Introduction

It is John’s last day of work. Finally, social security had approved his application for retirement and he brings home all the material he had had on his desk for ages. There is no one home to make him company but he feels the relief of not needing to commute for more than two hours every day, in pain from his arthritis. Not too distant is also the last day of work for Mary who taught her last class as a high school teacher. There is no one waiting for her at home either, she became a widow a few years ago. Now she will have more time to take care of her grandchildren. John and Mary both live in a single person household and are retired, it is very likely that from today on their level of social exclusion increases. The labour status, combined with other aspects like health or household structure, can be important determinants of social exclusion (SE). They represent different levels of resources, needs and limitations, and they are frequently attached to specific phases of life. The life course perspective of SE, therefore, focuses on the roles typically played in different stages of life and how the change in conditions may influence the existence of SE. The central forms of organization of individual adult lives in nowadays’ societies are work and family. Being out of work and having no family support leaves the individual in danger of becoming socially excluded, although some people find alternative participation strategies. The institutional rules developed by the welfare states generate more or less predictable events that mark transitions between life phases, and that are frequently structured around chronological age.

In this chapter, we explore the recently updated data from the European Social Survey (ESS), Round 9 – 2018, combined with data from Round 1 – 2002, and we look at individuals in late life to analyse how SE evolved for their cohort along time. The purpose is to gain a life-course perspective since we look at individuals of the same birth cohort, people who were born between 1945 and 1953 and hence were between the ages of 49 and 58 in 2002, and between the ages of 65 and 74 in 2018.Footnote 1 People in the same cohort, who were born during the same period, are exposed to similar socio-cultural context in the same phase of life. The cohort-based trajectory approach, by looking at individuals of the same cohort, aims to understand how such cohort experiences the later part of the life-course, with an emphasis on work participation. The changes in SE that affect the 1945–1953 birth cohort are both a result of period and age effects, which we do not attempt to disentangle. The two sets of observations typically correspond to two different stages of life: the first being late working life and the second being retirement. Nevertheless, to reinforce the life course perspective, we separate individuals in the same group according to their employment status and consider that only those in the same age group and in the same labour situation are in the same stage of life.

We are also interested in gaining an insight into the evolution in the situation of individuals of a certain age group, that is, in understanding if older individuals became more or less socially excluded between 2002 and 2018, in Europe.

SE is a multidimensional concept, broader than just poverty or economic exclusion, that aims at expressing to what extent people have the opportunity to participate in society. Such participation may be affected by deprivation experienced in different wellbeing domains, which may interact and reinforce each other. The ESS has information relevant to analyse domains like Social Relations, Civic Participation, Neighbourhoods and Community, and Health and Well-Being. The ESS also includes demographic characteristics of the respondents. We use this information to measure SE and each SE domains, that can shed some light on the relationship between the labour market status and SE in the later stages of life.

Background

Conceptual Framework

SE has been in the European discourse for some decades (Torres, 2018; Bradshaw, 2004; Peace, 2001). Apparently, it was adopted as a concept to replace the more stigmatizing term ‘poverty’ (Peace, 2001). This explains why in the beginning, SE was conceptually very close to financial/material deprivation. In the Lisbon Summit of 2000, the European Commission embraced the objective of creating an inclusive Europe, which led to the creation of national action plans against poverty and SE. The fight against poverty and SE was reinforced in the Europe 2020 strategy, through the open method of coordination, issuing guidelines combined with timetables for achieving goals set for the member states, accounting for regional differences (Schoukens et al., 2015).

The right to be protected from poverty and SE is recognized in Article 30 of the Revised European Social Charter, a human rights protection instrument of the Council of Europe.

In the international organizations’ documents, the mention of SE used to come together with the mention of poverty. “Poverty and social exclusion” was the common expression. Naturally, this means that they are not the same, otherwise only one word would be used. At the same time, this constant association may centre the focus on the economic aspects of exclusion. The adoption of a 2030 European agenda that has the Millenium Development Goals as its core, changed this a bit. Now the eradication of poverty is one of the goals. There is no specific goal that mentions SE/inclusion. Nevertheless, it may be argued that all of them are part of a global objective of inclusion and to support that view, we have the announced purpose of “leaving no one behind” in the implementation of the development agenda.

In the literature, the concept of SE gradually changed, with the addition of several contributions to its definition so that, although not completely consensual, it has converged to a common ground. Nowadays, it is accepted that ‘SE’ is multidimensional, it is relational and implies some agency – one is excluded from something, by someone/something – and it is dynamic or processual – it takes place in time and its current duration affects its future course (Atkinson, 1998; Tsakloglou & Papadopoulos, 2002; Béland, 2007; Silver, 2007; Levitas et al., 2007; Walsh et al., 2017; MacLeod et al., 2019).

Several definitions of SE can be found in the literature, with slightly different emphases, but one of the most cited is the one by Levitas et al. (2007) (p. 86) “SE is a complex process operating across several dimensions or domains. It involves both the lack or denial of resources, rights, goods and services, and the inability to participate in the normal relationships and activities, available to the majority of people in a society, whether in economic, social, cultural or political arenas. It affects both the quality of life of individuals and the equity and cohesion of society as a whole.” One advantage of this definition is that it is suited to empirical work, providing hints about how to operationalize its measure. One disadvantage is that it does not suggest how to incorporate the dynamic and the agency elements.

Since this chapter focuses on SE in later life, it is worth considering in what respect can it be different from exclusion in earlier phases of the life cycle. There are already some studies that address this issue (Barnes et al., 2006; Scharf & Keating, 2012; Scharf, 2015; Kneale, 2012; MacLeod et al., 2019; Walsh et al., 2017; Van Regenmortel et al., 2018).

As a relative concept, it is necessary to reflect to whom we want to compare the individuals whose situation we analyse (Scharf & Keating, 2012). All the domains of SE may be applied to older people, it is the probability of being excluded that may be different, particularly the risk of accumulating several forms of exclusion. And the probability of being excluded from various domains may be higher because of drivers that are particularly associated with older age: ageism, disruption to individuals’ networks, physical limitations and health problems.

Measuring Social Exclusion

Empirical studies that treat SE as a multidimensional concept identify several domains of exclusion. For each domain, several indicators may be found. For example, Burchardt et al. (2002) view four areas of SE: (i) production, (ii) consumption, (iii) political engagement, and (iv) social interaction. Bradshaw (2004) separate three components of SE: (i) labour market, (ii) services, and (iii) social relations. Jehoel-Gijsbers and Vrooman (2008) use a conceptual model considering only two dimensions of SE, each with two subdimensions: (i) an economic/ distributional dimension, comprising material deprivation, and access to social rights, and (ii) a relational/socio-cultural dimension, comprising social integration and cultural integration. Levitas et al. (2007) also use domains and subdomains, but they disaggregate more: (i) resources, (ii) participation and (iii) quality of life are disaggregated into ten subdimensions. Kneale (2012) considers seven domains of exclusion: (i) financial products; (ii) common consumer goods; (iii) local amenities; (iv) social relationships; (v) cultural activities; (vi) decent housing and public transport; and (vii) civic activities and access to information. Walsh et al. (2017) identify, from a review of the literature: (i) neighbourhood and community, (ii) services, amenities and mobility, (iii) social relations, (iv) material and financial resources, (v) sociocultural aspects, and (vi) civic participation. Van Regenmortel et al. (2018) use most of the same domains as Walsh et al. (2017) but identify eight dimensions of exclusion; they do not isolate a sociocultural dimension, but they add exclusion from decent housing, ageism, and digital exclusion.

Sometimes there is some confusion between determinants and domains. For example, the type of participation in the labour market can be viewed as a determinant, as a domain, and even as the meaning of SE. The argument that it does not make sense to include the labour market participation as a dimension of the SE of older people, because 65 or 66 years on, most individuals are not in the labour market (Van Regenmortel et al., 2018) is, in our view, a result of this confusion. The participation in the labour market may be responsible for several types of exclusion – not only from financial resources, but also from social relations – that are very relevant in late life. So, it is not because most of population in this life stage is out of the labour market that the labour market status should not be viewed as a dimension of SE, but because it should instead be considered a determinant, which is exactly Van Regenmortel et al. (2018)’s position with respect to health.

Data Source and Samples

We use data from two waves of the European Social Survey (ESS) – 2002 (Round 1) and 2018 (Round 9).Footnote 2 The comparability of different years creates limitations to the study: the number of comparable countries, 15 in totalFootnote 3 (Table 11.A1), reduces the number of observations, and some variables are unavailable for both roundsFootnote 4. Focusing on the SE in the later stages of life, the relevant population group is composed by the individuals born between 1945 and 1953, those who belong to the age group of 65–74 years old in 2018.Footnote 5 The data from ESS do not correspond to true panel data because it is not possible to follow the same respondent of ESS across the different waves or rounds.

Table 11.1 describes the sample characteristics concerning the main activity by gender. In 2002, for the birth cohort 1945–1953, paid work represents the largest share of main activity (65%), and in 2018 being retired is the main activity (82%), reflecting the advance in the life course. In both years the ratio men/women in paid work is higher than one and around one among those in retirement. Housework and looking after children or others are activities with a low participation of men, changing the ratio men/women from 0.1 in 2002 to 0.2 in 2018.

Table 11.1 Sample ESS rounds 1 (2002) and 9 (2018), Birth cohort 1945–1953, 15 European countries, labour market status by gender
Full size table

Methodology: The Social Exclusion Indexes

Following the methodology adopted by Macleod et al. (2019), we consider four SE domains: Social Relations (SR), Civic Participation (CP), Neighbourhood and Community (NC), and Health and Well-being (HW). Each domain is measured in a scale of 0–4. The higher the score, the higher the level of SE in that domain, that is, the lower the level of social inclusion.

We compute two SE indicators:

  • The index SocExc 1 is obtained as the average of two domains: Social Relations and Civic Participation.

  • The index SocExc 2 is obtained as the average of four domains: Social Relations, Civic Participation, Neighbourhood and Community, and Health and Well-being.

The reason for calculating SocExc 1 and SocExc 2 separately, is that some authors consider Neighbourhood and Community and Health and Well-being as drivers of SE, not as domains of SE (Pratteley et al., 2020; Van Regenmortel et al., 2018), so the SocExc 1 is probably a more consensual measure of SE. The SE indexes, by construction, vary within the interval 0 and 4, being 4 the maximum of SE. Table 11.2 presents the description of each domain used to construct the SE indexes. Table 11.A2 gives the descriptive statistics of the variables included in each domain revealing the diversity of values. With few exceptions, the values generally remain stable from one year to the other, or have small changes (last column of Table 11.A2).

Table 11.2 Description of variables by social exclusion domain
Full size table

The empirical analysis has two parts. Firstly, following the methodology of Macleod et al. (2019), the SE Indexes are built, using data for the birth cohort 1945–1953, for 2002 and 2018. Next, we examine the evolution across time for the same birth cohort, by main activity and by gender. Here, we intend to capture the evolution of SE for this group of individuals. Secondly, we compare the two age groups (49–58 and 65–74), within both years (2002 and 2018). In this case, the two groups belong to different birth cohorts. We are interested in comparing the situation of the older group with that of a co-existing younger group. We call this a ‘non-cohort analysis’.

For the domain SR, the original data about Social Relations are transformed adopting the methods summarized in Table 11.2. The outcome is a measure from 0 to 4.

The component CP has a very large scope: it includes political activities as well as cultural and leisure activities. Given the data available for the two ESS Rounds (2002 and 2018), the CP measure is computed from the 8 binary variables listed in Table 11.2. After recodificationFootnote 6 and recalibration of the eight variables, the scale of CP is also from 0 to 4.

The score of the NC domain is obtained from four original variables after recoding (Table 11.2).

The HW component is also obtained from the combination of four variables: two variables about satisfaction with life and happiness, and two others related with healthFootnote 7 (Table 11.2).

Results

Our results are of two types: (i) one that explores the life cycle perspective, looking at the birth cohort of 1945–1953 in both years, and in particular comparing the population in paid work in 2002 (65% in year 2002) with the population retired in 2018 (82% in year 2018) (Table 11.1), and (ii) another that observes people of the same age and of the same situation in the labour market, in two different periods, or alternatively, for people of different age groups or different labour market statuses, in the same period.

Social Exclusion of Birth Cohort 1945–1953

The SE results for the birth cohort of 1945–1953 disaggregated by main activity can be seen in Figure 11.1 and in Table 11.3. Additionally, in Table 11.4, differences between men and women for each domain are detailed.

Fig. 11.1
A graph compares the labor status and social exclusion by age, for 2002 and 2018. Data are approximated. In 2002, The highest is for paid work at 72 percentage, and the lowest is retired with 10 percentage. In 2018, The highest is the retired at 90, and the lowest is paid work with 10.

Social exclusion (SocExc 1) and labour market status by age in paid work and retired; birth cohort 1945–1953; 15 European Countries; 2002 and 2018. (Data source: ESS R1 2002 and ESS R9 2018)

For simplicity, for each age, only shares of population in Paid work or in Retirement higher than 10% are shown (left axis). The SocExc 1 is calculated for all the individuals of same age, not only for the retired or in paid work

Full size image
Table 11.3 Social exclusion (SocExc 1) and labour market status; paid work, retirement and housework and care; birth cohort 1945–1953; 15 European Countries
Full size table
Table 11.4 Social exclusion by domain, labour market status and gender gap; in paid work, retired and housework and caregiver; birth cohort 1945–1953; 15 European Countries
Full size table

Figure 11.1 illustrates the evolution of SocExc 1 by age, for the birth cohort 1945–1953. In 2018, when the retired individuals represent the highest share of the population, the gene

The SocExc 1 index increases by 6.3 p.p. (from 1.92 for people in paid work in 2002 to 2.04 for people retired in 2018) and the SocExc 2 index increases by 3 p.p. (from 1.34 to 1.37). In general, the SocExc 1 of the retired is higher than the SocExc 1 of those who are in the paid work, but the difference is very small, when using SocExc 2 (Table 11.3).

Also noteworthy is the result for the houseworkers and caregivers who at older ages have the highest level of SE (Table 11.3). Houseworkers and caregivers, the third category of labour status group, represent respectively 12% and 7% in 2002 and 2018, and are mainly women (93% in 2002, and 82% in 2018) (Table 11.1).

Table 11.4 presents the results for the four domains of SE, and within each domain, the results are shown by three labour market status (in paid work; retired; and nonpaid houseworker or nonpaid care worker). The measures of SE vary between 0 (minimum) and 4 (maximum). Civic Participation is the domain with higher SE, showing values higher than 3 in both years and little differences between men and women. The gender gap in each labour market status (the last 3 lines of the table for each domain) is measured by the division of the SE value for men by the SE value for women. When men and women have the same level of SE, the value of that division is 1. If the ratio is lower than 1, it means that, compared with women, men have a lower level of SE. In both years (2002 and 2018) independently from the labour market status, women have higher SE compared with men, in three of the four domains under study. The exception is Social Relations in the year 2002, when the group of those who are in paid work as well as houseworker and care show values of 1.03 and 1.1. respectively. In general, Neighbourhood and Community is the only domain for which SE is not higher for people in retirement in 2018 than for people in paid work in 2002.

Social Exclusion Non-cohort Analysis

A non-cohort analysis (Table 11.5) reveals that, within the same year and in both age groups, the retired individuals have always a higher SocExc 1 and SocExc 2, compared with the individuals in paid work. Considering the older age group (65–74 years old), the level of SE is higher for retired people than for people in paid work, and even higher for houseworkers and caregivers, whatever the year and the SE index that we consider. In the younger age group, the retired individuals are those in worse position, possibly reflecting situations of early retirement caused by problems that interact with the probability of SE. Comparing the SE of people in the same age group and the same work status in the two different years, it is clear that there is a general improvement.

Table 11.5 Social exclusion (SR;CP) and labour market status by age group 15 European Countries
Full size table

Figures 11.A1, 11.A2 – based on data from 2002 and 2018, respectively – illustrate the relevance of the labour force statuses for the level of SE by age, measured by the index SocExc 1.Footnote 8 The secondary axis (right axis) measures the SocExc 1 for all labour force statuses. In 2002 (Figure 11.A1), the two age groups (49–58 and 65–74) exhibit very different patterns. The first is largely composed by those who are in the labour market and has a SocExc 1 lower than 2; the second, mostly comprised by those who are retired, exhibits a much higher SocExc 1. In 2018 (Figure 11.A2), the profile of the two main activities has some differences compared with 2002: a higher level of participation in paid work and a longer stay in the labour market as the average age of retirement increases. In 2002, the SocExc 1 increases from 2.12 at 65 years old to 2.33 at 74 years old, and the correspondent values for 2018 are 2.03 and 2.05.

The SE situation improved particularly for the older age group (65–74) (Table 11.5).

We can see (Table 11.4) that the only domain for which SE is not higher for people in retirement in 2018 than in paid work in 2002 is NC. In general, being in paid work is associated with lower SE.

Conclusion

Social exclusion is a multidimensional concept that captures several forms of disadvantages, which combine to function as barriers towards the full participation of people in society. It can affect people differently according to their position in the life course. In this chapter, we construct measures of social exclusion to analyse its evolution in later life paying particular attention to its intersection with labour market participation and with gender. A multidimensional approach to social exclusion is adopted. Each of the two measures used assesses more than one dimension of social exclusion, although deflecting away from the competing concept of monetary deprivation. The dimensions incorporated into the analysis are: Social Relations, Civic Participation, Neighbourhood and Community and Health and Well-being.

We use the ESS data for 2002 and 2018, and focus on two groups of people: one, aged 49–58, that is mainly in paid work, and the other, 65–74, whose large majority is in retirement. We carry out two types of analysis on these data. One, the birth cohort-based approach, aims to understand how a cohort (those born between 1945 and 1953), experiences the later part of the life course. The other stems from what we called a ‘non-cohort analysis’, which looks at people of the same age group and of the same labour market status in two different years (2002 and 2018), or alternatively, at people of different age groups or different labour market statuses, in the same year.

Our findings significantly point to the protective role of labour market participation against social exclusion, with a systematic association of higher levels of social exclusion with retirement or with housework or caregiving, rather than with being in paid work, despite not considering directly measures of monetary resources. This conclusion is obtained regardless of the social exclusion measure adopted and the analysis applied. It is worthwhile noting how caregiving – as well as housework, – which is in itself a valuable form of social participation, when performed as the main activity in old age, interacts with several domains of social exclusion, potentially leading to a disadvantaged outcome.

Evolving in the life course, most of the individuals change from being in paid work to being retired, becoming more affected by social exclusion. This is the main result of the cohort approach. Additionally, social exclusion of the people in the same labour market status, aged 49–58 in 2002, seem to have slightly reduced when they become 65–74 in 2018.

The non-cohort analysis provides a positive outlook of the evolution of social exclusion in Europe: people of the same age group and in the same labour market status, are slightly better in 2018 than they were in 2002.

Gender disparities materialize in the higher levels of social exclusion of women in almost all domains. Although this deserves a dedicated study, the low participation of women in the labour market in some European countries, and their dominant presence in the category of housework and caregiving, respectively with better and with worse social exclusion scores, can help explain such result. The different levels of social exclusion by gender suggest a need for different measures and services to counteract the gender imbalance.

Comparing the two analytical tools that are used in this chapter, the social exclusion measure that includes four domains (Social Relations, Civic Participation Neighbourhood and Community, and Health and Well-being) is always lower than the measure that only includes the domains of Social Relations and Civic Participation. The reason seems to be related to the fact that Civic Participation is the area where the worse results are obtained, which echoes frequently heard concerns about the withdrawal of European citizens from civic engagement. And Civic Participation contributes one-in-two parts to the narrower measure, but only one-in-four parts to the broader one.

Our study has some limitations. The indicators used for each domain were data-driven, and the scarcity of available information precluded the inclusion of a material deprivation dimension, which is sometimes the only one used to assess social exclusion. Also, part of the individuals in society that may exhibit higher levels of exclusion are not covered in the survey: institutionalized people, and homeless people. Nevertheless, due to its comprehensive nature, social exclusion is a helpful organizing construct to the understanding of social stratification. The capture of the individual perception of one’s situation, crucial to the idea of social exclusion, is also an added value of this analysis.