1 Introduction

The relationship between social capital and material wellbeing on life satisfaction within the family unit is largely unexplored despite the fact that it embraces the fields of sociology, psychology and economics. Indeed Helliwell and Barrington-Leigh (2010) and Donati (2003, 2007) argue that there is very little work done in the area even though the family is a core societal institution and thus worthy of research. The goal of this study is to assess the impact of social capital and material wellbeing on life satisfaction within the family. We use a unique dataset from Italy, designed by a multidisciplinary research team, which allows us to directly assess family wellbeing and indirectly assess social capital characteristics. Further, as compared to the standard measure of life satisfaction which asks “all things considered, how satisfied are you with your life as a whole these days?” our information about life satisfaction is delimited to the domain of relational wellbeing within the family unit (Helliwell and Barrington-Leigh 2010; Donati 2003, 2007).

Social capital can be defined as the “…features of social life—social networks, associated norms and trustworthiness—that enable participants to act together more effectively to pursue shared objectives” (Putnam 1995: 664–665; see also Durlauf and Fafchamps 2005; Franke 2005; Schookner 2002; Glaeser et al. 2002; Van Der Gaag and Snijders 2005). While there are competing definitions of social capital, a common thread by way of an acknowledgement of social relations through network interactions is always present. Reimer et al. (2008), for example, describe social capital as “…the social networks and their associated norms that may facilitate various types of collective action” (page 258). Coleman (1988) describes community social capital as the social relationships that exist among people and the relationships they have with institutions in the community.

An important distinction within the social capital literature lies in the difference between “bonding” (within group interaction) and “bridging” (interactions which reach outside the group) social capital. It is this distinction that has resulted in a growing understanding that not all social capital may be positive. In particular, if individuals’ interactions and trust in others remain within a fairly narrow range of contacts with similar individuals (bonding), increased levels of social capital may actually serve to increase distrust between groups, thereby exacerbating inequities, constraining actions and increasing social exclusion (Babacan and Babacan 2007; Cheong 2006; Jaynes 2007; Portes 1998; Wakefield and Poland 2005). For example, bonding on the part of the majority can act to exclude minorities from membership recruitment, which results in lower civic participation and social capital formation on the part of minority group members (Bloemnraad and Ramakrishnan 2006). Bridging social capital is more difficult, requiring people to go outside their comfort zone, but potentially resulting in greater information flows across groups.

Italian sociologists envision the family as a “firm” par excellence producing primary social relations and merit goods that benefit family members, parents, friends and institutions (Donati 2003, 2007; Prandini 2007; Tronca 2007). The extent of the benefit is proportional to the proximity of the relational circle to the family core. These relational goods and services are traditionally seen as positive generating bonding social capital in the form of trust and support within the family, as well as the wider kinship network. The family unit is also engaged in civic activities through participation in associations committed to meeting common social goals. Thus, the level of social capital held by a family unit may differ depending on how it combines the nuclear, kinship or civic dimensions of social capital and in the amount and quality of time invested in weaving together and maintaining nuclear, community or civic ties.

In this study we assess the importance of social capital as a determinant of four dimensions of subjective wellbeing within the family: satisfaction with the rules adopted in the decision process, the marital relationship, the parental relationship with children, and with the time spent with the children. We operationalize the different definitions of social capital by distinguishing three main dimensions: membership or degree of civic participation capturing the degree of social inclusion of the family, trust of people with both family and kinship ties describing a Putnam type of network capital, interactions with others of both the bonding and bridging type, and the degree of altruism as expressed by participation in public petitions and charitable donations. As a proxy for the intensity of the interactions with others we use information about the amount of time shared with family members during breakfast, the time invested in talking with each other about personal problems and the amount of time spent for helping family members in accomplishing their daily tasks. Other controls include income, the location in the macro-regions of Italy, level of education, marital status, and family composition. We find that while socioeconomic characteristics and income in general do not have an impact on family wellbeing, social capital matters for family life satisfaction with different facets depending on the dimension of subjective wellbeing of interest.

We first revisit the existing literature in order to derive an informed specification of our working model presented in Sect. 3. The subsequent section presents the data and empirical method used in the paper. Section 5 discusses the results and derives policy implications that are summarized in the conclusive section.

2 The Literature

Social capital is an important dimension of social inclusion. An individual who lacks relationships and interactions within or outside the family is likely to be socially excluded thus aggravating the disadvantage, particularly if economically deprived (Berghman 1995; Bradshaw et al. 2004; Burchardt et al. 2002; Robila 2006; Sirovátka and Mareš 2006). From a policy perspective social inclusion presumes that a fundamental objective of society is to enable all members to participate fully as valued, respected and contributing members (United Nations 2005). It describes a means by which everyone, regardless of their experiences and circumstances, may gain fair access to the key social and economic resources required to achieve their potential, including access to health and education services and employment, available within a family or a community (Atkinson et al. 2002; Levitas 1998, 2003; Monnickendam and Berman 2008; Siemiatycki 2005; Toye and Infanti 2004). Thus an individual who is rich in physical capital, but poor both in the quantity and quality of social interactions may not be satisfied with life.

Much of the literature on family wellbeing, inclusion and social capital is driven by researchers in social work and sociology. These studies are often qualitative in nature and point to differences across family types, often highlighting the place of women and single parent households (see for example Castillo and Fenzi-Crossman 2010; Edwards 2004; Furstenberg 2005; Stone 2001; Stone and Hughes 2002). These researchers assess the dynamics of family structure on social capital formation and wellbeing. Furstenberg (2005), for example, looks at the prerequisites for assessing how social capital is accumulated across family structures. In particular, he argues that children quickly become players in the accumulation of social capital. Thus children can act as the go-betweens in more isolated communities (i.e.: recent immigrants) creating ties with the mainstream social worlds through school and friendship networks. Others (Castillo and Fenzi-Crossman 2010; Donati 2010; Hogan 2001) look at the impact of family structure, interactions within the family and networks on outcomes for children. Looking at interactions, Hogan finds that involving fathers in children’s activities has a positive impact on children’s schooling. Donati (2010) observes that the presence of children generates both costs and benefits, mainly of a relational nature. Such bonding capital stemming from the presence of children contributes significantly to the level of subjective wellbeing of the household. Castillo and Fenzi-Crossman (2010) assess the relationship of non-marital fathers’ informal social networks and social capital of children. Similar to Hogan, they find that while informal social networks have a positive impact for children, formal networks do not.

There is limited economic or quantitative research linking family structure to social capital or wellbeing. Ravenera and Rajulton (2010), for example, look at social capital by family structure, arguing that the type of family (i.e. two parents and children versus single parent families) can affect the size of networks or interactions, which in turn can have an imopact on total social capital and wellbeing.

Helliwell and Barrington-Leigh (2010) define subjective wellbeing as divided into three main categories, which tap into human experience in different ways—life evaluations, positive emotions and negative emotions. Bartolini et al. (2013), taking a more economistic perspective, argue that subjective wellbeing is largely explained by four forces acting in different directions: income growth, decreasing relational goods, decreasing confidence in institutions, and social comparisons. These four groups of variables, they contend, allow us to explain the whole variation in subjective wellbeing.

White (2009) ties the three types of wellbeing together, arguing that they are essentially related within a triangle associating subjective, material and relational wellbeing. Within this construct, material wellbeing is associated with food, shelter, and economic items. Relational wellbeing involves social interactions, and subjective wellbeing concerns people’s perceptions of their situation. Relational wellbeing is most closely related to the interactions central to social capital, whereas material wellbeing is a more traditional assessment of wellbeing within the economic literature. Subjective wellbeing is most closely tied to the ideas espoused by Helliwell and Barrington-Leigh (2010).

3 Data and Method

The above review suggests that subjective wellbeing is multidimensional and not solely based on income. Rather, subjective wellbeing is a product of material wellbeing, social capital, and personal circumstance. Our goal is to parse apart these dimensions of wellbeing in order to determine the degree to which they are distinct and additive.

Our data are drawn from a survey sponsored by the Italian International Center of Family Studies (CISF). This nationwide survey was conducted in 2009 using computer assisted telephone interviews by Coesis Research.Footnote 1 The sample of 4,017 interviews is a representative sample of Italian families from the universe of households with land-based or cellular telephone service. The survey was designed to by a multidisciplinary group of sociologists, psychologists and economists study the material and subjective wellbeing of Italian households and the cost of children. Survey modules cover information on a wide range of topics including family background, disposable household income, social capital, and satisfaction with many aspects of family life.

Satisfaction with family life is assessed using four indicators. First, a measure assesses how satisfied a respondent is with the way decisions are made in her/his family. The second indicator of family life satisfaction measures satisfaction with the respondent’s partner relationship. The third indicator measures how satisfied the respondent is with the relationship with children. The fourth indicator assesses how satisfied the respondent is with the time that s/he and children spend together.

Given the aim of our study, we include only observations in which the respondent is a member of a family, either with or without children. Specifically, when analysing satisfaction with the way family decisions are made we only look at households comprising couples with and without children or an adult with children (sample size is 1,955 observations). When assessing satisfaction with the relationship with children and with the time spent with children, we include only households with children (sample size is 1,835 observations). When assessing the level of satisfaction with the partner we include only couples regardless of the presence of children (sample size 1,554 observations).Footnote 2 Table 1 shows the distributions of the four measures of life satisfaction. In the Appendix we report a full description of the associated questions asked in the survey.

Table 1 Descriptive statistics of satisfaction with family life, socioeconomic characteristics and social capital
Full size table

Social capital attributes are measured at the household levelFootnote 3 and include membership in civic organizations, as well as trust and interactions with others. The set membership in civic organizations taps inclusion or degree of family isolation and comprises information about (1) whether an individual within the family is a member of one or more associations, (2) whether they participate in voluntary activities, (3) whether individuals participate in social activities, and (4) whether they participate in neighbourhood meetings. The variables capturing trust are (1) trust in family members, (2) trust in friends, and (3) generalized trust. The set interactions with others taps relational well-being within both bonding and bridging realms and comprises (1) mutual help from friends, (2) help from family members, (3) breakfast time spent with family, (4) time spent with family members to talk through problems, (5) time spent for helping family members in their daily tasks, (6) participation to public petition, and (7) donation of money (altruism).Footnote 4

As pointed out by Ferrer-i-Carbonell and Frijters (2004) psychologists and sociologists describe responses to self-reported happiness or life satisfaction or subjective wellbeing surveys as cardinal and comparable, thus allowing the estimation of OLS regressions on levels of happiness. Cardinality implies that the relative difference between satisfaction responses is clearly perceived by all individuals and is linked to the relative difference in welfare (W i  − W j ) = δ(LS i  − LS j ) by a positive function δ known up to a multiplicative constant, where W is the welfare function and LS is life satisfaction of individuals i and j. On the other hand, economists maintain that individuals have a common opinion of what subjective wellbeing is recognizing that if LS i  > LS j , then W i  > W j . Therefore, they only assume interpersonal ordinal comparability and adopt ordered latent response models. In the case of single cross-sections, the authors report that the two econometric methods should not matter for the results.

We study the relationship between family life satisfaction (subjective wellbeing), socioeconomic status and social capital characteristics by estimating the following reduced equation model using ordinary least squares

$$LS_{i} =\upalpha +\upbeta{\kern 1pt} X_{i} +\upgamma{\kern 1pt} SC_{i} +\upvarepsilon_{i}$$
(1)

where LS i is the level of an individual’s satisfaction i in the four different dimensions of family life, and α, β, γ are parameters to be estimated. The vector X i refers to a set of socioeconomic characteristics that may affect the dependent variable LS i . The vector X i comprises variables at the individual and household level, such as age, gender, marital status, education, presence of children by age groups, and quintiles of household equivalent income. The vector SC i comprises variables related to the social capital of the household, and ε i is the latent error term capturing unobservable heterogeneity.

We estimate two types of regression models differing for the variables used as controls. We first estimate a baseline model controlling only for socioeconomic variables and then a model adding social capital characteristics of the family. In doing so, we can test the explanatory power of social capital in explaining the four measures of satisfaction with family life.Footnote 5

The four dependent variables are coded as 0, when the individual is not satisfied, and 10, when the individual is fully satisfied. Table 1 provides descriptive information on the distributions of the life satisfaction indicators. The majority of respondents report being fully satisfied with all four aspects of family life: 87 % of the respondents are fully satisfied with the way family decisions are made, 91 % of the respondents are fully satisfied both with the relationship with their partner and their children, while 75 % of the respondents are fully satisfied with the time spent with their children. These figures are in line with expectations because the sample is comprised of randomly selected households where the proportion of families deprived under both material and relational dimensions is likely to be negligible.

Table 2 provides definitions and descriptive statistics of the socioeconomic variables used in the empirical analysis.Footnote 6 We aggregate Italy in five macro-regions (North West, North East, Centre, South and Islands) and whether the respondent lives in an urban area or non-urban. We control for a wide number of respondent characteristics and household attributes including age, gender, education, marital status, presence of children and number of income earners. The average respondent is 49 years old and about 74 % of respondents are female. In our sample 70 % of households are single-earner families where the breadwinner is likelihood to be the husband. This may explain the overrepresentation of female respondents. Marital status is grouped in married, unmarried, separated/divorced and widowed. Because of the selection criteria, almost 4 fifths of respondents are married and only a small percentage are unmarried or divorced, 8 and 6 % respectively.

Table 2 Summary statistics of socioeconomic characteristics
Full size table

We control for the presence of children using three dichotomous indicator variables equal to one if there is at least one child aged 0–5 (child 0–5), if there is at least one child aged 6–14 (child 6–14), and if there is at least one child 15–17 (child 15–17). The set of socioeconomic variables also includes education of both the respondent and her/his partner when present. Looking at Table 2, we note that there is not much variation between the educational attainment pattern of the respondent and her/his partner. For instance, for many respondents and their partners the highest level of schooling attended is high school, 42 and 32 % respectively. Whereas 14 % of respondents have completed university, the corresponding proportion for partners is 10 %. In the empirical analysis, we control for household equivalent income aggregated into quintiles.Footnote 7 Equivalent income is thus modelled by five dichotomous indicator variables. The first quintile corresponds to an equivalent monthly income between €185 and €649, the second quintile between €650 and €810, the third quintile between €811 and €1,017, the fourth quintile between €1,018 and €1,287, and the last quintile if the equivalent income is above €1,287. The reference category is individuals who belong to the first (poorest) quintile of household equivalent income.

As discussed earlier, social capital characteristics are assessed using fourteen indicators grouped into three classes: membership, trust and interactions with others. In our sample, 17 % of respondents are members of one or more associations, 3 % participate in voluntary family activities, 20 % in social activities and 16 % have participated in neighbourhood meetings over the last year (Table 3). Looking at the table, we see that on average respondents have high levels of trust for her/his family members, but trust decreases both for friends and for people in general.

Table 3 Summary statistics of social capital variables
Full size table

Interactions with others are assessed using seven variables. We use variables tapping the propensity to offer or receive help by family members or friends, and how much time the respondent and her/his family spend together during breakfast and dealing with family challenges. On average, respondents are moderately satisfied with their interactions with others.Footnote 8 These interactions are characterised as bonding activities—they are largely within group. Bridging activity crosses traditional groups and allows for interactions with people outside the immediate circle of friends and relatives. We assess bridging activity through questions on networks and by assessing ‘softer’ bridging activity such as information about participation to public petitions and charitable giving. Over the last year, 75 % of respondents donated money and 18 % signed a public petition.

4 Results

We begin by estimating a baseline version of Eq. (1) without social capital and then turn to models that include also social capital. Table 4 presents results from regressions assessing four aspects of satisfaction with family life—satisfaction with the way family decisions are made, satisfaction with partner, satisfaction with children and satisfaction with time spent with children—without controlling for social capital.Footnote 9

Table 4 Satisfaction with life and socioeconomic variables in italy
Full size table

Overall the explanatory power of these models is not high with adjusted R2 values of between 0.014 and 0.036. However, the models do point to some interesting findings. As can be seen, macro-region has no significant impact on any of the measures of satisfaction. In line with previous research on the economics of happiness (see for example, Bartolini et al. 2013) our estimates show the existence of a U-shaped relationship between age and satisfaction with family life. This relationship has a significant impact for satisfaction with relationship and time with children. As compared to males, female respondents display greater satisfaction with time spent with children but gender does not play a role in our other models. Marital status does impact satisfaction in a number of models. As compared to being married, being unmarried is negatively correlated with satisfaction with family decisions, though statistically significant at only 10 %. People who are divorced are less satisfied with both the relations with children and time spent with them (coefficients of −0.54 and −0.65 respectively). Those who are widowed are also less satisfied with time spent with children (coefficient of −0.75).

As compared to having no young children in the home, having young children increases satisfaction in all the models by about 0.20 points. However, having older children does not have a significant impact on satisfaction in any of the models.

Socioeconomic indicators, usually relevant determinants of life satisfaction (Clark et al. 2008; Layard et al. 2012), do not have a strong impact on satisfaction with the family. Level of schooling has no significant impact on satisfaction in any of our models. However, having two income earners (as compared to one or no earners) in the household results in significantly lower levels of satisfaction with time spent with children (coefficient of −0.32). In the same way, income does not appear to be a driver of satisfaction. The one exception is with satisfaction with spouse. As compared to being in the bottom income quintile, being in the second or top income quintile results in a significantly lower level of satisfaction with spouse (coefficients of −0.19 and −0.33 respectively).

There is an important distinction to keep in mind when comparing our outcomes with the results of other happiness studies. Our life satisfaction questions are not overall measures of the type “In general, how happy would you say you are? Or how satisfied are you with your life as a whole?” Rather, they ask about the level of satisfaction specific to the relational dimension of subjective wellbeing associated with intra-family interactions within the couple and between parents and children. Because our indicators of subjective wellbeing are confined to non-material aspects, it is not surprising to find that the wellbeing derived from intra-family relationships does not depend on absolute incomes. Based on our evidence, we may answer Easterlin’s question (1974) asking whether, at a point in time within a society, the wealthy are on average happier than the poor by asserting that, for Italian society, wealthy households are not happier than poor households in terms of relational wellbeing.

Focusing on satisfaction with relationships between partners reveals that being very poor, (that is in the lowest quintile of the income distribution), or being very rich does significantly affect the partnership. Having too little or too much money is a relevant determinant of the fragility of a couple. The policy implications of this fact are of course different. While for households in poverty it may be a society’s responsibility to remove such circumstances, this is not the case for rich households.

What we take from this initial model is that the standard human capital characteristics such as education and income are not drivers of satisfaction with family. Indeed, with few exceptions, human capital does not impact subjective wellbeing.

In order to test this further, models in Table 5 add characteristics associated with social capital—membership in organizations, trust, and interactions with others. The first thing of note is that adding these social capital characteristics greatly increases the explanatory power of the models, which now range from an adjusted R2 value of 0.137–0.204. In particular, the explanatory power of the model concerning satisfaction with how decisions are made increases almost eight fold (from 0.014 to 0.204).

Table 5 Satisfaction with life, socioeconomic variables and social capital in italy
Full size table

The second finding of note is the fact that effects of the previously included variables are remarkably stable. With few exceptions, the impact of demographic and socioeconomic characteristics on our measures of satisfaction is the same in both magnitude and level of significance. This combined with the increase in explanatory power of the models suggests that social capital characteristics (trust, interaction and participation) have strong impacts on subjective family wellbeing.

Turning to the social capital variables, we see trust in family members has a strong positive impact across all models (ranging from a coefficient of 0.22–0.39). As is to be expected, trust in family members has the strongest impact on satisfaction with spouse. Trust in friends and the ability to get help from friends have a small but significant effect on satisfaction with the family (coefficient of 0.072 and 0.029 respectively) but do not really impact satisfaction with spouse or how decisions are made. The ability to get help from friends does have a positive impact on satisfaction with time spent with children (coefficient of 0.069) suggesting that family members may be able to draw on support from friends thereby increasing satisfaction with time spent with family.

The ability to get help from family members and time spent eating breakfast with family members does not appear to have a substantive impact on subjective family wellbeing. However ability to get help from within the family has a small (but significant) impact on overall satisfaction. The ability to talk through problems does however have a significant impact on all the measures of subject family wellbeing (ranging from 0.053 to 0.111) suggesting that open communication is a pathway to familial wellbeing. Somewhat surprisingly, help from family members outside the household is not a strong predictor of family wellbeing.

The preceding variables are often viewed as bonding characteristics—these characteristics bring within group members closer together. We also tested five measures of social capital related to membership and participation that are often seen as bridging (making connections outside the group). With the exception of membership in an organization, which is strongly negative across all variables, other bridging activity has no impact on subjective family wellbeing. The impact of membership is however quite powerful reducing for example overall satisfaction with the family by −0.22 points for every increase in the scale. This suggests that membership in organizations, rather than adding a dimension to family relations, is seen as taking time away from the family.

5 Conclusions

In this study we have assessed the impact of social capital and two forms of wellbeing (material and relational) on subjective wellbeing within the family. A distinctive feature of our study is the ability to separate out the impact of relational dimension from subjective wellbeing associated with intra-family interactions.

Using this approach allows us to assess the impact of the material and non-material aspects of wellbeing on subjective wellbeing and ask if within the family, at least, “Does money buy happiness?” Based on our evidence, we may qualify the answer to Easterlin’s question (1974) asking whether the wealthy are, in general, happier than the poor by asserting that, for the Italian society, wealthier households are not happier than poor households as far as intra-family relations are concerned. In the same way, higher levels of schooling do not buy happiness—subjective wellbeing for more educated families are no higher than for families with lower levels of schooling. This evidence may be partly explained by the fact that Italian households are endowed with comparable stocks of human capital.

Income, on the other hand, is a relevant determinant of the fragility of the Italian couple. But the effect is bimodal. Satisfaction with relationships between partners is significantly worse for very poor or very rich households. The policy implications of this fact are of course different. While for households in poverty it may be a society’s responsibility to remove such circumstances, this may not be the case for rich households. The causes behind the fragility of the very poor and very rich are likely different and should be analysed in a dedicated study if it is of social interest to undertake effective prevention policies.

Finally, it appears that family wellbeing is all about bonding as compared to bridging social capital. Interactions within the family (both nuclear and extended) are important for wellbeing, but interactions outside the family do not appear to have an impact.

Our results are especially relevant if read through the lenses of the recent recession. Economic crises not only contract the material dimension of wellbeing, but also may critically weaken both the quality of relations and the strength of bridging and bonding ties. In periods of economic recession, the private and public investment on social capital may be a successful copying strategy because of the low economic costs associated with the investment on proximity relations and reciprocity activities. On the other hand, families that are more endowed with social capital may have higher resiliency.