Volunteer activity and the demands of work and family.

The purpose of this paper is to analyze the determinants of volunteer activity through the theoretical perspective of a household production function that encompasses both the "warm glow" and the productive aspects of volunteering for various groups in the employment relationship. In the empirical work, emphasis is placed on how volunteering varies by characteristics of work and family. Especially important will be our estimate of the effect of working time and the balancing of familial obligations on the likelihood of volunteering. Comparisons will be made with the existing Canadian literature on the topic. (5)

VOLUNTEERING AND THE HOUSEHOLD PRODUCTION FRAMEWORK

Given the emphasis on the relationship between work and family, the empirical results will be interpreted through the lens of the theoretical framework of the household production function where households value goods, leisure and charitable activity as a collective good. Charitable activity is "produced" via household inputs of money and volunteer time. This highlights the importance of substituting money for time within the household and over different periods in the lifecycle, and it highlights that households may economize on scarce household time both by "consuming" fewer charitable activities that involve time (substitution in consumption), and by substituting money for time in "producing" charitable activity (substitution in production).

The household production function perspective incorporates the fact that the decision to engage in charitable activity can be based on altruistic reasons--to help a cause--with different people having different amounts of altruism as well as different causes. Or it can be more of an investment decision with the expectation of some private return in terms of such factors as reputation, resume building, networking, experience or a reciprocal favour. The household production function approach also emphasizes that volunteering is a family decision affected by family income as well as by the opportunity cost of time of different family members and the extent to which different family members allocate their time to labour market versus household activity. Families that require more of their time at home (e.g., for the care of very young children) are less likely to have time for outside charitable activities, especially in the form of unpaid volunteer time. On the other hand, some unpaid work may be associated with the activities of other family members, such as school or club or team activities of children as they grow older.

The household production function perspective also reminds us that variables such as a person's wage rate can have opposing effects on volunteering. Other things being equal, high wage persons may be less likely to volunteer because of the high opportunity cost of their time. However, the "production function" aspect of the theory highlights that their high wage also means that they may be more "productive" in certain types of volunteer activity and hence may volunteer more, or be pressed into volunteer service where their skills are important (Freeman 1996).

In essence, the household production function perspective provides a theoretical framework that is useful for interpreting the empirical relationships between volunteering and characteristics of work and family. Rather than using it in advance to set out the expected relationships between volunteering and each of the explanatory variables, it will be used to interpret the empirical relationships that emerge and that will be discussed subsequently.

DATA AND ECONOMETRIC PROCEDURES

The econometric analysis is based on Cycle 9 of the Canadian General Social Survey (GSS) of 1994. That data is ideally suited for analyzing the relationship between volunteering and work and family since Cycle 9 provides information on whether people volunteered and if so, for how many hours. It also has information on a wide range of work and working time characteristics (including union status and attitudes towards work) as well as family characteristics that can influence volunteer activity. It has measures that can capture economic concepts such as the opportunity cost of time as well as non-labour income or wealth. As well, it has measures that can reflect whether individuals have "fixed-effect" unobserved preferences that simply make them more prone to volunteer. The GSS also has a wide range of personal and demographic characteristics that can be important control variables, and that yield interesting information in their own right.

Information with respect to voluntary activity from the GSS is based on the survey question:

The measure of activity refers to formal activity and does not include, for example, informal activity not provided through an organization including family member care. (6)

In the GSS data set, the hours of volunteer activity were provided only in categories. They were converted here into continuous numbers by using the midpoint of each of the nonzero categories. That is, 1 to 4 hours was assigned 2.5 hours; 5 to 9 hours was assigned 7 hours; and 10 hours or more was assigned 15 hours. This category had the smallest number of respondents, which should make the assignment of hours in that group less consequential. Respondents in the categories "did not know" and "not stated" were omitted from the analysis.

Because of the very crude measure of the hours of volunteer activity, our empirical work focuses on estimating the probability of engaging in any volunteer activity; that is, on a binary coded dependent variable coded 1 if the person engaged in any volunteer activity, and zero if they did not. Tabulations (Appendix table 1 available on request from the authors) of the average hours of volunteer activity by the different explanatory variables yielded remarkably little variation in hours of volunteering across the different characteristics of the respondents. There was little deviation from the average hours of volunteering of 5.8 hours (e.g., 4.2 hours for employees whose spouse was ill or on maternity leave, to 6.4 hours for employees in public administration). For these reasons, our empirical analysis focuses on estimates of the probability of volunteering.

As is appropriate given the binary nature of the dependent variable, logistic regression is employed. Since the logit coefficients by themselves do not directly give the change in the probability of volunteering associated with a unit change in the explanatory variable, such marginal effects are calculated, evaluated at the mean probability of volunteering. (7)

In most cases, the coding of the variables is straightforward and readily corresponds to the variable name as used in the tables. In some cases, however, the definition is less obvious, and sometimes imputed values had to be calculated. (8)

THE DETERMINANTS OF VOLUNTEER ACTIVITY

Our discussion of the empirical results will focus on the effect of the different explanatory variables on the probability of volunteering as given in column 3 of Table 1. The magnitude of these effects should be interpreted relative to the average probability of volunteering as given by the mean value of the dependent variable of .212--that is, 21.2 percent of the respondents volunteered in the week prior to the survey week.

The Effect of Earnings and Wealth Measures

The earnings and wealth measures have the expected effects based on the household production function perspective. That is, wealthier people volunteer more as indicated by the positive coefficient on each of the measures of wealth household income (less the respondent's own employment earnings); being in receipt of interest income; and being a home owner (although the later is statistically insignificant). The effect of being in receipt of interest income (and hence of having investments) is especially large, with such persons being about .05 more likely to volunteer--a large amount relative to the average probability of .21. In essence, volunteering appears to be a "normal good"--something we do more of if we can afford to do so. The same result also emerges in the only other comparable Canadian econometric study to include an income measure (Vaillancourt 1994).

Higher wages have no significant impact on the probability of volunteering. This likely reflects the net effect of income and substitution effects working in opposite directions. That is, higher wages increase volunteer activity by enabling people to afford to volunteer (income or wealth effect) but they also reduce volunteer activity by raising the opportunity cost of time spent volunteering (substitution effect). The coefficient on our wage measure is a gross wage effect, capturing both the income and substitution effects. The small negative magnitude of the net impact suggests that these income and substitution effects roughly offset each other. This highlights, however, that the pure substitution component is negative--that is, the higher opportunity cost of time leads to reduced volunteer activity. (9)

The Effect of Workplace and Working-time Characteristics

Working time arrangements can have substantial impacts on facilitating or inhibiting volunteer activity--especially important given the changing nature of work-time arrangements often associated with dual earner families. Relative to the regular day shift, a split shift increases the probability of volunteering by .08, a substantial amount relative to the average probability of .21. Split shifts tend to be "split" between early morning and the end of the day, leaving much of the day free for volunteering. In contrast, the uncertainty associated with rotating shifts reduces the probability of volunteering. Somewhat surprisingly, flextime does not seem to facilitate volunteering, perhaps because slight variations in start and end times do not substantially free up time for volunteering.

Print Get the Mag Weekly Updates