Abstract
This study uses the International Social Survey Programme (ISSP) 2013 ‘Family and Changing Gender Roles’ module (N = 1773) to examine cross-country differences in the relationship between women’s part-time work and work–life conflict and job satisfaction. We hypothesize that part-time work will lead to less favorable outcomes in countries with employment policies that are less protective of part-time employees because the effects of occupational downgrading counteract the benefits of increased time availability. Our comparison focuses on the Netherlands and Australia while using Germany, the United Kingdom, and Sweden as benchmarks. Part-time employment is prevalent in all five countries, but has the most support and protection in the Dutch labor market. We find little evidence that country of residence conditions the effects of part-time work. Overall, the results suggest that part-time work reduces work–life conflict to a similar extent in all countries except Sweden. The effects on job satisfaction are negligible. We discuss the implications for social policies meant to stimulate female labor force participation.
Keywords
Introduction
Work demands and parenting ideologies both intensified in recent decades, making the combination of work and family responsibilities highly challenging for contemporary families (Bianchi and Milkie, 2010; Forsberg, 2009; Hays, 1996). Many parents, mothers in particular, cope by ‘scaling back’ in the work domain (Becker and Moen, 1999; Bianchi and Milkie, 2010; Jacobs and Gornick, 2002; Pettit and Hook, 2005; Thurman and Trah, 1990; Uunk et al., 2005). Some parents scale back by limiting the overtime they work, but many others switch to a part-time job.
Despite its popularity, part-time work is contested. It is often associated with occupational downgrading and disproportionately reduced earnings and career opportunities (e.g. Bardasi and Gornick, 2008; Gornick and Meyers, 2003). Prior research on the subjective effects of part-time work yielded inconclusive results and suggested that they vary across countries (Booth and Van Ours, 2009; Chalmers et al., 2005; Treas et al., 2011). This study aims to find out whether the subjective effects of part-time work are moderated by its national prevalence, protections, and quality. It contrasts the association between part-time work and (a) work–life conflict and (b) job satisfaction in the Netherlands and Australia, using Sweden, Germany, and the United Kingdom as benchmarks.
The Netherlands and Australia are prime examples of ‘part-time countries’. In prevalence, they ranked first and second in the most recent Organization for Economic Cooperation and Development (OECD) comparison with 74.7 percent of the Dutch and 55.4 percent of the Australian female workforce employed part-time (Abhayaratna et al., 2008: 221; OECD, 2010). However, they represent different types of welfare states, and this has implications for the protections and quality of part-time work. Dutch employees are legally protected and can adjust their working hours in their current job, whereas Australian employees often have to change jobs to reduce their hours, usually by transitioning to lower status sectors. Germany, the United Kingdom, and Sweden are useful benchmarks because they represent typical conservative, liberal and Nordic regimes (Ferragina and Seeleib-Kaiser, 2011) and are characterized by varying levels of part-time work prevalence and protection.
The main goal of this study is to investigate whether or not the consequences of part-time work on women’s well-being are contingent on the policy context. In other words, do women gain more from part-time work in policy contexts in which protections and the quality of part-time jobs is high? Or do institutional structures play a relatively small role? Our analyses use data from the International Social Survey Programme (ISSP) 2013 ‘Family module’ and center on two quality-of-life outcomes. First, we examine work–life conflict. Since many women scale back in order to free up time for their personal life, it is important to know whether part-time work does indeed have this intended effect. Second, we investigate the impact on job satisfaction to examine whether ‘gains’ in the personal domain are accompanied by ‘losses’ in the work domain.
We argue that part-time work is not uni-dimensional but can differ in nature and that its effects may vary as a consequence. The few prior studies which viewed the nature of part-time work as relevant when studying its effects are sparse in their statistical underpinnings (e.g. Plantenga, 1996; Workplace Research Centre (WRC), 2011; Yerkes, 2009: 704). To our knowledge, international comparative research has not yet addressed differences in the nature of part-time work. By investigating the role of institutional structures, this study follows the advice of Treas et al. (2011) to ‘understand what about countries makes women’s full-time employment a more or less satisfying experience’ (p. 126).
Although formal institutions, such as labor market characteristics and country-level policies, shape the context in which part-time work is embedded, it is likely, there are large differences within countries as well. The quality of part-time work may vary across sectors (Centraal Bureau voor de Statistiek (CBS), 2010), and employers’ support for part-time work may vary across organizations. Moreover, feelings of work–life conflict and job satisfaction depend on an individual’s frame of reference (Diener et al., 1999). For example, as compared to Australian women, women in the Netherlands may have higher expectations of part-time jobs because they feel more entitled to solid work conditions. If women indeed adapt their expectations to their context or meso-level factors are at play, cross-national differences in the effects of part-time work may be small or non-existent.
We proceed in four steps. We first provide a short review of the literature on part-time work and its consequences. Second, we set out our main conceptual idea and explain why and how we expect that the consequences of part-time work are contingent upon the country context. Third, we apply these general expectations to the countries in our study. Finally, we analyze whether the association between part-time employment and women’s perceived work–life conflict and job satisfaction indeed varies across countries. In doing so, we specifically investigate the role of occupational status.
Prior research on part-time work and its consequences
Definition and prevalence
Definitions of part-time work have varying cut-off points. The OECD defines jobs of one to 30 hours a week as part-time (http://stats.oecd.org/wbos), but other statistical bodies (including the Australian Bureau of Statistics (ABS) and the Dutch statistical institutes Sociaal en Cultureel Planbureau (SCP) and CBS) set the threshold at 35 hours. In the Netherlands, it is also common to distinguish between short-, medium-, and long-hours part-time jobs, although the thresholds are set at varying points (e.g. SCP, 2008; SCP and CBS, 2011).
Part-time work is much more prevalent among women than it is among men. Although attitudes vary across countries (Boye, 2011), it is widely considered harmful to young children if their mother works full-time, especially when high-quality childcare is unavailable (Treas and Widmer, 2000). Women anticipate future family responsibilities by choosing jobs with fewer hours, even before they actually have children (SCP and CBS, 2011). Part-time employment among men is low (in 2006, male part-time rates varied between 9.9% in Germany and 27.2% in the Netherlands, OECD, 2014), and although it is not frequently studied, it appears that men chose to work part-time for different reasons than women (e.g. Dekker et al., 2000; SCP and CBS, 2011). In the Netherlands, for example, the presence of children has only a minor impact on men’s working hours (SCP and CBS, 2011).
The quality of part-time work
Part-time jobs differ in nature and quality, as well as in hours. The literature addresses several indicators to assess the quality of a (part-time) job, including access to training, health insurance and pension benefits, career options, pay, security, working time arrangements, and job satisfaction (Abhayaratna et al., 2008; Chalmers et al., 2005; Dekker et al., 2000; Kalleberg et al., 2000; Plantenga, 1996; SCP, 2008; WRC, 2011; Yerkes, 2009: 704). Often, the employment conditions of full-time jobs are taken as a benchmark (e.g. Bardasi and Gornick, 2008; Chalmers et al., 2005; Román et al., 2004; Thurman and Trah, 1990). Yet, few studies actually measured the quality of part-time jobs, even in countries where part-time employment is relatively prevalent.
The studies that do include specific measures generally confirm the idea that part-time employees have lower quality jobs (Abhayaratna et al., 2008; ABS, 2008, 2009; Chalmers et al., 2005; Lyonette et al., 2010; Vosko, 2010; WRC, 2011: 33–34), although some authors add important nuances (e.g. Abhayaratna et al., 2008; Bardasi and Gornick, 2008). Moreover, studies on the Netherlands suggest that employees having had a part-time job in the past does not necessarily harm their careers (Román et al., 2004) and that average educational level and occupational status of part-time employees in the public (but not the private) sector is higher than their full-time counterparts (CBS, 2010).
The consequences of part-time work
Work–life conflict
Some scholars argue that part-time work is an ideal middle road that combines the benefits of work with the ability to meet family and personal demands (Booth and Van Ours, 2009). This draws implicitly on the role strain hypothesis that work and family demands compete for one’s time, energy, and attention (Grönlund and Öun, 2010). Compared to full-time work, part-time hours leave more time for family and personal needs and limit negative spill-over effects. Moreover, by working part-time, women conform to gender norms that still oppose full-time maternal employment (e.g. Booth and Van Ours, 2009; Uunk et al., 2005).
On the other hand, research finds that female part-time work reinforces rather than challenges the traditional gender division of domestic labor (Gornick and Meyers, 2003; Stier and Lewin-Epstein, 2000), and mothers employed part-time work long paid and unpaid hours because they spend as much time on childcare as full-time homemakers (Craig and Mullan, 2009). Indeed, role expansion theory argues that because it provides less economic, human, and social capital, part-time work has fewer positive work-to-family spill-over effects (Booth and Van Ours, 2009). Interestingly, Grönlund and Öun (2010) find that compared to full-time workers, part-time employees are more likely to experience role conflict and role expansion, but are less likely to experience both at the same time. This supports the idea that part-time work is a ‘double-edged sword’.
Job satisfaction
Booth and Van Ours (2009) find that Australian women who work part-time are more satisfied with their working hours and lives than full-time employed women, but do not rank higher in terms of job satisfaction. In contrast, in the United Kingdom, part-time employment increases working-hour and job satisfaction, but not life satisfaction (Booth and Van Ours, 2008). This may indicate that the supposed detriments of part-time work outweigh the benefits differently in different country contexts.
Conceptual idea
We expect that cross-national institutional differences in part-time work are reflected in differences in part-time workers’ work–life conflict and job satisfaction. Specifically, we expect that the consequences of part-time work are contingent upon the prevalence and protection of part-time work.
Prevalence
The countries in our study all have average or high rates of part-time female employment, but there is sufficient variation to explore the moderating role of part-time rates. In countries where part-time work is relatively common, part-time employed women may conform more to ‘societal ideals’ and role expectations. Moreover, because institutions such as schools and childcare centers are more adapted to part-time work (e.g. with regard to their opening hours), part-time employed women have more opportunities to care for their children (Bianchi, 2000). As a result, we could expect women employed part-time to experience less work–life conflict and feel more satisfied with their work hours than their full-time employed counterparts. In contrast, in the ‘dual earner family policy model’, where state support enables women to combine long work hours with family responsibilities (Boye, 2011), the beneficial effects of part-time work may be limited.
Protection
We expect that the quality of part-time work is higher in countries with more protective labor policies because employees do not have to sacrifice job quality for fewer hours. Policies that entitle part-time employees to the same benefits as full-time employees ensure the quality of part-time work (Boulin et al., 2006; Gornick and Meyers, 2003; Jacobs and Gornick, 2002). Moreover, the ability to transition from full-time to part-time hours within the same job safeguards quality because an employee remains entitled to the same benefits. When such rights are absent, part-time and full-time jobs are more likely to differ in terms of sector, status, and employment conditions.
When part-time work is not associated with occupational downgrading, it enables women to have the best of both worlds (Treas et al., 2011). Reduced working hours frees up time to do necessary household chores, care for family members, meet friends, and develop personal interests (Grönlund and Öun, 2010). If the employment conditions of part-time and full-time workers are similar, both groups can develop their professional skills and have a fulfilling career. However, if part-time jobs are not well protected, the adverse effects of a lower quality job may attenuate the beneficial effects of part-time hours, undermining well-being and job satisfaction. Moreover, work–life conflict is less likely to be counterbalanced by the benefits of role expansion if the quality of work is low (Grönlund and Öun, 2010). Thus, women in low-protection countries may experience higher levels of work–life conflict because they feel less comfortable rejecting an employer’s request to work overtime. Summarizing, we expect that the beneficial effects of part-time work on work–life conflict and job satisfaction are stronger in countries where labor and social legislation is more protective.
Part-time work in the Netherlands, Australia, Germany, the United Kingdom, and Sweden
Below we sketch the differences in the prevalence and protection of part-time work and apply the general expectations set out above to the specific countries in our study.
Prevalence of part-time work in Australia, the Netherlands, Germany, the United Kingdom, and Sweden
Table 1 gives an overview of relevant labor market statistics in the five countries. Female labor force participation is similar across the countries and highest in the United Kingdom (71.5%) and Sweden (78.7%). Unfortunately, there is no internationally comparative data available on mothers only. Part-time work is less prevalent in Germany, the United Kingdom, and Sweden than in Australia and the Netherlands. Percentages are still considerably higher than the OECD average (20.4%), however, with the share of women working less than 35 hours per week varying between 38.1 percent in Sweden and 47.6 percent in Germany. The relative ranking does not change with alternative working-hour cut-off points. Average female working hours are lowest in the Netherlands (24.3 hours per week) and highest in Sweden (34.1 hours per week). The Netherlands’ average reflects the ‘short-hours’ nature of part-time jobs there: in 2006, 33.4 percent of employed women worked fewer than 20 hours per week. The proportion of short-hours part-time jobs is also relatively high in Australia and Germany, but much lower in Sweden and the United Kingdom. The Netherlands is also exceptional in that the percentage of childless women working part-time is over 20 percentage points higher than in the other countries.
Part-time rates and work hours in Australia, the Netherlands, Germany, the United Kingdom, and Sweden in 2006 and 2012.
OECD: Organization for Economic Cooperation and Development; LFS: Labour Force Survey; FTPT: full-time part-time; EU: European Union.
OECD Online Database: labor force participation rate, LFS by sex and age, and indicators (accessed 31 January 2014).
Part-time work is defined as <35 hours per week. Source: Abhayaratna et al. (2008).
OECD Online Database: incidence of FTPT employment, common definition (accessed 31 January 2014).
OECD Online Database, usual hours worked by weekly hour bands (accessed 31 January 2014).
In the EU Labor Force Survey, part-time employment is measured as perceived by the respondent. In the Netherlands, employees are considered as part-time if their usual hours are fewer than 35 hours. In Sweden, the same criterion is applied to the self-employed. Source: EU Labor Force Survey, Online Data Base.
For Australia: coupled mothers only. Source: Abhayaratna et al. (2008: 106).
Percentages for Sweden are from 2008.
Labor market policies
We now briefly describe the labor market policy context in the five countries as we expect this to drive differences in the nature of part-time work. Three relevant institutional factors are the following: (a) the level of collective bargaining, (b) rights to adapt work hours, and (c) anti-discrimination legislation. We refer readers who want more background details to the social policy literature. The International Labor Organization (ILO) report on decent work (Boulin et al., 2006) is especially useful, as is the literature on welfare states (Castles, 1994; Esping-Andersen, 1999; Ferragina and Seeleib-Kaiser, 2011; Korpi, 2000) and family policies (Boye, 2011; Gornick and Meyers, 2003; Hegewisch and Gornick, 2011; Lewis, 2009).
Collective bargaining
The level and history of collective bargaining are likely to affect the protection of part-time work (Thurman and Trah, 1990). Strong market-orientation in Australia and the United Kingdom means the role of collective bargaining is limited, whereas employee representatives is strongly involved in developing employment policies in the Netherlands and Sweden (Anxo et al., 2006). The level of consensus between employees and employers is also relatively high in the latter countries. For example, in the Netherlands, protective provisions for part-time employees are included in collective bargaining agreements since the early 1990s (Román et al., 2004; Yerkes and Visser, 2006). In contrast, German employer organizations continue to resist legislation governing part-time work (Yerkes and Visser, 2006).
The right to adapt work hours
The ILO considers the employees’ ability to adapt their hours to their needs as a requirement of decent work (Boulin et al., 2006: v–vi). In Australia, employees have no legal entitlements to reduce their working hours in existing jobs. This directly contrasts with the other countries, which have all developed relevant policy, in part driven by the 1997 European Union (EU) directive on part-time work (Directive 97/81/EC). Provisions vary, however. In the United Kingdom, legislation on part-time employment exists, but it is controversial and limited in scope. The right to request a change in hours applies exclusively to employees who have worked for more than 26 weeks for one employer and bear the main responsibility for children aged 6 years or younger, or for disabled children under the age of 18 years. Exceptions enable certain firms to reject such requests (Yerkes and Visser, 2006). In contrast, ‘[t]he Netherlands has designed what is probably the most comprehensive state effort to increase high quality part-time work’ (Gornick and Meyers, 2003: 165). The Wet aanpassing arbeidsduur (Working Hours (Adjustment) Act), enacted in 2000, allows employees to request variable working hours. Employers may only reject this request if they can prove it would seriously harm the organization. The content and roots of Swedish legislation are similar to those of Dutch legislation. Swedish employees have the right to adjust their working hours, which provides ‘large opportunities for households to adapt their working time to various situations and commitments over the life course without significant income losses’ (Anxo et al., 2006: 114). In the early 2000s, Germany introduced legislation that increased access to part-time work and made it possible to adjust one’s working hours. However, the scope and impact of German legislation are more limited, and part-time work has not reached the same ‘normalized’ status as in the Netherlands (Yerkes and Visser, 2006).
Anti-discrimination legislation
A third way to safeguard quality is to make it unlawful to distinguish between part-time and full-time employees (Thurman and Trah, 1990). In Australia, anti-discrimination policies do not cover work schedules. Although some legislation exists in the United Kingdom, protection against discrimination is lower than in other European countries (Fagan, 2001). In the Netherlands, the Wet onderscheid arbeidsduur (Act Prohibiting Discrimination by Amount/Volume of Working Hours), introduced in 1996, stipulates that employers cannot put part-time workers at a disadvantage. This act ensures ‘equal treatment in areas negotiated by the social partners, such as (above minimum) wages, holiday pay and entitlements, overtime payments, bonuses and training’ (Plantenga et al., 1999: 108). When part-time employees feel discriminated against or consider themselves at a disadvantage, they can appeal to the Netherlands Institute for Human Rights. The protection of Swedish part-time employees is equally high (Anxo et al., 2006). To our knowledge, there is no specific policy in Germany. Moreover, the quality of part-time work in Germany is partly undermined by ‘mini-jobs’ in which employees can earn up to €450 per month but which are extremely precarious and unprotected (Yerkes and Visser, 2006)
Expectations
We expect that part-time employment would be most beneficial in institutional contexts where it is widespread and its quality is ensured. Applied to the countries in our study, this would imply part-time work is most beneficial in the Netherlands where part-time work is both very prevalent and strongly protected. Although Australia comes closest to the Netherlands in prevalence, its protective policies are on the other end of the scale, and thus, we expect the positive effects of part-time work to be reduced. Similarly, in the United Kingdom, part-time employment is relatively prevalent, but protection policies are limited (although more extensive than in Australia). The prevalence in Germany and Sweden is relatively low, but whereas the level of protection in Sweden is comparable to the Netherlands, labor market policies in Germany are much more contested. Summarizing, we expect part-time employees to benefit most in the Netherlands, least in Australia and Germany, Sweden, and the United Kingdom take a middle position.
We restrict our study to women because men’s part-time work is rare, is often driven by different motives, and has different outcomes than women’s part-time work (Román et al., 2004; SCP, 2008). Because domestic demands on women with and without partners differ, we study partnered women only. We control for the commonly considered antecedents of work–life conflict (children, educational level, e.g. Byron, 2005; Grönlund and Öun, 2010) and job satisfaction (children and age, e.g. Booth and Ours, 2008; Easterlin, 2006). Unfortunately, we cannot include schedule flexibility (Byron, 2005) or health (Booth and Ours, 2008; Diener et al., 1999). We also control for gender ideology as this is likely to affect women’s employment status and well-being. Finally, we consider the role of occupational status because we expect the impact of part-time work to vary across countries because women hold different jobs and not because they respond differently to the same types of job. Thus, cross-country differences should be partly explained when occupational status is included in the analysis.
Methods
Data
The analytical models use data from the most recent available wave of the ISSP ‘Family and Changing Gender Roles’ module (2013). The response rates were 34 percent in the Netherlands, 59 percent in Australia, 42 percent in Germany, 58 percent in the United Kingdom, and 73 percent in Sweden (ISSP, 2013). The low Dutch response rate was average for Dutch survey data (Stoop, 2005). Overall, the samples were nationally representative (ISSP, 2013). Highly educated respondents were overrepresented in Australia and Germany. The percentage of employed respondents in Australia was also slightly higher than the country average. No additional information was available for the United Kingdom or the Netherlands.
We limit the full sample of 3744 Dutch, Australian, Swedish, and German women to those with a steady life partner. Second, we select the respondents in paid employment. Third, because we wish to compare the results for the different outcome measures, we exclude respondents with one or more missing values. The exception is occupational status, for which we impute missing values using singular imputation (see section ‘Analytical strategy’). This leaves a final sample of 1773 women (348 Australian, 301 Dutch, 264 German, 340 Swedish, and 520 British).
Measures
Dependent variables
Work–life conflict is measured by two items: (a) ‘In the past three months, how often [have you] come home from work too tired to do the chores which needed to be done?’ and (b) ‘In the past three months, how often has it been difficult for [you] to fulfil family responsibilities because of the amount of time [you] spent on [your] job?’ (1 = several times a week to 4 = never). The new measure was constructed by taking the average (α = .70) and recoded so that higher values correspond with more conflict. Job satisfaction is measured by the single item ‘How satisfied do you feel with your main job?’ (1 = very dissatisfied to 7 = very satisfied). We recode the variable so that higher values reflect more satisfaction.
Independent variables
We distinguish between short-hours part-time jobs (<25 hours per week), long-hours part-time jobs (25–34 hours per week), and full-time jobs (35+ hours per week). We use respondents’ reported working hours to construct this variable. To capture work quality, we include occupational status, using the International Socio-Economic Index (ISEI) coding to transfer the four-digit International Standard Classification of Occupations (ISCO) 1988 code (Ganzeboom and Treiman, 2003). Higher values indicate a higher status.
Control variables
The presence of minor children in the household (0 = no, 1 = yes) and marital status (1 = cohabiting or married) are included as dummy variables. Age, age-squared, and educational level (‘What is your highest level of education?’ 0 = no formal education to 5 = university degree completed) are included as well. Finally, our indicator of gender-egalitarian attitudes is a replication of Fuwa’s (2004) measure (α = .78, five items, e.g. ‘A pre-school child is likely to suffer if his or her mother works’, 1 = strongly agree to 5 = strongly disagree). Higher values indicate more progressive attitudes. We do not include any partner variables because almost all partners worked full-time, and we cannot include a measure for excessive hours because Australian respondents only provided information on partner’s employment status and not his actual working hours.
Analytical strategy
To test the assumption that job quality is higher in some countries than others, we use a t-test and χ2 test to compare the job-quality indicators (occupational status and supervisory position). We then run separate analyses for each country and outcome variable, using ordinal logistic regression because the levels of the dependent variables are ordinal (it is unlikely that the distances between the categories are similar). We use the same control variables across the models, so that differences between models can be attributed to the outcome measure. We first estimated the basic model (including controls and employment status) and then entered the quality indicator, occupational status. We report the results of the final model. In order to test whether the effects differ significantly across countries, we ran a seemingly unrelated regression with the stored results (suest in Stata) and test the equality of the coefficients. A disadvantage of this method is that it is based on the problematic assumption that the error variances of the ordinal logistic regression models do not vary across countries. This is a common issue when the coefficients of nonlinear probability models are compared (Allison, 1999; Mood, 2010). In order to check whether residual heteroskedasticity affects our results, we run an ordinal generalized linear model on the pooled data (Allison, 1999; Williams, 2011). This model includes interactions of dummies for countries and the employment variables, and allows the error variances to vary across the countries. Thus, we assume that the effects of the control variables do not differ across counties. This admittedly restrictive assumption identifies the interaction of country and employment while accounting for residual heteroskedasticity across countries. We conduct a likelihood ratio (LR) test to test the hypotheses of error homoskedasticity assumed by the suest approach.
As noted above, a substantial percentage of the respondents had a missing value on occupational status (9.2%). Because we do not want to exclude such a substantial proportion of our sample, we imputed this variable using singular imputation. According to Scheffer (2002), this is an acceptable strategy if the pattern of missing values is unrelated to our dependent variable (and thus ‘Missing at Random’) and the percentage of missing values is 10 percent or lower. Both conditions apply. It is important to note, however, that respondents with part-time jobs were more likely to have a missing value on occupational status. We would have preferred multiple imputation, a more advanced method of handling missing values, but it was not possible to combine the imputed data with our analytical strategy. The imputed variable was based on the partner’s occupational status, the respondents’ age and educational level, and family income. Finally, we checked for multicollinearity, but this was not an issue.
Results
Descriptive results
Mean levels of work–life conflict and job satisfaction are similar across countries. It is important to note that only 10 respondents (none German or Swedish) reported being completely dissatisfied with their work. Additional t-tests (results not shown) show German and (especially) Dutch women report significantly lower levels of work–life conflict than women in other countries. Job satisfaction is lowest in Australia, and the difference between Germany and Sweden is significant. χ2 tests (results not shown) suggest that the proportion of short-hours jobs is highest in the Netherlands and Australia. The prevalence of long-hours part-time jobs is similar in the Netherlands, the United Kingdom, and Australia. The control variables show few remarkable results: the proportion of women with children varies between 40 percent and 60 percent, and the women have an average age of around 40 years (Table 2).
Descriptive results.
SD: standard deviation.
0 = no; 1 = yes.
Contrary to expectations, the data do not conclusively show country differences in job quality. Australian women report the highest occupational statuses and Swedish women the lowest. The results of t-tests (not shown) suggest a significantly lower position for Swedish and British women as compared to Australian and Dutch women, but no noteworthy differences between the other countries (note that the standard deviations (SDs) are relatively high). We ran the same analyses on the sub-sample of part-time employees to identify differences in the quality of part-time work. The results (not shown) suggest that part-time employed women in Sweden and the United Kingdom work in lower status jobs than their counterparts in the Netherlands, Australia, and Germany. Supervisory positions are remarkably infrequent in the Netherlands. There are few differences between the other countries.
Multivariate analyses
Because we expected that the effects of part-time work would become weaker once we controlled for occupational status, we originally took a stepped approach and entered it in a separate model. However, the differences between the models were negligible. We therefore report and discuss the models that include occupational status.
Work–life conflict
Models 1a through 5a (Table 3) show that compared to full-time jobs, short-hours part-time jobs are associated with less work–life conflict in all countries except Sweden (where the effect is marginally significant). For example, the log-odds of work–life conflict are 0.981 lower for Australian women with a short-part time job than for Australian women with a full-time job. The differences between full-time jobs and long-hours part-time jobs are smaller but also significant in the Netherlands and the United Kingdom. A comparison of coefficients across the countries shows the effect sizes of part-time work dummies are largest for the Dutch and British sample. The cross-country differences do not reach significance, however, which could be explained by the high standard errors. Our original stepped approach found no change in the effects of part-time work as a result of including occupational status. However, the variable does yield a direct effect in the United Kingdom, where work–life conflict is higher among women with a higher occupational status. Summarizing, there was little support for our expectation that part-time work would reduce work–life conflict more in countries where part-time work is more prevalent and protected.
Multivariate ordinal logistic regression explaining work–life conflict by country (unstandardized log-likelihood coefficients).
DF: degree of freedom; SE: standard error.
0 = no; 1 = yes.
McFadden’s pseudo-R2.
p ≤ 05, **p ≤ 01, ***p ≤ 001, + p ≤ 0.10, and two-tailed test of statistical significance.
Job satisfaction
In Models 1b through 5b (Table 4), only one effect is found to be significant, and then only marginally. Australian women in long-hours part-time jobs report lower levels of job satisfaction than their full-time counterparts. This effect is weaker in other countries and even positive in the Netherlands. A comparison of the coefficients (based on the seemingly unrelated regression with the stored results) shows the difference between Australia and the Netherlands is significant. Moreover, the effect of long-hours part-time work between the Netherlands on the one hand and Sweden and the United Kingdom on the other hand is marginally significant.
Multivariate ordinal logistic regression explaining job satisfaction by country (unstandardized log-likelihood coefficients).
DF: degree of freedom; SE: standard error.
0 = no; 1 = yes.
McFadden’s pseudo R2.
p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, +p ≤ 0.10, and two-tailed test of statistical significance.
Additional analyses
The results of the ordinal generalized linear model (results not reported) suggest that if we allow the error variances to vary across countries, this does not result in a significant improvement of the model fit for work–life conflict (LR(4) = 2.44, p = 0.656), but that it does for job satisfaction (LR(4) = 20.14, p = 0.001). Moreover, the inclusion of the interactions of dummies for countries and the employment variables does not improve the model fit for either work–life conflict (LR(8) = 5.91, p = 0.657) nor job satisfaction (LR(8) = 5.48, p = 0.705). This implies that the results for work–life conflict are not affected by heteroskedasticity and that the findings for job satisfaction should be interpreted with care. Thus, we cannot conclude that part-time work affects work–life conflict and job satisfaction differently in different countries.
Conclusion and discussion
This study compared Australia, the Netherlands, Germany, Sweden, and the United Kingdom to find out whether the subjective effects of part-time work are moderated by its national prevalence, protections, and quality. Part-time employment is relatively common in these countries, but there are substantial differences between their labor markets. The first, descriptive, aim was to compare the policy background in the five countries and derive expectations on the implications for the nature of part-time work. The second aim was to test empirically whether associations between part-time work, work–life conflict, and job satisfaction vary across the countries. The social policy literature suggested that it is likely that the nature of part-time work depends on the policy context of a country. However, our evidence suggests that the impact of part-time work on quality of life is only marginally contingent on the country context.
Our review of the social policy literature suggested that the quality of part-time work relates to social legislation. For example, in the Netherlands, part-time jobs are ‘slimmed-down’ full-time jobs because employees are entitled to reduce their hours in their current job. The relevant laws are not restricted to certain groups, not opposed by employers, and anti-discrimination legislation is extensive (Yerkes and Visser, 2006). Moreover, employers and trade unions played a large role in developing this legislation, and it was already common practice for employers to allow work-hour adjustments (e.g. Plantenga, 1996; Yerkes and Visser, 2006). In contrast, Australian women are often forced to transition to lower quality jobs if they want to work part-time, because labor market regulation is weak and employers do not have legal or normative obligations to allow women to reduce their hours (Abhayaratna et al., 2008; Chalmers et al., 2005). The literature suggests that the quality of part-time work in Germany, the United Kingdom, and Sweden is positioned between these two extremes. Legislation in Germany and the United Kingdom is more limited in scope than in the Netherlands (e.g. restricted to specific groups) and part-time work is much less grounded in their history or supported by trade unions and employers. Sweden offers a high level of protection, but the incidence of part-time work is relatively low, which may lead to normative sanctions.
After our review of the literature, we tested our central hypothesis that consequences of part-time work would vary across countries. The results suggest that the impact of part-time work on job satisfaction is almost non-existent and that there are few differences between the countries. The impact of part-time employment on work–life conflict was more substantial and strongest among those with short-hours (24 hours or less). Part-time employed women reported lower levels of conflict in all countries except Sweden.
Our analyses do not support the idea that the beneficial effects of part-time work vary greatly across countries. Although part-time work is highly prevalent in the Netherlands and Dutch legislation the most protective, the impact on work–life conflict was not much larger in the Netherlands than in the other countries. There is some evidence that Dutch women in long-hours part-time jobs are more satisfied with their job, but this evidence is tentative. The finding that the benefits of part-time work for work–life conflict are relatively low in Sweden is surprising considering its relatively high degree of job protection and suggests that the prevalence of part-time work is a more relevant factor. Our data suggest that the occupational status of Swedish women in part-time jobs is relatively low, which may result from the fact that they are in a more exceptional position than part-time workers in the other countries. Moreover, because high-quality childcare is more widely available and used in Sweden, women’s presence at home may be less crucial than in other countries. It is also possible that Swedish women have relatively less to gain from part-time work because it is less gendered and reinforces gender role expectations to a lesser extent than in other countries.
Overall, we found few cross-national differences in the effects of part-time work, and there are at least three possible explanations for this. The null findings may imply that (a) our assumptions on variations in the quality of part-time work were incorrect, (b) the quality of part-time work is not as relevant for its effects as we expected, or (c) our empirical analyses were not able to capture the effects. We briefly reflect on these possible explanations and their implications for our theoretical arguments and the broader literature.
First, our assumption that part-time jobs are of better quality when it is better protected may have been incorrect. Although the institutional arrangements are likely to put Dutch part-time jobs on a par with full-time jobs in terms of employment conditions, the empirical evidence is tentative at best. Moreover, in Sweden, a country with extensive legislation, occupational status of part-time work was relatively low. Possibly, country-level policies have little meaning for every-day life. Employers play a key role in determining their employees’ work conditions, and discrimination of part-time employees may be subtle and accepted, despite legislation. For example, Van der Lippe (2004) showed that many Dutch managers consider a full-time week a precondition for a supervising function. As such, a useful pathway for future research may be to investigate the role of the employer.
Second, it is possible that our assumption that the quality of part-time jobs differs across countries holds, but that these differences do not condition the effects on work–life conflict and job satisfaction because women ‘adapt to circumstances’ (Diener et al., 1999). For example, if women lower their expectations and ambitions when they transition to part-time, Australian women may not evaluate their jobs as ‘lower quality’ than their Dutch counterparts. Satisfaction measures are subject to cognitive dissonance reduction and people rate their actual situations against their expectations, as well as against real standards (Diener et al., 1999; Huppert et al., 2009). Similar mechanisms may apply to work–life conflict, because the standards by which experiences are judged are likely be affected by the country context (Fagan, 2001). Alternatively, contradictory country effects may cancel each other out. For example, lower expectations and aspirations among women in the Netherlands may cancel out the benefits of high protection. Future research could use multi-level analyses to test cross-level interactions and disentangle the effects of country characteristics.
Third, the quality of the data may be insufficient to capture cross-country differences. The available job-quality indicators were limited, and the data were cross-sectional rather than longitudinal, so we could not control for selection effects. Women with certain background characteristics, such as traditional parenting ideology, may be more likely to self-select into part-time work (Boye, 2007). If differences between part-time and full-time employees also explain variations in the outcomes (e.g. affluent women may be more likely to work part-time and report low levels of work–life conflict), this may have biased the results. The degree of self-selection is likely to vary across countries. Stier et al. (2001) argued that labor force participation is more selective in liberal welfare states than in social-democratic regimes because parents must be more committed to work when government support is limited, and formal childcare is more exclusive. Other factors may play a role as well. For example, because the resistance to full-time employment for mothers with young children is very high in the Netherlands, it may be limited to very specific households. Unfortunately, the work of detecting and accounting for selection lies beyond the scope of this study. If selection effects were present, we are likely to have overestimated the results. With regard to selection into ‘bad’ or ‘good’ part-time jobs, a longitudinal panel study would make it possible to test whether women in countries with less labor market protection indeed transition into lower quality jobs when they switch to part-time work and whether they choose or are forced into these jobs (Forsberg, 2009).
Future research could extend this study in several other ways. First, it could examine the quality of part-time work in more detail. Considering part-time work’s popularity, it is remarkable how little we know about the employment conditions. The finding that part-time employed women in Sweden reported a lower occupational status and experienced as much work–life conflict as their full-time counterparts warrants specific attention. Second, despite being the most recent available, the ISSP data are 10 years old, and the sample size was relatively small. The incidence of part-time work changed very little between 2006 and 2012, but there have been some policy changes. For example, in the Netherlands, government support for formal childcare was substantially reduced in 2012, and in Australia, a national paid parental leave scheme was introduced in 2011. A replication of this study with newer data could clarify the implications of recent reforms. Third, it would be interesting to contrast the consequences of scaling back to part-time work with other ‘household strategies’, such as multitasking and tag-team parenting (Forsberg, 2009).
Encouraging female labor force participation is an important policy objective worldwide (OECD, 2010). This study suggests that part-time jobs vary on a number of dimensions, both within and between countries, and that it is important to acknowledge the quality of part-time work when discussing it as a strategy for reconciling work and family responsibilities (Crompton and Lyonette, 2006). By examining a range of possible outcomes of part-time work and comparing countries with varying labor markets, we have improved our knowledge of the potential role of job quality. That the effects were limited is disappointing, but nevertheless important as it suggests that formal policies do not automatically safeguard its quality or ensure its effects are beneficial. This study took the first step toward analyzing empirically the relevance of part-time work quality, but also made clear that richer data are needed. We provide a starting point for follow-up research examining how the quality of part-time work can be assessed, how formal and informal institutions interact, and how women’s expectations respond to actual employment conditions. The popularity of part-time work in Australia, the Netherlands, and other countries means that we must consider the conditions needed to make part-time work a successful strategy.
To conclude, although there are large cross-national differences in the institutionalization of part-time work, we found no evidence that these differences are reflected in the consequences of part-time work. This is an important finding because it is not in line with the prevalent contention that part-time work is of a higher quality in some countries than in others. Part-time work seems to have the same effects in countries with varying levels of protection and prevalence. We responded to Treas et al.’s (2011) encouragement to study how country characteristics affect the subjective consequences of part-time and full-time jobs, but our findings suggest that the key to understanding these consequences may not be located at the country level. Most likely, the relevant conditions for satisfying and fulfilling part-time work are located on the sector- and job-level instead.
Footnotes
Acknowledgements
The authors would like to thank Tanja van der Lippe for facilitating this project and the anonymous reviewers and Jeroen Weesie for their useful comments and advice.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
