Explaining differences in women's working time in European cities Open Access

The male breadwinner model, based on a division of labour between an employed husband and a stay-at-home wife who cares for their children, is becoming less common; but has not disappeared completely. In many countries, significant numbers of women work part-time. This is problematic from a gender equality perspective, since the part-time employment of women possibly contributes to the persistent gender gap in income and the dependence of women on their spouses or male partners, who are working full-time. Part-time work also increases the risk of poverty for women in the event of divorce and old age (Saraceno and Keck 2011). Moreover, part-time employment often involves lower pay, fewer benefits, poorer work conditions, and less social security than full-time employment (Hegewisch and Gornick 2011).

It is important to note that the proportion of women working part-time and the number of part-time hours worked differ substantially between countries and social groups, as do the work conditions and social rights related to part-time work (Fagan and O’Reilly 2000). The article addresses the question: How is it possible to explain the differences in the number of women's work hours? Few studies have thus far analysed why some women take part-time work while others enrol in ‘standard full-time employment relationships’ in a comparative perspective (Pfau-Effinger 1993). The aim of this article is to contribute to the scientific discussion about the factors conditioning women's work-time patterns with reference to the number of work hours per week in formal employment. The innovation here is our application of a complex explanatory framework including women's orientation towards specific cultural ideas about work and family (in brief women's cultural work–family orientation), together with individual characteristics, the main features of their family status, their trust in public childcare institutions, economic hardship, and their work conditions. As the empirical basis for our study, we use survey data collected from women living in five European cities, each representing an urban environment that particularly supports women's employment. This survey was conducted in the context of the international EU project FLOWS – Impact of Local Welfare Systems on Female Labour Force Participation and Social Cohesion (www.flows-eu.eu/).

The article is structured as follows: We start with an overview of the current theoretical debate and empirical research in the field followed by our main hypotheses. The methodological approach and data are then introduced, followed by discussion of our findings. The article is rounded off with some concluding remarks.

The differences in women's employment behaviour and work time are a much-debated issue in comparative social sciences. In this section, we introduce theoretical approaches for explaining different work-time patterns between women. The dominant theoretical approaches include micro-level factors such as economic rationality or the diversity in women's preferences. Other approaches stress macro-level factors such as cultural family models, family policies, and socio-economic factors.

The empirical research related to such approaches focuses primarily on individual characteristics such as the educational attainment of women (and their partners), age, health status, women's work conditions, and women's work orientation. Particular emphasis has also been placed on household composition, such as marital status, the household's financial situation, care responsibilities for children or frail relatives, and the sharing of housework (see also Thévenon 2013). The following part introduces the different theoretical approaches in greater detail.

Particularly in economics or sociological rational choice theory, it is often argued that the integration of women into the labour market differs in relation to their family status and that their employment decisions are aimed at maximizing utility. This argument was mainly developed by Gary Becker (1981) in the context of the New Home Economics approach. Becker also argues that ‘intrinsic differences between the sexes’ are relevant, since women are ‘biologically committed to the care of children’ (1981: 37–40).

The feminist counterargument is that the gendered division of labour does not have biological roots, instead being the result of power relations and that class and gender inequalities interact differently in different political economics (Cook 2011). In addition, Becker's theoretical approach has been criticized for treating the private household as a monolithic block in which women and men pursue the common aim of household efficiency, neglecting the different interests held by individual women and men (Katz 1997).

Another problem with this approach is its inability to explain why women with similar characteristics in similar situations nevertheless make different decisions regarding their employment (McRae 2003). According to Hakim's preference theory (2000), differences in individual preferences help explain differences in employment behaviour. Hakim (2000) argues that, in principle, there are three groups of women who prioritize matters of family and career differently.

Hakim's theoretical concept does not clarify where the differences in women's preferences come from. Why should the preferences of women differ while men's preferences do not? This approach was also criticized for neglecting the constraints women possibly face in terms of inadequate public childcare provision and the gendered division of labour in society (McRae 2003; Leahy and Doughney 2006; Kangas and Rostgaard 2007) and that women's preferences may change over the life course (Crompton and Harris 1998; McRae 2003; Leahy and Doughney 2006; Schober and Scott 2012). Moreover, it ignores the role of the demand side for the number of hours women work (Estevez-Abe 2009).

Comparative welfare state research argues that welfare state policies regarding the provision of public care for children and the elderly are a main factor explaining the cross-national and cross-regional differences in women's work hours (e.g. Lewis 2002, 2006; Gornick and Meyers 2003; Hegewisch and Gornick, 2011; Bünning and Pollmann-Schult 2016; Duvander and Ellingsæter 2016).

An alternative approach introduced by Pfau-Effinger (1993, 1998) argues that women (and men) do not only act on the basis of institutional factors, but also on the basis of cultural ideas and models. This theoretical approach designates that the employment behaviour of women (and men) is influenced by the prevailing cultural ideas in the society in question regarding the ‘desirable’ or ‘ideal’ work–family relationship.¹ At the same time, cultural differences between different social groups and regions within given societies can occur (Duncan and Ewards, 1997; Mandel, 2009). According to this approach, differences in women's employment behaviour must be explained by cultural differences and the interaction between cultural, institutional, and socio-economic factors. Welfare state institutions and socio-economic factors may help or prevent women from realizing their respective cultural orientations towards specific cultural ideas about work and family. For instance, mothers oriented towards the labour market may be prevented from participating due to the lack of childcare, but informal resources, including undeclared care work, can compensate for this lack (Pfau-Effinger 2012).

Pfau-Effinger (1998, 2012) has developed a theoretical classification approach for comparative research that distinguishes between different ideal types of ‘cultural family models’ combining cultural ideas related to different dimensions of the work–family relationship. These include cultural ideas regarding the ‘ideal’ gendered division of labour between parents (e.g. both parents acting as full-time employed breadwinners or men serving as primary breadwinners while women remain home to care for young children) and cultural ideas relating to how children should ‘ideally’ be cared for (e.g. parental care, care provided by the extended family, or extra-familial care). Different cultural family models combine these two dimensions differently. For contemporary European societies, these mainly consist of (1) the male breadwinner/female part-time care model, (2) the dual breadwinner/extended family care model, and (3) the dual breadwinner/state care model. Cultural family models can be contested between different actors and are subject to change.

The assumption that culture contributes to the explanation of differences in women's employment behaviour is also supported by empirical studies (e.g. Jensen 1996; Mandel 2009; Hummelsheim and Hirschle 2010; Uunk 2015); however, empirical analyses based on a complex explanatory framework that systematically includes a cultural dimension are rare.

Numerous studies also stress the impact of individual characteristics such as age and education on the differences in women's employment behaviour within societies. Several authors show how education levels affect female employment rates and their work-time patterns (Gustafsson et al.1996; Vlasblom and Schippers 2006; Grunow et al.2012; Bieri et al.2016).

Family-related factors are also stressed as being relevant for women's employment and female work-time patterns, including their partners’ income and the number of children (e.g. McGinnity 2004; Matysiak and Steinmetz 2008). It was also shown that the more children a married woman has, the more likely she is to reduce her work hours or drop out of the labour market entirely (e.g. Baumgartner 2003). This could be explained by the lack of affordable public childcare, which hinders women with a cultural orientation towards employment in participating in the labour market.

In this paper, we aim to explain differences in women's work hours, defined here as ‘work-time patterns’ in five different cities in which the labour market structures are relatively favourable for women in terms of the employment system depending on a relatively high share of services (Kuronen et al.2014). Beginning with the existing literature, we have identified several variables that may help to explain the different work-time patterns observed among women.

We hypothesize that women's cultural orientation towards specific cultural ideas about the ‘ideal’ work–family relationship (in brief ‘women's cultural work–family orientation’) is important for their work-time patterns. At the welfare-institution level, we consider family policies towards paid maternity/paternity leave and childcare coverage. Moreover, if women trust the public childcare system, the share of women working full-time might be higher, since more of them would use public childcare. We assume that the work-time patterns among women can differ on the basis of individual characteristics such as education level, age, and health status. We also consider demand-side factors (availability of part-time work) and – from a utility maximization perspective – women's relative wages and work conditions. We include in our explanatory framework the employment sector in which women work, together with their status in relation to managerial tasks and whether or not the job involves irregular work hours.

For mothers, child-related factors are relevant (number of children, age of youngest child).

This article analyses the work-time patterns of working women in five European cities: Aalborg (Denmark), Jyväskylä (Finland), Hamburg (Germany), Bologna (Italy), and Leeds (England). These five cities are embedded in four different national welfare regimes but have in common the important characteristic of an urban labour market that particularly supports female employment (e.g. Jensen and Lolle 2013; Kuronen et al.2014). All of the cities comply with the EU 2010 employment target, meaning that women's employment rate in all three cities is above the 60% threshold. In 2011–2012, the female employment rate in the five cities was (with national figures in brackets): Aalborg 68% (70%), Jyväskylä 67% (67%), Hamburg 70% (68%), Bologna 64% (47%), and Leeds 64% (65%) (cf. Ranci et al.2014). As becomes clear, the differences in the employment rates of women between the selected cities are rather small. The differences between the cities in terms of the proportion of women working part-time, however, are more pronounced: 25% in Aalborg, 30% in Bologna, 38% in Hamburg, 20% in Jyväskylä, and 25% in Leeds.

The data collection was conducted in late 2012 and early 2013 (for an extensive description of the data, see the introduction to this edited volume). The data were primarily collected via telephone interviews based on a random sampling of women aged 25–64 in the five European cities. Approximately 800 interviews were conducted in each city, the response rates being: 28% in Bologna, 21% in Jyväskylä and Hamburg, 24% in Leeds, and 50% in Aalborg. In this paper, unemployed women and women outside the labour market have been excluded from the data set, leaving 2483 total observations in the data. The data are skewed towards older age groups and women with tertiary education. However, since the data were primarily subject to regression analyses and the covariates in question were included in the model, this was considered a minor problem. In regression analyses, the coefficients will be estimated irrespective of the actual number in each category (Winship and Radbill 1994).

The data have been merged into a common pool, rendering it possible to model the differences between the cities. The merging of data, however, is based on the assumption that the cause of an effect, including the number of children, has the same direction in all contexts (cities), which may not actually be the case. The analysis therefore consists of two steps: first we analyse the pooled data and second the individual cities are reanalysed in order to check the implicit assumption of homogeneity in the effects. These latter analyses are not presented in the article. In cases of difference, the interactions between the relevant variables have been included in the analysis of the pooled data.

The primary dependent variable in the article is the self-reported number of hours worked per week among employed and self-employed women. It is based on the following survey question: ‘On average, how many hours per week do you work, including both paid and unpaid overtime?’. The dependent variable is continuous,² and the ordinary least squares method is applied.³

The following focus points have been singled out as the most likely predictors for the weekly work hours among women: (a) women's cultural work–family orientation, (b) individual and family characteristics, (c) trust in public institutions, (d) position in the vertical and horizontal division of labour, including working irregular hours, and (e) city of residence. Table 1 provides an overview of the variables from the survey data, and descriptive statistics can be found in Appendix.

The results of the regression analysis are presented in Table 2.

Table 2 indicates a significant relationship between women's cultural work–family orientation and work hours. Employed women who ‘agree’ or ‘strongly agree’ that ‘a woman should be prepared to reduce her work time in favour of her family’ work fewer hours per week than do those who disagree with the statement. Still, this statement implies that family and work are reconcilable; women may adjust their work hours to the needs of the family or (perhaps) make use of public childcare. The second ‘cultural orientation’ question is more categorical: ‘A man's job is to earn money; a woman's is to look after the home and children.’ However, the employed women who are oriented towards the traditional cultural male breadwinner model do not work fewer hours than average.⁴ Nonetheless, it remains unquestionable that cultural orientations, designating that woman must give greater primacy than men to the family, have a significant effect on women's work hours.

Age has an effect on women's weekly work hours, although only for the older age groups (55–59 and 60–64 years of age). Especially, women aged 60–64 work significantly fewer hours per week than the reference group (women 50–54). For the remaining age groups, no differences in work hours can be identified. The results may be an indication of flexible retirement patterns or a cohort effect referring to how older women have worked part-time all their lives. An in-depth register-based analysis of women's work hours in the cities included in this paper has found that the cohort effect predominates vis-à-vis the age effect (Pfau-Effinger et al.2014).

It is often assumed that a relationship exists between education level and work hours. As seen in Table 2, the results corroborate this assumption, although the result for ‘lower secondary’ and ‘upper-tier, upper secondary’ (model 2) are only significant at the 10% level. We also find that managerial responsibilities have a mediating effect on the education–work hours nexus. If managerial responsibilities are removed from the model (results not shown), the effects of education on women's work hours become more pronounced, indicating that the level of education leads to different vertical positions in the occupational structure, which, in turn, influences work hour levels.

As expected, health has an impact on work hours. Women reporting ‘very good’ health work more than those with ‘good’ health. In addition, we find that those reporting ‘fair, bad, or very bad’ health do not work less.⁵

Marital status has a major impact on work hours; single women work more hours per week than those living with a partner, married or not.

In model 1 (Table 2), the number of children living in the household has a negative impact on women's work hours. Moreover, having children of less than 3 years of age reduces work hours further, although this result is only significant at 10%. This illustrates how women's work-time pattern is synchronized with the stages in the (family) life course – women cut their work hours due to family and childcare when at a child-rearing stage.

While trust in public childcare institutions was expected to affect work time, we cannot empirically support this expectation.

Turning to the horizontal position of women on the labour market, we find some differences by sector of employment. Women working in the industrial sector work more hours than those employed in ‘health and social work’, ‘wholesale and retail trade’, ‘education’, and ‘other services’. This might indeed suggest that structural and regulatory factors affect female employment behaviour.

Working irregular hours prolongs the number of hours worked per week, as do managerial responsibilities. Women with managerial responsibilities work significantly more hours per week than women without such responsibilities.

Finally, there are major differences between the cities in the analysis. Women in Leeds work markedly fewer hours per week than women in Aalborg, and women in Hamburg (model 1) and Bologna also work less than women in Aalborg, whereas Jyväskylä and Aalborg women work about the same.

In order to better understand the possible effects of women's cultural work–family orientation on work hours in the different cities, the following section will focus on contextualizing the findings from the analysis of the survey data in relation to city-specific structural information regarding the welfare state arrangements and general cultural orientation.

Economists often argue that the hourly wage offered to women is the major factor behind female labour supply decisions (e.g. Smith 1989). Differences in women's relative wages, however, cannot thoroughly explain the differences in employment patterns among women. Women's relative wages – the so-called gender pay gap – vary from 75 in Leeds to 82 in Jyväskylä (Flaquer et al.2014). The fact that these differences are minor raises questions about whether or not the cross-city differences in women's weekly work hours are indeed an outcome of utility-maximizing calculations.

Instead, women's cultural work–family orientation may function as an overall framework for their employment behaviour. Table 3 shows how the main cultural family models differ between the cities in the study. The survey findings indicate that women orient themselves in their employment behaviour towards the different cultural family models in the respective cities. Six out of 10 women in Leeds, Bologna, and Hamburg – where the male breadwinner/female part-time care model dominates – ‘agree’ or ‘strongly agree’ that ‘[a] woman should be prepared to reduce her working time in favour of her family’, while this is only the case for four in 10 women in Aalborg and Jyväskylä, where the dual breadwinner/state care family model prevails (cf.  Appendix). Differences in women's cultural orientation cannot fully explain the differences in female employment patterns in the five cities; for example, women in Hamburg are marked by a strong cultural orientation towards the male breadwinner/female part-time care model, while at the same time exhibiting the highest employment rate of the five cities. However, women's cultural work–family orientation can explain many of the differences in the work-time patterns among women with children in the respective cities.

We have tested for the effect of the number of children by including the interaction between the city dummies for Hamburg and Leeds, and number of children (see Table 2, model 2), both highly significant. When including these interaction variables in the model, the number of children no longer has a significant effect, indicating that the number of children only has a negative effect for female work hours in Hamburg and Leeds. This has further consequences: When the effect of the number of children is modelled correctly, the differences between Hamburg and Aalborg are no longer statistically significant, meaning that childless women in Hamburg and Aalborg conform to similar work-time patterns when controlling for all of the other variables included in the model.

Assessing the institutional environment and factors conditioning the work hours of mothers requires an evaluation of the interaction among (1) parental leave arrangements, (2) coverage and price of childcare, and (3) take-up rates, as outlined in Table 3.

Huge discrepancies clearly exist between opportunities and practices (coverage versus take-up rates), especially in Hamburg and Jyväskylä. In both cities, only about one-third of all children under age 3 are in formal childcare, and only about one-third of the women with children in this age group are employed, even if every child has an individual right to full-time public childcare (see Table 3). Pfau-Effinger and Smidt (2011) have previously found corroborative results, which indicated that the main reason for these discrepancies is that women in Hamburg with children under age 3 prefer to care for their children themselves, as their cultural work–family orientation is influenced by the cultural male breadwinner/female part-time carer model that is dominant in West German culture, which has also been shown by Pfau-Effinger and Smidt (2011). In Finland, the enrolment of children under age 3 is also relatively low. This finding seems at first glance to be in opposition the general orientation of women in Finland towards the cultural dual breadwinner/state care model. However, on the basis of EVS and Eurobarometer data, Pfau-Effinger and Euler (2014) show that in Finland, most people support the cultural idea of gender equality, and it is also strongly believed that parental care is the best form of care for children under age 3, which contributes substantially to explaining why the employment rate for women in Finland with children under age 3 is relatively low. Repo (2010) even talks about a ‘new maternity’ in this context.

It has been argued in this article that it is not possible to treat the number of women's work hours, defined here as ‘work-time patterns’, one-sidedly as a result of utility maximization. Our analyses support the fact that work hours are also culturally and socially structured. These findings support our theoretical assumption that there is a significant relationship between women's cultural work–family orientation and work practices. Women with a more family-oriented cultural orientation work fewer hours.

Individual characteristics were found to matter for female work-time patterns. Women with lower self-reported health, women's age, marital status, and having children are also relevant characteristics for female work-time patterns. Highly educated women work more hours than do women with little or no education. When controlling for managerial responsibilities, however, education impacts less. Jobs with irregular work hours and managerial tasks are associated with increased work time.

As expected, family characteristics also matter; being married, having children, the age of the youngest child, and the number of children are all associated with fewer work hours. This shows that part-time work plays an important role in how women reconcile childcare and employment.

It is somewhat surprising, however, that the female employment rate is relatively independent of the price, coverage, and take-up rates of childcare. In all five cities, female employment rates are similar (and high), while price, coverage, and take-up rates differ. This may be due to demand-side factors. All five cities are marked by a service economy conducive to female employment, partly because part-time jobs are more available in the service sector than in industry.

It must also be emphasized that the differences in work-time patterns between the five cities primarily concern working women with children. In Hamburg, for example, childcare coverage is high but take-up rates are low, indicating that working mothers decide to work short hours despite affordable childcare institutions due to strong orientation towards the cultural male breadwinner/female part-time care model. In contrast, mothers in Aalborg make great use of childcare institutions, which ties in with an orientation towards the cultural dual breadwinner/state carer model.

Conversely, women without children exhibit almost identical work-time patterns in the cities included in this study. Thus, family characteristics and cultural work–family orientations matter significantly for the work-time patterns displayed by women with children. This shows that future research is needed to achieve a better understanding of the mechanisms driving the relationship between (a) the opportunities provided by institutions and (b) cultural orientations for different categories of women.

No potential conflict of interest was reported by the authors.

Per H. Jensen is professor of Social Policy at Center for Comparative Welfare Studies (www.ccws.dk), Aalborg University, Denmark. He has published widely in the fields of comparative welfare state analysis, formal and informal work, elder care, comparative labour market analysis, early exit/retirement, and the sociology of family and gender relations.

Rasmus Juul Møberg is associate professor in the Department of Sociology and Social Work at Aalborg University, Denmark. His research interests include comparative welfare studies, work-life balance, and female labour market participation.

Ralf Och is a research assistant and PhD candidate at Hamburg University. His research interests lie in comparative welfare state research, particularly in the fields of labour market and care policies, institutional theory, governance of social policy and comparative methods in social science.

Birgit Pfau-Effinger is professor in the Department of Social Sciences at the University of Hamburg, Germany, and professor at the Department of Political Science and Public Administration at the University of Southern Denmark, Odense, Denmark. Her research focuses on cross-national differences in the development paths of the cultural and institutional context of the work-employment relationship, and how cultural and institutions together impact on women's employment and care.