«IZA DP No. 8764 PAPER Voluntary Activities and Daily Happiness in the US J. Ignacio Gimenez-Nadal DISCUSSION Jose Alberto Molina January 2015 ...»
In order to better understand the greater daily happiness of volunteers vis-à-vis nonvolunteers, we explore how pleasant single activities are perceived. Table 2 shows a list of 26 activities ranked by the average value of the net-affect (see Appendix A1 for categorization of these activities). The activities at the top are the most enjoyable, while the activities at the bottom can be considered the least enjoyable. Together with non-basic childcare, religious activities, and out-of-home leisure, voluntary activities are in the group of activities that are most enjoyable, with average net-affect and u-index of 3.483 and 0.194, respectively, with this being consistent with prior results (Kahneman et al., 2004; Krueger, 2007;2009; White and Dolan, 2009; Robinson, 2014).
which increases the daily happiness of individuals. We address in Section 5 the extent to which this difference in voluntary time can explain the difference in daily happiness.
However, the differences in daily happiness are raw differences, not taking into account that socio-demographic differences may partially or totally drive the difference in daily happiness between volunteers and non-volunteers. In order to net out the effect of voluntary activities from the effect of other socio-demographic characteristics, we use several explanatory variables aimed at capturing differences in household and personal characteristics across respondents.
We include gender (male), age and its square, dummy variables for university and secondary education, dummies for working full- and part-time, a dummy to control for the presence of children under 18 in the household, a dummy to indicate whether the respondent is
married/cohabiting, household income, and dummy variables for region of residence (ref.:
West). 7 Secondary education is defined as having high school level, while university education is defined as having some college, a college degree, or more. Household income refers to the combined income of all family members during the last year and includes wages, net income from business, farm or rent, pensions, dividends, interest, Social Security payments, and any other money income received by family members who are 15 years of age or older. Household income is coded with income brackets with the following values: 1 “Less than $5,000”, 2 “$5,000 to $7,499”, 3 “$7,500 to $9,999”, 4 “$10,000 to $12,499”, 5 “$12,500 to $14,999”, 6 “$15,000 to $19,999”, 7 “$20,000 to $24,999”, 8 “$25,000 to $29,999”, 9 “$30,000 to $34,999”, 10 “$35,000 to $39,999”, 11 “$40,000 to $49,999”, 12 “$50,000 to $59,999”, 13 “$60,000 to $74,999”, 14 “$75,000 to $99,999”, 15 “$100,000 to $149,999” and 16 “$150,000 and over”. We consider the mid-point of each interval, and $150,000 for the last interval, and we apply the log of the value of each interval to allow for non-linearities in the effect of income.
Furthermore, we include information on whether the activity of reference was done in the presence of others, with being alone as our category of reference. The reason is that the existing literature shows that activities done in the presence of others provide greater daily happiness compared to actitivities done alone. Also, volunteering often involves spending time with others, which is emotionally beneficial. Thus, we include dummy variables to control for whether the activity was done in the presence of household children, the spouse/partner, any other household adult, other close friends, or co-workers. Alternatively, we control for the time spent during the day with others at the diary level.
We also include day-of-week dummies (ref.: Friday) to control for the fact that the time restriction may become more binding during the week, as people who work normally must accomplish their work responsibilities on weekdays, and thus voluntary activities may be more See Appendix Table A2 for summary statistics of the variables in our sample personally enriching during the week, or more abundant at weekends. Figure 1 shows the distribution of voluntary activities in our sample for the seven days of the week, and it can be seen that around 25% of voluntary activities are done on Sundays. This is consistent with the previous hypothesis, that time restrictions may become more binding during the week, and people volunteer more during the weekends. This may also be related to religious participation, as many voluntary activities are done on Sundays at church, with this being explored later.
We find that, in comparison with individuals who do not do any voluntary activity during the reference day, those who do present a lower proportion of men, are older, a higher percentage have a university education, a smaller percentage work full-time as opposed to parttime, a higher percentage have at least one child under 18 and live in couple, the household income is higher, and they spend more time with children, the spouse/partner and friends, while spending less time with co-workers. These differences show the importance of controlling for all these characteristics in analyses that follow.
Finally, Kahneman and Krueger (2006) show that the level of tiredness increases during the day. That is, in late hours, individuals report being more tired than in earlier hours. This may affect the differences between individuals if the selection of activities for those who volunteer was different compared to those who do not. To control for this time effect, we include in our analysis the time band of the day when the activity was done, and its square, measured in 1-hour time bands (e.g., 12-1 am, 1-2 am…).
4. Relationship between volunteering and daily happiness In this Section, we study the relationship between voluntary activities and daily happiness, by analyzing the relationship between feelings reported by the respondent and participation in voluntary activities on the same day. The large number of randomly selected episodes provides us with a solid framework for the analysis of this relationship. We estimate Ordinary Least Squared (OLS) models of the happiness measure (net-affect and u-index) associated with episodes as follows: 8 (1) where represents the happiness measure of individual “i” in episode “j”, is a dummy variable that indicates whether respondent “i” engaged in any civic/voluntary activity (1) or not (0) during the day. Thus, Voluntaryi takes value “1” if we observe positive time devoted to voluntary activities in the diary of respondent “i”, and value “0” if we do not observe In our regressions, we will assume that happiness measures are cardinal, an interpretation that is common in the literature on well-being (Ferrer-i-Carbonell and Frijters, 2004).
time devoted to such activities. We have also estimated our models including the time devoted to voluntary activities throughout the day, measured in hours per day. According to the previously hypothesized relationship between feelings and voluntary activities, we should expect for the net-affect (larger values indicate greater differences between happiness and negative feelings), and for the u-index (happiness overcomes the feelings of stress, tiredness, sadness, pain). represents a vector of socio-demographic characteristics, while represents the error terms.
The window length used in the ATUS, as in other time use surveys, may lead to measurement errors in the volunteering behavior of individuals, since individuals are asked what they did on the previous day, and it may well be that individuals do voluntary activities weekly or monthly, but did no voluntary activities on the day before the survey. Thus, the shortness of the reference period for time diary studies potentially limits their usefulness for estimating the distribution of activities across populations, and the relationship between activities and daily happiness. Frazis and Stewart (2012) show that OLS models are preferred in the analysis of time-allocation decisions, compared to tobit models, as the latter yield biased results. Gershuny (2012) argues that traditional diary studies can still produce accurate estimates of mean times in activities for samples and subgroups, at least in the short-run. Foster and Kalenkoski (2013) compare estimation results from OLS and tobit models on childcare time, and they obtain almost identical results. Thus, we rely on OLS linear models.
We further estimate a Random-Effects (RE) linear model to take into account the scaling effect of individuals (Kahneman and Krueger, 2006). This implies that individuals may have a different conception of what the scale of measurement really refers to, or may interpret the scale
differently. We estimate the following equation:
(2) represents the feeling of individual “i” in episode “j”, and αi represents the individual where effect. The time variation needed to estimate a panel data model is given by the fact that each respondent has 3 episodes, and each episode may be of a different nature (e.g., leisure, personal care, housework), which gives sufficient variation to estimate the model.
Results Table 3 shows the results of estimating Equations (1) and (2) on participation in voluntary activities throuought the day. Columns (1) and (3) refer to OLS models for the net-affect and uindex respectively, considering all activities. Columns (2) and (4) refer to RE models for the net-affect and u-index respectively, considering all activities. We observe a positive association between the net-affect and participation in voluntary activities on the diary day, and a negative association between the u-index and participation in voluntary activities on the diary day, with these associations being statistically significant at standard levels. Participation in voluntary activities on the day of the diary is associated with an increase and a decrease of 0.198 and
0.029 in the net-affect and the u-index, respectively, representing an increase and a decrease of 7 and 13 percent in the overall values. When we consider posible scaling effects, we observe that in the RE models, participation in voluntary activities on the day of the diary is associated with an increase and a decrease of 0.285 and 0.035 in the net-affect and the u-index, respectively, representing an increase and a decrease of 10 and 16 percent in the overall values.
Hence, we find that for the net-affect and for the u-index, which is consistent with the hypothesis that voluntary activities are positively related to the daily happiness of individuals.
In the analyzed relationship, it could be that those who devote time to voluntary activities obtain more daily happiness from their voluntary activities, as seen in Table 2, but in the other activities they report similar levels of daily happiness. Thus, we now estimate the same four models, excluding all episodes that were voluntary activities (we exclude 459 episodes).
Columns (5) and (7) refer to OLS models for the net-affect and u-index respectively, excluding episodes of voluntary activities. Columns (6) and (8) refer to RE models for the net-affect and u-index respectively, excluding episodes of voluntary activities. We observe that, for both the net-affect and u-index, we obtain lower, but still statistically significant, positive and negative associations with participation in voluntary activities on the day of the diary. Thus, the positive association between voluntary activities and increased daily happiness is still present, even if we exclude episodes of voluntary activities from the analysis. The results point toward voluntary activities increasing the happiness obtained during non-voluntary activities, which would be consistent with the consumption motive of volunteering, as it seems to increase the contemporaneous utility of individuals.
Table 4 shows the results when we estimate our models considering the duration of voluntary activities. In this case, we use an indicator of participation in voluntary activities that measures the hours spent in voluntary activities during the diary day. Here results are less clear cut, as we only obtain statistically significant coefficients of voluntary time in the case of the net-affect when we include voluntary activities in the regressions. Neither the u-index in all cases, nor the net-affect when we exclude voluntary activities, present a statistically significant association with time in voluntary activities. Thus, while participation in voluntary activities seems to be positively related to daily happiness, the time in voluntary activities is not, which can be explained by the fact that activities present diminishing marginal utility, at some point Gerhuny (2013), indicating that “prolonged exposure to highly enjoyable daily activities does not always foretell higher levels of cumulative subjective well-being, which is associated with balanced use of time rather than increased participation in individual activities” (Zuzanek and Zuzanek, 2014, pp. 1). Thus, for the rest of our analysis we will focus on participation in voluntary activities only.