داوطلبی، بهزیستن ذهنی و سیاست های عمومی

Abstract We apply matching estimators to the large-scale British Household Panel Survey (BHPS) data set to estimate the impact of volunteering on subjective well-being. We take into account personality traits that could jointly determine volunteering behaviour and subjective well-being. We find that the impact of regular volunteering on subjective well-being is positive and increasing over time if regular volunteering is sustained. In a quantile analysis, we find that this effect seems to be driven by reducing the unhappiness of the less happy quantiles of the well-being distribution for those who volunteer regularly. We test the robustness of our findings and discuss their relevance for public policy.

1. Introduction Homo oeconomicus has become a rare breed according to the findings of behavioural economics. Non-selfish behaviour as another driving force of human (economic) activity has made its way into economists’ thinking and modelling, and it is relevant in many respects: examples include private fund-raising, organ donations, intergenerational transfers, the funding of public goods as well as support for charities. But altruistic behaviour does not only benefit the recipient. Studies have established a positive relationship between individuals’ volunteering activities and their own health (Post, 2005). A similarly positive relationship was found for volunteering and subjective well-being (e.g., Brooks, 2006, Borgonovi, 2008 and Meier and Stutzer, 2008). In the present paper, it is this latter relationship we are interested in. In the empirical part of our paper, we analyse how volunteering is related to individuals’ subjective well-being (or “happiness”). We apply our analysis to a large-scale British sample (using data from the British Household Panel Survey for a time span of 15 years). Comparing our results for a different country to the results of Meier and Stutzer (2008) is useful in itself to see whether the volunteering-happiness relationship may differ across countries or cultures. But apart from this, we contribute to the literature in several further ways. First, we aim at addressing some unresolved questions regarding the confounding role unobserved variables may play in conventional estimations of the effect volunteering has on subjective well-being (Dolan, Peasgood, & White, 2008, p. 104). For example, personality traits of an individual may (jointly) influence the degree of volunteering as well as self-reported subjective well-being or lead to self-selection of subgroups into altruistic behaviours: especially “extravert” and “agreeable” persons,1 who tend to be either more outgoing or cooperative, might self-select into volunteering behaviours, biasing any regression estimates not accounting for this self-selection. To address these problems we first analyse the impact of volunteering on subjective well-being with reference to different personality traits. This has only recently become possible for big household panel data sets, as personality inventories have not been elicited in them up to a short time ago. Second, we offer an econometric account of the causal impact of volunteering on subjective well-being by making use of “propensity score matching” estimators (Rubin, 1974, Imbens, 2004 and Caliendo and Kopeinig, 2008). Propensity score matching is an econometric technique that pays special attention to the information on the distribution of covariates in the treatment versus control groups (so as to allow us to compare individuals that have similar values for all covariates). If there is no substantial overlap in the two covariate distributions, multivariate regression estimates rely heavily on extrapolation, and can therefore be misleading (Imbens, 2004 and Ichino et al., 2008, p. 312–13). Matching estimators are preferable because more care is taken to establish an appropriate control group. They also require no assumptions on functional forms (Hussinger, 2008, p. 730). While widely used in other subfields, to our knowledge, matching estimators have only recently been introduced to the analysis of subjective well-being and its causes and correlates (e.g., Binder & Coad, in press). Our third contribution lies in examining the extremes of the subjective well-being distribution via quantile regressions (Koenker and Bassett, 1978 and Koenker and Hallock, 2001). In heterogeneous distributions, regression methodologies that focus on means might seriously under- or overestimate effects or even fail to identify effects at all (Cade & Noon, 2003). Applying quantile regressions in our context, we are able to assess whether the (average) effect of volunteering on happiness that is usually found in the literature might actually be driven by the effect that volunteering has for especially happy or unhappy individuals (the extremes of the distribution). Finally, we provide a discussion of the implications of our findings for public policy. If volunteering increases the well-being of all individuals involved, can increased volunteering lead to a happier society? Considering the increasing debt level in ageing OECD countries, would it be wise to encourage more volunteering to combat financial gaps and labour shortages, and if so, what policy means are available? These questions become more urgent, the less the (welfare) state is able to provide for the less well-off of its citizens, and we will therefore discuss prospects and pitfalls associated with them. The paper is organized as follows. We discuss the theoretical background regarding the volunteering-happiness relationship in Section 2. The empirical part of our paper consists of Section 3, where we discuss our econometric strategy, the BHPS data set as well as our findings and robustness checks. We then give particular attention to the implications our analysis might have in the context of public policy in Section 4 before concluding in Section

4. Conclusion In this paper we have explored the relationship between volunteering and subjective well-being. We have found that for our British sample, regular volunteering significantly increases well-being even when taking into account stable personality traits, social networks and trust, all of which could potentially confound this relationship. We have given an estimate of the causal effect of volunteering on individuals’ well-being and shown that this effect increases over time for individuals who continue their volunteering efforts. This effect does not seem to be well captured by typical regression methodologies focusing on the mean effects of the explanatory variables: expanding our analysis to the full subjective well-being distribution, we find no longer an association between volunteering and subjective well-being for the upper quantiles of the distribution. In other words: volunteering can be conjectured to be protective for the volunteers insofar as it allows them to mitigate their own unhappiness. Happy individuals, on the other hand, seem to derive their happiness from other sources.