European feelings of deprivation amidst the financial crisis

Abstract

As European governments have embraced the credo of austerity, the perennial discussion whether welfare states erode the quality of social networks has taken on a more prominent position on political and social science research agendas. While non-believers of this so-called ‘crowding out’ thesis argue that social networks flourish well in welfare states, believers argue that welfare provisions render social networks irrelevant in mobilizing resources. Using the 2010 wave of the European Social Survey, we analyse the extent to which both the welfare state and social networks have prevented deprivation, as well as the extent to which the functional quality of social networks in inhibiting impoverishment differs as a function of welfare state generosity. Both the ‘crowding out’ and the ‘crowding in’ theses are supported: resources are less mobilized through networks in more generous welfare states precisely because encompassing welfare provisions reduce deprivation significantly, lowering the functional quality of social networks.

Keywords

crowding-out hypothesis European Social Survey multi-level analysis social capital welfare state worldwide financial crisis

Introduction

Whether welfare states erode the quality of informal social networks – ties between people as distinct from formal networks of voluntary involvement in organizations (Pichler and Wallace, 2007) – is a recurrent theme of social science research agendas. The topic gained renewed importance as a result of the consequences of the worldwide recession initiated in 2008. The economic crisis has caused national governments across the European continent to consider and eventually implement welfare retrenchment. Yet, in order to comprehend the unintended consequences of such retrenchment programmes, an important and largely unsolved puzzle is whether such informal bonds between people can replace the role of decreasing welfare provisions in combating poverty. Put differently, if welfare states are dismantled, to what extent would informal social networks be able to prevent socio-economic deprivation?

Academic discussions on this issue have been framed mainly on communitarian and neo-liberal inspired critiques of generous welfare programmes, the argument being that encompassing social benefits and services ‘crowd out’ social networks that otherwise would have remedied social needs (Etzioni, 1995; Fukuyama, 2001). Fukuyama (2001: 18) explicitly opposes state intervention in the form of welfare provisions as ‘activities that are better left to the private sector or to civil society’. Alternatively, opponents of this ‘crowding out’ thesis reject the idea that residents of more developed welfare states are socially isolated, and suggest the ‘crowding in’ argument, which posits that welfare state effort actually redistributes resources (van Oorschot and Arts, 2005; van Ingen and van der Meer, 2011) and facilitates conditions that are necessary for informal social networks to flourish (Kumlin and Rothstein, 2005; Rothstein and Stolle, 2008).

Present debates on the relative strength of welfare effort and informal social networks are obscured due to the fact that both proponents and opponents of the ‘crowding out’ thesis make reference to different aspects of informal social networks conflating the size (how well connected individuals are) and quality (the extent to which connections embed resources) of informal social networks (Flap and Volker, 2001: 300). Based on cross-national research, proponents of the ‘crowding in’ thesis point to the positive effects of welfare provisions on the size of social networks and the frequency of informal social contacts; proponents of the ‘crowding out’ thesis, on the other hand, are mainly concerned with the quality of social networks, in particular with the degree to which they are functional for people coping with their needs irrespective of the role of the welfare state. The aim of this article is to bridge these two views by departing from the instrumental account of social capital, described as resources that accrue from social ties between people (Bourdieu, 1986; Coleman, 1988; Lin, 1999; Flap and Volker, 2001). We combine the insight that network size may differ across European welfare states with the question whether cross-national variation in the functional quality of networks reflects differences in European welfare state spending. In other words, when resources are embedded in networks there is the possibility that they are more likely to be mobilized in weaker welfare states for the reason that, in larger welfare states, encompassing welfare provisions are already effective in preventing deprivation.

Clearly, the question whether informal social ties can take over the role of the welfare state cannot readily be answered in an empirical and direct fashion by the lack of comparative longitudinal or panel data for a time period that would be sufficiently long to have variation in both welfare state effort and informal social network characteristics. Alternatively, it is possible to provide new insights from a cross-sectional perspective by analysing the role of informal social networks relative to the size of the welfare state in preventing deprivation. In particular, in current recession the opportunity arises for a detailed analysis of this issue, as citizens across the continent bear the consequences of the crisis. In the absence of a comparative longer-term longitudinal or panel study, the 2010 European Social Survey, fielded two years after the collapse of Lehman Brothers, offers a unique and interesting alternative with which to analyse the relationships among welfare state effort, informal social networks and socio-economic deprivation. In this 2010 snapshot, respondents of European countries have been asked retrospectively about having been financially deprived since the outbreak of the crisis in 2008. The data allow us to assess the extent to which people who have less or more access to social networks have seen their financial situation degrade, if at all. By performing multi-level analysis, the leading questions in this comparative cross-national strategy are: (1) what is the relation between having frequent informal social contacts and the experience of deprivation?, (2) what is the relation between welfare state effort and self-reported deprivation?, and (3) does the relation between having informal social networks and the experience of deprivation depend upon welfare state effort?

Literature review

Quality of informal social ties

Whether welfare state intervention makes informal social networks¹ (i.e. ‘having contact to various groups of people […] and the intensivity […] of encounters within those groups’ (Pichler and Wallace, 2009: 321)) irrelevant or ‘crowds them out’ is a recurrent theme in social science research (Day and Devlin, 1996; Andreoni and Payne, 2003; van Oorschot and Arts, 2005; van Ingen and van der Meer, 2011; Stadelmann-Steffen, 2011). Despite research outcomes from various angles, this puzzle can only adequately be addressed if social networks, often referred to as ‘social capital’, are approached from an instrumental or functional account. In his seminal contribution, Bourdieu (1986: 248) described social capital as ‘the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition’. This instrumental approach towards social capital clearly reconciles the ‘social’ aspect of it – how well connected individuals are – with the ‘capital’ aspect – whether networks embed resources. A series of empirical studies evidenced that being well connected indeed leads to a number of positive externalities: for instance, the odds of finding a job are a function of network density, because job offers tend to spread through social networks (Granovetter, 1973; Flap and Boxman, 2000); also, being involved in networks has positive effects for subjective well-being (Helliwell and Putnam, 2004) and health (Kawachi et al., 1999).

In the socio-economic realm in particular, studies have aimed at disentangling the extent to which resources are mobilized through networks, i.e. what the relationship is between informal social networks and the prevalence of deprivation.² Not unsurprisingly, many of these outcomes originate from an interdisciplinary approach that connects students of development research with social capital theorists (Woolcock, 1998; Narayan, 1999; Fukuyama, 2001; Saegert et al., 2001), as they question how poverty can be combated effectively in societies where state-regulated welfare provisions are mostly absent. Although these studies have taken place in different institutional and cultural contexts, they converge in the finding that having extensive and intensive informal social contacts generates an important set of resource assets for individuals to rely on, as they are effective in reducing deprivation and socio-economic stress.

While the negative association between informal social networks and deprivation seems to be less of an object of dispute, the causal mechanisms are debated much more (Cattell, 2001), with the issue being whether network size and density are more important or less important than network composition (e.g. Woolcock, 2002). Illuminating in this respect is the seminal dichotomy of bonding or homogeneous vs. bridging or heterogeneous networks (Putman, 2000; Narayan and Cassidy, 2001; Paxton, 2002).³ Whereas the former refers to ties between similar people and usually involves rather intense and frequent interactions (e.g. family or friendship networks), bridging ties are bonds cutting across social cleavages of, for example, social class, ideological dispositions or race, usually involving less intense and frequent contacts. Weighing the relative importance of bonding and bridging networks, studies tend to show that both are beneficial in reducing deprivation. Often, this is exemplified with the argument that bonding networks help people ‘get by’, while bridging networks facilitate ‘getting ahead’ (de Souza Briggs, 1998; Narayan, 1999; Putnam, 2000).

Apart from the difference between bonding and bridging ties, an additional qualification is the fact that access to informal social networks is unequally distributed across social classes (Finsveen and van Oorschot, 2009; van Ingen and van der Meer, 2011) – a phenomenon Pichler and Wallace (2009) explicitly connect with Bourdieu (1986). Whereas poorer people having stronger bonds with family and friends (bonding social capital) are able to draw upon network resources of social support in cases of emergency, they are less likely to participate in more heterogeneous or bridging social ties that would enable them to escape their situation (Narayan, 1999).⁴ These two qualifications, namely the difference between bridging and bonding social ties and inequalities in informal social network access, are necessary if we are to comprehend how social bonds between people bring about socio-economic resources.

Welfare states and the quality of informal social ties

On the basis of this brief literature review on the quality of informal social networks – indicating that they are functional in relieving hardship – the neo-liberal and communitarian critique that the erosion of people’s social bonds has a bearing on socio-economic consequences needs to be taken seriously. Introducing the welfare state in explaining the link between informal social networks and socio-economic resources, a first stream of studies has approached this puzzle from the perspective of the size of social networks, leaving the resources aspect largely untouched. These studies generally come to the conclusion that the welfare state does not ‘crowd out’ social networks (van Oorschot and Arts, 2005; van Oorschot et al., 2006; Norris and Davis, 2007; Pichler and Wallace, 2007). With the exception of family ties, which are more common in Southern welfare states, informal social ties tend to be more common in the more developed welfare states of Europe than in the least developed. Hence, welfare states seem rather to ‘crowd in’ social networks than to crowd them out.

As for the mechanisms behind this ‘crowding in’ effect, one is that welfare states provide structural conditions that lubricate social bonds (van Oorschot and Arts, 2005; van Ingen and van der Meer, 2011). As welfare states actively redistribute resources, especially income and time, people are better able to sustain informal social contacts. Another mechanism implies that universalist social policies, which are more widespread in generous welfare states, embody the norms of solidarity, equality and impartiality that foster pro-social attitudes, which in turn facilitate social cooperation (Kumlin and Rothstein, 2005; Rothstein and Stolle, 2008). Empirical studies, for instance, show that interpersonal trust is greater among people living in more comprehensive welfare states (van Oorschot and Arts, 2005; Kaariainen and Lehtonen, 2006; Larsen, 2007).

Alternatively, research on volunteering – an activity that, according to Stadelmann-Steffen (2011), is in direct competition to social welfare provisions – shows that unpaid voluntary work is more common in less generous welfare states, leveraging the ‘crowding out’ thesis. Yet, Stadelmann-Steffen (2011) additionally discovered that lower income groups – groups that benefit most from social welfare provisions – are more likely to volunteer in more generous welfare states. The results on volunteering suggest that ‘crowding in’ and ‘crowding out’ can be considered as two sides of the same coin: whereas welfare provisions ‘crowd out’ social bonds of the higher social strata, they ‘crowd in’ network capacities of lower social groups that are most dependent on social welfare provision.

Yet, looking just at the size of informal social ties in different welfare states leaves the issue of their quality largely untouched, i.e. whether under varying degrees of welfare provisions informal social ties vary in bringing about socio-economic resources. The scarce evidence on this issue is not conclusive. Gelissen et al. (2012) show that in higher developed welfare states there is a tendency for people to say they have more access to informal care and services, e.g. being able to borrow money from their network members. However, despite reference to possible access to financial resources, the study does not explicitly focus on any resulting relief from socio-economic deprivation.

Alternatively, one cross-national study has sought to explain the effect of welfare state characteristics on the relationship between social networks and deprivation using the 2003 Quality of Life Survey (Böhnke, 2008). In this comparative research, Böhnke selects a subsample of poor respondents and analyses which contextual features impact on their social disintegration (which was operationalized by a scale comprising several dimensions: frequency of social contacts with friends and neighbours, the availability of support in emergencies, and the evaluation of social integration). While she found that social isolation and deprivation were positively correlated, she also discovered that public images of poverty and religious traditions contribute to explaining cross-national differences in the social isolation of the poor, while this is not so in the case of welfare state generosity and welfare-state regime type. Nevertheless, despite no clear relation between welfare state expenditure and social integration of the poor, she explicitly appeals to the crowding-out thesis by arguing that ‘being poor in the richest countries of the EU, which have the highest level of social protection, also translates to a large extent into social exclusion’ (Böhnke, 2008: 147).

Hence, based on the scarce evidence, recent research outcomes suggest that the literature can be qualified in several ways. First of all, in line with the instrumental account to social networks, to analyse their functional quality, apparent access to resources needs to be considered as the outcome, which contrasts with Böhnke (2008), who used income as a criterion to select a subsample of the poor and analyse their social integration, and Gelissen et al. (2012), who explained variation in network-based access to forms of informal help. This signifies that the present economic shock, as Europeans are hit hard by the worldwide crisis, provides a unique opportunity to analyse how resources – or the lack of them – accrue from vertical networks (welfare provisions) and from horizontal networks (informal social bonds). Second, recent innovations in statistical modelling facilitate the analysis across the range of least connected to most connected individuals, making it feasible to test whether there are differences between countries regarding the extent that informal social networks have prevented deprivation from occurring. If there are cross-national differences, it can be assessed whether they are a function of welfare state effort.

Despite the absence of conclusive evidence on the relation between welfare effort, informal social networks and access to socio-economic resources, we expect the following relationships: (1) informal social contacts reduce the experience of deprivation; (2) the experience of deprivation is lower in more developed welfare states; (3) because the experience of deprivation is lower in more developed welfare states, residents of less developed welfare states are more likely to mobilize their networks to prevent deprivation occurring. In other words, in less developed welfare states the negative effect of social networks on deprivation will be stronger than in more encompassing welfare states.

Data and methodology

We analysed the 2010 European Social Survey (ESS Round 5, 2011) in order to disentangle the relationship between people’s informal social ties and access to socio-economic resources across different European welfare states. This comparative biennial survey project was carried out in more than 25 countries, but data availability meant that we include only 22 in our analysis, adding up to 40,246 respondents (or on average 1,829 respondents per country).⁵

Dependent variable

In this manuscript, access to socio-economic resources serves as dependent variable and is proxied by self-reported socio-economic deprivation since the outbreak of the 2008 financial crisis. This variable is a means scale composed of three items tapping the extent to which the respondent, in the past three years, (1) had to manage on a lower household income, (2) had to draw on savings or get into debt to cover ordinary living expenses, (3) had to cut back on holidays or new household equipment. These three measures have been offered with a seven-point response scale ranging from 0 (‘not at all’) to 6 (‘a great deal’). According to conventional scaling techniques (see Appendix Table A1),⁶ these three items form a unidimensional scale that meets statistical tests. According to our insights, it is the first time that this variable has been analysed in cross-national research. As previous research efforts using this variable are absent, this article might also serve as a first test of the validity of this scale of socio-economic deprivation since the outbreak of the financial crisis.

Independent variables

In order to test whether informal networks’ access to resources is lower in more extensive welfare states, our indicator for informal social ties is the survey item, ‘How often do you meet socially with friends, relatives or work colleagues?’, which was offered with a seven-category response scale ranging from 1 (‘never’) to 7 (‘every day’). This survey question has been the object of several studies on social capital (Pichler and Wallace, 2007, 2009; Böhnke, 2008). On the one hand, we need to be aware that this single ESS item for informal social bonds is rather crude, since it does not measure bonds with the various network agents of friends, relatives and work colleagues separately. This makes it difficult to relate research outcomes with the distinction between bonding and bridging ties, a typology of crucial importance in the understanding of the networks as providers of resources (de Souza Briggs, 1998; Narayan, 1999). On the other hand, as this measurement encompasses various network agents it nonetheless serves as a relevant indicator for network access, implying that our analyses add to existing studies that used an index comprising this item along with others (Böhnke, 2008). In addition, to weigh the validity of the ESS social networks indicator, we have cross-validated it with two other components of social capital present in previous ESS waves, i.e. associational involvement and social trust (see Appendix), showing as expected that the three are positively interlinked. So, even though our measurement is fairly general, it gives an insight into how well connected individuals are.

The second independent variable is welfare state generosity, which is operationalized by social expenditure per head of the population (in 1,000s euro) for 2009, as the Eurostat (2012) database did not yet have information available for 2010, which was the year the ESS was fielded.⁷

Control variables

The effect of informal social contacts, welfare state generosity and their interaction on self-reported deprivation will be controlled for relevant variables. As our measure of self-reported deprivation has not been analysed previously, our choice of control variables is inspired by research on perceived socio-economic insecurity (for a review, see Mau et al., 2012). At the individual level, we first control for the curvilinear effect of age, as it can be assumed that middle-aged people in particular are more impoverished as a result of the financial crisis. The mechanism is that up until retirement individuals take a more vulnerable position in the labour market, as employers might find the young workforce more flexible and less costly. As both women and residents of foreign origin are vulnerable groups in the labour market, it can be expected that they have encountered more deprivation than men and native residents. As impoverishment is expected to be a function of a number of socio-economic respondent features, we control for levels of education (ISCED-coded), work status, being or having been long-term unemployed, income level and being welfare-dependent. Work status has been categorized into having paid work (reference category), being unemployed, being a student, being retired or other status. A dummy indicating whether one has been unemployed for longer than three months is included. We include objective income as well, but because this variable is hampered by a high level of non-response we use means substitution, implying that respondents with a missing value (20.0 per cent of the sample) receive the mean value on this variable. A dummy indicating item non-response on the income variable is included in the model as well. The last socio-economic characteristic regards welfare state dependency, asked by whether one’s main source of income is welfare benefits. In addition, as religiosity, proxied by church practice, is linked to social capital (Putnam and Campbell, 2010) and is related to economic insecurity (Immerzeel and van Tubergen, 2013), we include it in our control model.

At the country level, two relevant control variables are added, namely the unemployment rate and economic growth for the year 2010, i.e. the year of surveying, as obtained from the Eurostat database. The correlation between unemployment rates and social expenditure per capita, is –0.52, which is high but does not conflate the results; the correlation between economic growth and social expenditure per head of the population is 0.05. Lastly, the correlation between unemployment rates and economic growth is –0.22.

Methodology

The assumption that deprivation is explained simultaneously by individual and welfare state characteristics (Gelman and Hill, 2006; Hox, 2010) requires the use of multi-level modelling using MLwiN 2.27 (Rasbash et al., 2013). This technique accounts for the clustered nature of the ESS data – individuals within countries – and enables estimating national-level effects on individual outcomes as well as on relationships between dependent and independent variables. As the intra-class correlation is 11.14 per cent,⁸ controlling for the country clustering is necessary. The independent variables have been grand-mean centred. For model specificities, we have added the –LogLikelihood to compare nested models, and R2 coefficients at both levels, calculated on the basis of Snijders and Bosker (1999).

In what follows, we start with simple bivariate descriptives and a straightforward random intercepts model that regresses socio-economic deprivation on social network involvement. In subsequent steps we estimate more advanced random slopes models in which we evaluate whether the effect of social networks on deprivation varies across countries, and, if so, whether such a differential effect can be explained by welfare state generosity.

Results

Bivariate association

Table 1 shows that, across the board, a widespread impoverishment of European societies has not taken place since the outbreak of the financial crisis. The average deprivation scale score is 2.30 (SD of 1.91) and is substantially lower than the mid-scale point of 3. Nevertheless, there is considerable cross-national variation, with residents of Nordic societies being least deprived, whereas residents of Ireland, Bulgaria and Greece top the list of countries where the population report highest impoverishment.

Table 1.

Descriptive statistics of and correlations between deprivation and frequency of social networks across Europe.

		Deprivation		Social networks
Country	N	Mean	Std Dev	Mean	Std Dev	Correlation
Ireland	2,219	3.59	1.55	4.41	1.58	–0.08***
Bulgaria	2,349	3.03	1.51	4.78	1.73	–0.07**
Greece	2,592	3.02	1.80	3.86	1.70	–0.08***
Czech Republic	2,250	2.97	1.83	4.61	1.48	–0.17***
Slovenia	1,321	2.79	1.91	4.59	1.55	–0.21***
Hungary	1,515	2.72	2.01	3.64	1.71	–0.07**
Estonia	1,746	2.64	1.80	4.24	1.59	–0.08**
Slovakia	1,734	2.57	1.84	4.59	1.62	–0.11***
Spain	1,843	2.56	2.03	5.31	1.49	–0.05*
United Kingdom	2,241	2.47	1.98	4.96	1.56	–0.07***
France	1,685	2.40	1.84	5.14	1.45	0.05
ESS	40,246	2.30	1.91	4.84	1.61	–0.14***
Cyprus	1,001	2.28	1.74	4.29	1.65	–0.02
Portugal	1,979	2.27	1.77	5.86	1.55	–0.13***
Poland	1,603	2.22	1.94	4.27	1.56	–0.10***
Finland	1,840	1.86	1.69	5.10	1.37	–0.01
Germany	2,914	1.81	1.86	4.81	1.47	–0.11***
Belgium	1,671	1.54	1.81	5.23	1.36	–0.06*
Switzerland	1,468	1.43	1.71	5.22	1.29	–0.06*
Sweden	1,461	1.39	1.77	5.39	1.38	0.02
Netherlands	1,772	1.34	1.71	5.40	1.27	–0.03
Norway	1,513	1.24	1.59	5.55	1.27	0.04
Denmark	1,531	1.17	1.69	5.47	1.25	0.04

*p < 0.05; **p < 0.01; ***p < 0.001.

Regarding the cross-national distribution of informal social networks, patterns confirm previous studies (e.g. Norris and Davis, 2007; Pichler and Wallace, 2007), with residents of Portugal and the Nordic countries having frequent informal social interactions, whereas residents of Eastern Europe are poorly connected. So, in general, the often discussed North–South divide in social capital (Delhey and Newton, 2005; Norris and Davis, 2007) is supplemented with an exceptional position of the Iberian countries, where many informal contacts with friends and family members take place (Pichler and Wallace, 2007). Related, and not unimportant, the effect parameter of welfare state spending on informal social networks is significant and positive (b = 0.096, se = 0.026), conforming the traditional ‘crowding in’ thesis: the size of networks is larger in more generous welfare states.

Regarding the quality of informal social ties, Table 1 displays the bivariate correlation between informal social networks and self-reported deprivation. In line with the hypothesis, the pooled ESS data show a significant negative correlation of –0.14, implying that better-connected individuals report less impoverishment. The country correlations are mostly negative; with a few exceptions no association is present (the Nordic countries, as well as in The Netherlands, France and Cyprus). Clearly, negative correlations dominate the pattern and are strongest in a number of Eastern European societies where deprivation is high and the frequency of contacts is relatively low; also in Portugal, Belgium and Switzerland, where frequent contacts are common, the association is negative.

In the next step we assess whether the overall relation found is spurious to other relevant variables, and whether the effect also differs in a statistically significant manner across the countries where the ESS has been fielded. In addition, and only if these associations differ significantly across countries, we analyse whether the size of the welfare state explains this differential effect.

Individual level

Table 2 gives the results of analyses in which we regressed experience of deprivation on degree of being connected to informal networks without (Model 1) and with (Model 2) inclusion of our series of control variables. The uncontrolled Model 1 in Table 2 shows a sizeable effect of –0.083, meaning that the difference in having become deprived between those with infrequent social contacts and those with frequent social contacts is almost –0.5, which is about a quarter of the standard deviation in socio-economic deprivation of 1.91. The impact of social networks on preventing deprivation is thus fairly substantial. In Model 2 we can see that this bivariate effect is only very weakly mediated by the control variables, as the effect parameter is still –0.067. In other words, a European who is averagely connected (a score of 0 on the networks scale, as the variable is grand-mean centred), all else being equal, has a value of 2.36 on the deprivation scale (the intercept). Doing the same mathematics for a socially isolated and well-connected European person, we can observe that the isolated individual has a deprivation score of 2.62, while the individual with frequent social contacts has a deprivation score of 2.22. Thus, across the 22 analysed countries of the ESS, the quality of networks cannot be understated, as they reduce self-reported deprivation substantially.

Table 2.

Individual-level determinants of socio-economic deprivation.

	Model 1: Frequency		Model 2: Individual control variables
	Param	Std Err	Param	Std Err
Fixed effects
Intercept	2.243***	0.135	2.363***	0.140
Social networks	−0.083***	0.006	–0.067***	0.006
Age			0.047***	0.003
Age squared			–0.001***	0.000
Woman (Ref: Man)			0.121***	0.017
Foreign origin (Ref: Native)			0.175***	0.024
Education level			–0.009	0.005
Work (Ref: Employed)
Unemployed			0.415***	0.039
Student			–0.204***	0.041
Retired			–0.026	0.037
Other			0.168***	0.031
Been long-term unemployed			0.464***	0.021
Income			–0.180***	0.004
Non-response on income			–0.177***	0.022
Welfare dependent			0.147***	0.030
Religiosity			–0.013*	0.006
Random effects
Individual variance	3.200***	0.023	2.789***	0.020
Country variance	0.401**	0.121	0.303**	0.092
R² individual level	0.50%		13.28%
R² country level	8.62%		30.91%
-2LogLikelihood	161,148.399		155,610.246

*p < 0.05; **p < 0.01; ***p < 0.001. Entries represent the result of two separate multi-level regression models.

Before we proceed with the influence of welfare state generosity on the association between social networks and self-reported deprivation, we briefly review the control. The model confirms existing work on socio-economic insecurity (see, e.g., Mau et al., 2012), i.e. we find that self-reported deprivation since the start of the economic crisis tends to be higher among various types of vulnerable groups. As expected, middle-aged individuals have encountered more impoverishment, as well as women and people of foreign origin. There is no effect of education, meaning that the deprivation is equally experienced among all levels of education, all else equal. Of more interest is the effect of work status, as the unemployed have become significantly more impoverished than those on the labour market, while students noticed less financial drops. Those that have been long-term unemployed also encountered more hardship, as well as those that are welfare dependent. Not surprising either is the finding that the low-income groups in particular have felt a drop in their financial assets. Despite recent evidence on the link between economic insecurity and religiosity (Immerzeel and van Tubergen, 2013), we find no association.

Country level

The final step of the analysis is to disentangle the influence of the national context on individual differences in self-reported deprivation, as well as its interaction with informal networks. The results of our analyses are presented in Table 3, where all models are controlled for the individual level covariates of Model 2 in Table 2. In a first step, only a random slopes model for the effect of informal social contacts on self-reported deprivation is given (Model 3). The relevant information of this model can be found in the ‘random effects’ part. The coefficients show a significant slope variation (0.004) combined with a significant negative co-variation between the country slopes and the country intercepts (–0.022), meaning that in countries where deprivation is higher the effect of social networks on self-reported deprivation is more negative than in countries where deprivation is low. In other words, in more impoverished societies networks are more likely to be mobilized in order to prevent deprivation than in societies where experienced deprivation is lower.

Table 3.

National-level determinants of socio-economic deprivation.

	Model 3: Random slopes		Model 4: Country covariates		Model 5: Cross-level interaction
	Param	Std Err	Param	Std Err	Param	Std Err
Fixed effects
Intercept	2.349***	0.143	2.352***	0.101	2.345***	0.100
Social networks	–0.057**	0.015	–0.058**	0.015	–0.057***	0.011
Social expenditure			–0.067**	0.019	–0.091**	0.020
Unemployment rate			0.042*	0.018	0.043*	0.018
GDP growth			–0.078*	0.028	–0.079*	0.028
Expenditure* networks					0.013**	0.003
Random effects
Individual variance	2.781***	0.020	2.781***	0.020	2.781***	0.020
Country variance	0.320***	0.097	0.095**	0.029	0.890**	0.027
Slope variance	0.004**	0.002	0.004**	0.001	0.002*	0.001
Cov intercept* slope	–0.022*	0.010	–0.009	0.005	–0.005	0.004
R² individual level	13.53%		13.53%		13.53%
R² country level	27.05%		78.10%		79.46%
-2LogLikelihood	155,526.623		155,639.307		155,489.611

*p < 0.05; **p < 0.01; ***p < 0.001. Entries represent the result of three separate multi-level regression models controlling for the variables of Model 2 in Table 2.

More informing is the extent to which this differential effect can be explained by welfare state expenditure. First of all, we need to take a closer look at the direct effects of social expenditure per capita, as well as of the unemployment rate and economic growth control variables (Model 4). As expected, self-reported deprivation is positively associated with unemployment rates: respondents of countries with high unemployment levels experienced more impoverishment than residents of countries with low unemployment rates. Economic growth, on the other hand, shows negative effects, implying that respondents of countries that caught up better in times of crisis have a less impoverished citizenry. Regarding welfare state effort (Model 4), we see that, in line with the hypothesis, higher welfare spending goes along with lower levels of self-reported deprivation. The unstandardized effect of –0.067 means that between Bulgaria (the country with the lowest expenditure of 500 euros per person) and Norway (with an expenditure of almost 12,500 euros per person), the difference in deprivation is three-quarters of a scale point on the 0–6 scale. Thus, whether one lives in the most or least generous European welfare state makes a remarkable difference in reporting a drop in financial resources since the start of the economic crisis.

Introducing then the cross-level interaction effect of social expenditure on the relationship between informal social networks and self-reported deprivation, we can see in Model 5 a significant positive interaction effect (0.013), which means that higher welfare spending tends to cushion the negative effect of informal social ties on self-reported deprivation. In other words, having frequent meetings with friends, relatives or colleagues has prevented people from becoming deprived generally across Europe, but especially in the least developed welfare states. Resources that are embedded in social networks are thus more likely to be mobilized in smaller welfare states, precisely because generous welfare states are already quite effective in reducing impoverishment.

Figure 1 summarizes the results of Model 5 of Table 3 showing the predicted regression lines (plus confidence intervals) for the effect of informal social networks on self-reported deprivation for the least (Bulgaria) and most generous (Norway) welfare state. As Figure 1 shows, deprivation is more reported in smaller welfare states. Additionally, whereas the plot displays a weak upward trend for Norway,⁹ there is a strong negative effect of informal social networks on deprivation in Bulgaria. Here, the predicted self-reported deprivation is almost one scale point (on a 0–6 scale) lower for most-connected residents compared to least-connected residents. In sum, as we expected, resources are more likely to be mobilized through networks in the least generous welfare states, in the sense that there they cushion against deprivation more. So it seems that both proponents and opponents of the ‘crowding out’ thesis represent a piece of the truth. In higher developed welfare states, a higher frequency of informal social contacts is reported (size aspect), whereas in lower developed welfare states resources in such informal social networks are more likely to be mobilized in order to prevent deprivation (quality aspect).

Figure 1.

The moderating effect of welfare state effort on the relation between informal social networks and self-reported deprivation. The graph displays the predicted values (plus confidence intervals) of self-reported deprivation (Y-axis) regressed on the frequency of informal social networks (X-axis) for the most generous (Norway) and least generous (Bulgaria) welfare state.

Conclusion

To sum up, what is there to say about how welfare state effort affects socio-economic resources through informal social networks in the present age of austerity? If governments implement budget cutbacks, including welfare retrenchment, can we expect social bonds between friends, relatives and colleagues to take over the same role that welfare states play in keeping people from being deprived? The answer is complex. On the one hand, self-reported deprivation in times of crisis varies greatly across European countries. Not surprisingly, this variation can largely be explained by differences in welfare state effort: people report lower levels of impoverishments in more extensive welfare states. While this might sound as stating the obvious, the research strengthens the message that reducing the welfare state can increase impoverishment among national populations, not only measured by objective poverty rates, but, as we have shown, also by subjective deprivation indicators.

On the other hand, despite neo-liberal and communitarian arguments that welfare retrenchment might not necessarily be excessively harmful, as informal social networks will take over the interventionist role of the welfare state, our analysis does show ambiguous outcomes in that respect. First of all, the success of informal social networks in preventing deprivation is less compared to the success of the welfare state, refuting the criticism of state intervention. Second, it nonetheless needs to be underscored that the functional quality of informal social ties differs across welfare states. Crudely speaking, while we diagnose that informal social bonds between people are of quality, as they embed resources that prevent people from becoming impoverished, the mobilization of these resources is less prevalent in more encompassing welfare states. The fact that these resources are less mobilized can only be logically explained because encompassing welfare states already do an excellent job in preventing impoverishment, making the role of social networks less necessary. Predictions inform us that, on average, the best-connected respondents in the least generous welfare state (i.e. Bulgaria) have become more impoverished than the most isolated respondents in the most generous welfare state (i.e. Norway). The present research thus shows that the traditional ‘crowding out’ thesis, which says that the size of social networks would be lower in more developed welfare states, needs to be amended. Our study also shows that, in terms of preventing deprivation, the quality of informal social networks is lower in more generous welfare states, and that this is precisely because the developed welfare state is already effective in redistributing resources, making intervention from friends, relatives or work colleagues less urgent.

In our interpretation, the findings predict less room for optimism in the present economic downturn, as a vicious circle initiated by welfare state retrenchment might occur. On the one hand, there is a direct impact of retrenchment on deprivation, since with restriction of welfare provisions certain social risks may no longer be (sufficiently) covered. And, on the other hand, welfare state retrenchment shifts the emphasis from vertical (state) to horizontal (informal social ties) relations. With this, welfare state retrenchment might affect deprivation indirectly as well if it leads to a reduction of social networks. This is not imaginary, because several studies, including ours, have shown that informal social networks are less common in more limited welfare states.

Regarding our findings, four important remarks need to be made that require more attention in future research. The first concerns the crudeness of the independent variable, i.e. informal social networks. Due to a lack of refined social network indicators in the ESS 2010, the present research was limited to an indicator that measures frequency of informal social contacts. As we know from previous studies, bonding with family is more common in less generous welfare states, while connections with friends are more likely to occur in more generous welfare states. Insofar as we discovered that informal social networks are better able to reduce deprivation in less generous welfare states, one interpretation might be that it is especially family ties that keep people out of deprivation, implying that an absence of deprivation-reducing effect in generous welfare states might not only be due to effective welfare provisions, but also to the fact that families are less integrated in encompassing welfare states. Evidence from a number of less generous welfare states suggests that the nuclear family is a significant supplier of socio-economic resources in times of crisis (Buchmann and Kriesi, 2011; Moreno, 2012). Despite the important qualification that family networks have strong and consistent negative effects on deprivation, this does not jeopardize the finding that welfare state effort outweighs the effect of informal social networks in combating poverty.

The second remark regards the causal direction of the networks–deprivation link. Theoretical arguments borrowed from the social capital literature have justified an analysis of the effect of informal social networks on the access to socio-economic resources, i.e. self-reported deprivation. On the other hand, both poverty scholars (Böhnke, 2008; Ruggeri Laderchi et al., 2003; Townsend, 1979) and public policy analyses (European Council, 2010) touch upon the reversed causal relationship. According to the European Council (2010), social inclusion must be promoted ‘in particular through the reduction of poverty’. Applied to the present study, such a perspective would say that it might be possible that due to impoverishment people have lost their social networks, but that impoverished residents of generous welfare states have not lost as much of their contacts as residents of limited welfare states. This reversed causal interpretation nevertheless does not violate the interpretation that welfare state retrenchment is not always preferred, as welfare state effort makes deprived people better able to sustain their social contacts.

Third, the interpretation of our findings in terms of the functional quality of social networks is hampered by so-called floor effects: precisely because the welfare state is so effective in redistributing resources making deprivation less likely to occur, our assessment that the quality of social networks in generous welfare states is low is restricted to aspects of preventing deprivation, which is of course the core interest of the welfare state. However, this does not imply that networks are not of value in encompassing welfare states in other ways, as they probably mobilize resources other than those that prevent deprivation, such as moral support, a social base for belonging, etc. Future research could investigate this.

Fourth and final, as networks help to protect against deprivation, especially in less generous welfare states, future reflections on possible perverse effects in the mechanisms between social networks and financial deprivation need to be considered, too. To explain this, we recall the distinction between bonding networks that make people ‘get by’ and bridging networks that make people ‘get ahead’ (de Souza Briggs, 1998). It can be expected that especially higher socio-economic groups, and not those groups at the bottom of society, have access to bridging types of network resources (van Ingen and van der Meer, 2011; Woolcock and Narayan, 2000). Therefore, it might be that with retrenching welfare state spending, network inequalities within the socio-economic most vulnerable societies increase, given the unequal access to bridging networks. This path needs to be further explored, while keeping Worldbank economist Deepa Narayan’s (1999) words in mind that: ‘Interventions are also required to foster bridging ties’.

Footnotes

Acknowledgments

We thank the editors and anonymous reviewers for their thoughtful and detailed comments. We are also indebted to Violet Benneker, Paul de Beer, Mirjam Fischer, Ferry Koster, Marie Pichault, Tom van der Meer and Kirsten van Kaathoven. A previous version of this article was presented on the Dutch ESPAnet Research Day.

Funding

This research was made possible thanks to support from the Research Council – Flanders (FWO) and the Dutch Organization for Scientific Research (NWO).

Notes

References

Andreoni

Payne

(2003) Do government grants to private charities crowd out giving or fund-raising? American Economic Review 93(3): 792–812.

Böhnke

(2008) Are the poor socially integrated? The link between poverty and social support in different welfare regimes. Journal of European Social Policy 18(2): 133–150.

Bourdieu

(1986) The forms of capital. In: Richardson

(ed.) Handbook of Theory and Research for the Sociology of Education. New York: Greenwood, 241–258.

Buchmann

Kriesi

(2011) Transition to adulthood in Europe. Annual Review of Sociology 37: 481–503.

Cattell

(2001) Poor people, poor places, and poor health. The mediating role of social networks and social capital. Social Science and Medicine 52(10): 1501–1516.

Coleman

(1988) Social capital in the creation of human capital. American Journal of Sociology 94(S): S94–S120.

Day

Devlin

(1996) Volunteerism and crowding out: Canadian econometric evidence. Canadian Journal of Economics 29(1): 37–53.

De Deken

Kittel

(2007) Social expenditure under scrutiny. The problems of using aggregate spending data for assessing welfare state dynamics. In: Clasen

Siegel

(eds) Investigating Welfare State Change: The ‘Dependent Variable’ Problem in Comparative Analysis. Cheltenham: Edward Elgar, 72–105.

Delhey

Newton

(2005) Predicting cross-national levels of social trust. Global pattern or Nordic exceptionalism? European Sociological Review 21(4): 311–327.

10.

De Souza

Briggs X

(1998) Brown kids in white suburbs. Housing mobility and the many faces of social capital. Housing Policy Debate 9(1): 177–221.

11.

Etzioni

(1995) The Spirit of Community. London: Fontana Books.

12.

European Council (2010) Conclusions – 17 June 2010. Brussels: General Secretariat of the Council.

13.

Eurostat (2012) Eurostat statistics database. Available at: http://epp.eurostat.ec.europa.eu/portal/page/portal/statistics/search_database (accessed 25 October 2012)

14.

Finsveen

van Oorschot

(2009) The welfare state and social capital inequality. An empirical exploration using longitudinal European/World Values Study data from 13 Western Welfare States. European Societies 11(2): 189–210.

15.

Flap

Boxman

(2000) Getting started: The influence of social capital on the start of the occupational career. In: Lin

Cook

Burt

(eds) Social Capital: Theory and Research. New York: Aldine de Gruyter, 159–184.

16.

Flap

Völker

(2001) Goal specific social capital and job satisfaction. Effects of different types of networks on instrumental and social aspects of work. Social Networks 23(4): 297–300.

17.

Fukuyama

(1995) Trust. The Social Virtues and the Creation of Prosperity. New York: Free Press.

18.

Fukuyama

(2001) Social capital, civil society and development. Third World Quarterly 22(1): 7–20.

19.

Gelissen

JPTM

van Oorschot

Finsveen

(2012) How does the welfare state influence individuals’ social capital? Eurobarometer evidence on individuals’ access to informal help. European Societies 14(3): 416–440.

20.

Gelman

Hill

(2006) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge: Cambridge University Press.

21.

Granovetter

(1973) The strength of weak ties. American Journal of Sociology 78(6): 1360–1380.

22.

Helliwell

Putnam

(2004) The social context of well-being. Philosophical Transactions of the Royal Society B: Biological Studies 359(1449): 1435–1446.

23.

Hox

(2010) Multilevel Analysis. Techniques and Applications. London: Routledge.

24.

Immerzeel

van Tubergen

(2013) Religion as reassurance? Testing the insecurity theory in 26 European countries. European Sociological Review 29(2): 359–372.

25.

Kaariainen

Lehtonen

(2006) The variety of social capital in welfare state regimes. A comparative study of 21 countries. European Societies 8(1): 27–57.

26.

Kawachi

Kennedy

Glass

(1999) Social capital and self-rated health: A contextual analysis. American Journal of Public Health 89(8): 1187–1193.

27.

Knack

Keefer

(1997) Does social capital have an economic payoff? A cross-country investigation. Quarterly Journal of Economics 112(4): 1251–1288.

28.

Kumlin

Rothstein

(2005) Making and breaking social capital. The impact of welfare-state institutions. Comparative Political Studies 38(4): 339–365.

29.

Larsen

(2007) How welfare regimes generate and erode social capital. The impact of underclass phenomena. Comparative Politics 40(1): 83–101.

30.

Lin

(1999) Building a network theory of social capital. Connections 22(1): 28–51.

31.

Mau

Mewes

Schöneck

(2012) What determines subjective socio-economic insecurity? Context and class in comparative perspective. Socio-Economic Review 10(4): 655–682.

32.

Moreno

(2012) The transition to adulthood in Spain in a comparative perspective: The incidence of structural features. Young 20(1): 19–48.

33.

Narayan

(1999) Bonds and Bridges. Social Capital and Poverty. Washington, D.C.: World Bank.

34.

Narayan

Cassidy

(2001) A dimensional approach to measuring social capital. Development and validation of social capital inventory. Current Sociology 49(2): 59–102.

35.

Norris

Davis

(2007) A continental divide? Social capital in the US and Europe. In: Jowell

Roberts

Fitzgerald

Eva

(eds) Measuring Attitudes Cross-Nationally. Lessons from the European Social Survey. London: Sage, 239–264.

36.

Paxton

(2002) Social capital and democracy: An interdependent relationship. American Sociological Review 67(2): 254–277.

37.

Pichler

Wallace

(2007) Patterns of formal and informal social capital in Europe. European Sociological Review 23(4): 423–435.

38.

Pichler

Wallace

(2009) Social capital and social class in Europe: The role of social networks in social stratification. European Sociological Review 25(3): 319–332.

39.

Putnam

(2000) Bowling Alone. The Collapse and Revival of American Community. New York: Simon and Schuster.

40.

Putnam

Campbell

(2010) American Grace. How Religion Divides and Unites Us. New York: Simon and Schuster.

41.

Rasbash

Browne

Healy

Cameron

Charlton

(2013). MLwiN Version 2.27. Bristol: Center for Multilevel Modeling, University of Bristol.

42.

Room

(1995). The Measurement and Analysis of Social Exclusion. Bristol: Policy Press.

43.

Rothstein

Stolle

(2008) The state and social capital: An institutional theory of generalized trust. Comparative Politics 40(4): 441–459.

44.

Ruggeri Laderchi

Saith

Stewart

(2003). Does it matter that we do not agree on the definition of poverty? A comparison of four approaches. Oxford Development Studies 31(3): 243–274.

45.

Saegert

Thompson

Warren

(2001) Social Capital and Poor Communities. New York: Russell Sage Foundation.

46.

Snijders

TAB

Bosker

(1999) Multilevel Analysis. An Introduction to Basic and Advanced Multilevel Modeling. Thousand Oaks, CA: Sage.

47.

Stadelmann-Steffen

(2011) Social volunteering in welfare states: Where crowding out should occur. Political Studies 59(1): 135–155.

48.

Townsend

(1979) Poverty in the United Kingdom. A Survey of Household Resources and Standard of Living. Berkeley, CA: University of California Press.

49.

Van Ingen

van der Meer

(2011) Welfare state expenditure and inequalities in voluntary association participation. Journal of European Social Policy 21(4): 302–322.

50.

Van Oorschot

Arts

(2005) The social capital of European welfare states. The crowding out hypothesis revisited. Journal of European Social Policy 15(1): 5–26.

51.

Van Oorschot

Arts

Gelissen

(2006) Social capital in Europe. Measurement and regional distribution of a multifaceted phenomenon. Acta Sociologica 49(2): 149–167.

52.

Woolcock

(1998) Social capital and economic development. Toward a theoretical synthesis and policy framework. Theory and Society 27(2): 151–208.

53.

Woolcock

(2002) Social capital in theory and practice: Where do we stand? In: Isham

Kelly

Ramaswamy

(eds) Social Capital and Economic Development: Well-Being in Developing Countries. Cheltenham: Edward Elgar, 18–39.

54.

Woolcock

Narayan

(2000) Social capital: Implications for development theory, research, and policy. World Bank Observer 15(2): 225–249.