THE INFLUENCE OF SOCIAL CAPITAL ON WILLINGNESS TO PAY FOR THE ENVIRONMENT AMONG EUROPEAN CITIZENS Open Access

The conservation of the environment is a major challenge for the European countries. Environmental problems may range in scale, from local to international problems necessitating cooperation of citizens and actors for their resolution. In this context there is a growing literature analyzing the interaction between society and the environment. The present article will focus on one specific attitude of European citizens: their willingness to contribute money for the environment and the influence of social capital components on their decision.

Willingness to pay (WTP) for the environment has been extensively analyzed in the literature. From the perspective of environmental economics WTP of citizens is estimated in order to price environmental goods and investigate factors influencing their decision. The main factors identified in the relevant literature may be classified in three categories (see Mitchell and Carson 1989; Bateman and Turner 1993; Hanley and Spash 1993). A first set of factors refers to demographic data such as age, gender and the number of household members. Secondly, economic factors, such as individual or household income and thirdly factors which refer to individuals’ attitudes towards the environmental good being valued. In addition, the influence of political factors such as the involvement of the state in the provision of environmental goods (Meyerhoff and Liebe 2006) and the idea that the state should be responsible to pay for expenditures on environmental issues (e.g., Jones et al.2008a) have been proposed. Other studies have explored the influence of materialist and post-materialist attitudes (Gelissen 2007), the level of political interest (Torgler and Garcia-Valinas 2007), free-riding behaviour¹ and issues of collective action (Mitchell and Carson 1989; Jones et al.2008a). Finally, contextual variables in country-level have also been investigated (Gelissen 2007).

The aim of this study is to further explore this field by proposing the influence of social capital components (namely ‘generalized’ and ‘institutional trust’, ‘social networks’, ‘social norms’ and ‘civic participation’) on the tendency of individuals to contribute money for the environment. Although there is no widely accepted definition, it is assumed that social capital influences individuals’ tendency to act in a collective manner (Putnam 1993) and in this context, it has been introduced to environmental issues. In particular, it is assumed that higher stocks of social capital may facilitate environmental management significantly (Pretty 2003) mainly due to the former's connection with individual behaviour (Pretty and Ward 2001; Rydin and Pennington 2001; Pretty 2003; Torgler and Garcia-Valinas 2007).

Consequently, the main hypothesis to be tested in the present study is that individuals with a higher ‘stock’ of social capital will have a greater tendency to contribute money for the environment. In addition, the influence of several other variables, both at the individual and contextual level, will be assessed. For this purpose, we use data from 24 European countries that participated in the European Values Survey (1999). The article is divided into five main parts. Firstly, an analysis of the main connections of social capital with environmental issues is presented in order to justify the primary hypothesis tested in the article. Secondly, the data and methodology used are presented and next, the results of the study are analyzed. In the final section a discussion of the findings is provided followed by the main conclusions of the article.

Social capital is one of the most influential terms in the social sciences of the last decade. According to several analysts, social capital refers to social characteristics, which promote collective activity towards the common good (Coleman 1990; Putnam 1993). Components of social capital may include social and institutional trust, social networks, civic participation and the level of compliance to social norms (Coleman 1990; Putnam 1993, 2000; Woolcock and Narayan 2000; Bowles and Gintis 2002). The wide expansion of the theory of social capital has also been accompanied by significant criticism regarding issues of definition, its components and consequences (see Levi 1996; Newton 2001; Uslaner and Conley 2003; Pawar 2006). Despite this critique, the theory has been widely used mainly due to its connection with public policy issues (e.g., Newton 1999; Whiteley 2000; Wallis and Dollery 2001) and its promising insights for studying collective and individual behavior (Coleman 1988; Zhao 2002; Poortinga 2006).

Regarding the connection of social capital to environmental issues, it has been recently argued that in communities with higher stocks of social capital there is a tendency among citizens to act in a collective manner for the conservation of natural resources (Pretty and Ward 2001; Pretty 2003; Rydin and Pennington 2003). A primary link derives from the fact that the management of natural resources necessitates cooperation and participation of community members. The transaction costs and consequently the level of participation may be influenced positively from elements of social capital, mainly referring to social trust and the density and type of social networks (Woolcock 1998; Fukuyama 2001; Pretty and Ward 2001; Lubell 2002).

Another important issue focuses on the need of pursuing the common good when managing natural resources. The tendency of citizens towards this direction is connected with the level of trust developed among citizens. In particular, environmental behaviour of individuals is influenced by the belief that, in general, the members of the community act in the interests of the common good (Pretty 2003; Rydin and Holman 2004). In this context, it is argued that higher stocks of social capital minimize the occurrence of ‘free-riding’ behaviours (Coleman 1990) a fact which is connected to the general tendency of compliance with social norms in a community (Pretty 2003).

The level of trust towards institutions is equally important (Beierle and Cayford 2002). Natural resources, as public goods, are mainly managed by the state through public policies. Consequently, trust towards institutions involved in these procedures and the belief in their legitimacy, influences citizens’ compliance with environmental regulations and consequently their environmental behaviour. This assumption has also been pointed out through several studies of WTP, where citizens’ refusal to pay for the environment was justified by their distrust towards governmental management (e.g., Damigos and Kaliampakos 2003; Jones et al.2008a).

A final component of social capital connected with environmental behaviour refers to formal social networks and civic participation. The term ‘formal social networks’ refers to the density of citizens’ participation in organized social groups, such as sport clubs and non governmental organizations (see Paxton 1999; Newton and Norris 2000; Narayan and Cassidy 2001; Beugelsdijk and van Schaik 2005; van Oorschot et al.2006). A main function of these networks (which mainly refer to ‘bridging’ social capital – see Putnam 2000: 22; Kaariainen and Lehtonen 2004) is that they facilitate the flow of information and thus may influence the level of awareness on environmental issues.

Regarding civic participation, it mainly refers to the level of political interest. It may be assumed that citizens who are interested in political issues are more informed and concerned on environmental issues. To that extent, numerous studies have substantiated the supportive function of social networks for activism in general (Rosenstone and Hansen 1993; Knoke 1994; Verba et al.1995; Klandermans 1997; Downton and Wehr 1998), and environmental activism in particular (Tindall 2002; Wakefield et al. 2006).

From the above, analysis it follows that social capital is related to environmental behaviour. However, limited studies have explored the influence of social capital, as a multidimensional concept, on a specific environmental behaviour in general and specifically WTP (e.g., Torgler and Garcia-Valinas 2007). Thus, the present study will explore the influence of social capital on the tendency of European citizens to contribute money for natural resources.

Apart from the individual determinants of environmental behaviour, broader, societal, influences also come into play. Thus, previous research on environmental behaviour, especially the trait which deals with socio-psychological constructs – i.e., the role of values, beliefs, etc. (Dietz et al.1998) – has demonstrated that behaviour is also shaped by nation-level cultural values. Thus, Oreg and Katz-Gero (2006) 27-countries’ analysis has shown that incorporating country-level values, particularly ‘post-materialism’ (Inglehart 1977), as contextual antecedents, influences individual environmental attitudes and behaviours. Furthermore, national differences can also be accounted for on the basis of different, actual, environmental conditions. A number of studies have supported the correlation between environmental concern and objective environmental problems, albeit predominantly at the local level (Inglehart 1995; Blake 2001; Scruggs 2002; Bamberg 2003). Finally, differences between countries could be related to different national policies, and to citizens’ perceptions of them. Thus, Guerin et al. (2001) multilevel analysis of the determinants of recycling behaviour in European countries, showed that individuals’ ‘perceptions of the effectiveness of the public bodies regarding their action to protect the environment’ as well as contextual ‘national state of policies implemented in the sector of waste management’ do increase individual (Guerin et al.2001: 207) and public participation in recycling activities (Guerin et al.2001: 211).

More related to our research, are the findings that there are national differences concerning social capital in Europe (Kaariainen and Lehtonen 2006; van Oorschot et al.2006) and also that national, contextual-level variables (such as GDP and environmental conditions) do influence an individual's willingness to pay for the environment (Gelissen 2007). Thus, in the subsequent analysis we shall also check for possible influences of macro-level variables to the individual willingness to pay.

In order to conduct the analysis, data were obtained from the European Values Survey (wave of 1999) for 24 European countries² (Table 1). A set of variables were selected as indicators of individual social capital while for assessing the WTP the question ‘Would give part of my income for the environment’ was used. The analysis was conducted in two main stages. In the first stage we estimated the individual social capital, whereas in the second stage the exploration of the possible relation of the estimated social capital and other societal and contextual factors with the willing of individuals to contribute money for the environment was investigated.

The literature of measuring social capital has increased significantly in the last decade (e.g., Narayan and Cassidy 2001; Grootaert and Bastelaer 2002; van der Gaag and Snijders 2005). In the present study several variables commonly used in relevant surveys were selected, although there is no widely acceptable method for measuring social capital. In addition, for the measurement of individual social capital, Confirmatory Factor Analysis (CFA) was conducted, which permits testing factor analysis modelling hypotheses with a specific factor structure (e.g., van Oorschot et al.2006).

The variables selected (all in ordinal scale) were organized in four factors. The first factor (NETWORKS) aimed at measuring formal social networks and consisted of two variables distinguishing passive and active participation (as proposed by Beugeldsijk and van Schaik 2005).The first variable (PASS) displayed whether the individual was a member to at least one of a list of organizations presented in the questionnaire of the study, and the second (ACT) whether the individual was a volunteer to at least one of these organizations.

The second factor (TRUST) deals with social and institutional trust (see Paxton 1999; Newton and Norris 2000). For the measurement of this factor five variables were included. The first one (STRUST) was the commonly used question of social trust (see Paxton 1999; van Oorschot et al.2006): ‘Most people can be trusted or you can't be too careful’. The remaining questions examined the level of trust on different types of institutions: the police (POLICE), the national parliament (PARL), the civil services (CIVIL) and the justice system (JUST).

A third factor (NORMS) referred to social norms and examined the perception of individuals towards actions connected to the common good. Three indicative actions were chosen where individuals declared the level that they regarded them as justifiable (claiming governmental benefits, BENF; cheating on taxes, TAX; and someone accepting a bribe, BRIBE) (see also van Oorschot et al.2006).

A final factor (CIVIC) created was ‘civic participation’ indicating the level of interest for public issues (e.g., Narayan and Cassidy 2001; Hjollund and Svensen 2003; van Oorschot et al.2006). Three variables were included: whether the individual regarded politics as important in life (IMP), how often the individual discusses political matters with friends (DISC) and the frequency of following politics in the news (NEWS).

Prior to estimating individual social capital the relation of social capital variables with WTP was investigated and statistically significantly interrelations were observed among all variables of social capital and the WTP of individuals (P<0.05).

In order to proceed to social capital measurement, a social capital score for each individual was calculated, using the European Values Survey data. Various approaches and models may be utilized for analyzing ordinal categorical variables in the latent analysis framework. Among them, Muthen (1984) and Jöreskog (1994) proposed estimation methods for handling ordinal polytomous data. Thus, Confirmatory Factor Analysis (CFA) was conducted for the estimation of individual social capital using the selected variables mentioned in the previous section and the methodology of ordinal-scale variables (see Jöreskog and Moustaki 2006).³

The estimated CFA model of the study is presented via the path diagram of Figure 1. In this path diagram squares represent observed variables, whereas circles represent latent variables. Moreover, numerical values along each arrow correspond to factor loadings for each observed variable on the corresponding first-order latent variable and factor loadings of each first-order latent variable on the single second-order latent variable. Loadings appearing on the path diagram are the standardized factor loadings. Chi-square statistic for testing model's adequacy was 2911.27 (df = 61, P<0.001) and accordingly, the null hypothesis that the model provides a good fit to the data was not rejected.⁴ Since the CFA model fulfilled the requirement of strong associations between observed and latent variables, individual factor scores were derived of the second-order CFA model and used as estimates of individual social capital.

Through individual factor scores the mean score for each country included in the analysis was estimated (see also Kääriäinen and Lehtonen 2006). A reverse on the scales used in the EVS was conducted in order higher scores to indicate higher stocks of social capital.

In addition, differences among European regions were explored, by dividing countries in five main categories.⁵ A first category refers to the Scandinavian countries participating in EVS including Iceland. A second category refers to the Anglo-Saxon counties (or Liberal), including Ireland and Great Britain and Northern Ireland. Thirdly, South European (or the Latin Rim) countries (Spain, Italy, Greece, Portugal). A different category was created for the Conservative (or Bismarckian) countries (Austria, Belgium, France, Germany, Luxembourg and The Netherlands). Finally all post-socialist countries were included in the fifth category.

The results of the measurement are in accordance with the relevant literature regarding regional differences of social capital (e.g., Kääriäinen and Lehtonen 2006; van Oorschot et al.2006). The highest stock of social capital is observed in the Scandinavian countries (mean score: 2.208), followed by the Anglo-Saxon countries (mean score: 2.078). The lowest mean score is presented in post-socialist countries (mean score: 1.965). Yet we should note that for the purpose of the present study informal networks were not included in the analysis, which are regarded dense in the Latin Rim (Scheepers et al.2002; Kaariainen and Lehtonen 2006; Jones et al.2008b).

The question selected for the investigation of WTP among European citizens was phrased as: ‘Would give part of my income for the environment’, with four possible answers ‘Strongly Agree’, ‘Agree’, ‘Disagree’ and ‘Strongly Disagree’. Regarding the total sample of European citizens, the majority (45.3 percent) ‘agreed’ in giving part of their income for the environment, while 13.8 percent ‘strongly agreed’. Nevertheless, a significant percentage of individuals ‘disagreed’ with the WTP question (27.7 percent) and 13.2 percent ‘strongly disagreed’. After estimating individual social capital and frequencies of WTP, the connection between them was investigated. It is observed that variations between average social capital scores for the four categories of WTP exist and higher scores of social capital correspond to higher WTP.

Nevertheless, it can not yet be concluded whether these differences are statistically significant. For this purpose, further statistical analysis must be utilized. In addition although measurements have been collected at an individual level, we also have to take into consideration possible variations among the 24 European countries of our dataset. In this context, a multilevel-type analysis was conducted in order to investigate the connections of WTP with social capital, while controlling for the variance between individuals and the variance among different countries. With only two sources of variation (level 1: individuals; level 2: countries), the fit of a generalized estimating equation (GEE) model (Liang and Zeger 1986) allowing two sources of variation was selected. For modelling the ordinal categorical response, the multinomial distribution and the cumulative logit were used as the model's link function. Specifically, two multilevel models were assumed using WTP as the dependent variable (Table 2). A summary of the results of the fitted models is presented in Table 3, where the Wald-type test results indicate the significance for each effect separately.

In the first model social capital scores were introduced as the only explanatory variable (level 1). According to the results of the model fit (Table 2), social capital is a significant predictor for WTP (beta coefficient = 0.67, P<0.01). However, it is necessary to additionally examine the influence of other socio-economic factors.

By fitting the second model it is examined whether social capital remains statistically significant even after controlling for the influence of other individual variables. These include: Gender (reference category: females), Age (recoded into three intervals with reference category citizens over 50), Education level (reference category: Middle) and Income level (reference category: Medium). The results of the second model fit indicate that after the inclusion of these variables social capitals’ effect remains significant (beta coefficient = 0.801, P<0.01).

Regarding the connection of WTP with demographic data it is observed that women are more willing to contribute money compared to men (beta coefficient = –0.116, P<0.05). Furthermore, younger respondents (15–49) are more willing to pay for the environment compared to older respondents (50 + ) (beta coefficient = 0.366, P<0.01), whereas lower and middle educated people are also less willing to pay for the environment when compared to citizens of higher educational level. Finally, low income individuals are less willing to pay for the environment (beta coefficient = –0.155, P<0.1), while there is no statistical difference between medium and high income citizens (P>0.05).

Although the aim of the present article was to observe the influence of social capital on WTP in individual level, it is interesting to additionally observe the influence of other types of variables from an aggregate perspective. National variations are of particular interest: ignoring either the within-country or the between-country variation in the data may lead to serious problems in analysing such data, for instance may cause standard errors of regression coefficients to be underestimated. Also, measuring only one source of variation, for example only variation among individuals, one loses significant information attributed to the between-countries differences. Moreover, the possible within-country correlation in the data should be considered since one could assume that measurements taken within the same country are more likely to be correlated, compared to measurements taken from different countries. Thus, we employ the five-fold country categorization of ‘Regions’ (or, welfare state-regime) we have already referred to in a previous section (reference category: Post-socialist region).

Also, similar to the study of Gelissen (2007: 402–3) four additional context-level covariates/factors where included in the previous model. For enhancing the comparatibility of our results, we use the same – yet not all- contextual factors employed by Gelissen (2007). Namely, ‘GDP growth rate’, measured by the GDP per capita annual growth rate (%) for the period 1990 to 2001 (UNDP 2003: 278). Next, ‘population density’, measured as people per square kilometre, in 2000 (Nationmaster.com 2008). Lastly, two indicators measuring actual environmental conditions, from the Environmental Sustainability Index (World Economic Forum 2001): ‘Air quality-based on urban SO2 concentration, urban NO2 concentration, and urban total suspended particulates concentration – and water quality – based on dissolved oxygen concentration, phosphorus concentration, suspended solids, and electrical conductivity’ (Gelissen 2007: 403). The theoretical reasons for choosing these factors are described in detail by Gelissen (2007: 395–6, 402–3) and can be summarized in a nutshell as follows: it would be easier for publics in wealthier states to allocate money for the environment while their willingness would be further enhanced by the existence of acute, objective, environmental problems – which are more easily identified and perceived in densely populated countries.

The results are presented in Table 4. We observe that the significances of social capital and the other demographic indicators remain relatively unaffected, even after the inclusion of the five country-level covariates/factors. Among these, only two have been found to be significant predictors in explaining WTP for the environment. In particular, population density is positively related to WTP (see Table 4), indicating that countries with higher density are more willing to pay. In addition, the welfare state-regime variable, dividing Europe in five main regions is also statistically significant in explaining variation of WTP (see Table 4). However, not all regions differ between them, in relation to WTP: only Anglo-Saxon and Conservative countries are statistically significantly less willing to pay for the environment compared to all other regions.

In the present study we found a strong connection between social capital and the decision of willing to contribute money for the environment, at the individual level. Individuals with higher stocks of social capital are more willing to give money even when taking into consideration other important socio-economic factors. Consequently it should be assumed that elements constituting social capital influence pro-environmental behaviour.

Regarding the different components of social capital included in our study it can be claimed that all elements have a significant influence as it was emphasized in the relevant literature. This implies that social trust may indeed minimize incidents of ‘free-riding’ and promote the tendency of individuals to act for the common good. Similarly, the influence of institutional trust may be connected with the belief that these institutions are responsible for the management of natural resources in the context of national and European environmental policies. This fact is in accordance with previous research findings: for example Sonderskov (2008) has demonstrated that aggregate generalized trust has a positive effect on national environmental group memberships. Regarding the importance of social networks, it was observed that citizens who participate in at least one organization are more willing to pay for the environment. This may be attributed to the role of networks on the flow of information for environmental issues and thus in the increase of environmental awareness. Again, this finding is congruent with other research: Guerin et al. (2001) have demonstrated that the higher the participation in environmental organizations, the more the population is likely to recycle. We have also found that citizens who had an increased interest in politics presented a higher WTP.

A second important finding of our analysis is the fact that the majority of European citizens are willing to contribute money for the environment. Nevertheless statistical significant differences were observed among European regions. At the theoretical introduction we have discussed how national and context-level variables might influence individual behaviour. Our results have shown that a number of these variables, such as actual environmental conditions and GDP growth rate, have no impact on WTP, while others, such as population density, did. In addition regional differences were observed. Although a detailed explanation of these differences exceeds the scope of this article, we consider them (which are after all welfare state-regime differences) to be due to these macro-level variables which where not included in our model, namely political system and policy styles variables. Nevertheless, these regional (welfare regime) differences should not obscure the main finding of our paper. That is, that even in a model compromising both micro and macro/contextual level variables, the individual WTP is by far mostly influenced by the individual social capital.

Finally, some issues for further research must be underlined. One main issue refers to the measurement of social capital and the fact that there is no widely acceptable measurement technique. On the contrary there is significant differentiation among the variables included in several studies measuring social capital. In the present analysis we attempted to use the most commonly used variables. However, there is a need for a common measurement tool which may be used for comparative purposes among different surveys. Another suggestion in this direction is the further exploration of institutional trust on environmental attitudes. In particular, trust towards ‘experts’ and other entities (e.g., NGOs) involved in the process of environmental policy and management must also be investigated. Similarly, it is important to distinguish generalized and particularized trust (Uslaner and Conley 2003) and observe their different influence on environmental attitudes. Finally, it would be interesting to investigate social norms related to environmental issues as well as the influence of other type of networks included in bonding and linking social capital. A final major issue emerges regarding the aggregate and individual level analysis. In particular, through this article it is obvious that the explanation of macro phenomena necessitates an approach from a different perspective. It may be assumed that theories regarding the influence of cultural values (e.g., cultural theory, Ellis and Thomson 1997) along with an exploration of local environmental conditions and the effectiveness of state environmental policies may contribute significantly to this discussion.

Despite these limitations the present article indicated that components of social capital in individual level influence a certain environmental behaviour. Consequently such social factors may be included and emphasized in future studies estimating the WTP of individuals and in general environmental behaviours. There is an increasing need of a more holistic approach in order to understand environmental attitudes and consequently be able to promote public participation and increase environmental awareness, essential prerequisites for sustainable communities.

An earlier draft of the present paper was presented at the 8th Conference of the European Sociological Association, 3–5 September 2007, Glasgow. The authors gratefully acknowledge the financial support of the MSc program ‘Environmental Policy and Management’ (Department of Environment, University of the Aegean) in allowing them to attend the conference.

Nikoleta Jones holds a PhD in Environmental Policy. She is currently a post-doctoral researcher at the department of Environment, University of the Aegean. Her main research interests include social capital and environmental policy, environmental sociology, voluntary policy instruments and valuation of environmental goods.

Chrisovaladis Malesios is a Mathematician and holds an MSc in Statistics. He is currently a PhD student in the Department of Statistics, Athens University of Economics and Business. His main research interests are statistical analysis of bibliometric indices.

Iosif Botetzagias is a Lecturer in Environmental Politics & Policy at the Department of Environment, University of the Aegean, Greece. His research interests include Green political parties, environmental NGOs and New Social Movements. He is currently involved in an international project studying policy networks dealing with climate change.