Taking Its Toll: The Influence of Paid and Unpaid Work on Women's Well-Being

Abstract

This paper examines gender differences in the impact of paid and unpaid productive activities on well-being. Using recent Canadian data, we examine the time spent by prime-age women and men (25 – 54) on paid work, childcare, eldercare, household work, volunteering, and education, and then assess its impact on stress and work-life balance. Using multivariate analyses, we show that women's greater hours of unpaid work contribute to women experiencing more stress than men, and of that work, hours spent on eldercare and housework are more stressful than those spent on childcare. We also examine the influence of job characteristics and spouses' paid and unpaid work time on stress. Neither spouse's unpaid work nor most job characteristics alleviate stress, once work hours are controlled. However, the evidence suggests that women, more so than men, use strategies such as self-employment to improve work-life balance.

Keywords:

• Stress
• work-life balance
• unpaid work
• women's health
• caregiving
• intra-household inequality

ACKNOWLEDGMENTS

We gratefully acknowledge the financial support of the Canadian Institute for Health Research (CIHR) through the “Healthy Balance Research Programme: A Community Alliance for Health Research on Women's Unpaid Caregiving.” We also thank our colleagues in the Healthy Balance program for their support and extremely useful comments. An earlier version of this paper was presented to the Healthy Balance Equity Reference Groups and has benefited from comments received at that time.

Notes

JEL Codes: I1, J16, J22

Thus, of course, individuals without telephones are excluded, though sampling weights based on the census take this into account.

We also carried out all of the analyses discussed here using a question that asks: “How often do you feel rushed? Would you say that it is: (1) Every day? (2) A few times a week? (3) About once a week? (4) About once a month? (5) Less than once a month? (6) Never.” The major qualitative lessons learned were essentially the same. However, there are interesting subtleties of difference, which we plan to examine further in future versions of this paper.

“Full-time/full-year” means that the respondent reported 30 or more “usual weekly hours” of paid work for at least 50 weeks in the previous year. “Not in the labor force” means that the respondent reported no weeks of paid work. “Part-time or part-year” includes individuals usually working less than 30 hours per week or fewer than 50 weeks in the survey year.

We focus upon usual weekly hours, since previous work (e.g., Marshall 1993; Shelley Phipps, Peter Burton, and Lars Osberg 2001) has suggested that weekly hours of work rather than annual hours are most important for measuring time stress. This relates to the fact that most caregiving cannot be postponed until paid work responsibilities are less onerous. Moreover, this is particularly true for traditionally “female” tasks (preparing meals for the children) as compared with traditionally “male” tasks (mowing the lawn).

The public-use version of the GSS provides income information in twelve intervals. Each household is assigned the income level from the midpoint of the appropriate interval, with the exception of households in the highest income category (over 100,000 1998 Canadian dollars). These households are assigned the median income of all households in the Survey of Consumer Finance with incomes over 100,000 dollars ($138,675). Poverty status is then calculated according to the Statistics Canada Low-Income Cut-Offs for residents in urban areas with populations between 100,000 and 500,000. We have experimented with other ways of controlling for household income (e.g., including income as a continuous variable in linear form and in quadratic form). None of our substantive results appears to be affected by these modeling choices. Hence, for simplicity, we report only specifications with the dichotomous poverty status variable.

Specifically, respondents were asked, “Are you limited in the amount or kind of activity you can do at home, at work, or at school because of a long-term physical or mental condition or health problem?”

We also know region of birth for both the respondent and the respondent's parents. We experimented with both variables with little success given extremely small cell counts.

The base case uses mean hours for all women. Dummy variables are all set equal to zero. We also conducted all the simulations reported here from a base that uses mean hours for all men. Estimated marginal effects were essentially no different, and so in the interests of space we do not report them here.

The co-author of this paper with three children in the 5 to 12 category finds this result extremely reasonable!

We also experimented with poverty-work hours interaction variables, but these were not particularly important, especially for women.

We also attempted to conduct separate analyses for single respondents, since we have a particular interest in the stresses associated with paid and unpaid work for lone mothers. Unfortunately, sample sizes were too small to provide reasonable estimates.

Unfortunately, the unpaid work questions that the respondent answered about his/her spouse are not identical to those answered by the respondent about him/herself.

Respondents were asked about the “main activity” of their spouse in the last seven days. Answers included: working at a paid job or business; looking for paid work; going to school; household work/caring for child; retired; and “other” (including maternity/paternity leave and long-term illness). To avoid classifying spouses on maternity leave together with those experiencing long-term illness, we have grouped those answering “other” but with a child aged less than 1 year together with those who are not in a “long-term illness.”

Of course, the two figures are not directly comparable as they reflect different samples of married men/women (i.e., the ones who are themselves survey respondents versus the ones whose wives/husbands are survey respondents). However, they do represent the same individuals in the population. One explanation for the higher hours reported by respondents themselves is that they include overtime while spouses answer in terms of a standard workweek.

Husbands with a long-term illness are reported to do more weekly hours of in-home housework (6.3 hours) than husbands on average (4.7 hours). However, hours spent on “other” housework, which includes yard work, working on the car, etc., are less for husbands with a long-term illness. Women with ill or disabled husbands are also less likely to have children at home.

Women with disabilities in our Equity Reference Groups found these findings plausible, arguing that taking on the care of a disabled partner requires men to step far outside their traditional responsibilities and hence such caregiving is particularly stressful for men.

We have grouped these options into the three categories as described above as a result of sample size considerations.

For example, women who work irregular shifts report, on average, 24.3 hours per week compared to 37.8 hours per week for those who work regular workdays. Men who are self-employed work 54.7 hours per week on average, compared to 42.5 hours per week for the non-self-employed. Men with multiple paid jobs report 58.9 hours per week, compared to 43.9 for those with just one paid job.

Additional information

Notes on contributors

Martha MacDonald

JEL Codes: I1, J16, J22