«IZA DP No. 8764 PAPER Voluntary Activities and Daily Happiness in the US J. Ignacio Gimenez-Nadal DISCUSSION Jose Alberto Molina January 2015 ...»
The joint null hypothesis is that the instruments are valid instruments (i.e., uncorrelated with the error term), and that the excluded instruments are correctly excluded from the estimated equation (Hayashi, 2000). Under the null hypothesis, the test statistic is distributed as a χ2 (1). In our case, the values of the Sargan’ statistic are 0.103 and 1.452 for the net-affect and u-index, respectively, which leads us to accept the null hypothesis of the validity of instruments at standard levels (p=0.749 and p=0.228 for the net-affect and the u-index, respectively). We observe that the coefficients for participation in voluntary activities are not statistically significant at standard levels in any of the columns in Table 6. Thus, once we take into account reverse causality, we find that participation in voluntary activities has no causal effect on the happiness of individuals.
Finally, we have done an analysis where we consider only those who volunteer during the day. If the anticipation of volunteering makes people happy, we should find that volunteers achieve greater daily happiness during activities prior to volunteering. If volunteers are happier as a result of volunteering, we should obtain greater daily happiness via post-volunteering activities. Thus, we compare the levels of happiness of before and after the voluntary act, controlling for whether the non-voluntary activity of reference was done after any voluntary activity (reference) or before. After estimating equations (1) and (2), we find no evidence of differences between activities done before and after voluntary activities (results available upon request), which we interpret as evidence that voluntary activities do not affect daily happiness, but happier individuals are more likely to participate in voluntary activities, a result that is consistent with the IV results.
7. Voluntary activities: “time-composition” and “personality” effects.
Given the previously reported positive association between voluntary activities and daily happiness, in this Section we decompose the difference in daily happiness between volunteers and non-volunteers in two components: the “time-composition” and “personality” effects. The former captures the difference in the daily happiness that can be attributed to differences in the distribution of activities during the day. To the extent that different activities provide different levels of daily happiness to individuals, including voluntary activities that have been shown to be ranked among the five most enjoyable, the difference in daily happiness between volunteers and non-volunteers can be explained because those who volunteer during the day may differ in how they spend their time, compared to those who do not volunteer. The latter captures the variations in daily happiness that can be attributed to differences in happiness obtained while engaged in similar activities. It could be that volunteers and non-volunteers report different levels of daily happiness when engaged in the same activities during the day. Such differences are defined at the individual level, as they depend on inter-personal differences in same activities, and thus we call this the “personality” effect.
We first analyze differences in the time devoted to different activities and the associated net-affect and u-index, broken down by participation in voluntary activity, shown in Table 7.12 The net-affect and u-index are computed at the episode level, while the time devoted to each of the activities is computed at the diary level (i.e. for each individual, we sum the time devoted to the reference activity during the day, and compute the overall time devoted to this activity by all individuals in the reference sample). The time devoted to the different activities is measured in minutes per day. The classification of activities corresponds to the basic classification of Aguiar and Hurst (2007) and Gimenez-Nadal and Sevilla (2012). Though sleeping is not reported, we computed the time devoted to it to see if there is any difference between the two groups of individuals.
Activities are sorted following the ranking in the net-affect shown in Table 2, although we place voluntary activites in the first position. We compute the difference between the two groups in the net-affect, the u-index, and the time devoted to each activity, and we compute the p-value of that difference (t-type test of equality of means), where a p-value lower than 0.05 indicates that the difference between the two groups is statistically significant at standard levels.
Regarding the time devoted to the different activities, we find that volunteering individuals take up 102.733 minutes during the day, and that, compared to individuals who do not volunteer, they devote more time to out-of-home leisure (8.881 more minutes), religious activities (0.115 more minutes), teaching child care (1.790 more minutes), basic child care (7.477 more minutes), cooking and meals (4.011 more minutes), shopping (3.502 more minutes) and housework (7.187 more minutes). On the contrary, individuals who volunteered during the day of the survey devote less time to sports/exercise (4.609 fewer minutes), gardening/pet care (4.189 fewr minutes), leisure travel (2.524 fewer minutes), TV watching (58.645 fewer minutes), commuting/work related activities (9.045 fewer minutes), main work (109.288), job search (2.648 fewer minutes) and own medical care (4.010 fewer minutes). Thus, in comparison to individuals not doing voluntary activities, those who volunteer spend around 120 more minutes on voluntary activities, out of home leisure, religious activities, supervisory child care and basic childcare, while they employ 177 fewer minutes doing market work, activities related to work, and TV watching. According to Table 2, the former group of activities produce higher levels of individual happiness, while the latter group of activities produce lower levels of happiness. Thus, this “composition” effect may explain the differences in the average daily happiness of those who devote time to voluntary activities.
If we look at differences in the net-affect and the u-index between the two groups, we find that for both measures, individuals volunteering during the day of the survey obtain a higher level of happiness from commuting/work related activities (differences in the net-affect and uindex are 0.406 and -0.084) and housework (differences in the net-affect and u-index are 0.744 and -0.122), while they obtain less happiness from supervisory child care (differences in the netaffect and u-index are -1.583 and 0.146). Thus, it seems that volunteers obtain a different level of daily happiness while engaged in commuting/work related activities, housework, and supervisory child care, pointing to a “personality” effect: when engaged in similar activities, those who devoted time to voluntary activities during the day of the survey feel happier than those who did not.
To disentagle the extent to which each of the effects contributes to the previously reported differences in daily happiness between the two groups, we follow Knabe et al. (2010) by decomposing the difference in our two measures of daily happiness by a simulation. First, we calculate how the average experienced utility of all volunteers would change if they did no voluntary activities, under the assumption that they experience the average utility of a nonvolunteer in all activities, but maintain the time schedule they had when they were doing voluntary activities (i.e. well-being as counterfactual). 13 The difference between the experienced utility while doing voluntary activities and its value after this hypothetical change corresponds to the “personality” effect; the remaining difference in the experienced utility associated with not volunteering can then be assigned to the “time-composition” effect.
The decomposition of these two effects, using the well-being measures as counterfactual, is shown in Panel A of Table 8. The net-affect for those who devote time to voluntary activities, using the net-affect values of those who do not volunteer, is 2.907 (see Table A4 for details of calculations), which when compared with the real value of 3.113 of the net-affect of this group, indicates that the “personality” effect accounts for 0.179 of the 0.385 difference in the netaffect. Thus, of the 0.385 difference in the net-affect between the two groups, the “personality” effect explains 46 % (0.179 out of 0.385), while the “time-composition” effect explains 54% (0.206 out of 0.385) of the difference. We find that the u-index for those who volunteer, using the u-index values of those who do not, is 0.245, which when compared with the real value of
0.194 of the u-index of this group indicates that the “personality” effect accounts for -0.020 of the -0.071 difference in the u-index. Thus, of the -0.071 difference in the u-index between the two groups, the “personality” effect explains 28 % (-0.020 out of -0.07) while the “timecomposition” effect explains 72% (-0.051 out of -0.07) of the difference. If, rather than considering individuals who volunteer as the reference group, we consider those who do not, and use the affective measures of individuals who do volunteer as the counterfactual (Panel B), we observe that for the net-affect and the u-index, the “personality” effect accounts for 0.042 and -0.018 of the difference between the two groups, respectively, representing 11% and 25% of the difference.
In sum, we find that the difference in daily happiness between volunteers and nonvolunteers can be decomposed into two components: the “personality” and “time-composition” effects. While the “time-composition” effect is large, as it explains between 54% and 89% of the difference in daily happiness between the two groups, the “personality” effect on other activities is smaller but still significant, as it explains between 11% and 46% of the observed difference in daily happiness during the day.
Here we must consider that the time devoted to sleep is slightly different, although the difference is not statistically significant at standard levels, between the groups. Rather than considering the total time devoted to the activity, we multiply by the net-affect/u-index of reference for each activity, and then divide the sum of the product by the waking time, to obtain a measure of experienced utility during the day. We directly divide the average time devoted to the activity divided by the waking time (820 and 880 minutes per day for volunteers and non-volunteers, respectively) and multiply for the corresponding affective measure. Since we do not observe experienced utility ratings for voluntary activities for non-volunteers, we assume that volunteers maintain their original values.
We offer a decomposition of the two channels that may help to explain the difference in daily happiness between volunteers and non-volunteers, on a daily basis. However, this decomposition only provides lower and upper limits of the part each effect is able to explain. In the case of the “personality” effect, while 11% may seem a small effect, 46% is significant.
Additionally, we do not know what factors comprise this “personality” effect (social networks, extraversion, motivations…). Additionally, the “time-composition” effect shows that volunteers devote less time to market work activities, which may indicate that they also work less.
However, as shown in Section 5, this difference does not depend on the day (working or nonworking).
8. Conclusions Working for no pay is a widespread economic activity whose significance is not yet fully understood. In the United States, around 50% of all adults do some kind of volunteering, equivalent to five million full-time jobs (Anheier and Salamon, 1999). Thus, understanding the reasons why individuals do such work has been the focus of a significant amount of research. In this paper, we analyze voluntary activities, with a focus on how participation of this type is associated with individual happiness.
Using the Well Being Module of the American Time Use Survey (ATUS) 2010, we find that participation in voluntary activities is positively associated with the daily happiness of individuals, as individuals who volunteer report higher values of the net-affect, and lowervalues of the u-index. These results are maintained when we control for the scaling effect of individuals, and when we exclude episodes of voluntary activities from the analysis, and the variations are not driven by participation in religious activities. Using cross-state variations regarding tax deductions for charitable contributions as our main instrument, we apply an IV estimation to deal with the issue of reverse causality, and our results indicate that happier individuals are more likely to participate in voluntary activities. Finally, we decompose the difference in daily happiness between volunteers and non-volunteers in a “time-composition” and “personality” effect, and find that the “personality” effect explains between 11% and 46% of the difference in happiness between volunteers and non-volunteers during the day.
To the extent that happier individuals are more likely to volunteer, it may be of interest for employers, given existing research that has found a positive relationship between happiness and worker productivity (Freeman, 1978; Bockerman and Ilmakunnas, 2012; Oswald et al., 2014).
Studying the extent to which volunteers are more productive in the firm posits an interesting line of future research. If volunteers are more productive workers, employers may consider this voluntary work as a marker for productivity. However, volunteers devote less time to market work, an effect that appears to be negative in terms of productivity. It would be of interest to see which effect is dominant. Furthermore, the “personality” effect may account for a significant part of the difference between volunteers and non-volunteers. Analyzing the extent to which volunteers differ in their non-observable characteristics (motivations, social context, social networks, trust…) may be of interest, as in this paper we offer both lower and upper bounds of the effect. Additionally, the fact that volunteers devote comparatively less time to watching TV may help media channels to focus their commercial campaigns and programming according to audience demographics. We leave these issues for future research.