Job Satisfaction, Work Time, and Well-Being Among Married Women in Japan

Abstract

This paper examines the relationship between paid work time and other time use of working married women with children in Japan and two aspects of well-being: job satisfaction and stress. The study demonstrates that rather than the amount of daily paid work time, both the gap between actual and desired work time and the intrinsic utility derived from paid work as an activity appear to be the key time-related variables affecting Japanese women's job satisfaction. The paper also shows that paid work time has multiple spillover effects on stress. It discusses the tradeoffs that married women with children in Japan make to stay in employment and the consequences for employer strategies and public policy.

Keywords:

• Job satisfaction
• stress
• time use
• well-being
• work time
• JEL Codes: J2, J22, J28

Acknowledgments

We thank the staff of the Social Science Data Archive of the Information Center for Social Research, Tokyo University for their assistance with the data. We have received financial support from the Tezukayama Education Foundation, Japan Bank Research Foundation, and the Japan Economic Research Foundation. For many insightful comments we particularly thank Kuramitsu Muramatsu, Fumio Ohtake, Kerstin Schneider, and Masanori Tahira. We are also indebted to the anonymous reviewers and staff of the Journal of Feminist Economics for many valuable comments and suggestions. Any errors or omissions in the paper are the responsibility of the authors.

Notes

¹According to the 2001 Japanese national time-use survey (carried out once every five years), once unpaid work was taken into account, married working women had an average work day of nine hours and two minutes and married working men of eight hours and twelve minutes (Bureau of Statistics, Management and Coordination Agency, Japan 2001).

²In the 2001 national time-use survey in Japan, working married women averaged more than four hours of housework a day, but working married men averaged only twenty-eight minutes of housework per day (Bureau of Statistics, Management and Coordination Agency, Japan 2001).

³Where medical data is available, the reliability of self-reported stress can be checked against objective measures. Maruyama and Morimoto (1996) found that respondents with higher subjective stress levels do in fact have less well-functioning immune systems, a sign of experiencing higher stress levels.

⁴All the survey questions and other details can be viewed at the Social Science Japan Data Archive of the Information Center for Social Science Research at Tokyo University website: http://ssjda.iss.u-tokyo.ac.jp/en/index.html (data set no. 46).

⁵See Japan Institute of Life Insurance (1992) for the original survey report. The Institute regularly carries out nationwide statistical surveys and has deposited a number of data sets in the archive.

⁶See Equal Employment, Children and Families Bureau, Ministry of Health, Labour, and Welfare, Japan (2005: Figure 12). The rate began to decline after 1998 as labor market conditions worsened.

⁷Measurement of time use based on recall is regarded by social scientists as an inferior method compared to the alternative of time diaries (Michelle Budig and Nancy Folbre 2004), although the burden on respondents of the latter make it unlikely that the time diary method will be used in surveys other than those that have a particular focus on time use. A problem is that recall-based measurement is known to be plagued by over-reporting of activities so that total time adds up to more than the time actually available (Thomas Juster and Frank P. Stafford 1991). Our data avoid this problem by the restriction on the sum of total hours placed on respondents. A further issue, discussed at length by John P. Robinson and Geoffrey Godbey (1999: 59), is that people who feel rushed may report work hours, both paid and unpaid, as being considerably longer than their actual hours. If this is the case then there will be some downward bias in our estimated parameters.

⁸To ensure consistency with our theoretical framework we exclude as inconsistent the small handful of respondents who gave their desired paid work time as zero, on the grounds that as long as labor is not forced, this time allocation is within the possibility set.

⁹When people are asked about what activities make them feel happy or fulfilled, paid employment receives a relatively high ranking. For example the Economic Planning Agency, Japan (1992) survey of national life preferences asks, “When do you feel happy?” The response, “when working,” ranks fourth overall after “when talking with family,”“when enjoying leisure,” and “when engaged in hobbies” and ahead of “when spending time with friends” and “when relaxing and not doing anything in particular.” These rankings are similar to those Thomas Juster (1991) found in the United States. The existence of intrinsic work utility is an important labor supply issue in need of greater attention (Bruce E. Kaufman 1999).

¹⁰In Japanese kezaiteki-yutori or literally “economic room.”

¹¹Age, its square, and educational levels are standard variables in socioeconomic analyses but are not included in the estimation equations reported below as these variables were not statistically significant in either equation.

¹²In a simple example with a single explanatory variable x, suppose that p₁ is the probability of being satisfied with one's job and p₂ the probability of being reasonably satisfied. Normalizing the logistic procedure on the first response we fit an estimation model such that: log(p₁/1-p₁) = α₁ + βx, and log(p₁+p₂/1-p₁-p₂) = α₂ + βx and so on. Note that the intercepts increase in size across the estimated equations, but they have a common slope parameter.

¹³The desired time-use equations estimated by OLS included variables such as actual hours and income levels as a proxy for the wage, intrinsic work utility variables, attitudes to work and housework, dummies for help from spouse with housework, husband's employment status, presence of children who are not yet of school-going age, and the amount of leisure spouses spend together as a proxy for happiness in marriage.

¹⁴The estimation output is available on request.

¹⁵We included a third equation for desired time use in this test. The output is available on request.

¹⁶Although the estimated parameters of the control variables are not reported in detail in Table 4, we note that among the paid employment status dummies, self-employed married women with children (both agricultural and non-agricultural), have parameter values much larger than the others. The higher job satisfaction of the self-employed women is consistent with previous research on entrepreneurial satisfaction (David G. Blanchflower and Andrew J. Oswald 1998; Shigemi Yahata 1998). We can only speculate, but since we are controlling for work time and overwork, these results may in part reflect the importance for job satisfaction of flexibility in the distribution of work time over the day, being able to take days off, and/or work autonomy.