doi:10.1016/j.ssresearch.2004.04.006 
Copyright © 2004 Elsevier Inc. All rights reserved.
Is there a flexiglass ceiling? Flexible work arrangements and wages in the United States*1
Kim A. Weeden
Department of Sociology, Cornell University, 323 Uris Hall, Ithaca, NY 14853, USA
Available online 10 June 2004.
References and further reading may be available for this article. To view references and further reading you must
purchase this article.
Abstract
Employment contracts that offer employees flexibility in their work schedules or locations are often touted as politically and economically viable solutions to contemporary employees' needs to balance work, leisure, and family. However, little is known about their career consequences or their impact on aggregate levels of gender inequality in labor market outcomes. This paper outlines possible sources of an association between flexible work and wages, then quantifies this association using data from the 2000 and 2001 Current Population Surveys (CPS). Results indicate that flexible-work employees earn wages that are at least equal to, and often higher than, their fixed-schedule and fixed-location counterparts. These wage premiums are greater in nonmanual occupations, but do not vary substantially by gender or parental status. The results contradict the notion of a “flexiglass ceiling” in wages, but also suggest that flexible work arrangements do little to reduce the gender gap in pay or the motherhood wage penalty.
Table 1. Means/standard deviations or distribution of flexible work, human capital, demographic, and employment variables, by gender and sample
Note. See text for data sources. The data are unweighted.
Table 2. Mean hourly wages in 2001 of full-time employees, by work arrangement, parental status, and gender
Notes. N(men)=4861; n(women)=3937. The data are weighted by the CPS outgoing rotation group weight, rescaled to retain original sample size.
Table 3. Estimated coefficients from OLS regressions of (logged) wages on flexible work schedules, human capital, demographic, and labor market traits of full-time employees, by gender
Notes. N(men)=4861, n(women)=3937. Coefficients that differ significantly (p<.05) by gender are outlined. Robust standard errors are reported in parentheses. +p<.10, *p<.05, **p<.01, and ***p<.001, two-tailed test.
Table 4. Estimated coefficients from OLS regressions of (logged) wages on flexible work locations, human capital, demographic, and labor market traits of full-time employees, by gender
Notes. N(men)=4861, n(women)=3937. Coefficients that differ significantly (p<.05) by gender are outlined. Robust standard errors are reported in parentheses. +p<.10, *p<.05, **p<.01, and ***p<.001, two-tailed test.
Table 5. Estimated flexible work coefficients for full-time employees, by gender and occupation category
Notes. N(men)=4861; n(women)=3937. Estimated coefficients that differ significantly (p<.05) by gender are outlined. Robust standard errors are reported in parentheses. Models also fit the human capital, demographic, labor market, and detailed occupation variables (see Model 2, Table 4). Too few cases are available to estimate the flexplace coefficient for female operatives and laborers; +p<.10, *p<.05, **p<.01, and ***p<.001, two-tailed test.
Table 6. Interaction of flexible work arrangements and union or public sector employment, by gender
Notes. N(men)=4861; n(women)=3937. None of the coefficients differ significantly (p<.05) by gender. Robust standard errors are reported in parentheses. All models fit the main effects of the human capital, demographic, labor market, and detailed occupation variables, the coefficients of which differ trivially from those reported for Models 2a and 2b (see Table 3 and Table 4). +p<.10, *p<.05, **p<.01, and ***p<.001, two-tailed test.
Table 7. Estimated coefficients from models regressing (logged) wages on flexible work schedules, human capital, demographic, and labor market traits, and lagged wages of full-time employees, by gender
Notes. Coefficients that differ significantly (p<.05) by gender are outlined. Robust standard errors are reported in parentheses. +p<.10, *p<.05, **p<.01, and ***p<.001, two-tailed test.
Abstract
Purchase PDF (276 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (4)
Purchase PDF (66 K)
