PROFILING BARRIERS TO SOCIAL INCLUSION IN IRELAND THE RELATIVE ROLES OF INDIVIDUAL CHARACTERISTICS AND LOCATION

The study uses data from participants in the Department of Rural and Community Development’s Social Inclusion and Community Activation Programme (SICAP), administered by Pobal, to examine who is most likely to experience at least one of five barriers to social inclusion. The barriers are (a) belonging to a jobless household, (b) being a lone parent, (c) having a disability, (d) being homeless or affected by housing exclusion and (e) belonging to an ethnic minority.


LIST OF TABLES
Ordinary least squares model results of individuals experiencing barrier(s) by lot, 2016 ............. 31 Table 12 Probit models of social barrier information with population density added to the model, 2016 .. 33  Table A1 Probit Executive Summary | i

EXECUTIVE SUMMARY
The Social Inclusion and Community Activation Programme (SICAP) represents a major component of Ireland's community development strategy, led by the Department of Rural and Community Development (DRCD). 1 Pobal managed the initial set-up and design phase of the social inclusion programme and has responsibility for a wide range of programme functions. The SICAP programme budget for 2016 was €35.8 million. SICAP aims to tackle poverty, social exclusion and inequality through local engagement and partnerships between disadvantaged individuals, community organisations and public sector agencies.
In order for SICAP to succeed in meeting its goal, it is important that the programme can successfully target the individuals who are most in need of assistance. The purpose of this study is to provide an in-depth analysis of the extent to which SICAP clients experience potential barriers to economic and social inclusion. We focus on five key potential barriers, which reflect some of the groups explicitly targeted under SICAP and have also been shown internationally to represent significant challenges to inclusion. They are: (a) belonging to a jobless household, (b) being a lone parent, (c) having a disability, (d) being homeless or affected by housing exclusion and (e) belonging to an ethnic minority.
The aim of this study is to inform policy by shedding light on a number of issues: 1. individual characteristics that are most common among individuals reporting potential barriers to social inclusion; 2. the extent to which the incidence of potential barriers varies according to spatial dimensions such as the area level of deprivation, population density or urbanicity; 3. potential implications of the findings for the future administration of the SICAP programme.

KEY FINDINGS AND ISSUES
We show that the incidence of potential barriers reported to SICAP Programme Implementers (PIs) will heavily reflect the demographic profile of the areas within which the PIs operate. In 2016, the potential barrier most commonly reported by SICAP beneficiaries was belonging to a jobless household, which was reported by over 40 per cent of SICAP clients. Almost 12 per cent of beneficiaries were lone parents, 7 per cent reported having a disability, 4 per cent were homeless or i i | P r o fi l in g Ba r ri er s to So c ia l In clu s io n i n I r ela n d affected by housing exclusion and 4 per cent belonged to an ethnic minority.
In order to help match the design of policy interventions to the specific needs of the target groups, multivariate models were estimated to capture the characteristics most commonly associated with each of the potential barriers to inclusion. Generally, gender has a relatively low influence on the risk of experiencing most potential barriers, with being female typically raising the risk by 2 to 3 percentage points. The clear exception to this is being a lone parent, as females are 25 percentage points more likely to report this potential barrier when compared to males. Age appears to be a particularly important risk factor for belonging to a jobless household, being a lone parent or reporting a disability.
In particular, SICAP beneficiaries aged under 25 are approximately 20 percentage points less likely to be members of jobless households or lone parents. Being in the 56+ age category raises the probability of reporting a disability by 10 percentage points relative to the 16-24 age group. Higher levels of educational attainment are associated with a lower probability of reporting all potential barriers. In particular, beneficiaries educated to above Leaving Certificate level (NFQ Level 5) are at least 10 percentage points less likely to belong to a jobless household, be a lone parent or have a disability. Nationality plays an important role for housing difficulties and disability, with European Union (EU) nationals less likely to report most potential barriers.
The Pobal Haase Pratschke (HP) Deprivation Index for Small Areas (SA) is a key component of SICAP's Resource Allocation Model (RAM), which is designed to determine the amount of funding allocated to particular geographical areas under SICAP. This Pobal HP Deprivation Index (SA) was found to raise the likelihood that individuals will report being a lone parent and/or belonging to a jobless household. This suggests that persons facing these potential barriers are more heavily concentrated within more deprived areas. Nevertheless, the Pobal HP Deprivation Index (SA) measure was not a predictor of disability, housing problems or ethnic minority in beneficiaries, suggesting that persons experiencing such potential barriers are more evenly distributed across areas with varying levels of deprivation.
With respect to whether an individual was domiciled in a rural or an urban location, after controlling for all other factors, beneficiaries living in urban locations were between 4 and 10 percentage points more likely to report each of the five potential barriers. The level of urban disadvantage was highest for lone parents and jobless households. The results suggest that urban environments influence an individual's risk of experiencing potential barriers in ways that are unrelated to the level of deprivation in the area where they live. The models suggest that individuals living in urban locations experience additional difficulties beyond the level of deprivation, or population density, in their immediate area.
Executive Summary | iii Our analysis of the data at PI level did not suggest that PIs in deprived areas tended to have a more disadvantaged clientele, as measured by the percentage of total recipients reporting one or more potential barriers. However, the average incidence of reported barriers among PIs in urban areas is 18 percentage points higher than the rural equivalent. This confirms the analysis from the individual level data that urban environments are associated with a greater prevalence of social risk factors that are not related to area levels of deprivation or population density. It is not evident that the higher concentration of disadvantaged clients among PIs located in urban locations is sufficiently captured within the current SICAP funding model.
There is evidence that PIs focus on targeting individuals from areas with higher levels of disadvantage. Nevertheless, it is observed that 40 per cent of participants qualify for assistance without reporting any of the five identified barriers and more than two-thirds of SICAP participants come from small areas that are not classified as disadvantaged (with a Pobal HP Deprivation Index (SA) of greater than -10). These findings support the view that existing funding arrangements reflect historical funding patterns rather than a strict implementation of SICAP's RAM, which is designed to skew funding heavily towards areas with high levels of social disadvantage. In the event of increased funding being assigned to the SICAP programme, the allocation of such additional

Introduction
This study uses data on individuals accessing assistance under Ireland's social inclusion and community development programme to measure the relative impact of individual and spatial factors on various barriers to social inclusion. The information used for the study comes from data captured for the 2016 Social Inclusion and Community Activation Programme (SICAP). SICAP is a large-scale national programme aimed at tackling poverty, social exclusion and long-term unemployment through local engagement and partnerships between disadvantaged individuals, community organisations and public sector bodies. Assistance under the programme is targeted towards individuals and local community groups meeting particular criteria.
This study seeks to address a number of specific research questions related to the following barriers to social inclusion:  belonging to a jobless household;  being a lone parent;  having a disability;  being homeless or affected by housing exclusion;  belonging to an ethnic minority (members of the Travelling Community, Roma, refugees, asylum seekers).
These five particular barriers were chosen for further analysis as they were present in the IRIS dataset and these issues have been shown internationally to represent significant challenges to economic and social inclusion (see Chapter 2). They also encompass the majority of the SICAP target groups (see Chapter 3).
With regard to these particular forms of social disadvantage, this study addresses the following questions.
1. To what extent do personal characteristics and location vary with the particular barrier to social inclusion being reported?
2. Do the characteristics of individuals reporting various forms of barriers to social inclusion vary according to the associated area deprivation level or does targeting make such spatial factors irrelevant? 4. Does the intensity of reported social barriers vary spatially?
From a policy perspective, the research provides evidence relating to the individual profiles most commonly observed among individuals reporting particular barriers to social inclusion and also examines the extent to which spatial factors, specifically the area-level deprivation and urban/rural controls, remain relevant within a highly targeted policy context. Furthermore, the study examines the extent to which personal characteristics and area-level deprivation impact the relative risk of individuals reporting each of these barriers. Given that these barriers reflect select target groups, the research will allow bodies such as government departments and Pobal to further tailor programmes and resources to the needs of individuals experiencing different forms of disadvantage. Given that SICAP expenditure is not exclusive to areas of high deprivation, we are also interested in the extent to which SICAP can effectively identify individuals experiencing barriers in more affluent areas. While it is recognised that the occurrence of barriers to social and economic inclusion may be lower in more affluent areas, effective implementation suggests that Programme Implementers (PIs) should still be accessing individuals facing these difficulties irrespective of area-level deprivation.
The report is structured as follows. Chapter 2 reviews the literature on the key factors associated with the specific forms of disadvantage being considered; Chapter 3 provides detail on the SICAP programme, its funding mechanism and targeting strategy; Chapter 4 discusses the data and methodology; Chapter 5 addresses the results from the specific research questions; and a summary and conclusions are provided in Chapter 6.
In 2010 the European Commission, as part of the Europe 2020 strategy for smart, sustainable and inclusive growth, set a target to reduce the number of people subject to poverty and social exclusion by 20 million. In addition, it was proposed to raise the employment rate of those aged 20-64 to 75 per cent. In any society, ensuring that all individuals have the opportunity and support to prosper, both economically and socially, is a priority for government. However, in order for the Irish government to succeed in meeting this challenge it is necessary to understand exactly what the main barriers to economic and social inclusion are within Ireland. This report discusses the five key barriers outlined above: jobless households, lone parents, homeless or those affected by housing exclusion, disability and ethnicity. While we do not assert that these are the only, or even the primary, barriers to inclusion within an Irish context, they do reflect some of the principal constraints that underpin Ireland's most prominent strategy against social exclusion, i.e. SICAP.

JOBLESS HOUSEHOLDS
At least half of those living in jobless households in the majority of EU countries are either income poor or deprived 2 (de Graaf-Zijl and Nolan, 2011). Ireland has one of the highest rates of jobless households in the EU. The proportion of persons aged 0-59 living in jobless households in the second quarter of 2017 was 11 per cent (Central Statistics Office (CSO), 2017). 3 In 2010, the percentage of households that were jobless stood at 22 per cent, which was double the EU-15 average of 11 per cent (Watson et al., 2012). Those living in jobless households were most likely to be lone parents, to have few or no education qualifications, to have a disability and to live in households where no member has ever worked. The combination of these factors makes those living in jobless households particularly susceptible to poverty and social exclusion. There is also some regional disparity, with almost 23 per cent of jobless households located in the Border region, but only 14 per cent in the Mid-East region (O'Rorke, 2016). Nickell (2004) argues that a decrease in demand for low-skilled labour explains the rise in workless households within the UK; however, few causal explanations have been forwarded to explain the high occurrence in Ireland.
In 2015, the European Commission issued a Country Specific Recommendation to Ireland which referred to the household joblessness and low work intensity issue. 4 Almost 1 in 5 (19 per cent) of people in Ireland in 2015 belonged to Very Low Work Intensity (VLWI) households, which measures people aged 0-59 years only living in households where adults worked less than 20 per cent of the total work potential over one year. The European Commission suggested that in order to combat the high degree of jobless households in Ireland the government should slowly reduce benefit payments, so that individuals are not discouraged from working due to a loss of benefits once they enter employment. The government recently launched Pathways to Work 2016-2020, which aims to increase employment within jobless households by making work more rewarding with the help of the Housing Assistance Payment and the Single Affordable Childcare Scheme. However, there are two worrying trends with jobless households: firstly, those in jobless households are just over half as likely to enter employment as an individual from a working household (Watson et al., 2016); secondly, during the recovery after the recent Great Recession, the proportion of jobless households obtaining a job did not keep pace with the general rise in employment, suggesting that getting those from jobless households into employment may require more than just creating jobs. The low transition into employment and the fact that other members in the household are less likely to have ever worked suggest that joblessness may become culturally embedded within the household, which can lead to a high degree of intergenerational transmission of jobless households. Headey and Verick (2006), using Australian data, found that people who lived in jobless households at age 14 were more likely to be welfare dependent and in poverty later in life.

LONE PARENTS
Lone parents consistently rank as the group most vulnerable to poverty and social exclusion in Ireland (Watson et al., 2016). In 2014, 58 per cent of lone parents in Ireland were at risk of poverty or social exclusion, compared to only 35 per cent in Sweden (European Institute for Gender Equality (EIGE), 2016). Their high poverty rates may be partly driven by low employment rates. In 2016, only 56.4 per cent of lone parents were in employment compared to 74.4 per cent of twoparent households (CSO, 2017). It is not clear what the mechanisms driving low employment levels are, but it could be that lone parents prefer to stay at home and look after their children, or alternatively, it may be the case that they just cannot find a suitable job or affordable childcare. It may also be due to the fact that 43 per cent of lone parents in Ireland have at most lower secondary education, which is higher than the EU average with the corresponding proportion, for example, in Slovakia at just 7 per cent (Chzhen and Bradshaw, 2012). Low education levels mean that the jobs available to lone parents often do not pay enough to justify forgoing welfare payments. Consistent with this hypothesis, Zagel (2014), using German and British data, finds that lone parents with at least tertiary education are more likely to be employed than their lower educated counterparts. This poverty trap is made substantially worse in Ireland by the high level of childcare costs. The cost of childcare in Ireland is one of the highest in Europe, representing about 45 per cent of the average wage (EIGE, 2016). High childcare costs have been noted by the European Commission, in a country-specific recommendation to Ireland, as a substantial barrier to female labour market participation. In addition to barriers to employment, lone parents are more likely to suffer from mental stress compared to those with a partner (Watson et al., 2016).

HOMELESS OR AFFECTED BY HOUSING EXCLUSION
The level of homelessness in Ireland has increased substantially over the past year. In January 2017 there were 1,172 homeless families in Ireland, with that figure rising to 1,530 by November 2017. Of the homeless families in November, 946 (62 per cent) were single-parent families (Department of Housing, Planning and Local Government (DHPLG), 2017a). The rise in homelessness is thought to be driven both by a sharp increase in rents after the recession and by the low supply of housing in Ireland. Not having a home constitutes one of the most severe consequences of poverty and leads to a vicious cycle of economic and social exclusion. Those who do not have a permanent address will struggle to find employment, as employers are reluctant to hire individuals with no fixed address. In addition, there can be difficulties associated with basic banking and the ability to obtain loans and, therefore, individuals may be forced into accessing black market finance and spiralling debt. Homeless individuals will find it difficult to feel part of their community and are also likely to experience many forms of prejudice (Phelan et al., 1997). Some other individuals in society face housing difficulties and live in constant worry of losing their home; for example, Focus  found that almost one-third of individuals worry about and/or struggle to pay their rent every month. This is not without due cause, as 1694 homes were repossessed in 2016 -the highest number to date (Central Bank, 2017). In July 2016, the government launched the Rebuilding Ireland initiative, an action plan for housing and homelessness with the target to ensure that everyone has access to a home (DHPLG, 2017b).

DISABILITY
Across the EU-28 almost 40 per cent of individuals with a disability face risk of poverty or social exclusion (EIGE, 2016), while for Ireland this figure is even higher at almost 50 per cent (Watson et al., 2016). Disabled individuals may be particularly susceptible to social exclusion if they cannot participate in local events due to inadequate facilities or the fact that they are reluctant to join a group for fear of being discriminated against. Gannon and Nolan (2006) find that disabled individuals are significantly less likely to engage in social activities as measured by participation in a club, evenings out, or meeting individuals on a regular basis. Individuals with a disability are only about half as likely to be in employment as those who do not have a disability (Watson and Nolan, 2011). This may be partly driven by low education levels. For example, 43 per cent of individuals with a disability had not progressed beyond primary education compared with 19 per cent of all adults (CSO, 2007a). Over one-third of individuals with disabilities who were not employed indicated that they would like to work if the conditions were right (CSO, 2012). Many individuals may not actively search for employment, for a variety of reasons including fear of losing welfare payments, lack of knowledge of suitable jobs and/or employer discrimination (CSO, 2007b). Disabled individuals also face direct monetary costs in terms of paying for disability-related goods and services and thus require extra income in order to have the same standard of living as an equivalent nondisabled person. Cullinan and Lyons (2015) find that the extra economic cost in Ireland is large (35.4 per cent of income or €207 per week) and varies according to the severity of the disability; they note that measures of poverty which use income as a proxy will tend to underestimate the level of poverty for a disabled person.
In 2015, in order to provide a more inclusive society for individuals with a disability, the government launched the Comprehensive Employment Strategy for People with Disabilities 2015-2024 (Department of Justice and Equality, 2015). The initiative is designed to support disabled individuals who are sufficiently able and would like to work. The strategy is focused on making work pay, 5 engaging employers, building skills, providing coordinated support and promoting job retention and re-entry. There is also a target to increase the number of individuals employed in the public sector who have a disability from 3 per cent to 6 per cent.

ETHNIC MINORITIES
Those with a different ethnicity or cultural background such as Travellers, Roma, refugees or asylum seekers may face particular barriers to social and economic integration. This is due to a number of factors including low education levels, a nomadic lifestyle and prejudice or discrimination. The levels of education of these groups is particularly low. For example, only 1 per cent of Travellers have a college degree compared to 30 per cent of non-Travellers (Watson et al., 2017). In 2014, Eurostat found that 40 per cent of adults born in a country outside the EU-28 were at risk of poverty or social exclusion; another report found that nine out of ten Roma men and women were living in poverty (EIGE, 2016). On average across EU countries, just 21 per cent of Roma women and 35 per cent of Roma men were in employment in 2014, while 42 per cent did not have adequate housing (FRA, 2014).
In June 2017, the Irish government launched the National Traveller and Roma Inclusion Strategy 2017-2021 (Department of Justice and Equality, 2017), which aims to improve the lives of the Traveller and Roma communities in Ireland. These groups are also vulnerable to discrimination when applying to rent private housing. For example, EUMC (2005) found that in several countries within the EU, including Ireland, minorities are sometimes explicitly asked not to apply. These factors are likely to increase feelings of isolation and exclusion among the affected groups and also force them into sub-standard accommodation.
Asylum seekers and refugees are particularly vulnerable to poverty and social exclusion due to the traumatic and psychological distress endured and poor health. Almost half of refugees and asylum seekers coming to the EU are between the ages of 18 and 34, while 25 per cent are children (Benifei, 2016). There is a clear need to provide training and education in order to improve labour market integration levels among these groups. Barrett and Duffy (2008) found evidence of a lack of integration of immigrants into the Irish labour market, which they suggest may be due to language skills or the non-transferability of qualifications. It is likely that Roma people, refugees and asylum seekers will also face these barriers to economic integration. In addition, non-documented or illegal ethnic minorities are particularly vulnerable to labour exploitation. Their precarious resident status may lead them to accept low pay, long hours and poor working conditions (Arnold et al., 2017). Some small-scale interviews have been carried out in Ireland with refugees; a report by Galway City Council on 23 one-on-one interviews found that 'asylum seekers and refugees did not feel integrated in this country, either because there were few opportunities to meet Irish people or because of language barriers. Some felt discriminated against' (Stewart, 2006). 8 | Pr of i l in g B a r r i er s t o So c ia l In clu s io n i n I r ela n d CHAPTER 3 The SICAP programme SICAP aims to tackle poverty, social exclusion and inequality through local engagement and partnerships between disadvantaged individuals, community organisations and public sector agencies. The programme is funded and overseen by the Department of Rural and Community Development (DRCD) 6 and Pobal. It is co-funded by the European Social Fund (ESF), including a special allocation under the Youth Employment Initiative (YEI), with a total programme budget for 2016 of €35.8 million. It was rolled out on 1 April 2015 and ran until 31 December 2017.
The goals of SICAP are as follows (Pobal, 2016): 1 Strengthening local communities: to support and resource disadvantaged communities and marginalised target groups to engage with relevant local and national stakeholders in identifying and addressing social exclusion and equality issues (Social inclusion and capacity building); SICAP is the successor programme to the Local and Community Development Programme (LCDP). It is a national programme that is led in each county by a Local Community Development Committee (LCDC). Following a public procurement process, contracts for the implementation of the new programme have been awarded by LCDCs to 45 Programme Implementers (PIs) covering 51 geographic areas (Lots).
The distribution of funds is partially determined by the Resource Allocation Model (RAM), which was developed for Pobal based on the Pobal HP Deprivation Index (SA). This deprivation index provides a method of measuring the relative affluence or disadvantage of a particular geographical area using data compiled from various censuses. It is used as a key resource to enable a targeted approach towards tackling disadvantage, by providing local analysis of the most disadvantaged areas nationwide. Percentage data for each area are provided under a range of categories such as unemployment, educational attainment and population change.
There is no detail regarding the extent to which the RAM has been implemented under the current SICAP programme. However, a description of how the predecessor LCDP was available is given by Haase and Pratschke (2013). The RAM distributes resources across 51 Lot areas. The RAM is a spatial tool that is designed to allocate resources to Local Development Companies (LDCs) based on three key criteria. The three factors at the heart of the RAM are ( It is clear that if SICAP funding was to be allocated solely on the basis of the RAM, this would generate higher rates of total expenditure per head of population in some deprived areas. Information was available on total expenditure by 51 Lot areas and previous analysis confirmed a positive relationship between the level of deprivation and the Lot expenditure per head of population (McGuinness et al., 2016). We find that approximately 30 per cent of SICAP participants live in small areas that are classified as disadvantaged, whereas approximately 14 per cent of the population nationwide live in small areas that are classified as disadvantaged. While this suggests that SICAP funds are disproportionately targeted towards areas of social disadvantage, we would expect the share allocated to disadvantaged areas to be somewhat greater than 30 per cent if distributed purely on the basis of the RAM (Model 3). Our understanding is that the distribution of funding is based largely on historical patterns and that the RAM application to date has been primarily to ensure that funding reductions were distributed away from areas of social disadvantage.
In terms of the programme administration, PIs' spending should be split evenly, with a 5 per cent leeway (28-38 per cent), across the three goals: (i) social inclusion and capacity, (ii) lifelong learning and (iii) employment. For individuals 10 | Pr o fi l i n g Ba r ri e r s t o So ci a l In c l u si o n i n Ir e l a nd to be eligible for programme assistance under any of these goals, they should qualify under one or more of the 11 target groups:  children and families from disadvantaged areas;  lone parents;  new communities (including refugees and asylum seekers);  people living in disadvantaged communities;  people with disabilities;  Roma;  the unemployed (including those not on the Live Register);  low income workers/households;  Travellers;  young unemployed people from disadvantaged areas;  young people aged 15-24 who are not in employment, education or training (NEETs).

Data and methodology | 11
CHAPTER 4 Data and methodology The Integrated Reporting and Information System (IRIS) is an administrative data capture system that is used by SICAP PIs when registering individuals for SICAP supports. IRIS contains information about individuals, such as age, gender, education and economic status of all SICAP participants; in addition, data are recorded on the activities and training provided to individuals under the programme. Table 1 shows the individual-level characteristics such as gender, age and education most common among SICAP participants and the degree to which they vary across the lifelong learning and employment goals. Of all SICAP participants in 2016, 55 per cent were male and 45 per cent were female. Males were more likely to be assisted with employment, while females were more likely to be assisted with lifelong learning opportunities. SICAP is primarily focused on people who experience disadvantage and social exclusion and who are of working age (15-65 years). 7 A majority (53 per cent) of all participants are between the ages of 25 and 45 years. One-fifth of participants are aged between 46 and 55, while only 11 per cent are over the age of 55. Individuals between the ages of 25 and 55 are most likely to be assisted with employment, while the very young (aged 15-24) and those approaching retirement (aged 56-65) are more likely to receive lifelong learning supports. In terms of education, more than two-thirds of SICAP participants have at most secondary education, i.e. have obtained a leaving certificate or less (NFQ ≤5). More specifically, 32 per cent have below an upper-secondary education (NFQ <4) and 37 per cent an upper secondary-level education (NFQ 4 & 5). The share of SICAP participants with third-level education (NFQ 6-8) is 27 per cent, and only 4 per cent have postgraduate education (NFQ 9 & 10).
In terms of how education characteristics vary across goals, not surprisingly, those with at least a third-level qualification (NFQ >5) are more likely to receive employment supports, as opposed to lifelong learning. Those with less than secondary education are more likely to be assisted with lifelong learning, while there is not much difference across programme goals for those with secondary education. Overall, there are more individuals supported with employment than with lifelong learning opportunities.
A number of target groups, along with individuals who are living in disadvantaged areas, have been prioritised (Pobal, 2016). Specific target groups are:  children and families in disadvantaged areas;  lone parents;  NEETs (young people aged 15-24 years who are not in employment, education or training);  new communities (including refugees/asylum seekers);  people living in disadvantaged communities;  people with disabilities;  Roma;  the unemployed (including those not on the Live Register);  Travellers;  low income workers/households;  young unemployed people living in disadvantaged areas.
With regard to the IRIS data, while these 11 target groups are specified under the programme, we have adopted in this analysis a focus on five tangible measures of barriers to social and economic inclusion: (i) belonging to a jobless household, (ii) being a lone parent, (iii) having a disability, (iv) being homeless or affected by housing exclusion and (v) ethnic minority. This reflects both the objectives of SICAP and the broader international policy literature discussed in Chapter 2.
In terms of the data, it is important to recognise that a proportion of individuals will qualify for assistance under SICAP without reporting any of these specific barriers. For instance, individuals living in disadvantaged communities (as measured by the Pobal HP Deprivation Index (SA)), young people who are NEETs, those who are unemployed, members of low-income households and new communities will all qualify for assistance under the programme. In the data for 2016, around 40 per cent of SICAP participants qualify for assistance without reporting any of our identified five barriers and this group will constitute our reference category. This reference category will consist of young people and parents living in disadvantaged communities, members of low income households, the unemployed and members of new communities not reporting any of the five barriers. 8 Using this reference category enables us to assess the extent to which various risk factors and characteristics impact the relative risk of experiencing each of the five barriers in a way that allows for the highest level of comparability across each risk category. Finally, it is important to note that while some of the SICAP target groups are not reflected explicitly in the barriers, they are reflected in our models as control variables; for example, spatial disadvantage measures, age, NEETs, unemployment durations and migrant status.
The objective of the study is to examine the extent to which personal characteristics and area-level deprivation impact the relative risk of individuals reporting each of these barriers. Given that these barriers reflect select target groups, the research will allow bodies such as government departments and Pobal to further tailor programmes and resources to the needs of individuals experiencing different forms of disadvantage. For example, if we find that jobless households are predominantly experienced by young people with low levels of schooling in rural areas, this would require a different policy approach to targeting those with a disability if we find this issue is predominantly experienced among older people with higher levels of schooling in urban areas. 9 Furthermore, given that SICAP expenditure is not exclusive to areas of high deprivation, we are also interested in the extent to which SICAP can effectively identify individuals experiencing barriers in more affluent areas. While it is recognised that the occurrence may be lower in more affluent areas, effective implementation suggests that PIs should still be accessing individuals facing these difficulties irrespective of area-level deprivation. If we find that the personal characteristics of individuals reporting particular barriers are broadly similar across areas of high Nevertheless, despite the existence of targeting, we would expect that the profile of the client base will vary from area to area based on the particular demographic and social factors that are at play in a particular Lot. Given that employment is one of the three SICAP goals and that funding must be relatively evenly spread across all goals, we would expect PIs to have a broadly similar share of unemployed claimants as part of their client base. However, for other barriers not specifically related to the goals, such as the existence of lone parents, people with disabilities, homeless individuals, and persons from minority groups, we might expect to see more variation across Lots. This suggests that some PIs are likely to have more disadvantaged clients in their populations, which would suggest that the average intensity of interventions will also be significantly higher than for others. It is difficult to test this hypothesis using the existing data; however, we were able to cross-tabulate the proportion of SICAP clients at a Lot level who were lone parents or who had disabilities with the population-level incidences from the 2016 census (see Figures 1 and 2).  We can see that the share of clients who are lone parents in a particular Lot will strongly reflect the share of lone parents in the Lot population. While the relationship for individuals with disabilities is somewhat weaker, it is also positive. Therefore, while the different targeting strategies adopted by the different PIs will certainly affect the client base, it is also clear that the underlying demographic and social structure of the Lot area represents a driving influence on the client composition.
To answer our research questions, we make use of two different methods: probit models and dominance analysis. A probit model is a type of regression model whereby the dependent variable (variable of interest) is binary; for example, if the person has a disability or not. This outcome variable is regressed on several covariates (independent variables) and illustrates the association of each of the covariates with the dependent variable, while all other covariates are held constant at their mean values (for more information, see Aldrich and Nelson, 1984). We use a probit model to analyse the probability that an individual experiences each of the five barriers or not. We also use it to understand the effect of the various characteristics in explaining the probability that an individual will report a particular barrier (or multiple barriers rural area or not, 10 and length of time on live register. Age is categorised as: less than 25 years old, 25 to 35, 36 to 45, 46 to 55, or 55 to 65. The small-area deprivation index measures the relative affluence or disadvantage of an area using information from previous censuses (Haase and Pratschke, 2013). Education is composed of four categories denoting whether the individual has less than upper-secondary education (NFQ <4), upper-secondary education (NFQ 4-5), third-level education (NFQ 6-8) or postgraduate education (NFQ 9-10). The nationality variable is split into whether the individual is from an old EU member state, new EU member state, 11 or not from the EU. The length of time on the live register is less than 6 months, 6 to 12 months, 12 to 24 months, or more than 24 months. While the probit model illustrates the association of each of the variables in the model with the outcome variable while all other variables are held fixed, it does not clearly rank each of the variables in terms of the relative importance in explaining the outcome variable. For example, it does not clearly tell us which explanatory variable has the most predictive power in determining an individual's risk of experiencing a particular barrier. It is important for policy making to understand which characteristics contribute the most to this risk. Dominance analysis is a method that determines the relative importance of each explanatory variable in explaining the outcome variable (Budescu, 1993). It does this by calculating the proportion of the variation in the outcome variable that is explained for each model in which the explanatory variable is included. Dominance statistics are then calculated as a weighted average of the incremental contribution each explanatory variable makes to explaining the outcome variable. We use dominance analysis to understand the relative importance of each of the explanatory variables in determining the risk of experiencing each of our five barriers. This allows us to rank the explanatory variables in order of importance in predicting a particular barrier.

Results
This chapter includes an examination of the occurrence of individual barriers and the most common combinations of barriers reported by SICAP participants. We investigate the distribution of individual barriers (one or more) experienced across Lot deprivation levels and evidence of targeted approaches at Lot level. It also presents a comprehensive analysis of the risk of individuals experiencing barriers to full economic and social inclusion. Our multivariate regression analysis approach attempts to determine the complex relationships in order to explain how elements across a multitude of variables respond simultaneously to changes in other variables. This analysis examines the determinants (using personal characteristics and area-based measures of deprivation) of each individual barrier reported across goal type (lifelong learning and employment) and small areas with differing deprivation levels. In addition, this chapter examines the key characteristics of individuals reporting multiple barriers, as opposed to single barriers, to social and economic inclusion. Table 2 summarises the occurrence of the various social barriers at the individual level. The highest rate applies to jobless households (over 40 per cent) while the reported incidence of the other barriers is relatively low. As stated above, approximately 40 per cent of SICAP respondents did not report any of the five barriers included in this analysis. For each social barrier, approximately 14 per cent of cases exist where 'no response' was recorded or where the response was blank, 12 potentially creating problems around sample representativeness. To investigate this further, non-response was analysed with a probit model, relating non-response to gender, age, nationality, NEET status, urban/rural location, live register duration, smallarea deprivation levels and Lot area variables (see Table A1 in Appendix). Differences between Lot area and PI completion procedures regarding the recording of information on IRIS were found to have the largest effect on nonresponse, while individual characteristics appear to be much less important, with very small marginal effects. Furthermore, we have a large representative workable sample of 33,697 participants (71 per cent of all participants) when we restrict our sample to consider only those who responded to all the questions relating to the five barrier categories (see Figures 3 and 4). It is clear from Figures  3 and 4, and the results of our probit model, that our sample is highly representative of the SICAP population. Notes: jh, jobless household; lp, lone parent; pd, person with a disability; hah, homeless or affected by housing exclusion; trra, Travellers, Roma, refugees and asylum seekers. Table 3 shows the number of barriers reported by individuals accessing SICAP education and training programmes. Approximately 55 per cent of individuals report either one or two barriers, while 42 per cent report none of the five barriers. The table also shows that there is a low frequency of individuals reporting multiple barriers, with less than 3 per cent of individuals reporting three or more.  Table 4 shows the most common combinations of barriers reported by participants. In total, 31 combinations of the five specific barriers were reported; the top eight combinations in Table 4 account for 95 per cent of all participants.

TABLE 4 COMBINATIONS OF BARRIERS REPORTED BY SICAP PARTICIPANTS, 2016
Notes: Each individual is counted under one heading only. For example, the jobless household category (number 2) includes individuals who report being members of a jobless household but no other barrier. Table 5 describes the combinations of barriers by gender and reveals differences across males and females. While similar proportions of males and females (around 42 per cent) report no barriers and both report jobless household as the most common barrier, more males (40 per cent) than females (24 per cent) report being in a jobless household. However, females are more likely to report multiple barriers. For example, Table 5 shows that 12 per cent of females are in a jobless household and are lone parents, relative to 2.5 per cent for males.    Figure 5 shows the distribution of SICAP participants with respect to the level of disadvantage, as per Pobal HP Deprivation Index (SA), of the small area where they are living. This index is a method of measuring the relative affluence or disadvantage of a particular geographical area using data from various censuses. As discussed in Chapter 2, Pobal has developed its resource allocation model based on the Pobal HP Deprivation Index (SA) to inform what percentage of the total budget should go to which areas based on their associated levels of disadvantage. The scoring is given to the small area based on a national average of zero, and ranges from approximately -35 (the most disadvantaged) to +35 (the most affluent). One of the main advantages of using small areas is that they are standardised in size, with a minimum of 50 households and a mean of just under 100 households, thus effectively providing street-level information on the Irish population.

FIGURE 5 DISTRIBUTION OF SICAP PARTICIPANTS BY POBAL HP DEPRIVATION INDEX (SA), 2016
The distribution of the index across all small areas nationwide follows a bellshaped curve, with most areas clustered around the mean and fewer areas exhibiting extreme levels of affluence or deprivation. However, when we examine the locations of SICAP participants in 2016 ( Figure 5), the distribution is skewed to the left, implying that the majority of individuals accessing SICAP are from areas below the average of zero. Approximately 30 per cent of SICAP participants live in small areas that are classified as disadvantaged, and this compares to 14 per cent of the nationwide population in 2016 that are classified as disadvantaged (see Pobal, 2017). Nevertheless, more than two-thirds of SICAP participants come from small areas that are not classified as disadvantaged (with a Pobal HP Deprivation Index (SA) of greater than -10). This finding raises potential issues around the appropriateness of the funding approach and the extent to which it is successful in skewing the distribution toward areas with a greater level of need. The average number of individual barriers reported by participants does not appear to differ across Lot areas irrespective of their differing levels of deprivation ( Figure 6). For example, the average number of barriers experienced for participants is just below one for Lots with varying levels of deprivation scores (negative and positive). In order to further examine the PIs' focus at a Lot level on participants from relatively disadvantaged areas, we calculated the average Pobal HP score across all participants in 2016 at a Lot level and compared that to the assigned Lot level deprivation score. This reflects the fact that within Lots there will be a high level of variation in terms of the Pobal HP Deprivation Index (SA) of small areas within the Lots' particular catchment area. Figure 7 shows that in practically all Lots, some to a greater degree than others, provision is focusing disproportionately on those from areas with a higher level of deprivation than the average deprivation index assigned at Lot level.  In terms of our models, we aim to capture the characteristics most commonly associated with each of the various barriers to inclusion. It is important to stress that the data are representative of the SICAP client body; however, to the extent that SICAP is a national-level policy intervention that specifically targets social disadvantage, we can be relatively confident that the relationships identified will also reflect national trends. We begin by modelling each barrier to inclusion in Table 6 using a probit model framework, whereby the dependent variable takes a binary (0, 1). The models control for a range of factors including gender, age, educational attainment, unemployment history, small-area level of deprivation and a rural/urban control. The specification varies slightly depending on the barrier being modelled in order to avoid problems of collinearity; 13 for instance, unemployment controls are omitted in the jobless households model and nationality is dropped from the ethnic minority model. All models are estimated using a common reference category that consists of all SICAP clients reporting none of the five barriers (approximately 40 per cent of the sample).
It is apparent from the models that while there are some characteristics that are important for all or most barriers, the relative importance of factors varies by barrier; furthermore, some characteristics may have positive or negative associations across key independent variables. With respect to gender, being 13 This describes the situations whereby the predictor variables are highly correlated with the dependent variable.
Average Individual Small Area HP Index (Lot Level) 24 | Pr o fi l i n g Ba r ri e r s t o So ci a l In c l u si o n i n Ir e l a nd female raises the probability of reporting all barriers, with the exception of jobless household membership, where males have a higher relative risk. Generally, gender has a relatively low influence on the risk of experiencing higher number of barriers to inclusion, with being female raising the probability by 2 to 3 percentage points; the clear exception to this is the lone parent status, which indicates that females are 25 percentage points more likely to report this barrier than males.
With respect to age, older individuals are more likely to report most barriers, with the exception of the ethnic minority barrier, which appears to be more characteristic of individuals below the age of 25. Age appears to be a particularly important predictor for belonging to a jobless household, being a lone parent or reporting a disability. Members of jobless households and lone parents are approximately 20 per cent more likely to be aged 25 and older. People older than 55 (56-65) are more likely to have a disability by 10 percentage points compared to the 16-24 age cohort.
In terms of education, individuals with higher levels of educational attainment have a lower probability of experiencing all barriers; however, the effects are strongest for jobless households and lone parents, with persons educated above NFQ 5 (Leaving Certificate) being 10 or more percentage points less likely to report experiencing these barriers relative to those with no qualifications. With regard to nationality, Irish nationals are more likely to report being lone parents, having a disability or being affected by homelessness; conversely, individuals from non-EU and old EU countries are more likely to report belonging to jobless households relative to Irish natives. Non-EU nationals were almost 15 percentage points more likely to report housing difficulties compared to Irish nationals. Individuals who have experienced unemployment durations of six months or over were more likely to report being a lone parent or having housing problems, while ethnic minorities were more likely to have a history of long-term unemployment. Individuals with disabilities were less likely to report unemployment histories of any duration, which may reflect the fact that many may not be actively employed or seeking work.
Finally, while NEET status raised the likelihood of reporting all barriers other than lone parent status, it is a particularly strong predictor for belonging to a jobless household, with NEETs almost 16 per cent more likely to report this barrier relative to non-NEETs, which provides evidence of the inter-generational pervasiveness of joblessness.
The difference in risk of experiencing each barrier by migrant status is particularly interesting. Individuals from new EU member states are 7.5 per cent less likely to be from a jobless household, 9 per cent less likely to be a lone parent and almost 2 per cent less likely to be homeless/affected by housing exclusion. Conversely, individuals who are from outside the EU are almost 11 per cent more likely to be in a jobless household and nearly 15 per cent more likely to be homeless/affected by housing exclusion. Not surprisingly, we find that migrants in general are less likely to have a disability.
Turning to the location variables, we find the small-area HP deprivation level raises the likelihood that individuals will report being a lone parent and/or belonging to a jobless household, which suggests that persons facing these barriers are more heavily concentrated within more deprived areas.
Nevertheless, the small-area HP deprivation level measure was not a predictor of disability, housing problems or ethnic background, suggesting that persons experiencing these barriers are more evenly distributed across areas with varying levels of deprivation. With respect to the marginal impact of small-area HP deprivation level, individuals in a highly deprived area (with a small-area HP deprivation level of -20) are 16 percentage points more likely to report belonging to a jobless household and 12 percentage points more likely to report being a lone parent relative to those located in an affluent area (with a small-area HP deprivation level of 20).
Interestingly, whether a person lived in a rural as opposed to an urban location had a strong influence on their probability of reporting particular barriers after the small-area HP deprivation level had been controlled for, suggesting that urban/rural environments influence an individual's risk of particular barriers in ways that are unrelated to the level of deprivation in the area where they live. Specifically, after controlling for all other factors, individuals living in urban locations were between 4 and 10 percentage points more likely to report each of the specific barriers compared to those living in rural areas. The level of urban disadvantage was highest for lone parenthood and jobless households. The models suggest that individuals living in urban locations experience additional difficulties that exceed the level of deprivation present in their immediate area. Finally, it is worth noting that the probit models vary in the extent to which they can successfully predict an individual's probability of experiencing a particular barrier. For example, the pseudo-R 2 statistic, which measures the fraction of variation in the outcome variable (barriers) that is explained by the model, indicates that the jobless household model offers less predictive power than the other models.   Standard errors clustered at small-area level in parentheses. 'Ethnic minority' refers to members of the Travelling community, Roma, refugees, and asylum seekers. ***p < 0.01, **p < 0.05, *p < 0.1. Table 7 assesses the extent to which the characteristics that are correlated with the risk of experiencing a single barrier to inclusion differ from those that are correlated with experiencing multiple barriers. The table also shows the degree to which relationships vary depending on whether assistance was delivered under the SICAP employment or lifelong learning goals. Once again, all models were estimated against a reference category consisting of SICAP clients reporting none of the five specified barriers.
Relative to the reference category (individuals reporting no barriers), males were 6 percentage points more likely to experience a single barrier to inclusion but 15 percentage points less likely to report multiple barriers. The models also indicate that individuals without formal qualifications (below NFQ 4) were more likely to report multiple as opposed to single barriers relative to their counterparts with higher levels of educational attainment. There was no evidence that multiple barriers were more common in areas of high deprivation. Relative to the reference category, individuals in urban areas were 6.5 percentage points more likely to report a single barrier, but almost 12 percentage points more likely to report multiple barriers to inclusion, again pointing to additional difficulties associated with living in an urban area.
Individuals aged 25 to 45 also had a higher likelihood of reporting multiple barriers. With respect to SICAP goals some differences were apparent. Not surprisingly, individuals with higher levels of schooling were less likely to engage with lifelong learning programmes (goal 2) and more likely to receive employment assistance (goal 3). Conversely, males were somewhat more likely to receive assistance under goal 3 than under goal 2. When we look at the probability of experiencing multiple as opposed to single barriers across goals, similarly to when we do not split by goal, we find that being female and low levels of education have the largest impact.    Table shows average marginal effects with standard errors clustered at small-area level in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. Table 8 compares the characteristics of individuals reporting one or more barriers to inclusion in areas defined as deprived (small-area HP deprivation level ≤-10) with the most affluent areas (small-area HP deprivation level >10). This allows us to assess the degree to which PIs have been successful in targeting individuals with more challenging characteristics (low education and NEET status) irrespective of the area level of disadvantage. A finding that SICAP clients reporting barriers in more affluent areas had, for example, high levels of educational attainment, were older or were less likely to be NEET would raise questions regarding the extent to which the programme was successfully reaching individuals at the highest risk of social exclusion. The results in Table 8 indicate that the characteristics of SICAP clients reporting barriers to inclusion are broadly comparable across deprived and affluent areas with respect to education and NEET status, suggesting that targeting is consistent across all areas. In fact, it would appear that SICAP clients in more affluent areas have a somewhat more disadvantaged education profile than their counterparts living in deprived areas. Nevertheless, some differences do exist in the client profile, with clients reporting barriers in deprived areas more likely to be female and aged between 25 and 45. The models also confirm that across deprived areas, individuals located in urban areas have a higher risk of experiencing one or more barriers to inclusion. Dominance analysis provides a decomposition of the individual barrier models in order to measure the degree to which the correlation between various characteristics and barriers to inclusion varies across the five dimensions examined in the study (Table 9). In summary, we find that low levels of educational attainment are an important predictor of all barriers, while an urban location is a key factor for all barriers except of disability. Nationality plays an important role for housing exclusion and disability. However, we found that the small-area deprivation level was a strong predictor of just one barrier (jobless households), while NEET status is relatively strongly correlated with ethnic minority status only. More specifically, the characteristics most dominant among individuals reporting to belong to a jobless household are: living in a more disadvantaged small area, having no qualifications (compared to NFQ 6-8 attainment) and living in an urban location. With respect to lone parents, the most dominant characteristics are gender (female), a low likelihood of being longterm unemployed, 14 an urban location and a higher likelihood of holding no qualifications (relative to NFQ 6-8 attainment). Reported disability is most heavily correlated with having no qualifications (relative to NFQ 6-8 attainment), a lower likelihood of unemployment 15 and being an Irish national. Homelessness/housing exclusion are more heavily connected with an urban location, being a non-EU migrant and long-term unemployment status. Finally, ethnic minority status is more heavily associated with lower levels of educational attainment, an urban location, short-term unemployment and NEET status. Next, we estimate a model at Lot level to assess the degree to which the overall intensity of barriers among clients varies according to factors such as Lot size (measured by the number of clients) and spatial factors. While our estimates confirm that the general profile of disadvantaged clients is broadly similar across Lot areas with varying levels of deprivation, suggesting that targeting is consistent, the models tell us little about the degree to which the intensity of disadvantage varies by deprivation level. The analysis in Table 10 provides no strong evidence that the occurrence of barriers among client groups is higher in more deprived areas; in fact, at approximately 55 per cent, the rate of reported disadvantage is equivalent across PIs in the most advantaged and disadvantaged areas. While Table 10 mainly displays descriptive statistics, in Table 11 we estimate a more formal regression model in which we include a rural indicator, the number of clients and the Pobal HP Deprivation Index (SA) at the Lot level. While the model confirms that the intensity of deprivation does not vary according to the Lot area deprivation level, it does reveal that the average occurrence of reported barriers in urban areas is 18 percentage points higher than the rural equivalent. This model suggests that if we compare a rural PI in a disadvantaged small area and an urban PI in an affluent small area with a similar number of clients, the prevalence of reported barriers is likely to be considerably higher for the affluent urban PI.  Finally, it is possible that the urban impacts that we are detecting in our models relate to the pressures that arise from living in an area of high population density rather than an urban location per se. To test this, we re-estimate the social barriers model (Table 6) to include a measure of population density in order to examine its impact on the estimated urban/rural variable (Table 12). The results demonstrate that the disadvantages associated with an urban location are distinct from those that arise from living in an area with high population density. We found that residing in a highly populated area lowers the probability that individuals will belong to a jobless household, be affected by disability or belong 32 | Pr o fi l i n g Ba r ri e r s t o So ci a l In c l u si o n i n Ir e l a nd to an ethnic minority; perhaps, not surprisingly, individuals in highly populated areas were more likely to report being homeless or affected by housing exclusion. The estimated impacts of the rural variable remained largely unaffected by the inclusion of the population control variable, suggesting that an urban environment generates specific issues that raise the likelihood of individuals reporting these barriers to inclusion. 34 | Pr o fi l i n g Ba r ri e r s t o So ci a l In c l u si o n i n Ir e l a nd CHAPTER 6

Summary and conclusions
This report assesses the extent to which individuals participating in the SICAP programme experience barriers to inclusion. A number of important findings emerge with respect to the distribution of barriers, the key characteristics associated with the various forms of disadvantage and the role of spatial factors. In this study we examined five barriers which have been shown in the national and international literature to be highly damaging to individuals: membership of a jobless household, lone parent status, disability, homelessness or housing exclusion and being in an ethnic minority. While the incidence of barriers within a particular Lot is influenced by the targeting strategy of the Programme Implementer (PI), we found that the underlying demographic and social structure of the population in a particular area will be a key determinant in the composition of a Lot client base. Belonging to a jobless household was the most commonly reported barrier (over 40 per cent), while the reported occurrence of the other barriers was typically below 10 per cent. In terms of the overall composition of the SICAP client base, while it was clear that individuals from deprived areas were over-represented in the programme relative to their population shares, approximately two-thirds of SICAP clients come from areas not defined as deprived. However, our results also indicate that the characteristics of SICAP clients reporting barriers to inclusion are broadly comparable across deprived and affluent areas with respect to education and NEET status. This suggests that targeting is consistent across all areas and that PIs are effectively accessing and assisting highly disadvantaged individuals irrespective of the deprivation levels of the areas within their Lot.
Generally, gender has a relatively low influence on the risk of experiencing a particular barrier, with being female raising the probability by 2 to 3 percentage points; the clear exception to this is being a lone parent, as females are 25 percentage points more likely to report this particular barrier. Being aged 25 or over appears to be a particularly common feature of belonging to a jobless household, being a lone parent or reporting a disability. In terms of education, the pattern is consistent with higher levels of educational attainment associated with a lower probability of reporting all barriers; however, the impacts are highest for jobless households, lone parenthood and reported disabilities. With respect to spatial factors, we found that the small-area HP deprivation level raises the likelihood that individuals will report being a lone parent and/or belonging to a jobless household. Nevertheless, the small-area HP deprivation level measure was not a predictor of disability, housing problems or being in an ethnic minority, suggesting that persons experiencing these problems are more evenly distributed across areas with varying levels of deprivation.

Summary and conclusions | 35
Whether a person lived in a rural or an urban location had a strong influence on their probability of reporting particular barriers after the small-area HP deprivation level had been controlled for, suggesting that urban/rural environments influence an individual's risk of particular barriers in ways that are unrelated to the level of deprivation in the area where they live. The models suggest that individuals living in urban locations experience additional difficulties that transcend the level of deprivation present in their immediate area, and that these are over and above the impact that arises relating to the area's population density level. Our analysis ranked the various explanatory variables in terms of the strength of the association with the various barriers to inclusion. We find that low levels of educational attainment are an important factor for all barriers, while an urban location is a key factor for all barriers except disability. Nationality plays an important role for housing difficulties and disability. However, we found that the small-area HP deprivation level was a strong predictor of just one barrier (jobless households).
In terms of the risk factors associated with experiencing multiple barriers (more than one of the five barriers) to inclusion, females were 15 percentage points more likely to report multiple barriers than males. The analysis also indicated that individuals without formal qualifications (below NFQ level 4) were more likely to report multiple, as opposed to single, barriers, relative to their counterparts with higher levels of educational attainment. While there was no evidence that multiple barriers were more common in areas of high deprivation, relative to the reference category, we again found a strong urban effect with individuals in urban areas almost 12 percentage points more likely to report multiple barriers to inclusion relative to their counterparts in rural areas.
Our econometric evidence from the models presented shows that the intensity of deprivation does not vary according to the small-area deprivation level; however, the average occurrence of reported barriers among PIs in urban areas is 18 percentage points higher than the rural equivalent. This model suggests that if we compare a rural PI in a disadvantaged small area and urban PI in an affluent small area, both with a similar number of clients, the occurrence of the reported barriers considered in this study is likely to be considerably higher in an affluent urban PI. The analysis suggests that targeting under the SICAP programme is effective, with PIs effectively accessing and providing supports to disadvantaged individuals. We find that living in an urban area to be a more important determinant of disadvantage both in terms of individual risk levels and the intensity of barriers reported at the PI level. While low levels of education are an important risk factor associated with almost all barriers, other factors such as gender, nationality and labour market history are key determinants of others. Key differences in the characteristics of individuals facing particular barriers should be reflected in the design and delivery of interventions aimed at reducing the 36 | Pr o fi l i n g Ba r ri e r s t o So ci a l In c l u si o n i n Ir e l a nd impacts of particular forms of social exclusion.
Finally, with respect to the funding model, the analysis which shows that over two-thirds of individuals accessing SICAP supports do not reside in areas designated as disadvantaged supports the assertion that funding patterns are based heavily on historical factors rather than the full implementation of the Resource Allocation Model (RAM). We also find that for the barriers considered in this study, key elements of the RAM such as the small-area deprivation level or population density would not necessarily be effective in allocating funds towards the areas where these barriers were most pronounced. These findings suggest that the current funding allocation model does not take account of the fact that some Lot areas may have a more substantially disadvantaged clientele, requiring more intensive interventions than others. Going forward, some adjustments may be required to ensure a more effective allocation of resources towards Lots dealing with disadvantaged communities where barriers to social inclusion are particularly pronounced.