The investment model of volunteering in the EU-27 countries: volunteering, skills development and employability. A multi-level analysis Open Access

This article contributes to a better understanding of the social acceptance of the investment model of volunteering, based on the view that volunteering can enhance employability through the development of professionally relevant knowledge and competences (Roy and Ziemek 2000: 14).¹ Employability can be defined as the relative chance of finding and maintaining different kinds of employment (Brown et al. 2002). It has two dimensions: an absolute dimension, related to individual attributes such as skills and knowledge, and a relative dimension, related to the variable conditions of the labour market, which also affect the degree to which an individual can gain employment (Brown and Hesketh 2004). The paper focuses on the absolute dimension, which cannot tackle unemployment problems on its own but is an important element of the ‘employability mix’.

The article makes use of survey data from European Union (EU-27) countries, complemented with Eurostat, Hofstede and UNESCO data for national institutional variables, to explore the extent to which Europeans see volunteering as being conducive to the acquisition of knowledge and competences that facilitate employability. In doing so it responds to recent calls for greater empirical evidence on the value of volunteering (Tymon 2013). The lack of evidence in this area is unexpected given that a major motivator for volunteering is the opportunity to acquire work-related experience and skills that can help them in their careers (Eley 2003). While specific figures on the number of volunteers differ according to the survey and definitions employed, all studied confirm that the take-up of volunteering is substantial with between 100 and 150 million Europeans (around one in five to one in three) engaged in volunteering each year (see also GHK 2010).

This article focuses on social perceptions regarding knowledge and competence development through volunteering. This is crucial because social perceptions can be expected to mediate the effects of volunteering in social interactions, including hiring. The nature of this mediation, in turn, can affect individuals’ decision to volunteer (Friedland and Morimoto 2005). While the relation between perceptions and behaviour is undoubtedly complex, a first step in analysing such relationship is to clarify what those perceptions and those behaviours are in relation the area of interest. In addition, a better understanding of social perceptions on volunteering can help to inform the design and implementation of policy decisions in this area (Snyder and Omoto 2008). Thus, it is surprising that extant literature has neglected the analysis of the social assumptions that surround the outcomes of volunteering, which are embedded in the way in which volunteering is socially perceived and recognised.

In its empirical part the article analyses general social views – or the views of the general population – on the benefits of volunteering for labour market relevant competence development, and also the views of those individuals who have the potential to influence hiring decisions (IHP) – or likely ‘recruiters/employers’. Second, the relationship between socio-demographic variables and the acceptance of the investment model of volunteering is examined. Finally, we explore whether national differences in the acceptance of the investment model of volunteering are better explained by variation between countries’ institutional arrangements or cross-national demographic groups.

It should be noted that the focus on the investment model does not aim to deny the importance of factors other than employability for volunteers, for instance those related to altruism, social and personal development (Handy et al.2010; Holdsworth 2010). It does not deny that volunteering can have negative effects either (Eliasoph 1998), and does not suggest that all types of volunteering activities and engagement could be expected to have the same returns on competence development and employability. The article's findings are particularly relevant to young people. In a substantial number of countries, young people exhibit the highest level of volunteering (GHK 2010), and it is for this group that the signalling value of volunteering experiences are likely to count the most, given that they have little experience in the labour market (Katz and Rosenberg 2005; Hall et al.2006).

The paper is structured as follows: Section 2 takes stock of relevant literature on competence development and employability enhancement through volunteering and social views of volunteering. Section 3 explains the data and methods used for the analysis. Section 4 presents the findings, and Section 5 concludes.

This section outlines the main premises of the investment model of volunteering, and the socio-demographic factors that may affect the acceptance of this model, from which we derive a range of variables employed in our analysis. This is followed by a discussion of the empirical base of the relationship between volunteering and competence development and volunteering and employability, which underpin the premises of the investment model.

Volunteering has been theorised from a range of disciplines: economics, sociology, political science and psychology (see Hustinx et al.2010a). Such theorisation has focused, above all, on who volunteers and motivations to volunteer (Wilson 2000; Hustinx et al.2010b). Much research has been conducted on motivations to volunteer (Wilson 2000; Becker and Dhingra 2001; Bussell and Forbes 2002; Taniguchi 2006; Rotolo and Wilson 2007; Perry et al.2008; Holdsworth 2010). One of the main findings of extant research is that the motivation to volunteer is based upon a complex interplay between altruistic, individualistic and social factors (Haski-Levental et al.2009).

Regarding the outcomes of volunteering, its health and well-being benefits have been subject to much analysis (Meier and Stutzer 2008; Morrow-Howell et al.2009; Konrath et al.2012). Its wider benefits to the families of volunteers, organisations and the wider community have also been explored (Handy and Srinivasan 2004; Morrow-Howell et al.2009; Handy and Mook 2010). By contrast, the literature has paid much less attention to the labour market outcomes of volunteering or the specific mechanisms through which these may derive: such as increases in labour market relevant connections (Wilson and Musick 1999), empowerment (Cohen 2009) or the acquisition of competences and knowledge through the volunteering experience (Callow 2004; Souto-Otero et al. 2005). Moreover, as Smith et al. (2010: 69) and Holdsworth and Quinn (2010) note, most studies that have looked broadly at the motivations and benefits of volunteering have collected data only from active volunteers. This is also the case in studies that have explored increases in knowledge, skills or competences and employability through volunteering (Handy and Srinivasan 2004; MacNeela 2008; Surujlal and Dhurup 2008; Booth et al.2009; Souto-Otero et al.2016). This limitation has led to calls for greater empirical evidence on the value of volunteering (Tymon 2013).

This article is concerned, specifically, with the investment value of volunteering, reflecting the ‘consumerist view of volunteering’, as articulated by Kendall (2009). The ‘signalling value of volunteering’ and the investment model of volunteering relate volunteering to the expectation of a private benefit or return (Ziemek 2006). The investment model proposes that volunteers acquire knowledge, skills and competences, enhancing their human capital (Hustinx et al.2010b). Human capital development through volunteering can be ‘signalled’ (Spence 1973) to employers in recruitment processes, as applicants aim to offer employers convincing signals of their employability. Thus, the ‘signalling value’ theory proposes that volunteering can be used as a signalling device that provides recruiters with information on the potential of individuals (Katz and Rosenberg 2005). It should be noted that, while related, the signalling value of volunteering is different from the investment model –in fact for signalling to work it is not necessary that the volunteer develops professionally relevant knowledge and competences, as long as participation in volunteering conveys information about the likely ability of the applicant to employers.

In the absence of previous research on social and labour market acceptance of the investment value of volunteering, this section builds upon the literature on motivations to volunteer to identify relevant variables for our analysis. In doing so, the article examines whether the factors that influence social perceptions of the benefits of volunteering are similar to those that the literature has suggested are related to the adoption of the investment model as a motivation to volunteer.

The literature identifies several factors that affect whether the investment model is a motivation for volunteers. The literature has produced results regarding the relationship between gender and age and motivations to volunteer. Hustinx et al. (2010b) found that females were more likely to follow an investment model motivation to volunteer than males. They found age to be negatively correlated with motivations for volunteering related to resume building and signalling value: younger people think about the labour market value of volunteering more strongly than older people.

The literature has repeatedly found a positive association between levels of education (Bekkers 2006; Gesthuizen and Scheepers 2012) and the acceptance of the investment view. While there is evidence that in some countries the effect of education in volunteering is decreasing (Van Ingen and Dekker 2011), as explained by Handy et al. (2010) university applicants can benefit substantially from the signalling value of volunteering in university recruitment processes, and university students are significantly more likely to volunteer than other individuals in the same age group, which suggests that volunteering may have greater benefits for them. Following a similar logic in relation to occupational status, those individuals with more limited links with the labour market are seen to be more willing to accept the value of other forms of acquisition of professionally relevant knowledge and competences than through professional experience (Katz and Rosenberg 2005).

With respect to social positioning, Hustinx et al. (2010b) reported no effect of family income on the adoption of motivations to volunteer associated with resume building and signalling value. With respect to geographical positioning, given that networks in rural settings tend to contain ties of greater intensity and density and be based more on kinship and neighbourhood solidarities rather than other types of logics (Beggs et al.1996), we also expect the integrative function to be higher in rural areas than in urban areas.

Household composition – that is, the number of young children in the household – has been found to affect views on the purpose of volunteering. Through the membership of their children or school activities parents have higher chances to be asked to volunteer, and could be more receptive to the integrative role of volunteering rather than its investment value. The role of children directing their parents into volunteering activities of such integrative nature is strongest for children who are going to school but are yet to become independent (Rotolo and Wilson 2007; Van Ingen and Dekker 2011).

Participants in volunteering could be expected to be more likely to report the skills development benefits of participation (Hustinx et al.2010a) because they have first-hand experience of skills development and because some participants who were motivated to volunteer by the investment model would be likely to report this benefit.

Our analysis of the effect of political ideology is exploratory, as this variable has not often been included in previous studies of the determinants of volunteering. We expect that people from the left are less inclined to report the acquisition of competences as one of the main benefits of participation in volunteering, as they prioritise other benefits, associated with prosociality, social solidarity and cohesion (Van Lange et al.2012).

The factors identified are thus related to ascriptive elements regarding ‘who one is’ (gender, age) as well as factors related to ‘achievement’ (education, employment status) ‘where one is’ (geographical and social geographies), the contact one has with volunteering (volunteering intensity, household composition) and how one sees the organisation of society (political ideologies). While ethnicity and religion have also been found to affect views of volunteering (Musick et al.2000; Ruiter and De Graaf 2006), the Eurobarometer survey employed for this study did not include data on these variables.

It should be noted that the investment model is not accepted or rejected by individuals in an institutional vacuum. It can be institutionally induced as, for instance, increasingly educational institutions require students to take-up volunteering in exchange for credits. Companies may do the same for employees, in exchange for time-off from paid work (Basil et al.2009). So support for this model may not only reflect changes in the motivations of volunteers, but also institutional incentives and requirements. Handy et al. (2010) note that labour market signals often are context-specific, and this may affect the degree of acceptance of the investment model of volunteering.

Thus, in our analysis, we explore the importance of institutional factors in the acceptance of the investment model of volunteering. Specifically, we include in our analysis a measure of the competitiveness of the labour market, the specificity of the education system, the strength of the continuing vocational training system (both employed and training under active labour market policies for unemployed people) and culture influence the adoption of the model –see Section 3 for further details.

There are very few studies that focus on competence development through volunteering, and fewer on the extent to which this may affect the employability. As noted by Ellis Paine et al. (2013), most of these studies rely on the views of volunteers, specific social groups (women, refugees and higher education students) and rely on small sample sizes or anecdotal evidence (see also Booth et al.2009). They tend to suggest that competence development is one of the benefits of involvement in volunteering.

Ellis Paine et al. (2013) argue that the desire to learn new skills can be an important motivator for volunteering. Kamerade and Ellis Paine (2014) report that several studies suggest that volunteering enhances individual employability, through gains in aspects such as knowledge, skills, work attitudes, confidence, self-esteem (Hirst 2001; Nichols and Ralston 2011). Volunteers declare that volunteering activities enhance both ‘hard’ (business management, IT specific skills, etc.) and ‘soft’ (communication, teamwork, management and organisational skills, etc.) skills (Gerogy et al.2000; Hirst 2001; Peterson 2004; Cook and Jackson 2006; Nichols and Ralston 2011; Souto-Otero et al.2013). Vegeris et al. (2010) are more critical in arguing that volunteering does not provide volunteers with the types of skills that employers demand. Evidence on the extent to which these gains result in paid employment is weaker (Adams et al. 2011). The effects on employment seem to vary significantly depending on individual characteristics (age, gender and motivation to volunteer) and circumstances (frequency of engagement in volunteering, employment history, volunteer role and quality of the volunteering experience) (Hirst 2001; Ellis Paine et al.2013).

The literature has largely neglected the issue of skills and competence development through volunteering, has relied on small samples and has tended to focus on the views of volunteers, ignoring the views employers and society at large have regarding the labour market relevance of the skills acquired through volunteering.

The rich business literature on the use of biographical information in recruitment processes identifies three main factors that impact recruitment decisions: academic achievements, experience and extracurricular activities. However, little of this research has specifically included volunteering activities in their designs, as part of extracurricular activities. Much of this literature concentrates, restrictively, on graduate recruitment. Several of the studies discussed here are based on the presentation of real or researcher-produced CVs to individuals with recruitment responsibilities for their assessment (e.g. Brown and Campion 1994; Cole et al.2007; Chen et al.2011). The main conclusion is that employers do value extracurricular activities, including knowledge and competences achieved through volunteering, although the degree of value attached to those depends on certain conditions such as the number of volunteering experiences, their type and the way in which they are presented. Only one of the studies reviewed (Keenan and Scott 1985) reported that neither membership of clubs and societies nor being an office holder in societies had a high predictive power for final employment decisions.

Volunteering can affect the recruitment process as recruiters have implicit theories that associate experiences with skills sets. Brown and Campion (1994) report that recruiters associate participation in community activities with the development of interpersonal skills and high motivation, and associate ability and motivation with high performance. Thomas (2001), focusing on international volunteering experiences, reports that employers associate such experiences with adaptability, handling responsibility, stress management, self-assurance and problem solving.

Chen et al. (2011) include ‘volunteering for community activities’ as one of the three items making up an indicator for ‘extracurricular activities’ used in their study on recruiters’ decisions. They find that this indicator is of importance when explaining recruiters’ decisions to offer an interview to a candidate, particularly if in-depth information on the activities is provided. As it could be expected, the type of involvement in extracurricular activities (e.g. undertaking leadership roles or not) will affect the skills and competences developed and the value that employers attach to them (Walker 2010). As Chen et al. (2011) note, the effects of CV content on hiring recommendations are mediated by recruiters’ perceptions of person-job fit and person-organisation fit. Participation in certain extracurricular activities is interpreted as a signal of the personality of the young person (Tomlinson 2007). Moreover, applicants’ life experiences – such as volunteering – may enhance recruiters’ similar-to-me effects in relation to those applications (Tsai et al. 2011). Similar-to-me effects enhance the value that recruiters attach to applications (Tsai et al. 2011).

Some studies have started to look at the interaction effects between qualifications, experience and extracurricular activities – not only at their separate influence over recruitment decisions. Cole et al. (2007) make use of multilevel modelling to estimate such interaction effects and report that applicants ranked high in all three dimensions (qualifications, experience and extracurricular activities) unsurprisingly receive the highest employability rankings. But recruiters also judge applicants with low academic qualifications that rated high in work experience and extracurricular activities as being highly employable. Thus, one dimension can compensate for another, even if the weak dimension is qualifications. In fact, people performing highly in extracurricular activities can be seen to have a specialist skill set.

The literature thus suggests that employers value extracurricular activities generally, but has less to say regarding volunteering specifically. Moreover, much of the economic literature does not clarify whether the reported economic returns to volunteering are due to specific characteristics of those who volunteer, such as assumed personal qualities, or a result of improved job-matching based on the additional information volunteering provides to recruiters, rather than to competences the labour market may consider to be develop through volunteering. This is an aspect in relation to which the present study contributes, as it asks IHP to think abstractly about the benefits of volunteering for competence development, and provides an insight into their general assumptions about the value of volunteering for competence development. Most previous research has looked at individual motivations to volunteer, rather than society's views on its benefits, often focused on the views of students (rather than those of the general population or IHP), and tended to cover a very limited number of countries (often one or two).

The empirical analysis explores the acceptance of the investment model of volunteering in EU-27 countries, paying particular attention to variations between the general population and IHP, the demographic factors associated with the acceptance of that model, and to whether national differences in the acceptance of the investment model are better explained by institutional variations between countries. It addresses the following research questions:

Data are taken from the Special Eurobarometer 75.2 on volunteering and intergenerational solidarity, which contains nationally representative samples for the 27 EU countries. Fieldwork for the survey took place between April and May 2011, with responses from 26,825 European citizens aged 15 and over. Among other questions, the survey asked respondents about the main benefits of volunteering. Volunteering is a complex phenomenon that is not clearly delineated and can encompass a variety of activities, organisations and sectors. Some definitions provide concrete specifications as to the minimum time that needs to be devoted to qualify as volunteering, whether certain material rewards are allowed or not, whether relatives can be included as the beneficiaries and whether only activities with certain pre-defined intentions should be considered volunteering. A broad definition of volunteering was provided by Wilson (2000: 215), who referred to ‘any activity in which time is given freely to benefit another person, group or cause’. The ways in which questions about volunteering are asked in surveys can have implications for the results obtained (Lyons et al.1998). Volunteering was not defined for the respondents, and it is therefore likely that most had a broad definition of volunteering in mind. However, the lack of definition implies that representations of volunteering may differ from country to country, which may affect the results reported – a recurrent problem with multi-country research. The variable that measures participation in volunteering in the Eurobarometer survey differentiates between no, occasional and regular participation (again as judged by respondents), and this differentiation is utilised in the analysis of the importance of individual factors – the distribution of responses by countries regarding this variable is provided in the appendix (Figure A3). It should be noted that empirical research suggests that there is a large cross-cultural consensus in the public perceptions of who is considered ‘definitely a volunteer’ (Meijs et al.2003). While it would be beneficial to have data on the type of volunteering undertaken by volunteers, such information was not collected in the survey.

Respondents were asked to select two of seven options regarding the main benefits of volunteering: civic participation, strengthening of the fundamental values of solidarity in the EU, enhancing social cohesion, benefitting the European economy, environmental protection, self-fulfilment and personal development and ‘the acquisition of knowledge and competences which allow a good professional integration’. The last of these provides a measure of respondents who prioritised the individual returns of volunteering, in terms of competence and employability benefits, vis-à-vis other benefits. This measure provides a reasonable match with the premises of the investment model of volunteering. It should be noted that we do not wish to infer motivation from the data, so our claim is not that those individuals who reported that employability developments from volunteering consider that this is a key motivation for volunteering. The estimates provided below regarding the acceptance of the investment model of volunteering reported are a ‘lower bound’ development, as those surveyed may consider that volunteering contributes to the acquisition of professionally relevant knowledge and competences, but to a lower extent than to two other aspects.

Variations in respondents’ views of volunteering were explored making use of demographic and national covariates, selected on the bases of the literature review presented in Section 2. Table 1 presents a summary of the individual-level demographic covariates employed in the logistic regression analysis and associated expectations.

These included: age, gender, years of education, political affiliation measured on a scale of 1 (‘left’) to 10 (‘right’), previous participation in volunteering (none/occasional/regular), urban/suburban–rural residence, ‘level in society’ (which we also refer to as ‘class’) from 1 (‘lowest’) to 10 (‘highest’), number of children below 10 in the household and occupational status. Regarding education, respondents were asked the age at which they complete full-time education, which was used to create a binary of those who had completed at least 15 years of full-time education (i.e. those who could be expected to have completed all or most of secondary education). The survey included a question on ‘current occupation,’ asking respondents to select from 19 occupational categories – see TNS 2011: 53–4. This information was used to compute a new variable identifying individuals who are likely to make or influence decisions regarding hiring and employment (IHP). Those individuals who identified their occupations as ‘business proprietors’ or owners, employed professionals – which included employed doctors, lawyers, accountants and architects – directors and general/top managers, and middle managers were included in the ‘IHP’, or individuals who potentially could make hiring decisions, category. This new variable was utilised to assess the views of those individuals whose judgements on extracurricular activities have a greater effect on hiring decisions.

The central focus of the analysis is whether differences in the national context or – conversely – cross-national demographic groups are more closely related to the adoption of the investment model of volunteering. To this end, we also included national-level data on five key variables related to the labour market, education and training system and culture: national unemployment level, degree of specificity of the education system, strength of the continuing vocational training system (for people in employment and training under active labour market policies for people in unemployment), and the level of individualism. A highly competitive labour market, which exhibits high unemployment rates, may increase the need for employability-related investments, including volunteering. In those countries where the education systems mainly relies ongeneral secondary programmes and produces more generic skills that are demanded in the labour market (Hall and Soskice 2001) there may be a higher acceptance of the employability value of volunteering, as participation in volunteering could be expected to result in the development of generic skills, such as communication, leadership or organisational skills (Souto-Otero et al. 2013). We anticipate that a trade-off between the strength of the continuing vocational training system and the investment model of volunteering. Higher take-up of opportunities for professional training in the country may reduce the extent to which volunteering is associated with the development of professionally relevant competences. As other opportunities offered by continuing vocational training are used, these may be considered sufficient and/or more relevant for the development of such competences, and volunteering may be seen less often as an alternative provider of those competences, reducing the prioritisation of the investment model. We check this separately in relation to continuing vocational education and training for employed and training under active labour market policies for unemployed people. Regarding national culture, in more individualistic culture, the private benefits of volunteering may be emphasised more than in collectivist cultures.³

In order to investigate these relationships, the data were analysed using a combination of descriptive statistics (to identify broad patterns in the data and the distribution of variables) and inferential methods (logistic regression and multilevel logistic regression) to model the probability that an individual in the sample would prioritise the investment model of volunteering as an outcome of the set of demographic predictors, identify differences between IHP and the full set of respondents and test for differences across national contexts.

The analysis is structured in two parts: the first presents descriptive statistics on the prevalence of the investment model of volunteering across Europe, and the second presents the outputs of three logistic regression models. Each regression model is estimated on the complete data set and the subset of IHP, yielding a total of six sets of parameters.

Before undertaking statistical modelling of the data, we looked at how the investment model of volunteering was adopted across EU-27 countries and between the general population and IHP. Table 2, which gives the proportion of respondents who prioritised the employability benefits of volunteering, shows that over one-fifth (21%) of respondents in the general population reported to believe that the ‘facilitation of knowledge and competencies which allow a good professional integration’ is one of the two main benefits of volunteering. IHP exhibit very similar views to other respondents: also 21% reported to consider professionally relevant competence development as one of the two main benefits from volunteering.

Table 2 also shows that views vary considerably by country: from 14.1% of respondents in the Czech Republic prioritising the employability value of volunteering to 33.0% in Sweden. Sweden, Great Britain, Hungary and Romania are top. Czech Republic, Portugal, Malta, Ireland, Spain, Slovakia and Slovenia are bottom. The range for the prioritisation of the employability benefits of volunteering from IHP is wider: from a low of 8.6% in Finland (also low are Luxembourg, Portugal and Slovakia) to 32.3% in Latvia (which is followed by Great Britain, Romania, Bulgaria, Poland and Lithuania). Latvia, the Czech Republic, Bulgaria, Lithuania and Cyprus show the largest differences in the acceptance of the prioritisation of the employability benefits of volunteering between all respondents and IHP, in favour of IHP – this is, IHP reporting higher prioritisation of the employability benefits, than society at large. The opposite trend – with social prioritisation of the employability benefits of volunteering being above those of IHP – is particularly marked in Sweden, Finland, Hungary and Luxembourg.

Geographic patterns in the adoption of the prioritisation of the employability benefits of volunteering among all respondents are displayed in the appendix (Figure A1). Table A1 shows that the employability benefits of volunteering were more frequently reported together with ‘social cohesion’ benefits in the case of all respondents, and with ‘personal development’ in the case of IHP.

Logistic regression is a form of generalised linear modelling that expresses the probability of a binary outcome as a function of predictor variables and associated coefficients. These coefficients can be used to assess how they contribute to the probability of the outcome, including the magnitude of their effects and their statistical significance. Multilevel implementations of logistic regression account for nested or multilevel data, in our case the sampling of individuals within European nation-states (Pinheiro and Bates 2000). This approach allows for combined modelling of group variables (e.g. national unemployment) and individual variables (e.g. age, gender). Our binary outcome is the prioritisation of the professional benefits of volunteering in the Eurobarometer survey, and the predictors are the set demographic and national-level variables.

We model the data in three ways (Table 3): our first model uses a fixed-effects model to examine the relationship between demographic variables and prioritisation of the investment model of volunteering, without accounting for national differences. The second model extends the first by taking a multilevel approach: a random intercept is used to account for national differences (in other words, a separate intercept is calculated for each country, with parameter estimates that maximise within-country and between-country fit). Finally, we add five national-level predictors: unemployment levels, specificity of the education system, strength of continuing vocational training (for the employed and training under active labour market policies for unemployed people) and individualism, to test the extent to which aspects of the national labour market, education and training and culture are related to different patterns of prioratisation of the investment model of volunteering.

Model 1 expresses the probability of a respondent prioritising the investment model of volunteering as an outcome of the set of demographic variables described in Table 1. Results from all respondents show that several of the demographic variables are significantly related to the probability that an individual would prioritise the investment model of volunteering. The probability of prioritising the investment model of volunteering is higher for females, those who have completed at least 15 years of education and those in urban locations, and decreases with respondents’ age, all of which conform to our expectations. The probability increases with self-reported social positioning. The results for political views, participation in volunteering (except for ‘all/regular’), household composition and occupational status have the expected sign, but the results are not significant. However, for IHP, no demographic variables are significant, indicating that those who may actually make hiring decisions are more homogenous in their views of the employability benefits of volunteering.

Model 2 adopts a multilevel approach, adding a random intercept for national differences and retaining fixed effects for demographic variables. When these national differences are taken into account, many of the demographic relationships become non-significant. For the full set of respondents, the only significant predictors of mentioning the employment benefits of volunteering are age, with older respondents significantly less likely to mention the benefit, and occasional volunteering. The effect size is small for age, but it scales in years and can therefore alter the probability substantially. The effect of occasional volunteering – but not regular volunteering – may indicate that those individuals who are regularly involved in volunteering may have a less instrumental view of volunteering and as a result prioritise other benefits above skills development. In contrast to all respondents, among IHP no demographic variables are significantly related to mentioning the employment benefits of volunteering. For both the full set of respondents and the IHP subset, national variations in mentioning the employment benefits of volunteering are significant, although for IHP the variation is smaller. Thus, results suggest that the differences associated with demographic variables in Model 1 (i.e. gender, education, age, residence and social class) are better explained through national variation.

Model 3 extends Model 2 by including national-level covariates for unemployment, vocational specificity, strength of participation in continuous vocational training, and training under active labour market policies and individualism. For the general population model, age and social class are the only significant individual-level variables in Model 3. Regarding national variables, the effects of unemployment and vocational specificity are not significantly different from zero, but training under active labour market policies shows a significant, negative relationship to prioritisation of the investment model of volunteering. Individualism and the strength of training under continuing vocational training are only significant at the p < .1 level. These results thus provide some evidence of a trade-off between the take-up of professionally relevant training and prioritisation of the investment model. This is particularly clear in relation to unemployed people: thus, when those who are further away from the labour market have lower opportunities for professional training through active labour market policies the investment value of volunteering is more prevalent. As a corollary, higher take-up of opportunities for professional training in the country, in particular by the unemployed, reduces the extent to which the investment model is prioritised. For IHP no individual-level variables are significant; regarding national-level variables, the strength of training under active labour market policies is significant. There is thus also some evidence of trade-off between national provision of training to the unemployed through active labour market policies and the prioritisation of the investment model among IHP.

The regression results show that variation in the acceptance of the investment model is more attributable to differences between countries than cross-national demographic groups. This is evident in the standard deviation of random effects (Models 2 and 3), which indicate typical variation due to country differences. Typical between-country variations is larger than most of the cross-country demographic variables, and significant. However, for IHP the between-country variation is less than for the entire respondent sample (country variance at level 2 in the multilevel regression models is lower for IHP, which means that they vary less based on the country than people in the general population), and it becomes non-significant when national characteristics are taken into account (Model 3). IHP are more homogenous in their views, both at the individual level and at the national level.

For both multilevel models, we tested the influence of each country to see whether one particular country's observations disproportionately influences the model fit, a key concern in international survey research (Van der Meer et al.2010). This approach involves iterative fits of the model, omitting a different country on each iteration, with the expectation that a valid model will not differ significantly depending on the omission of a particular country (Nieuwenhuis et al.2012). Using procedures recommended by Van der Meer et al, we established only one case (Sweden) that might exert undue influence on results. However, recalculation of Model 3 with Sweden omitted (see appendix, Table A2), do not differ substantially from those presented in Table 3 in relation to country-level variables or individual-level variables – gender becomes non-significant whereas occasional volunteering, employment and urban residence significant. Thus, it appears that the reported national-level variation is not due to the presence of highly influential cases. We also tested for differences between volunteers and non-volunteers. The results show in general only small variations: gender and urban residence are significant for volunteers only, and social class for non-volunteers only (see appendix, Table A2).

For all models, the low pseudo-r² values – computed using Cox and Snell's (1989) methods – give some cause for concern. However, Hosmer and Lemeshow (2000) note that low r² values are typical for logistic regression (e.g. and that the approximated measures are based on comparisons to the predicted values of an intercept-only model rather than a true measure of goodness of fit). Thus, to determine the validity of the model, we examined plots of predicted probabilities versus observed outcomes using procedures described by Greenhill et al. (2011). These results are presented in the online supplement, and generally show a good fit with higher predicted probabilities corresponding to a greater frequency of observed responses (i.e. respondents prioritising the investment model).

While most studies on volunteering have focused on the individual motivations to volunteer and its determinants, this study has examined social views on the benefits of volunteering, on which there is a striking scarcity of empirical analyses (Dekker and Halman 2003). The specific focus has been on the acceptance of the investment model of volunteering across EU-27 countries, looking at the interaction between volunteering and the development of professionally relevant competences. The article explored the variations that exist between the general population and likely employers (IHP) in such acceptance, and how demographic characteristics relate to the acceptance of the investment model in both groups. We were also concerned with whether the differences found in levels of acceptance could be better explained by variations between countries or cross-national demographic groups.

There is very little difference in the extent of acceptance of the model between the general population and IHP: around a fifth of respondents in each group prioritised the employability benefits of volunteering over other benefits. There are, however, differences in how demographic characteristics relate to the acceptance of the investment model in these two groups.

We found significant age and class differences for the general population, but contrary to expectations, no significant results were obtained for gender, education, residence, household composition, employment status, previous participation in volunteering and political views. These findings suggest that the socio-economic factors affecting the importance given to the investment model of volunteering differ substantially from those affecting participation in volunteering (Hustinx et al.2010b). Regarding national-level variables, the weakness of alternative forms of upskilling for the unemployed, in particular, increases the prioritisation of the investment model, suggesting that volunteering may fill a gap when such opportunities are absent.

Moreover, we find that by contrast to the general population, IHP were a surprisingly homogeneous group. This is a positive result for those individuals who have volunteered/will volunteer with an employability benefit in mind. No demographic characteristics are significant for this group. Regarding national-level variables, only the strength of training for unemployed people is significant; like in the case of the general population it has a negative relationship with the prioritisation of the investment model of volunteering.

Our analysis pointed out that differences in the data are explained by variation between cross-national demographic groups, for the general population but not for IHP, and also variation between countries. The regression results show that variation in the acceptance of the investment model is more attributable to differences between countries than cross-national demographic groups. However, it should be noted that for IHP the between-country variation is less than for the entire respondent sample and it becomes non-significant when national characteristics are taken into account.

Much sociological literature has underlined the political and ‘good citizenship’ benefits of volunteering (Wuthnow 1998; Putnam 2000). But our findings reject the notion that social and labour market views of volunteering are based on pure ‘consumption’ or ‘public goods’ models of volunteering, in which utility from volunteering is derived from the act of giving in itself or the provision of valued public goods and services – see Roy and Ziemek (2000) for details. Instead, the results provide support to mixed models of volunteering (Etzioni 2000), based on ‘mutuality’ and reciprocal relations, which acknowledge the role of other factors beyond competence development and employability in volunteering, but do not neglect these benefits either: volunteers benefit from their work also in terms of their own education and training, which bestows some equality on the volunteering relationship. Our findings also have significant implications for academic queue models and human capital theory, and particularly for the articulations of these that relate the development of skills and competences restrictively to formal schooling. Instead, our results point towards the importance of an ‘economy of experience’ (Brown and Hesketh 2004), where human capital is no longer exclusively represented by academic credentials, but is also seen to be gained through various activities other than formal education, including volunteering experiences. For those individuals entering employment straight after school/university with little relevant work experience, the development of the experience side of human capital may be particularly important to enhance their employability (Holdsworth 2010). This, in turn, suggests that further consideration should be given to addressing inequalities of opportunity to take-up volunteering for people from different socio-economic backgrounds. It also suggests that there is greater scope to explore the knowledge, skills and competences that volunteering can help to progress – particularly where other human capital development activities would be less effective – and is perceived to progress, especially by employers. Kamerade and Paine (2014) go further to argue that targeted volunteering for employability programmes could be a cost-effective targeted human resource development activity, and Booth et al. (2009) for a skills-matching alliance between employers and volunteer organisations.

Further research should also explore other factors that drive national differences in the acceptance of the investment model of volunteering. In this respect, the analysis of the importance of differences in the nature of volunteering across countries (for example in terms of the formality/informality of arrangements, the volume of take-up of volunteering or the kinds of volunteering activities that are prevalent) would deserve particular attention, as possible further explanations to the national differences found.

No potential conflict of interest was reported by the authors.

Manuel Souto-Otero is Senior Lecturer in Social Sciences at the School of Social Sciences, Cardiff University. His research focuses on the link between education and work, social stratification and inequality, internationalisation in education, and policy evaluation. He has recently co-edited the ‘Palgrave International Handbook of Higher Education Policy and Governance’ and edited the book ‘Evaluating European Education policy-making: privatization, networks and the European Commission,’ also published by Palgrave.

Robin Shields is Senior Lecturer in Higher Education Management at University of Bath. His research focuses broadly on the globalisation of education; with publications on international higher education and education in international development. He is a specialist in quantitative methodologies and social network analysis.