................................................................................................................... 45 APPENDIX .............................................................................................................................................. 50 iv | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 LIST OF TABLES Table 4.1 Prevalence of barriers to social inclusion: Ireland, 2016 and 2022 ............................. 19 Table 4.2 Prevalence of barriers to social inclusion by deprivation quintiles: economic barriers, 2016 and 2022 .............................................................................. 20 Table 4.3 Prevalence of barriers to social inclusion by deprivation quintiles: social barriers, 2016 and 2022 ..................................................................................... 21 Table 4.4 Prevalence of barriers to social inclusion by deprivation quintiles: health barriers, 2016 and 2022.................................................................................... 22 Table 4.5 Prevalence of barriers to social inclusion by level of urbanisation: economic barriers, 2016 and 2022 .............................................................................. 23 Table 4.6 Prevalence of barriers to social inclusion by level of urbanisation: social barriers, 2016 and 2022 ..................................................................................... 24 Table 4.7 Prevalence of barriers to social inclusion by level of urbanisation: health barriers, 2016 and 2022.................................................................................... 26 Table 4.8 Results of OLS models, SA level, 2016 and 2022: economic barriers .......................... 30 Table 4.9 Results of OLS models, SA level, 2016 and 2022: social barriers ................................. 31 Table 4.10 Results of OLS models, SA level, 2016 and 2022: health barriers ............................... 32 Table 4.11 Testing of statistical differences between 2016 and 2022: economic barriers ........... 35 Table 4.12 Testing of statistical differences between 2016 and 2022: social barriers ................. 36 Table 4.13 Testing of statistical differences between 2016 and 2022: health barriers ................ 37 Table 4.14 Results of OLS specification varying the reference case for deprivation, 2016 and 2022, SA Level. ............................................................................................. 39 Table 4.15 Results of propensity score matching models, SA level, 2022, all barriers ................. 41 Table 4.16 Difference-in-differences interaction and OLS coefficient change comparators ........ 42 LIST OF FIGURES Figure 3.1 Six-way urban-rural area classification ........................................................................ 14 Figure 3.2 Six-way urban-rural classification using Census 2016 ................................................. 15 Figure A.1 Composition of the Pobal HP Relative Deprivation Index ........................................... 50 Executive summary | v EXECUTIVE SUMMARY Social inclusion is an increasingly important concept in policymaking, emphasising a more holistic measure of well-being beyond income and poverty indicators. The European Union defines social inclusion as ensuring citizens have the opportunities and resources necessary to participate fully in economic, social, and cultural life. This concept is central to European and Irish policy. This study examines how potential barriers to social inclusion in Ireland have evolved over time and space, using Census 2016 and 2022 data. Despite Ireland’s strong economic performance, substantial inequalities remain, particularly for people with disabilities and lone parents. Geographic concentrations of social disadvantage are evident, with barriers to inclusion having been found to be place dependent. This study focuses on numerous potential barriers to social inclusion: Unemployment, economic inactivity, low educational attainment, lone parenthood, being a carer, ethnic minority status, disability status and poor health. The majority, but not all, of these barriers decrease between 2016 and 2022. We find declines in the prevalence of unemployment, lone parenthood, and low educational attainment at the area level; this results in a degree of convergence between more and less disadvantaged areas. However, we see increases in poor health and disability, perhaps not surprising given the COVID-19 pandemic. And we also see higher shares of ethnic minorities. Examining the prevalence of potential barriers to social inclusion at the spatial level also finds notable change between 2016 and 2022. Declines in unemployment were most pronounced in rural areas and ‘independent urban towns’. This is particularly interesting given previous findings by Whelan et al. (2024) that those towns which tend to be further from cities are more likely to lack in economic opportunities. While economic inactivity was stable between the two time periods, it did increase in ‘independent urban towns’. Low education levels decreased across all area types. The share of ethnic minorities increased in all areas, with the largest increases in cities and rural areas with high urban influence. It is worth noting that these characteristics and attributes do not necessarily mean an individual is socially excluded but they do increase the likelihood of not being able to fully participate in society. These groups are also not homogeneous but display considerable differences between them which will impact how they interact with society. Furthermore, while these attributes may increase the likelihood of being socially excluded in Ireland, these groups also bring benefits to the economy and society e.g. healthcare provision provided by carers, the increase in the working age population due to inwards migration. vi | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 The decline in economic barriers suggests positive impacts from macroeconomic changes and government policies, although this is in light of higher costs of living, and we cannot account for job quality. The increase in health barriers indicates a need for improved healthcare planning and resource allocation. The study underscores the importance of understanding the dynamics of potential barriers to social inclusion over time. While some barriers have decreased, others have increased, reflecting the complex interplay of economic, social, and health factors. While recent policy changes and macroeconomic conditions may have supported these improvements in some social inclusion barriers, their resilience in the face of future challenges is uncertain. Policymakers should continue to focus on creating a fully inclusive society. At present, concerns around the long-term health impacts of COVID-19 are important particularly considering an ageing population. While lower proportionally in relative terms compared to the other barriers examined, the proportion of carers is also important as this will have impacts for the labour market, social security receipt as well as individuals’ well-being. Further research on the quality of work and who has moved in to work and why would also be insightful. Literature and policy context | 7 higher infection rates than more affluent areas throughout the pandemic period examined. In an earlier paper, Whelan et al. (2023) also revealed differential economic outcomes between deprived and affluent communities because of the COVID-19 pandemic. By examining the uptake of the Pandemic Unemployment Payment (PUP),6 the authors found that more disadvantaged communities were more likely to experience disproportionately high rates of pandemic-induced unemployment relative to more well-off communities. Disadvantaged areas therefore experienced more volatility in employment and more deprived areas were more susceptible to changes in government restrictions; employment disruption in deprived areas was greater than was the case in more affluent areas when restrictions were imposed, and fell faster when restrictions were eased. It is possible that these disproportionate economic and health impacts on deprived communities during the pandemic may also have been observed with respect to barriers to inclusion. Some more recent evidence suggests that specific subgroups (that experience barriers to social inclusion) experienced disproportionate adverse effects of the pandemic, though the evidence base for this is limited. Using the Survey on Income and Living Conditions (SILC), Roantree et al. (2024) provide descriptive evidence that lone parents experienced a sharper decline in life satisfaction throughout the pandemic when compared to other groups. Furthermore, lone parents have consistently had the lowest levels of life satisfaction when compared to other household types since the early 2010s. Complementary research to this comes from Byrne and Sassi (2023), who shed further light on the experiences of lone parents, specifically those living in the private rental sector,7 throughout the pandemic. In a series of interviews, the authors find that lone parents were particularly concerned about long-term residential security, despite the introduction of an eviction ban in March 2020. This was due in large part to their wanting their children to grow up in a secure environment. Furthermore, some lone parents outlined that the substandard quality of rental housing was brought to the fore during restrictions on physical distancing, due to more time being spent at home. McHugh and Walsh (2024) shed light on the interplay between the pandemic, remote working policies, gender and carers’ mental health. The paper uncovers several findings. First, the authors reported a striking gap in mental health status between carers and non-carers, with carers being far more likely to report poor mental health. The authors also found a substantial gender gap in worsening mental health for both carers and non-carers. Broadly, women were more likely 6 The PUP is Ireland’s principal welfare payment afforded to those whose employment was disrupted due to the public health restrictions imposed throughout the COVID-19 pandemic. 7 It is worth noting that lone parent families are far more likely to reside in private rental accommodation than other families (Hearn and Murphy, 2017; Russell and Maître, 2024). 8 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 than men to report poorer mental health since the start of the pandemic (Silverio, 2025). However, this gender difference was three times as large among carers relative to non-carers. Vallières et al. (2022) come to a similar conclusion for Irish adults. Much of the existing Irish research examines the outcomes of disadvantaged groups during the pandemic. Comparatively less research has focused on comparing both the incidence and outcomes of these groups preand postpandemic. One exception to this is a recent report by Alamir et al. (2024), who examined differential labour market impacts of the pandemic on those with disabilities and lone parents using Labour Force Survey (LFS) data. Generally, the Irish labour market experienced a strong recovery period post-pandemic, with little to no differential impacts between specific subgroups of the population, albeit with the caveat that there are stark differences pre-pandemic (e.g. lower labour force participation rates amongst those with disabilities, lone parents). 2.3 INTERNATIONAL EVIDENCE International evidence documenting the outcomes of subgroups of the population that experience potential barriers to social inclusion during the pandemic is varied. Brown and Ciciurkaite (2023) provide evidence of differential mental health and employment outcomes between those with disabilities and those without in the United States during the pandemic. Specifically, the authors highlight the role of discrimination and precarious employment in predicting psychological distress among those with disabilities, finding an elevated likelihood of experiencing distress among those with disabilities relative to those currently without. Additional evidence on employment outcomes (Maroto et al., 2021), and mental health (Ciciurkaite et al., 2022) further support evidence of a divide in employment outcomes on the grounds of disability due to the pandemic. Much of the international research regarding lone parenthood during the COVID-19 pandemic focuses on the case of single mothers. Broadly, the pandemic led to the closure of formal childcare services and the introduction of physical distancing restrictions. In many cases, parents could not rely on formal childcare or informal childcare (i.e. via family or friends) due to public health restrictions. This left many parents with an elevated care burden. The elevated care burden disrupted many single mothers’ ability to participate in both paid work (Radey et al., 2022; Wakai et al., 2023; Salin et al., 2023) and education (Trotter, 2023). Salin et al. (2024) conduct a survey to evaluate adequacy of resources, employment and policy responses for single mothers in Finland during the pandemic. The authors find that the added pressures of lockdowns led to single mothers experiencing substantial time barriers to employment (i.e. due to remote schooling, the absence Literature and policy context | 9 of formal childcare and the loss of support networks due to physical distancing restrictions). While relatively few studies examine how the COVID-19 pandemic might have impacted the prevalence of carers, the literature on the emotional and psychological effects is more plentiful (see Sousa et al., 2022; Lightfoot et al., 2021; Liberati et al., 2021 for examples). In a systematic review of the literature, Bailey et al. (2022) show that those engaged in informal care experienced significant emotional and mental distress throughout lockdowns in a range of countries. Specifically, the authors reveal a pattern of carers citing concerns around the provision of healthcare, ambiguous government messaging concerning physical distancing policies, the decline of their social outlets and the acute physical vulnerability of those in their care. That those who face potential barriers to social inclusion experienced disproportionate impacts as a result of the COVID-19 pandemic reinforces the need to better understand the prevalence and the spatial distribution of barriers to inclusion across Ireland in recent years. 10 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 CHAPTER 3 Data and methods 3.1 DATA The data used herein are primarily from the Irish Censuses for 2016 and 2022. These allow us to examine the presence of barriers to social inclusion amongst the entire Irish population and allow for comparability over time. The Censuses for Ireland collect data on a wide range of personal and sociodemographic characteristics as well as labour market and education information, presence of disability and health status, as well as information at the household level such as household size, occupancy type and access to cars/internet etc. The Census information is collected using paper forms sent to every household in Ireland every five years. However, the 2021 Census was delayed due to the COVID-19 pandemic and became the 2022 Census. The data collected contain a Small Area (SA) level indicator of which there are 18,641 in Ireland in 2016 and 18,919 in 2022. This is our primary level of analysis. Small areas have been used since 2011 and are the lowest level of geography used for data purposes, typically containing between 50 and 200 households. Given they are created based on population size and distribution they are redrawn with each new Census, hence the increase in the number of SAs between 2016 and 2022. As is to be expected, this causes some inconsistencies examining data at this level over time. We include Census data as controls in our econometric modelling to account for differences between areas. We utilise information on the area-level sex composition, the age structure (those aged 18 and under and those 65 plus), the share of the population who are ethnic minority groups (for the purposes of this study we consider anyone not White Irish to be an ethnic minority), and the share of people in poor health. For those in poor health, it is those who report as having either ‘very bad’ or ‘bad’ health in the Census. 3.1.1 Barriers to social inclusion examined The Census datasets are used to collect all the barriers examined. The barriers examined are based on the literature on social inclusion (e.g. Whelan et al., 2024), Pobal’s 12 target groups for the Social Inclusion and Community Activation Programme8 and the data which are available. Based on these considerations and 8 12 pre-defined target groups have been set for SICAP. These have been selected based on the socioeconomic context, the level of need in society, and government priorities. 1) People living in disadvantaged communities; 2) People Data and methods | 11 for the purposes of this study we therefore examine area-level rates of unemployment, economic inactivity, lone parent families, ethnic minorities, low educational attainment,9 being a carer, being in poor health or having a disability. We note that these are potential barriers to social inclusion and that an individual that faces any of these potential barriers, or a combination of these barriers, is not necessarily socially excluded but is at risk of or more likely to be excluded than their peers. For the purpose of this study, ethnic minority status is measured as anyone who does not report as ‘White Irish’10 (the most common ethnicity reported in the Irish Census). This definitional approach means we capture a range of ethnicities; ‘other ethnicity’ in this instance includes members of the Irish Traveller community. While there is considerable heterogeneity within this group, this approach allows us to pick up those who are different in some way from the majority population although this may be for different reasons (e.g. due to their race, language ability, accent etc). It is worth noting however that while minority status can be a potential barrier to inclusion it is not necessarily the case for all migrants that they are socially excluded. In Ireland, the migrant population have higher educational attainment and a higher employment rate than the Irish-born population (McGinnity et al., 2025). The higher educational attainment is particularly notable given Ireland has high education levels (Smyth et al., 2022). There has been progress in recent years in terms of the integration of some groups who previously did not fare so well in Ireland, in particular migrants from Africa (McGinnity et al., 2025). Issues remain however with access to English language provision and, not surprisingly, access to housing. Concerns also remain given the increased salience of immigration and, although overall attitudes towards migrants remain positive, there has been a slight downward trend in recent years (Laurence et al., 2024). The migrant population in Ireland plays a significant role in supporting the labour force participation rate (and therefore economic growth), given the migrant population is on average younger than the Irish-born population (Department of Finance, 2024). It is worth noting that while disability is seen in the literature as a potential barrier to inclusion, health is not seen as an equivalent barrier. However, health is used here as an additional proxy measure, as the relevant question for disability in the Census changes slightly between the 2016 and 2022 Censuses. We use two health measures: firstly, those who report as being in very bad health and, secondly, those impacted by educational disadvantage; 3) People living in jobless households or households where the primary income source is low-paid and/or precarious; 4) People who are long-term unemployed; 5) People with a criminal history; 6) Refugees; 7) International Protection Applicants; 8) Disabled People/People with Disabilities; 9) Heads of One-parent Families; 10) Travellers; 11) Roma; 12) Island residents (Pobal, 2024b). 9 Defined as those with no formal education or at most a primary level of education. 10 The options for ethnicity in the Census are ‘White Irish’, ‘White Irish Traveller’, ‘Other White’, ‘Black or Black Irish’, ‘Asian or Asian Irish’, ‘Other’, ‘Not Stated’. 12 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 who report as being in very bad health as well as those in bad health. The former method allows us to collect those who are sickest, and it is likely that given they have the worst levels of health they may also be impacted in their day-to-day life as a result. However, the group that report being in very bad health is relatively small, so we also utilise a broader measure of poor health. With regards the disability measure, in 2016 and 2022 the Census asks: do you have any of the following long-lasting conditions or difficulties: blindness or a serious vision impairment, deafness of a serious hearing impairment, a difficulty with basic physical activities, an intellectual disability, a difficulty with learning, remembering or concentrating, a psychological or emotional condition, a difficulty with pain, breathing, or any other chronic illness or condition? While the question itself remains consistent across Census waves, the responses differ. In 2016, the possible responses are ‘Yes’ or ‘No’ compared to ‘Yes, to a great extent’, ‘Yes, to some extent’, or ‘No’ in 2022. For the 2022 data, we combine the two positive responses to compare to the ‘Yes’ responses in 2016. However, given people may answer these differently based on the choices available to them, we supplement with additional data on health to see if the patterns are consistent or whether changes may be reflective of differences in possible responses. Specifically, there is a strong risk that individuals reporting some disability in the 2022 Census will have answered ‘No’ in the 2016 data, pointing to a higher incidence in 2022 as a direct result of the change in the questions response categories. Given we use an area-level dataset, it is impossible to account for compounding disadvantage amongst those who may face numerous potential barriers to social inclusion which is a caveat of this study. 3.1.2 Pobal HP Relative Deprivation Index We are interested in whether the incidence of potential barriers to social inclusion have changed over time, with an emphasis on the extent to which deprived areas and affluent areas may have experienced any change differently. The Pobal Haase Pratschke (HP) Relative Deprivation Index is used to account for differing levels of area-level deprivation. The Pobal HP Relative Deprivation Index is a composite measure generated using the Irish Census data with a view to providing an up-todate analysis of the geographic distribution of deprivation across Ireland.11 The measure is based on three key factors: demographic profile, social class 11 For more info on the Pobal HP Relative Deprivation Index see: https://www.pobal.ie/app/uploads/2018/06/The-2016Pobal-HP-Deprivation-Index-Introduction-07.pdf. Data and methods | 13 composition, and the labour market situation of an area. Various Census variables from each of these dimensions are used in the production of the index. Figure A.1 in the appendix shows a basic model of the index. The 2016 Pobal HP Relative Deprivation Index of small areas ranges from -39.3 (most deprived) to 40.5 (most affluent). For 2022, given the Pobal HP Relative Deprivation Index is re-generated using the latest Census data, it ranges between -56.1 and 29.4. As the Pobal HP Relative Deprivation Index is a continuous variable, we operationalise it here using categories which have been used in previous research.12 Previous research found the relationship with deprivation not to be linear; therefore using a continuous variable is not deemed appropriate (Whelan et al., 2024). We categorise all small areas into one of four groups: the most deprived, marginally below average, marginally above average and the most affluent. 3.1.3 Six-way urban-rural classification Using small area-level indicators, we match to the CSO six-way urban-rural classification of areas. This six-way classification system is detailed in Figure 3.1 for 2016. As it is based on population it also changes between 2016 and 2022,13 although the methodology and definitions remain consistent. The development of the six-way classification of areas is based on best practice used elsewhere (New Zealand and Canada for example) and is driven by the fact that areas regardless of whether they are rural or urban will not be heterogenous (CSO, 2019). Distance to services, employment opportunities and amenities will have significant impacts on standards of living for those in rural areas and, by using employment in more urban areas as a proxy for this, the six-way classification allows some of these differences to be considered. Whelan et al. (2024) argue that the six-way classification should be used, when possible, for research as their findings differ considerably from what they found when using a simple urban-rural binary. 12 This categorisation has been utilised in numerous other relevant studies (e.g. Whelan et al., 2024; Devlin et al., 2024). 13 https://www.cso.ie/en/census/census2022/census2022smallareapopulationstatistics/. 14 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 FIGURE 3.1 SIX-WAY URBAN-RURAL AREA CLASSIFICATION Type Definition Urban areas Cities Town/settlements with populations greater than 60,000 - using Census 2016 definitions/breakdowns. Satellite urban towns Town/settlements with populations between 1,500 and 49,999, where 20 per cent or more of the usually resident employed population's workplace address is in 'Cities'. Independent urban towns Towns/settlements with populations between 1,500 and 49,999, where less than 20 per cent of the usually resident employed population's workplace address is in 'Cities'. Rural areas Rural Areas with high urban influence Rural areas (themselves defined as having an area type with a population less than 1,500 persons, as per Census 2016) are allocated to one of three sub-categories, based on their dependence on urban areas. Again, employment location is the defining variable. The allocation is based on a weighted percentage of resident employed adults of a rural Small Area who work in the three standard categories of urban area (for simplicity the methodology uses main, secondary and minor urban areas). The percentages working in each urban area were weighted through the use of multipliers. The multipliers allowed for the increasing urbanisation for different sized urban areas. For example, the percentage of rural people working in a main urban area had double the impact of the urban centre has on its surrounding areas. The adopted weight for: Main Urban areas is 2; Satellite urban communities is 1.5; Independent urban communities is 1. The weighted percentage is divided into tertials to assign one of the three rural breakdowns. Rural areas with moderate urban influence As above Highly rural/ remote areas As above Source: CSO, see https://www.cso.ie/en/releasesandpublications/ep/p-urli/urbanandrurallifeinireland2019/introduction/. Figure 3.2 is a map showing the six-way urban-rural classification across Ireland at the SA level. The cities of Dublin, Cork, Galway, Limerick and Waterford are clear and are surrounded by rural areas with high urban influence interspersed with satellite urban towns (or what might be known as commuter towns). Outside of these hinterlands are rural areas with moderate urban influence as well as highly rural/remote areas. Highly rural and remote areas are particularly common in counties Donegal, Leitrim, Sligo, Galway, and Kerry. Independent urban towns are then interspersed throughout these most rural areas and in particular are located in the border region and in a spine up the middle of the country. Data and methods | 15 FIGURE 3.2 SIX-WAY URBAN-RURAL CLASSIFICATION USING CENSUS 2016 Source – CSO Ireland. 3.1.4 Homelessness or housing exclusion We recognise that homelessness and housing distress or housing exclusion are an important barrier to full social participation. This is particularly important given the ongoing housing crisis in Ireland. However, homelessness and housing distress are not examined in detail in this work, as they are hard to measure particularly at a spatially disaggregated level. For examining homelessness there are obvious issues with using Census data. Census data are based on where Census respondents are staying on Census night e.g. those who are staying in private emergency accommodation, supported temporary accommodation or family hub accommodation. The methodology used by the Central Statistics Office (CSO) to identify homeless individuals has evolved over time and has been created in 16 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 conjunction with the Census Homeless Methodology Liaison Group.14 However, it should be noted that this methodology may be imperfect for detecting housing distress and is unlikely to pick up those who are homeless and not in some form of accommodation (e.g. rough sleepers). Furthermore, the data are not available at a spatially disaggregated level which would be necessary for the analysis which is conducted in this work. On that basis we do not examine homelessness or housing distress but recognise that this is a limitation of this work. 3.2 METHODOLOGY 3.2.1 Descriptive analysis We begin with an extensive descriptive exercise to examine the prevalence of potential barriers to social inclusion, at the national level by deprivation categories, and then using the six-way urban-rural classification in 2016 and 2022. We then look at how the averages change based on deprivation categories and based on area type using the six-way classification. 3.2.2 Econometric analysis We then go on to examine the determinants of barriers to social inclusion at the area level in 2016 and 2022. This follows a similar methodology to what is used in Whelan et al. (2024), albeit it is at the individual level. Given our outcome variables (the potential barriers to inclusion) are continuous variables as they are at the arealevel we begin by using OLS regressions. These models take the form: Barrieri =β0 + β1Deprivationi + β2Area Typei + β3Controlsi +ϵi (1) where the outcome variable is the proportion of the population in each SA who experience that potential barrier to inclusion. The main variables of interest then are deprivation in the four categories as discussed above and the six-way urbanrural classification. Other area-level factors are then controlled for in the model. These are: share of females in an area; age structure of an area (share of young people and share of older people); share of ethnic minorities in a small area; and proportion of those in poor health in an area. The controls change slightly depending on the outcome in question; for example when the barrier being examined is poor health, health is not included as a control variable. As there could be potential collinearity between deprivation and other controls which also influence the likelihood of barriers being experienced, this may lead to confoundedness and therefore biased estimates on the variables of interest. To 14 For more information on how homelessness is captured in the Census see https://www.cso.ie/en/releasesandpublications/ep/p-cpp6/censusofpopulation2022profile6homelessness/backgroundnotes/. Results | 23 education over the COVID-19 period are reflective of long-term policy changes in Ireland. However spatial inequalities do remain, with higher prevalence of low educational attainment found in the most rural/remote areas, independent urban towns, and rural areas with moderate urban influence. TABLE 4.5 PREVALENCE OF BARRIERS TO SOCIAL INCLUSION BY LEVEL OF URBANISATION: ECONOMIC BARRIERS, 2016 AND 2022 2016 2022 Change PP % difference Unemployment Cities 7.0 4.5 -2.5 -36% Satellite urban towns 6.6 4.1 -2.5 -38% Independent urban towns 10.6 5.9 -4.7 -44% Rural areas with high urban influence 4.9 2.9 -2.0 -41% Rural areas with moderate urban influence 6.0 3.3 -2.7 -45% Highly rural/remote 7.8 4.2 -3.6 -46% Economic inactivity Cities 37.1 36.9 -0.2 -1% Satellite urban towns 34.4 35.4 1.0 3% Independent urban towns 39.0 40.4 1.4 4% Rural areas with high urban influence 38.2 38.6 0.4 1% Rural areas with moderate urban influence 40.4 40.5 0.1 0% Rural/remote 44.1 44.5 0.4 1% Low education Cities 9.4 7.5 -1.9 -20% Satellite urban towns 7.4 5.9 -1.5 -20% Independent urban towns 12.2 9.9 -2.3 -19% Rural areas with high urban influence 9.8 7.7 -2.1 -21% Rural areas with moderate urban influence 13.2 10.1 -3.1 -23% Rural/remote 17.2 13.2 -4.0 -23% Source: Authors’ analysis based on 2016 and 2022 Census. Table 4.6 displays the proportions of individuals in areas who face possible barriers to inclusion which are more social in nature. The share of ethnic minorities increased in all levels of urbanisation, albeit with increases largest in more urban areas. This is an interesting finding as, while migrants often move to areas where economic opportunities are greater and which are also associated with greater public services and amenities and existing migrant networks and support services (Centre for Cities, 2015), this has not been the case in Ireland with migrant populations fairly equally distributed across the country (Fahey, 2019). Cities and more urban areas also tend to have more liberal viewpoints and be more accepting of migrants (Luca et al., 2024). 24 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 In terms of lone parent households, these fell slightly below the two census periods although the magnitude of the difference is small for all areas. The proportion of carers increased for all levels of urbanisation from 2016 to 2022. More rural areas see higher proportions of carers than urban areas, although this may be reflective of the age structure of rural areas relative to urban areas, or representative of lower levels of formal health and social care support in these areas which therefore necessitates higher levels of family care. Seven per cent of the population are carers in rural/remote areas compared to 5.3 per cent in cities and satellite urban towns. TABLE 4.6 PREVALENCE OF BARRIERS TO SOCIAL INCLUSION BY LEVEL OF URBANISATION: SOCIAL BARRIERS, 2016 AND 2022 2016 2022 Change PP % difference Ethnic minority status Cities 8.5 11.9 3.4 40% Satellite urban towns 6.2 8.4 2.2 35% Independent urban towns 6.5 9 2.5 38% Rural areas with high urban influence 1.5 2.1 0.6 40% Rural areas with moderate urban influence 1.1 1.5 0.4 36% Highly rural/remote 1.4 1.7 0.3 21% Lone parent households Cities 11.9 11.7 -0.2 -2% Satellite urban towns 12.2 12.2 0.0 0% Independent urban towns 13.8 13.8 0.0 0% Rural areas with high urban influence 8.9 8.6 -0.3 -3% Rural areas with moderate urban influence 9.8 9.4 -0.4 -4% Rural/remote 10.7 9.9 -0.8 -8% Carers Cities 3.8 5.3 1.5 39% Satellite urban towns 3.6 5.3 1.7 47% Independent urban towns 3.9 5.5 1.6 41% Rural areas with high urban influence 4.5 6.6 2.1 47% Rural areas with moderate urban influence 4.6 6.7 2.1 46% Highly rural/remote 5.0 7.0 2.0 40% Source: Authors’ analysis based on 2016 and 2022 Census. In Table 4.7, we present the proportions of individuals facing barriers which are health-related by level of urbanisation. The share of individuals with disabilities has increased substantially between 2016 and 2022. However, as discussed, this may be partly due to the changes in the possible answers in the 2022 Census. Between 2016 and 2022 the proportion who report as disabled increased by at least 8 percentage points across all levels of urbanisation, although proportionately this differed due to varying levels of disability in 2016. Satellite urban towns have seen Results | 25 the largest increases from 12.4 per cent in 2016 to 20.9 per cent in 2022, an increase of 69 per cent. The most rural areas saw the smallest increases proportionally from 14.8 per cent in 2016 to 22.9 per cent in 2022, equivalent to a rise of 55 per cent. The area types that started with the lowest levels of disability experienced the largest increases. To get a grasp of whether disability may have materially increased over the same period we also examine the self-reported health question from the Census. While health and disability are not proxies for one another, we would expect them to be strongly positively correlated. The prevalence of poor health albeit at much smaller magnitudes than disability increased between 2016 and 2022 in all areas. While the prevalence is low and the percentage changes are low, the same cannot be said for the proportional change. In cities, very bad self-reported health prevalence increased from 0.3 per cent to 0.4 per cent (both rounded), a change that equates to a rise of 18 per cent. Rural areas with high urban influence saw an increase of 27 per cent over the period examined, while Satellite urban towns experienced a much lower increase of 8 per cent. When we combine bad and very bad health, all levels of urbanisation see increases between 2016 and 2022. However the increases are not as big when compared to what we have seen for very bad health only. Bad/very bad health increased by between 7 per cent and 17 per cent across the area types. Rural areas with high urban influence saw the largest increases between 2016 and 2022 regardless of what self-report health measure was used. As was seen for disability, these areas which had the largest increases had the lower baseline rates in 2016. Disability and both measures of self-reported health all increased between 2016 and 2022. The proportions are highest in the ‘independent urban towns’ regardless of which measure was used. Interestingly, cities and rural/remote areas have the next highest levels of disability and ill-health. Whelan et al. (2024) strongly support the importance of a more specific spatial urban/rural classification rather than the commonly used urban/rural dichotomy. These non-linear findings around the prevalence of poor health at the six-way urban-rural classification further confirm this finding. That the prevalence of disability differs to a high degree, which is likely to be in part driven by the pandemic but also partly by the change in the possible responses, is a finding in and of itself. There is a substantial international literature concerning the measurement of disability and the importance of the nuances with specific questions and the possible answers. There is a consensus within this literature that increasing the specificity of questions is key to ensuring reliable responses (Baker et al., 2004). It may be that the option of being disabled ‘a little’ or ‘a lot’ increases the likelihood of individuals to report as disabled rather than a 26 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 disability dichotomy of either disabled or not. The former with more detailed responses is also more in line with the various disability measurement methods which exist in surveys across the globe (e.g. Washington Group measures of disability, Global Activity Limitation Instrument and others). Furthermore, when measuring disability there are always concerns around justification bias; that is when individuals are more inclined to report as disabled to justify other behaviour such as not participating in the labour market (Oguzoglu, 2012). However, in saying that, the increases in disability are large and alongside the increases in poor selfreported health it would suggest that the disability rate changes are in part due to the change in the possible responses in the Census, but also partly due to actual changes in disability prevalence within the Irish population between 2016 and 2022. TABLE 4.7 PREVALENCE OF BARRIERS TO SOCIAL INCLUSION BY LEVEL OF URBANISATION: HEALTH BARRIERS, 2016 AND 2022 2016 2022 Change PP % difference Persons with a disability Cities 14.1 22.2 8.1 57% Satellite urban towns 12.4 20.9 8.5 69% Independent urban towns 15.8 24.3 8.5 54% Rural areas with high urban influence 11.7 19.7 8.0 68% Rural areas with moderate urban influence 12.8 20.9 8.1 63% Rural/remote 14.8 22.9 8.1 55% Very bad health Cities 0.3 0.4 0.1 18% Satellite urban towns 0.3 0.3 0.0 8% Independent urban towns 0.4 0.5 0.1 22% Rural areas with high urban influence 0.2 0.3 0.1 27% Rural areas with moderate urban influence 0.2 0.3 0.1 25% Rural/remote 0.3 0.4 0.1 23% Bad/Very bad health Cities 1.8 2.1 0.3 17% Satellite urban towns 1.4 1.6 0.2 14% Independent urban towns 2.2 2.5 0.3 14% Rural areas with high urban influence 1.2 1.4 0.2 17% Rural areas with moderate urban influence 1.4 1.5 0.1 7% Rural/remote 1.8 2.0 0.2 11% Source: Authors’ analysis based on 2016 and 2022 Census. Results | 27 4.2 ECONOMETRIC MODELLING We go on to formally model the determinants of the potential barriers to social inclusion at the area level. This is fully discussed in the Methodology section, but for brevity we utilise a series of OLS models whereby the main variables of interest are deprivation category and urban-rural area type. These are then supplemented with propensity score matching models to account for any observable differences between the various groups. We control for age structure and sex profile in all models and then other controls dependent on the barrier being examined. For example, we control for ethnicity in most models but not those where the outcome variable is the share of residents in an area who report as ethnic minorities, as this would generate biased results due to multicollinearity. We control for the share of a small area who report as being in bad or worse health in all models except those for which health is the outcome variable. This approach is similar to what is undertaken in Whelan et al. (2024). Table 4.8 displays the results of OLS models for the economic barriers. The dependent variables are the prevalence rate of each barrier at the small area level. Not surprisingly, for all economic-related barriers, the prevalence is highest in areas which are more deprived, albeit the magnitudes differ considerably. For example, in 2016 in the most deprived SAs, the unemployment (economic inactivity) rate was approximately 24 (7) percentage points higher than was the case for the most affluent SAs. The difference in terms of unemployment rate between the most deprived and most affluent areas attenuated between 2016 and 2022, with the difference falling from 24 percentage points to 14 percentage points. The same convergence was not seen for the other economic barriers. We also see higher rates of unemployment and economic inactivity amongst certain levels of urbanisation, but the magnitudes of these coefficients are of a much smaller order than is the case for deprivation. Interestingly, while unemployment and inactivity are highest in independent urban towns, low levels of education are lowest in this group with higher levels being found in the most rural areas. Areas with higher shares of ethnic minorities have higher levels of unemployment, lower economic inactivity and higher educational attainment. Areas with higher shares of people in poor health have higher levels of all economic barriers, although there are differences in the changes over time. The relationship between poor health and unemployment fell between 2016 and 2022 (0.69 to 0.38), while it strengthened between poor health and economic inactivity (0.30 to 0.54). This may suggest movements from unemployment to inactivity for those with health conditions. We test the significant of these changes and discuss them further in a later section. 28 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 In terms of the social barriers (Table 4.9), consistent with the descriptives the shares of the population who are ethnic minorities (not White Irish) are highest in the most affluent SAs, although there are changes between 2016 and 2022 in terms of the relationship between deprivation and ethnic minority shares. The relationship between the most deprived SAs and ethnic minority shares falls slightly between 2016 and 2022 (-4.8 to -4.4); we test this for statistical significance later. Relative to the most rural areas, all urban areas had higher shares of ethnic minority persons, while rural areas with high urban influence and rural areas with moderate urban influence had lower shares. Areas with higher proportions of young people (under 18) and older people (more than 65 years), and areas with higher shares of females, had lower proportions of ethnic minorities. Lone parent households are more likely in more deprived areas relative to the most affluent SAs by a substantial amount. More precisely, lone parent rates were 13 percentage points higher in the most deprived SAs relative to the most affluent in 2016. This relationship was linear across the deprivation categories with the marginally below (above) average group of SAs having lone parent rates 7 (4) percentage points higher than the most affluent in 2016. There was little change between 2016 and 2022. Lone parent rates were also higher in more urban areas (as was the case in Whelan et al., 2024). Areas with higher shares of females and higher shares of young people also had higher rates of lone parent households. There is also a positive relationship between poor health and the prevalence of lone parenthood. The results for 2016 and 2022 are consistent for this barrier. The share of carers in 2016 was highest in the marginal groups (0.35 for marginally above average and 0.31 for marginally below average) relative to the most affluent SAs. The relationship between deprivation levels and the prevalence of caring roles in 2022 differed from 2016 but again statistical testing of these changes is key. Descriptively though the most deprived SAs in 2022 had the lowest levels of carers within their areas, while the marginally above average had the highest, albeit we should note the magnitudes are small. The share of carers in an area based on level of urbanisation is highest in the most rural areas relative to all other areas. Not surprisingly, areas with more older people and with more people in poor health have higher shares of carers. Areas with higher shares of ethnic minorities have lower shares of carers. Table 4.10 presents the results of OLS models for the health barriers examined. Poor health and disability are all more likely in more deprived areas although the magnitude of the relationship varies but so does the prevalence. Disability prevalence is 6 percentage points higher in both 2016 and 2022 in the most deprived SAs relative to the most affluent SAs. There is a linear relationship between deprivation and disability prevalence. The same can be said for selfreported health. Disability and poor health are also correlated with urbanness. Results | 29 Areas with higher shares of females and older people (65+) had higher rates of disability. Not surprisingly, poor health is also positively correlated with disability prevalence. Areas with higher shares of ethnic minority groups had lower prevalence of disability. In terms of poor health, the results are somewhat similar. Areas with more older people have higher rates of poor health, but in terms of sex there is no statistically significant relationship between share of females and share of individuals in poor health (regardless of which measure is used). Again, the relationships between the variables and the outcome of poor health/disability change between 2016 and 2022, and this is examined in greater detail in the next section. 30 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 TABLE 4.8 RESULTS OF OLS MODELS, SA LEVEL, 2016 AND 2022: ECONOMIC BARRIERS Unemployment Inactivity Low Education 2016 2022 2016 2022 2016 2022 Pobal HP Relative Deprivation Index Most Deprived 23.77 *** 13.63 *** 6.98 *** 6.93 *** 15.31 *** 11.98 *** Marginally below average 11.11 *** 6.54 *** 4.21 *** 3.89 *** 7.6 *** 5.71 *** Marginally above average 4.86 *** 3.02 *** 2.56 *** 2.46 *** 3.06 *** 2.12 *** Most affluent Reference Reference Reference Reference Reference Reference Area Type Cities 0.65 *** 0.01 1.09 *** 0.11 -2.05 *** -1.96 *** Satellite urban towns 0.25 -0.02 0.03 -0.37 ** -2.91 *** -2.57 *** Independent urban towns 2.49 *** 0.99 *** 0.55 *** 0.63 *** -2.99 *** -2.45 *** Rural areas with high urban influence -0.29 * -0.3 *** -0.01 -0.37 ** -1.42 *** -1.34 *** Rural areas with moderate urban influence -0.58 *** -0.53 *** -0.3 -0.49 *** -0.65 *** -0.59 *** Rural areas/Remote areas Reference Reference Reference Reference Reference Reference Area-level controls Share of… Females -0.04 *** 0.03 *** 0.13 *** 0.24 *** -0.15 *** -0.12 *** Young people (<18) -0.07 *** -0.06 *** 0 0.09 *** -0.03 *** -0.03 *** Older people (65 plus) -0.23 *** -0.13 *** 0.68 *** 0.67 *** 0.15 *** 0.09 *** Ethnic minorities 0.08 *** 0.07 *** -0.16 *** -0.09 *** -0.06 -0.05 *** Bad, very bad health 0.69 *** 0.38 *** 0.3 *** 0.54 *** 0.76 *** 0.67 *** Constant 8.18 *** 3.41 *** 20.66 *** 11.68 *** 12.58 *** 10.56 *** N 18,641 18,919 18,641 18,919 18,641 18,919 Pseudo R 2 0.71 0.65 0.63 0.66 0.7 0.68 Source: Authors’ analysis based on 2016 and 2022 Census. Note: *** p<0.01, ** p<0.05, * p<0.1. Results | 31 TABLE 4.9 RESULTS OF OLS MODELS, SA LEVEL, 2016 AND 2022: SOCIAL BARRIERS Ethnic Minorities Lone Parents Carers 2016 2022 2016 2022 2016 2022 Pobal HP Relative Deprivation Index Most Deprived -4.79 *** -4.42 *** 13.28 *** 12.43 *** 0.11 ** -0.4 *** Marginally below average -0.59 ** -0.41 7.15 *** 6.86 *** 0.31 *** 0.05 Marginally above average 0.13 -0.82 *** 3.84 *** 3.64 *** 0.35 *** 0.28 *** Most affluent Reference Reference Reference Reference Reference Reference Area Type Cities 5.07 *** 7.23 *** 3.79 *** 2.97 *** -0.43 *** -0.55 *** Satellite urban towns 4.08 *** 5.14 *** 2.4 *** 2.33 *** -0.61 *** -0.72 *** Independent urban towns 8.4 *** 9.6 *** 1.93 *** 1.72 *** -0.48 *** -0.44 *** Rural areas with high urban influence -7.24 *** -6.5 *** -0.11 0.02 -0.35 *** -0.37 *** Rural areas with moderate urban influence -5.47 *** -5.23 *** -0.2 0.05 -0.28 *** -0.32 *** Rural areas/Remote areas Reference Reference Reference Reference Reference Reference Area-level controls Share of… Females -0.31 *** -0.34 *** 0.34 *** 0.37 *** 0.00 0.00 Young people (<18) -0.37 *** -0.43 *** 0.2 *** 0.21 *** 0.00 0.01 *** Older people (65 plus) -0.92 *** -1.02 *** -0.2 *** -0.22 *** 0.03 *** 0.04 *** Ethnic minorities -0.03 *** -0.03 *** -0.03 *** -0.05 *** Bad, very bad health 0.84 *** 0.71 *** 0.42 *** 0.57 *** 0.1 *** 0.07 *** Constant 50.63 *** 58.08 *** -15.76 *** -16.48 *** 4.4 *** 6.08 *** N 18,641 18,919 18,641 18,919 18,641 18,919 Pseudo R 2 0.43 0.49 0.52 0.50 0.18 0.26 Source: Authors’ analysis based on 2016 and 2022 Census. Note: *** p<0.01, ** p<0.05, * p<0.1. 32 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 TABLE 4.10 RESULTS OF OLS MODELS, SA LEVEL, 2016 AND 2022: HEALTH BARRIERS People with disabilities Very bad health Bad or worse health 2016 2022 2016 2022 2016 2022 Pobal HP Relative Deprivation Index Most Deprived 6.37 *** 5.78 *** 0.39 *** 0.5 *** 2.18 *** 2.58 *** Marginally below average 3.55 *** 3.37 *** 0.2 *** 0.22 *** 1.11 *** 1.26 *** Marginally above average 1.77 *** 1.65 *** 0.1 *** 0.1 *** 0.56 *** 0.62 *** Most affluent Reference Reference Reference Reference Reference Reference Area Type Cities 2.31 *** 1.84 *** 0.16 *** 0.11 *** 0.65 *** 0.61 *** Satellite urban towns 2.27 *** 2.44 *** 0.14 *** 0.07 *** 0.53 *** 0.52 *** Independent urban towns 2.05 *** 2.34 *** 0.11 *** 0.09 *** 0.52 *** 0.59 *** Rural areas with high urban influence 1.18 *** 0.55 *** 0.1 *** 0.08 *** 0.36 *** 0.34 *** Rural areas with moderate urban influence 0.54 *** 0.33 *** 0.04 *** 0.03 ** 0.13 *** 0.13 *** Rural areas/Remote areas Reference Reference Reference Reference Reference Reference Area-level controls Share of… Females 0.07 *** 0.18 *** 0 0 0 0 Young people (<18) -0.06 *** -0.15 *** 0 *** 0 *** -0.02 *** -0.03 *** Older people (65 plus) 0.21 *** 0.17 *** 0.01 *** 0.01 *** 0.04 *** 0.03 *** Ethnic minorities -0.02 *** -0.05 *** 0 *** 0 *** 0.01 *** 0.01 *** Bad, very bad health 1.46 *** 1.29 *** Constant 2.54 *** 7.87 *** -0.12 ** -0.02 0.07 0.56 *** N 18,641 18,919 18,641 18,919 18,641 18,919 Pseudo R 2 0.66 0.59 0.09 0.13 0.3 0.35 Source: Authors’ analysis based on 2016 and 2022 Census. Note: *** p<0.01, ** p<0.05, * p<0.1. Results | 39 TABLE 4.14 RESULTS OF OLS SPECIFICATION VARYING THE REFERENCE CASE FOR DEPRIVATION, 2016 AND 2022, SA LEVEL. Source: Authors’ analysis based on 2016 and 2022 Census. Note: *** p<0.01, ** p<0.05, * p<0.1. 2016 2022 Difference 2016 2022 Difference 2016 2022 Difference Unemployment Inactivity Low Education Pobal HP Relative Deprivation Index Most deprived relative to Marginally below average 12.58 *** 7.06 *** -5.52 *** 2.89 *** 3.19 *** 0.3 7.23 *** 5.82 *** -1.41 *** Most deprived relative Marginally above average 18.91 *** 10.50 *** -8.41 *** 4.78 *** 4.67 *** -0.11 12.05 *** 9.69 *** -2.36 *** Most deprived relative to Most affluent 23.77 *** 13.63 *** -10.14 *** 6.98 *** 6.93 *** -0.05 15.31 *** 11.98 *** -3.33 *** Ethnic Minorities Lone Parents Carers Most deprived relative to Marginally below average -4.11 *** -3.79 *** 0.32 5.58 *** 5.27 *** -0.31 -0.14 *** -0.36 *** -0.22 *** Most deprived relative Marginally above average -5.48 *** -3.75 *** 1.73 *** 9.70 *** 8.9 *** -0.80 *** -0.21 *** -0.63 *** -0.42 *** Most deprived relative to Most affluent -4.79 *** -4.42 *** 0.37 13.28 *** 12.43 *** -0.85 ** 0.11 ** -0.4 *** -0.51 *** People with disabilities Very bad health Bad or worse health Most deprived relative to Marginally below average 2.74 *** 2.34 *** -0.40 ** 0.17 *** 0.26 *** 0.09 *** 0.99 *** 1.21 *** 0.22 *** Most deprived relative Marginally above average 4.59 *** 4.24 *** -0.35 * 0.30 *** 0.40 *** 0.10 *** 1.66 *** 1.98 *** 0.32 *** Most deprived relative to Most affluent 6.37 *** 5.78 *** -0.59 0.39 *** 0.50 *** 0.11 *** 2.18 *** 2.58 *** 0.40 *** 40 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 4.3 ROBUSTNESS CHECKS Given that OLS regressions can be biased by collinearity between variables, we also use propensity score matching (PSM) techniques on the 2022 Census data to solve for this and to act as a robustness check. In our OLS models above, the reference category is the most affluent group of SAs and the three other categories are compared to this. However, in our PSM models the data are not sufficient to match areas in the most deprived group with areas in the least deprived group, given that they are markedly different by nature. To deal with this we readjust our results to allow us to compare to a different reference case. We subtract the OLS coefficient for the marginally below average group from the most deprived coefficient in each regression, and then compare this to the average treatment effect on the treated (ATT) from our PSMs for the most deprived group relative to the marginally below average deprived group. This allows for sufficient matching between the treated and untreated as per best practice. As can be seen in Table 4.15 our findings are robust regardless of which econometric technique is undertaken. That results are robust suggests that our previous OLS estimates are robust to selection bias. The results of our OLS differencing finds that the most deprived SAs have unemployment rates 7.1 percentage points higher than the marginally below average group; the comparable ATT is 6.9. Both of these are statistically significant at the 1 per cent level. Given the similarity between the OLS estimates and the PSM estimates we take that the OLS results across the board are not substantially impacted by biases. Results | 41 TABLE 4.15 RESULTS OF PROPENSITY SCORE MATCHING MODELS, SA LEVEL, 2022, ALL BARRIERS Unemployment Inactivity Low Education Ethnic Minorities Lone Parents Carers People with disabilities Very bad health Bad or worse health OLS Difference 7.09 *** 3.04 *** 6.27 *** -4.01 *** 5.57 *** -0.45 *** 2.41 *** 0.28 *** 1.32 *** ATT 6.87 *** 3.19 *** 5.77 *** -3.70 *** 5.27 *** -0.46 *** 2.34 *** 0.27 *** 1.29 *** Source: Authors’ analysis based on 2016 and 2022 Census. Notes: OLS Difference is the difference in coefficients between the most deprived SAs and the marginally below average SAs. The ATT is the average treatment on the treated of the most deprived relative to the marginally below average SAs. Full results of the PSM models are displayed in the Appendix. *** p<0.01, ** p<0.05, * p<0.1. 42 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 In our analysis we use a modelling framework to measure differences in the correlation between area-level deprivation and area-level barriers to social inclusion over time. We are not attempting to measure causality within our approach and, indeed, we recognise that our area-level measures of deprivation are derived from an area-level deprivation index that includes some of our dependent variables as components. We are simply attempting to assess if, for example, area-level unemployment is more or less correlated with general area-level deprivation over the period 2016 to 2022. Although we do not set out to measure causal relationships, we do not believe that our model coefficients are likely to be substantially biased by endogeneity. In particular, if the influence of a given component of the deprivation index – for example lone parenthood or unemployment – is constant over time, then any endogenous influences arising from the inclusion of this variable can effectively be treated as an area-level fixed effect within a modelling framework. If we re-estimate our models using a difference-in-differences approach, which explicitly eradicates any influences of area-level fixed effects, and compare the results with those generated by our previous models, we can get a sense to which our original estimates have been affected by such factors. The results of this are presented in Table 4.16. The difference-in-differences interaction term is comparable with the change in the OLS coefficient between 2016 and 2022 and show that our OLS estimates and those generated under the difference-in-differences model are in line with one another, suggesting that our original estimates are robust to the influences of time invariant unobserved factors and are not being particularly affected by endogeneity. TABLE 4.16 DIFFERENCE-IN-DIFFERENCES INTERACTION AND OLS COEFFICIENT CHANGE COMPARATORS 2016 2022 Change DiD Interaction Unemployment 15.7 *** 8.7 *** -7.00 *** -7.97 *** Inactive 3.7 *** 3.75 *** 0.05 *** 0.23 Low education 9.9 *** 7.88 *** -2.02 *** -2.59 *** Ethnic Minorities -4.5 *** -3.89 *** 0.61 0.19 Lone Parents 7.09 *** 7.87 *** 0.78 *** -0.64 *** Carers -0.54 *** -0.18 *** 0.36 *** -0.32 *** Disability 3.73 *** 3.2 *** -0.53 *** -0.53 *** Very bad health 0.25 *** 0.36 *** 0.11 *** 0.10 *** Bad/Very bad health 1.43 *** 1.75 *** 0.32 *** 0.30 *** Source: Authors’ analysis based on 2016 and 2022 Census. Conclusions and implications for policy | 43 CHAPTER 5 Conclusions and implications for policy Barriers to social inclusion in Ireland have evolved between 2016 and 2022, with some improving while others have worsened. Our findings indicate an overall decline in unemployment, lone parenthood, and low educational attainment at the area level, suggesting a degree of convergence between more and less disadvantaged areas. However, this progress is counterbalanced by increases in poor health and disability which highlights emerging risks. Economic inactivity has remained unchanged. The prevalence of ethnic minorities has also increased and while minority status can be a potential barrier to social inclusion it is worth noting that there is considerable heterogeneity amongst ethnic minorities in Ireland. Recent research has found the employment rate and education levels to be higher amongst foreign-born individuals in Ireland than the Irish born population (McGinnity et al., 2025). However, there remain issues in terms of language provision and housing access for migrants and concerns around the increasing salience of migration to Ireland (McGinnity et al., 2025). These trends must be understood in the broader context of significant macroeconomic shifts, including the COVID-19 pandemic and wider policy developments, which have influenced both individual and community-level outcomes. The decline in unemployment reflects the tight labour market that emerged prior to and following the pandemic. However, while lower unemployment rates are positive, they do not necessarily translate into better living standards, particularly given rising costs of living and potential job quality concerns. Similarly, while educational attainment has improved overall this is part of a long-term trend in Ireland. The longer-term effects of pandemic-related disruptions however remain uncertain, especially for disadvantaged youth and those with special educational needs. A particularly concerning finding is the worsening health outcomes at the area level, particularly in already disadvantaged communities. Previous research suggests that vulnerable and minority groups (e.g. particular ethnic groups) experienced disproportionate health impacts during the pandemic (Devlin et al., 2024), which may have long-term consequences. This raises important policy considerations for healthcare planning and resource allocation, as persistent health inequalities could undermine social inclusion efforts. Future research in this area could potentially explore how health-related barriers interact with other social and economic factors over time. 44 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 We note that the share of lone parent households is relatively constant over time but there are stark differences dependent on area. The share of lone parent households is substantially larger in more deprived areas and also in more urban areas, although the deprivation impact is the most significant. Again, this points to a need for place-based consideration of policy to reduce potential barriers to inclusion; in this instance childcare, early years education and employability support for lone parents, most likely to be women. The stability of economic inactivity rates warrants further investigation. Headline figures may mask underlying dynamics, particularly among those unable to participate in the labour market due to illness or disability. Changes in how disability was recorded in the 2022 Census further complicate analysis in this area, suggesting that more detailed subgroup assessments are needed to fully understand the evolving nature of labour market exclusion. While recent policies and macroeconomic conditions may have supported improvements in some social inclusion barriers, their resilience in the face of future challenges is uncertain. Ireland’s status as an open economy means that global economic fluctuations, trade shocks, and domestic policy changes will play a crucial role in determining whether these gains are sustained. Continued monitoring of social inclusion barriers, especially in disadvantaged areas and for vulnerable individuals, will be essential to ensure that progress is not reversed and that policy interventions remain effective. While there is evidence of some convergence across areas, the findings again highlight significant area-based differences in terms of barriers to social inclusion in Ireland. The findings therefore support the continued use of place-based approaches for policy aimed at tackling social inclusion. This is in line with findings by Whelan et al. (2024). Future research may explore intersections and compounding effects of multiple barriers, thereby revealing how layered disadvantages uniquely shape individual outcomes. Evaluating qualitative aspects such as job quality and long-term educational impacts, alongside incorporating lived experiences, can offer a more nuanced understanding of social inclusion beyond the broad patterns identified in this study. Detailed comparative analyses with international trends and focused studies of specific sub-populations will help capture additional complexities that quantitative data alone cannot reveal. Given the ongoing economic and political volatility, maintaining social inclusion as a policy priority remains essential for fostering resilience and equity in Irish society. References | 45 REFERENCES Alamir, A., McGinnity, F. and Russell, H. (2024). The equality impact of the Covid-19 pandemic on the Irish labour market, Jointly published Reports 6, Dublin: ESRI and Irish Human Rights and Equality Commission (IHREC), https://doi.org/10.26504/jr6. Bailey, C., Guo, P., MacArtney, J., Finucane, A., Swan, S., Meade, R. and Wagstaff, E. (2022). ‘The experiences of informal carers during the COVID-19 pandemic: a qualitative systematic review’, International Journal of Environmental Research and Public Health, 19(20), p.13455. Baker, M., Stabile, M. and Deri, C. (2004). ‘What Do Self-Reported, Objective, Measures of Health Measure?’, The Journal of Human Resources, 39(4), p. 1067-1093. Brown, R.L. and Ciciurkaite, G. (2023). ‘Precarious employment during the COVID-19 pandemic, disability-related discrimination, and mental health’, Work and Occupations, 50(2), pp.167-187. Byrne, M. and Sassi, J. (2023). ‘Making and unmaking home in the COVID-19 pandemic: a qualitative research study of the experience of private rental tenants in Ireland’, International Journal of Housing Policy, 23(3), pp.523-542. Centre for Cities (2015). ‘Immigration is an urban issue’. Available online from: https://www.centreforcities.org/blog/immigration-is-an-urbanissue/#:~:text=Migrants%20matter%20to%20cities.,jobs%20in%20KIBS%20in%2 0cities. Children’s Rights Alliance (2022). ‘Annual Report’. Available online from: https://childrensrights.ie/wp-content/uploads/2023/10/Childrens-RightsAlliance-Annual-Report-2022.pdf. Children’s Rights Alliance (2024). ‘Child Poverty Monitor 2024’. Available online from: https://childrensrights.ie/wp-content/uploads/2024/05/Child-Poverty-Monitor2024-4-1.pdf. Ciciurkaite, G., Marquez-Velarde, G. and Brown, R.L. (2022). ‘Stressors associated with the COVID-19 pandemic, disability, and mental health: Considerations from the Intermountain West’, Stress and Health, 38(2), pp.304-317. Convention on the Rights of the Child (1989). United Nations Treaty Series, 1577, pp. 3178. Available at: https://www.ohchr.org/en/instrumentsmechanisms/instruments/conventionrights-child. CSO (2019). ‘Urban and Rural Life in Ireland 2019’. Available online from: https://www.cso.ie/en/releasesandpublications/ep/purli/urbanandrurallifeinireland2019/introduction/. Darmody, M., Smyth, E. and Russell, H. (2020). The Implications of the COVID-19 Pandemic for Policy in relation to Children & Young People: A Research Review, ESRI Survey and Statistical Report Series Number 94. Available online from: https://www.esri.ie/system/files/publications/SUSTAT94_3.pdf. 46 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 Darmody, M., Smyth, E. and Russell, H. (2021). ‘Impacts of the COVID-19 Control Measures on Widening Educational Inequalities’, Young, 29(4), 366-380. Department of Children, Equality, Disability, Integration and Youth (DCEDIY) (2024). ‘Ireland’s European Child Guarantee National Action Plan’. Available online from: gov.ie - Ireland’s European Child Guarantee National Action Plan. Department of Finance (2024). ‘Population Ageing and The Public Finances in Ireland – Changing Demographics’. Available online from: https://assets.gov.ie/297016/3485f1c0-5e2c-4851-b534-100bbdb8b920.pdf. Department of Social Protection (2020). ‘Roadmap for Social Inclusion 2020-2025: Ambitions, Goals Commitments’. [online] Available at: https://www.gov.ie/en/publication/ca8bf-roadmap-for-social-inclusion-20202025/. Devlin, A. Whelan, A. and McGuinness, S. (2024). COVID-19 infection rates and social disadvantage in Ireland: An area-level analysis. ESRI Research Series 187. ESRI and Pobal. Available at: https://www.esri.ie/system/files/publications/RS187.pdf. Devlin, A., McGuinness, S., Bergin, A. and Smyth, E. (2023). ‘Education Across the Island of Ireland: Examining Educational Outcomes, Earnings and Intergenerational Mobility’, Irish Studies in International Affairs 34(2), 30-47. Disability Federation of Ireland (2024). ‘Submission to Department of Social Protection on Budget 2024’. Available online from: https://www.disabilityfederation.ie/budget24/’ EU (2021). ‘EU strategy on the rights of the child and the European child guarantee’. Available at: https://eur-lex.europa.eu/legalcontent/EN/TXT/?uri=LEGISSUM:4540916. Fahey, E., Russell, H., McGinnity, F. and Grotti, R. (2019). Diverse Neighbourhoods: An Analysis of the Residential Distribution of Immigrants in Ireland, Dublin: ESRI and the Department of Justice and Equality, https://doi.org/10.26504/bkmnext376. Haase and Pratschke (2017). ‘The 2016 Pobal HP Deprivation Index for Small Areas’. Available online from: http://trutzhaase.eu/wp/wp-content/uploads/The-2016Pobal-HP-Deprivation-Index-Introduction-07.pdf. Hearne, R. and Murphy, M.P. (2017). Investing in the right to a home: Housing, HAPs and hubs. Maynooth University, Dublin. Laurence, J., McGinnity, F. and Murphy, K. (2024). Attitudes towards immigration and refugees in Ireland: Understanding recent trends and drivers, Jointly-published Reports 5, Dublin: ESRI and Department of Children, Equality, Disability, Integration and Youth, https://doi.org/10.26504/jr5. Liberati, E., Richards, N., Parker, J., Willars, J., Scott, D., Boydell, N., Pinfold, V., Martin, G., Dixon-Woods, M. and Jones, P. (2021). ‘Remote care for mental health: qualitative study with service users, carers and staff during the COVID-19 pandemic’, BMJ open, 11(4), p.e049210. Lightfoot, E., Yun, H., Moone, R., Otis, J., Suleiman, K., Turck, K. and Kutzler, C. (2021). ‘Changes to family caregiving of older adults and adults with disabilities during COVID-19’, Gerontology and Geriatric Medicine, 7, p.23337214211002404. References | 47 Luca, D., Terrero-Davila, J., Stein, J. and Lee, N. (2024). ‘Progressive Cities: Urban-rural polarisation of social values and economic development around the world’, Urban Studies, 60(12), p. 2329-2350. Mackey, K., Ayers, C.K., Kondo, K.K., Saha, S., Advani, S.M., Young, S., Spencer, H., Rusek, M., Anderson, J., Veazie, S. and Smith, M. (2021). ‘Racial and ethnic disparities in COVID-19–related infections, hospitalizations, and deaths: a systematic review’, Annals of internal medicine, 174(3), pp.362-373. Maroto, M.L., Pettinicchio, D. and Lukk, M. (2021). ‘Working Differently or Not at All: COVID-19’s Effects on Employment among People with Disabilities and Chronic Health Conditions’, Sociological Perspectives, 64(5), p. 876-897. McGinnity, F., Carron-Kee, E., Alamir, A. et al. (2025). Monitoring Report on Integration 2024. Jointly-published Reports 11, Dublin: ESRI and Department of Children, Equality, Disability, Integration and Youth, https://doi.org/10.26504/jr11. McGuinness, S., Whelan, A., Bergin, A. and Delaney, J. (2018). Profiling barriers to social inclusion in Ireland: the relative roles of individual characteristics and location, ESRI Research Series 71, Dublin: ESRI, https://doi.org/10.26504/RS71. McHugh, E. and Walsh, B. (2024). ‘Mental Health, Caring, and COVID-19: Examining the Gendered Effects of Working from Home on Carers’ Mental Health’, The Economic and Social Review, 55(3, Autumn), pp. 457-489. Moynihan, R., Sanders, S., Michaleff, Z.A. et al. (2021). ‘Impact of COVID-19 pandemic on utilisation of healthcare services: a systematic review’, BMJ Open 2021;11:e045343. doi: 10.1136/bmjopen-2020-045343. Nugent, C. (2017). ‘The Affordability of Childcare in Ireland: Measuring Regional Disparities’, NERI Research inBrief. Available online from: https://www.nerinstitute.net/sites/default/files/research/the_affordability_of_c hildcare_in_ireland.pdf. Oguzoglu, U. (2012). ‘Is there a better measure of self-assessed disability?’, Applied Economics Letters, 19(14), p. 1335-133. Pobal (2024a). ‘Disability and Deprivation: Investigating the Relationship between Health Inequalities and Geographic Disadvantage using the Pobal HP Deprivation Index’. Available online from: https://www.pobal.ie/app/uploads/2024/09/Disabilityand-Deprivation_Investigating-the-Relationship-between-Health-Inequalitiesand-Geographic-Disadvantage-using-the-Pobal-HP-Deprivation-Index.pdf. Pobal (2024b). ‘Social Inclusion and Community Activation Programme Requirements 2024-2028’. Available online from: SICAP PRs v1.2_highlighted. Radey, M., Lowe, S., Langenderfer-Magruder, L. and Posada, K. (2022). ‘“Showing Everybody’s True Colors”: Informal networks of low-income single mothers and their young children during the COVID-19 pandemic’, Children and Youth Services Review, 137, p.106479. Reitsma, M.B., Claypool, A.L., Vargo, J., Shete, P.B., McCorvie, R., Wheeler, W.H., Rocha, D.A., Myers, J.F., Murray, E.L., Bregman, B. and Dominguez, D.M. (2021). ‘Racial/Ethnic Disparities In COVID-19 Exposure Risk, Testing, And Cases At The Subcounty Level In California: Study examines racial/ethnic disparities in COVID19 risk, testing, and cases’, Health Affairs, 40(6), pp.870-878. 48 | Barriers to social inclusion in Ireland: Change over time and space, 2016-2022 Roantree, B. and Doorley, K. (2023). Poverty, income inequality and living standards in Ireland: Third annual report, Jointly published Reports 4, Dublin: ESRI and Community Foundation Ireland, https://doi.org/10.26504/jr4. Roantree, B., Maître, B. and Russell, H. (2024). Poverty, Income Inequality and Living Standards in Ireland: Fourth Annual Report. Jointly published Reports 7, Dublin: ESRI and Community Foundation Ireland, https://doi.org/10.26504/jr7. Rosenbaum, P.R. and Rubin, D.B. (1983). ‘The central role of the propensity score in observational studies for causal effects’, Biometrika, 70(1), p.41-55. Russell, H. and Maître, B. (2024). Lone parent transitions, employment transitions and poverty outcomes. ESRI Research Series 193, Dublin: ESRI, https://doi.org/10.26504/rs193. Salin, M., Hakovirta, M., Kaittila, A. and Raivio, J. (2023). ‘Single earners and carers during lockdown: everyday challenges faced by Finnish single mothers during the COVID19 pandemic’, International Journal of Sociology and Social Policy, 43(13/14), pp.146-160. Silverio, S.A. (2025). ‘A Hidden Crisis: Women’s mental health after the pandemic’, The Psychologist. Available at: https://www.bps.org.uk/psychologist/hidden-crisiswomens-mental-health-after-pandemic. Sironi, A., Bauloz, C. and Emmanuel, M. (Eds.) (2019). ‘Glossary on Migration. International Migration Law, No. 34’. International Organization for Migration (IOM), Geneva. IOM (2019) IOM Glossary on Migration. Available online from: https://publications.iom.int/books/international-migration-law-ndeg34glossary-migration. Smyth, E., Devlin, A., Bergin, A. and McGuinness, S. (2022). A North-South comparison of education and training systems: lessons for policy, ESRI Research Series Number 138. Available online from: https://www.esri.ie/publications/a-north-southcomparison-of-education-and-training-systems-lessons-for-policy. Sousa, H., Frontini, R., Ribeiro, O., Paúl, C., Costa, E., Amado, L., Miranda, V., Ribeiro, F. and Figueiredo, D. (2022). ‘Caring for patients with end-stage renal disease during COVID-19 lockdown: What (additional) challenges to family caregivers?’, Scandinavian Journal of Caring Sciences, 36(1), pp.215-224. Sprong, S. and Maître, B. (2023). ‘Thematic Report on Poverty and Social Inclusion Indicators’. Available online from: https://www.esri.ie/system/files/publications/BKMNEXT434_0.pdf. Stapleton, A., Polakowski, M. and Quinn, E. (2022). The Integration of Non-EU Migrant Women in Ireland. ESRI Research Series 148. Available online from: https://www.esri.ie/system/files/publications/RS148_5.pdf. SVP (2024). ‘Minimum Essential Standard of Living (MESL) Report 2024’. Available online from: https://publicpolicy.ie/external-reports/minimum-essential-standard-ofliving-mesl-report-2024/. Trotter, L. (2023). ‘It has been an uphill battle from the get go’: The experiences of single parents studying at university in the UK during the COVID-19 pandemic’, Journal of Further and Higher Education, 47(8), pp.1042-1053.