Women's employment patterns after childbirth and the perceived access to and use of flexitime and teleworking.

This article sets out to investigate how flexitime and teleworking can help women maintain their careers after childbirth. Despite the increased number of women in the labour market in the UK, many significantly reduce their working hours or leave the labour market altogether after childbirth. Based on border and boundary management theories, we expect flexitime and teleworking can help mothers stay employed and maintain their working hours. We explore the UK case, where the right to request flexible working has been expanded quickly as a way to address work-life balance issues. The dataset used is Understanding Society (2009-2014), a large household panel survey with data on flexible work. We find some suggestive evidence that flexible working can help women stay in employment after the birth of their first child. More evidence is found that mothers using flexitime and with access to teleworking are less likely to reduce their working hours after childbirth. This contributes to our understanding of flexible working not only as a tool for work-life balance, but also as a tool to enhance and maintain individuals' work capacities in periods of increased family demands. This has major implications for supporting mothers' careers and enhancing gender equality in the labour market.

Introduction

Most studies that aim to explain women’s employment patterns after childbirth focus mainly on family structures, gender ideology, human capital, and bargaining power between the couple and/or national policy structures (e.g. Schober, 2013; Smeaton, 2006; Tomlinson, 2006a). The studies that examine working conditions, especially how flexible working may help women remain in the labour market post childbirth, focus largely on part-time work or reducing working hours only. Although part-time work can help mothers balance work with family demands, it also entails career sacrifices that can have long-term consequences (Yerkes, 2009). This is particularly the case in liberal welfare regimes like the UK, where part-time work is of poor quality (Anxo et al., 2007). For many women in the UK, moving to a part-time job entails a drop in their occupational status as well as wage levels (Connolly and Gregory, 2008; Manning and Petrongolo, 2008; Tomlinson, 2006b). This leads to severe negative consequences for women’s careers in terms of earnings and career progression over their life course, effectively increasing the gender gap in earnings (Sigle-Rushton and Waldfogel, 2007).

Another type of flexible working strategy that is frequently and increasingly being used to combine paid work with family demands is allowing workers more control over when and where they work (Eurofound, 2015). This article focuses on two distinct types of flexible work arrangements, namely: flexitime, which allows flexibility in work schedules, for example, in starting and ending times; and teleworking, which allows flexibility in the location of work, for example, the possibility to work from home on occasion. In 2009, on average across 27 European countries, 57% of all companies provided flexitime to its workers, an increase from 49% in 2004. For the UK, this number has grown to 71% from 58%, respectively (Chung, 2014). Given its role as a family-friendly arrangement, studies have shown that flexitime and teleworking can potentially increase the time parents spend with children (Craig and Powell, 2012; Noonan et al., 2007) and can reduce the conflict felt owing to the competing demands coming from work and family life, although this is not always the case (Allen et al., 2013; Michel et al., 2011). Employers also have a vested interest in flexible working as part of a high-involvement system. Flexible working has been linked to individuals working longer hours (Lott and Chung, 2016; Noonan and Glass, 2012) and more intensively (Kelliher and Anderson, 2010), which increases performance outcomes for companies and career progression/income for workers (De Menezes and Kelliher, 2011; Glass and Noonan, 2016; Leslie et al., 2012). It is still unclear, however, whether flexitime and teleworking can be used as substitutes for reduction in working hours after childbirth, that is, whether being able to adapt one’s work boundaries to fit around family demands enables mothers to stay employed and maintain their working hours.

We examine data from the UK, where the right to request flexible working has expanded quickly over the past decade and is being promoted as a major way to address work–life balance issues. Although many studies have observed some key determinants of women’s employment patterns after childbirth (e.g. Gustafsson et al., 1996; Schober, 2013; Smeaton, 2006), no study to our knowledge examined how flexitime and teleworking arrangements relate to mothers’ employment patterns after childbirth using longitudinal data. The dataset used is Understanding Society, 2009–2014 (Knies, 2015), a large household panel survey with data on access to and use of various types of flexible work arrangements. We use logistic regression models to examine whether access to and use of flexitime and teleworking increases women’s likelihood of staying in employment after childbirth as well as maintaining their working hours. The article is divided into five sections. The next section will define the key concepts used here and provide a theoretical background to the relationship between flexible working and women’s employment patterns after childbirth. Section two also provides some policy background concerning the family policy context of the UK and outlines the development of the right to request flexible working. Section three explores the data and method used for this article. The analysis follows and the article concludes with a summary of the findings and a discussion.

Background

UK family policy context and childcare availability

The influence of flexible working on work–life balance and labour market outcomes will depend on the institutional settings of the country (Piszczek and Berg, 2014). Institutional settings influence who has access to flexible work arrangements, the nature of the arrangements, and accordingly its work–family outcomes (Chung, 2016; Lott, 2015; Mills and Täht, 2010). The UK is a typical liberal welfare state (Esping-Andersen, 1990), and is traditionally a (modified) male breadwinner model country (Lewis, 1992). Work–family reconciliation was considered a private family responsibility till the late 1990s, with weak public financial support for families. However, some developments have been made since the late 1990s when the state accepted a role in the work–family policy area (Lewis et al., 2008). One of the first developments was in maternity leave: mothers can take up to 12 months off, 10 of which are paid but at a relatively low rate in comparison to other European countries. In 2016, statutory maternity pay, given only to those with more than one year of employment before the due date of the child, pays up to 90% of the average weekly earnings of the worker for the first six weeks. However, this drops to approximately £140 per week for the remaining 33 weeks. Some companies and sectors top up this pay, but this varies widely. Recently, the government has introduced a shared parental leave scheme, where fathers can take up the remainder of the statutory maternity leave when mothers do not, yet there is no earmarked leave for fathers beyond the first two weeks after childbirth.

In addition, public childcare provision remains rather restricted in the UK. Only since 2008 has part-time childcare been offered to children over three years of age (Lewis et al., 2008), at 15 hours per week for 38 weeks of the year but with plans for this to be doubled to 30 hours per week in late 2017. Moreover, the UK has one of the most expensive childcare costs of the OECD countries (OECD, 2011), which has further increased by a third between 2010 and 2015 (Rutter, 2015). Finally, there are serious shortages of childcare and after/out of school places across the UK, with more than half of all local authorities reporting shortages (Rutter, 2016). This is in stark contrast to some other European countries, such as Sweden, where access to formal childcare from the age of one is a formal right with restrictions on the maximum fee parents can be charged (Béland et al., 2014), or Germany, where affordable childcare provision for under threes has expanded greatly over the past decade (Oliver and Mätzke, 2014). In 2014, only 29% of children between the ages of 0 and two in the UK were in formal childcare, the majority being in part-time care (less than 30 hours a week), with only 4% of all children between 0 and two in full-time childcare. To put this in perspective, on average in the 28 EU countries (EU28), although only 28% of all 0–2s were in formal childcare, half of them were in full-time care (14% of the total). In Denmark and Sweden, the prototypes of childcare provision, 70% and 57% of 0–2s are in formal childcare, respectively, with the majority of them in full-time care (Eurostat, 2016a). Importantly, the UK was one of the few countries where the proportion of children in formal childcare has dropped over the years (Plantenga and Remery, 2015). Given the role early years childcare has in the prompt labour market re-entry of mothers after childbirth (Oliver and Mätzke, 2014), this lack of childcare provision in the UK is a major issue that hinders mothers’ return to work, especially on a full-time basis (Tomlinson, 2006a), with major consequences for women’s careers in a life-course perspective.

Women’s employment patterns after childbirth

In the UK, the employment rate of women is relatively high prior to childbirth, and the gender gap in employment rate and hours worked is relatively low compared with other European countries. For example, in 2015, on average in EU28, 78% of women between the ages of 25 and 49 without children were employed, whereas this was 85% for the UK. However, women in the UK are much more likely to drop out of the labour market or reduce their hours after childbirth compared with other European countries (Miani and Hoorens, 2014). For the EU28 average, 69% of women between the ages of 25 and 49 with children were employed in 2015, which is a nine percentage point drop compared with childless women. For the UK, despite still being slightly higher than the European average, a much sharper drop of 14 percentage points is found for the employment rate of women with children to 71% (Eurostat, 2016b). Further, in the UK in 2015, only 16% of all women between the ages of 25 and 49 without dependent children worked part-time, whereas this proportion more than triples for women in the same age group with children to 52%. This increase is much lower for the EU28 average, with an increase from 20% to 36% (Eurostat, 2016c).

A large and growing number of studies examine the key factors explaining changes in women’s employment patterns after childbirth, that is, their return back to employment and their working hours after childbirth (for the UK, for example, see Gustafsson et al., 1996; Schober, 2013; Smeaton, 2006; Tomlinson, 2006a). In short, the factors that can influence the working patterns of mothers include the family situation in terms of the number and age of children, existence of a partner, the gender attitudes of both partners, and the women’s and their partners’ human capital such as income, education and occupational levels. Many studies also note that family policy contexts such as the availability of (paid) leave and public childcare provision, and the availability of both formal and informal care networks, are major factors influencing women’s decision to drop out of the labour market or move to part-time employment (e.g. Tomlinson, 2006a). Despite the large number of studies, it is still uncertain whether providing women with more control over when and where they work can help them address family demands without reducing their working hours or dropping out of the labour market altogether. Before we move on to explain how these two distinct types of flexible working can help women maintain their labour market statuses, we examine its definitions and the UK policy development regarding flexible working.

Flexible working definitions

Flexible working includes employees’ control over when they work or where they work (Allen et al., 2013; Kelly et al., 2011). More specifically, flexitime entails workers’ ability to change the timing of their work (that is, to alternate the starting and ending times), and/or to fluctuate the numbers of hours worked per day or week – which may also include accumulating hours for days off. In the broader sense, flexitime can also include the following: annualized hours, where ‘working hours’ is not defined per day or week, but calculated throughout the year; and compressed hours, where workers maintain their working hours, usually full-time, but work fewer days; for example, four rather than five days. Teleworking allows workers to work outside of their normal work premises; for example, working from home. Although flexible working can also include workers having control over how much they work – for example, part-time work, term-time only and job-sharing – for the purposes of this article, when we refer to flexible working we concentrate on flexitime and teleworking.

UK right to request flexible working

The British right to request flexible working was introduced in 2003 ‘under the banner of enhancing parenting choice’ (Lewis et al., 2008: 272). In the context of a lack of other means for parents to address work–life balance issues, for example, through well paid leave and public childcare, this was a policy through which the then Labour majority government aimed to address women’s employment agenda without incurring significant costs for the government. Initially, the right was only available for parents of children under the age of six and children with a disability up to the age of 18. In 2007, this was extended to carers of adults, and parents with children below the age of 17, and finally extended to cover all workers as of June 2014. The right, however, is restricted to those who have been in continuous employment with their current employer for the past 26 weeks, and only one application can be made in the span of 12 months. Employers can reject this request on various business grounds (see ACAS, 2016). A recent survey of working mothers in the UK has shown that a quarter of mothers had their flexible working requests turned down, and this number increased to 35% for those who were currently on maternity leave (Workingmums, 2016).

According to the 2011 UK Work, Employment Relations Survey (WERS), the most frequently provided/used flexible work arrangements within the right to request after reduced working hours were flexitime (flexible scheduling of hours) and telework/home working (Figure 1). Thirty-five percent of all companies provided flexitime and 30% provided possibilities to work from home to at least some of its employees. Further, approximately one-fifth of all companies say they allow at least some employees the possibility to use compressed hours, job-sharing or term-time-only work arrangements. When examining the employee survey, approximately 30% of those surveyed used flexitime, 17% teleworked/worked from home on occasion, and less than 10% said they used the other types of flexible arrangements in 2011. This proportion is somewhat higher for those with caring responsibilities, especially for the former two (Wanrooy et al., 2013).

Figure 1

The proportion of companies providing flexible work arrangements, and percentage of workers taking up flexible work arrangements (source: Wanrooy, 2013, WERS 2011).

Percentage of workers using reduced hours not included in the graph.

Flexitime/teleworking and women’s employment pattern after childbirth

Based on Karasek and Theorell’s (1992) demand–control–support (DCS) model, Kossek et al. (2006) argue that, similar to other types of job autonomy, workers’ control over when and where they work can reduce work–family conflict, because workers can restructure work around family demands as needed. They argue that even perceived access is important, because workers expect to be able to use the flexibility to adapt work in the future when needed. Similarly, according to the work–family border or boundary theory (Clark, 2000; Piszczek and Berg, 2014), flexitime and telework provide workers with the flexibility and control over the temporal and physical boundaries between their work and home domains, allowing workers to adapt work to fit around family demands. It may also allow a certain level of blending, where work and family demands can be met at the same time. Such flexibility and control is crucial, especially in the UK context, where there is a general lack of affordable full-time formal childcare provision. Many women in the UK rely on a patchwork of childcare where they depend on a complex combination of various types of formal and informal care (Warren et al., 2010) that may not be reliable or predictable. Further, in many cases, childcare provision schedules (e.g. pre/school pick-up times at 3pm) may conflict with more typical work schedules (e.g. 9am to 5pm). In this sense, when there are strong borders or restrictions in the family domain, a better work–life balance can be facilitated by the flexibility and permeability of the work domain (Clark, 2000). This may be especially true for women, as they are more often than men the primary carers of children. Flexitime can allow work schedules to be fitted around family schedules, and may also allow the use of tag-team parenting to extend family time, allowing parents to care for children without reducing their working hours (Craig and Powell, 2012). Teleworking allows workers to address family demands by providing a possibility to blend/integrate the work and family domains, allowing parents to potentially combine childcare with work at the same time; for example, taking care of a sick child whilst working from home. In addition, workers with long commutes will have more time for childcare and/or work when they do not need to travel when working from home. In sum, we argue that flexitime and teleworking allow mothers to shape the boundaries between work and family domains, and potentially integrate them, enabling them to combine the two more easily. Thus, we expect that women are more likely to stay in employment and maintain their working hours after childbirth if they are able to control when and where they work.

In fact, several empirical studies provide evidence for women using flexibility in their work to adjust their work patterns to fit around family commitments (Craig and Powell, 2012; Maume, 2006; Singley and Hynes, 2005). Flexible working has also been linked to reduced labour turnover and turnover intentions of workers (De Menezes and Kelliher, 2011) via increased work-to-family enrichment (McNall et al., 2009). Some studies show that having control over where and when one works can relieve work–family conflict – that is, the conflict one feels from the demands coming from work and family life (e.g. Chung, 2011; Kelly et al., 2014), most notably during the transition into parenthood (Erickson et al., 2010). However, others argue that flexitime and telework have little or no impact on workers’ work–family conflict (Allen et al., 2013; Michel et al., 2011) or that they, particularly teleworking, can potentially increase work–family conflict (e.g. Golden et al., 2006; Kossek et al., 2006). One reason behind this is that flexitime and teleworking may allow workers, who would otherwise not have been able to, to remain in the labour market and maintain their working hours. This is in contrast to part-time work, where the reduction in working hours reduces the likelihood that work will interfere with family demands.

Of the two types of flexible work arrangements, we expect that telework may be somewhat limited compared with flexitime in enhancing women’s capacity to maintain work with family demands. Teleworking is not as commonly used as flexitime and deviates more from the presentism culture that still prevails in most work cultures. Thus, the use of telework may need more persuasion by employees for it to be allowed on a regular basis – which will be necessary for it to be a true option for mothers to integrate work with family life. This may be why scholars have noted that managers negotiate for or expect increased work intensity from employees in exchange for the ability to work from home (Bathini and Kandathil, 2017) and that teleworking leads to longer overtime hours (Glass and Noonan, 2016). In addition, the stronger tendency for teleworking to allow for a blurred boundary between work and family or blending of the two spheres may increase domestic responsibility for women. Some studies argue that teleworking can potentially increase traditionalization of division of labour by perpetrating further domestic responsibilities to mothers who work from home (Hilbrecht et al., 2008).

Access versus use

Finally, we need to address the difference between perceived access to and use of flexitime and telework. Perceived access may influence workers’ employment patterns as even if they do not use the arrangements at the moment, workers may expect to be able to use the arrangements when needed in the future (Eaton, 2003; Kossek et al., 2006). For perceived access to be translated into use, or perceived flexibility to be enacted, workers will further need a demand and desire to use the arrangements (Hochschild and Machung, 1989). In our study, we measure both workers’ perceived access to and use of flexitime and telework. In relation to our research aim, perceived access to these flexible work arrangements before childbirth alone may be sufficient to encourage women to stay in employment, expecting that they are able to take it up when needed/after childbirth. On the other hand, for mothers to retain their working hours, use of these flexible work arrangements may be more important in that they will need to be actively using the arrangements in order to adapt their employment patterns to their new family demands (childcare). In sum, we come to the following hypotheses:

Hypothesis 1: Access to (Use of) flexitime and telework will decrease the likelihood of mothers leaving their jobs after childbirth.

Hypothesis 2: (Access to) Use of flexitime and telework will decrease the likelihood of mothers reducing their working hours after childbirth.

Potential selection effect

The relationship between flexitime/teleworking and women’s employment patterns may not purely be owing to the use of these arrangements. Women’s access to these flexible work arrangements may reflect that they are in higher positions within the company and are more valued, more productive, and are the type of workers more likely to retain their labour market positions after childbirth. The introduction of flexitime or telework may be driven more by productivity goals rather than to address work–family issues (Kelliher and Anderson, 2010; Osterman, 1995). Providing workers with more control over their work may be used by employers to attract high-skilled workers who value such arrangements (Osterman, 1995) and provided to higher-status workers as a reward (Schieman et al., 2013). When employers are more interested in the enhanced performance outcomes gained from introducing flexibility in how and when people work, workers who are more likely to increase work intensity/hours or contribute to the organization are more likely to have access to it (Swanberg et al., 2005). In fact, studies indicate that higher-status workers, that is, higher-educated workers and workers in higher occupations and/or in supervisory roles, are more likely to have access to flexitime and teleworking (Chung, forthcoming; Golden, 2009; Noonan and Glass, 2012). Women in higher occupations have more potential for career progression, where long career breaks can entail a larger potential human capital depreciation and a larger financial capital loss. They are also more likely to have the financial capacity to afford private childcare. They are thus more likely to come back to work earlier as well as to maintain their working hours after childbirth (Smeaton, 2006). We can expect that women with access to/or using flexitime or telework are more likely to be the ones to maintain their employment status and hours, not only because they are able to do it owing to the control they have over their work, but also because they are likely to be the type of workers with more vested interest to do so. However, we expect that much of this selection effect will be explained away once we control for workers’ job authority, wage and education levels, as we do in this article. Also, we expect that this selection effect will be smaller when comparing women who have access and are using flexibility with women who have access but are not using flexibility than when comparing them with all women who are not using flexibility (see robustness checks).

Data and method

We used Understanding Society (Knies, 2015) waves 2 and 4 (2010/2011 and 2012/2013) to explore our research question. Understanding Society is a large household panel data containing (at wave 1) about 40,000 households in the UK (Knies, 2015). In waves 2 and 4, respondents were asked about their flexible work arrangements. We selected individuals who participated in both waves (also excluding proxy interviews), for whom it was possible to combine the household data with the individual data (over 99.9% of the cases in each wave), who had at least one child between waves 2 and 4 and at least one child living with them in wave 4 (for information about births between the waves, we used wave 3). We further exclude those currently on maternity leave, as they may decide to change their work arrangements only after going back to work rather than while they are on leave. We also exclude those not in paid employment in wave 2 or self-employed in either wave as flexible work arrangements are not relevant for these respondents. These selections led to a total of 523 women being included in this study.

Measures of employment and reduction in working hours

We examine two dependent variables in this study, namely whether the respondent is (still) employed in wave 4 and whether the respondent reduced her working hours in wave 4 compared with wave 2. Regarding employment, respondents were asked whether they did any paid work in the last week. If they indicated they did not do any paid work in the previous week they were asked whether they had a job that they were away from in that week. We combined these two questions to indicate whether respondents were still in employment (1 = yes, 0 = no). Second, to measure reduction in actual working hours we combined answers to the following variables: ‘Thinking about your (main) job, how many hours, excluding overtime and meal breaks, are you expected to work in a normal week?’, which is considered to reflect contractual working hours; and ‘And how many hours’ overtime do you usually work in a normal week?’, considered here as overtime hours. We set the maximum hours to 70 hours/week, though we tested the impact of this decision in a robustness check. Then we compared the answers given in wave 2 and in wave 4. Individuals who reduced their working hours by more than four hours/week were considered to have reduced their working hours. We chose four hours as the cutoff because there may be smaller fluctuations in people’s working hours that may also be owing to errors in reporting, and a belief that only a reduction of more than half a day a week can be considered a significant decrease. We also assessed different cutoff points for reduction of working hours as a robustness check. Flexible working can lead to an increase in work intensity and in overtime hours of workers (Kelliher and Anderson, 2010; Lott and Chung, 2016). Therefore, there may be women who are considered to have maintained their actual working hours even when they have reduced their contracted hours, owing to the increase in their overtime hours. When this overtime is unpaid and is not recognized by managers, contractual hours may be of more importance in relation to one’s career progression and income earned. Thus, we further compared the reduction in contractual working hours as a robustness check.

Measures of flexitime and telework

Our main explanatory variable is whether individuals are able to work flexibly and whether they are indeed doing so. To measure the perceived access to flexible work arrangements, we use the question ‘If you personally needed any, which of the arrangements listed on the card are available at your workplace?’, where available answers were: part-time working, working term-time only, job-sharing, flexitime, working a compressed week, working annualized hours, working from home on a regular basis, other flexible working arrangements, or none of these. The use of flexible work arrangements was derived by using the question ‘Do you currently work in any of these ways’, which was asked for those flexible work arrangements that the respondent stated were available at their workplace. Based on these questions we made six variables. We combined flexitime, compressed hours and annualized hours, to measure (i) access to and (ii) use of flexitime. Similarly, working from home on a regular basis was used to measure (iii) access to and (iv) use of telework. Finally, (v) access to and (vi) use of flexible work were constructed by combining flexitime and telework along with access to or use of ‘other flexible working arrangements’. Note that this does not include part-time working, working term-time only or job-sharing. Individuals who did not have access to a flexible work arrangement also were not able to use it, and thus were coded as not using this arrangement. We also did a robustness check with a reference category of having access to but not using the arrangement. The access to and use of flexible work arrangements in wave 2 were used to explain the respondent’s employment status in wave 4. We further combined information from both waves to explain women’s likelihood of reducing their working hours. This combined variable consists of the following categories: (i) the respondent had access to/used the arrangements in neither wave (reference category), (ii) in wave 4 only, (iii) in wave 2 only or (iv) in both waves.

Control variables

Based on previous literature, we included a number of control variables. They are: age of respondent in wave 4 (continuous), educational level in wave 4 (higher education degree or not), working hours of respondent in wave 2 (small part-time, large part-time, full-time, long hours), gender role attitudes in wave 2 (factor score based on three variables), occupational level (three categories), whether there was a union present at the workplace (yes/no), whether the respondent was working in the private sector (yes/no), log of wages of respondent corrected for working hours (continuous), age of youngest child in household in wave 4 (two or three categories depending on analysis), total number of children in household in wave 4 (three categories), whether the respondent was living with a partner in wave 4 (yes/no), whether this partner worked in paid employment in wave 4 (yes/no), or in self-employment in wave 4 (yes/no), the working hours of this partner in wave 4 (continuous), and whether the partner was using (any) flexible work arrangements – flexitime, telework or other – in wave 4 (yes/no). Details on these control variables can be found in Appendix 1, available online as a data supplement. Summary statistics of all variables can be found in Appendix Table A1–1, also available online as a data supplement.

Analytical strategy

Our main method consists of a series of logistic regression analyses. First, we assessed the influence of the access to flexible work, flexitime and telework in wave 2 on the likelihood of (still) being employed in wave 4. We then repeated these analyses for use of flexible work, flexitime and telework in wave 2. We treat anyone with missing values on any of the included variables as missing for the analysis. Because of this, these analyses were carried out on 335 respondents. In a robustness check we examined the impact of missing data (see robustness check section). Second, we examined the reduction in working hours. This is done on a sub-selection of mothers who were still in employment in wave 4. Therefore, the sample size for these analyses is 272. Again, we make a distinction between access and use. We then examined access to and use of flexible work arrangements in wave 2 as well as in wave 4. Flexible work arrangements may be most important for first-time mothers. Mothers who already had a child before our observation period may already have changed their work patterns and may not change them (much) again when having a subsequent child (Vlasblom and Schippers, 2006). Hence, we repeated all analyses described above on a subsample of first-time mothers only, defined here as women who gave birth between waves 2 and 4 and had one child in the household in wave 4. A variety of robustness checks are also performed and are discussed at the end of the Results section. All analyses are done in Stata 12.1.

Results

The first hypothesis was ‘Access to (Use of) flexitime and telework will decrease the likelihood of mothers leaving their jobs after childbirth’. We did not find much evidence for this hypothesis. Though in the right direction, access to flexible work arrangements (b = .17, p = .580), flexitime (b = .20, p = .548) and telework (b = .45, p = .383) were not significantly related to the likelihood of being employed in wave 4. Similarly, use of a flexible work arrangement (b = .45, p = .301), flexitime (b = .39, p = .481) and telework (b = .75, p = .296) were also not significantly related to the likelihood of being employed in wave 4. Full models including all control variables are in Appendix 2, available online as a data supplement. The second hypothesis was ‘(Access to) Use of flexitime and telework will decrease the likelihood of mothers reducing their working hours after childbirth’. Table 1 shows the results for access to flexible work arrangements and Table 2 shows the results for the use of flexible work arrangements.

Table 1.

Logistic regression of the likelihood of reducing working hours (by more than 4 hours/week) on access to flexible work.

	Flexible work		Flexitime		Telework		Flexitime & Telework
	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value
Constant	5.53	.010	5.54	.011	4.86	.024	4.64	.036
Main variables
No access to flexible work/time in either wave (ref)
Access to flexible work/time in wave 4 only	−.39	.462	.09	.860			.15	.776
Access to flexible work/time in wave 2 only	−.36	.458	.00	.994			.15	.755
Access to flexible work/time in both waves	−.70	.065	−.43	.257			.19	.675
No access to telework in either wave (ref)
Access to telework in wave 4 only					−1.36	.022	−1.47	.025
Access to telework in wave 2 only					−.39	.650	−.45	.608
Access to telework in both waves					−1.07	.029	−1.15	.035
Pseudo R2	.28		.27		.29		.29

Source: Understanding Society waves 2–4.

n = 272. The model controls for the respondent’s working hours, age, age of the youngest child in the household, total number of children in the household, gender attitude of the respondent, whether the respondent had a degree, whether she was working in the private sector, whether there was a union present, the occupational level of the respondent, her (log corrected) wage, whether she was living with her partner, whether this partner was employed or self-employed (compared with not employed), working hours and earnings of this partner, and whether the partner used flexible work. Detailed results are in Appendix 2, available online as a data supplement.

Table 2

Logistic regression of the likelihood of reducing working hours (by more than 4 hours/week) on use of flexible work.

	Flexible work		Flexitime		Telework		Flexitime & Telework
	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value
Constant	5.82	.011	5.63	.011	4.86	.023	5.57	.013
Main variables
No use of flexible work/time in either wave (ref)
Use flexible work/time in wave 4 only	−1.86	.012	−1.26	.157			−1.04	.268
Use flexible work/time in wave 2 only	−.77	.157	−.38	.498			−.34	.547
Use flexible work/time in both waves	−1.82	.001	−2.12	.001			−2.06	.003
No use of telework in either wave (ref)
Use telework in wave 4 only					−1.61	.080	−1.19	.264
Use telework in wave 2 only					−.13	.915	.32	.804
Use telework in both waves					−.78	.326	−.27	.740
Pseudo R2	.32		.30		.28		.31

Source: Understanding Society waves 2–4.n = 272. Same controls as Table 1. Detailed results are in Appendix 2, available online as a data supplement.

Table 1 shows that if women are able to work from home when necessary in wave 4 or in both waves, they are less likely to substantially reduce their working hours after childbirth. Surprisingly, as shown in Table 2, use of telework was not significantly related to the reduction of hours. The clearest result from this model was that if women used flexitime in both waves they were less likely to substantively reduce their working hours. Although there was a decrease in the likelihood of reducing hours for women who used flexitime in wave 4 only, this was not statistically significant and was less pronounced than for women using flexitime in wave 2 as well as wave 4.

To get an idea of the effect sizes, we examined the predicted probabilities based on the models including both flexitime and telework. Figure 2 provides the likelihood to substantially reduce working hours after childbirth for women with and without access to telework. For women with access to telework in wave 4 or both waves, we expect about 32–36% to reduce their hours. This is about 20% points lower than for women without access to teleworking (54%). Figure 3 compares the predicted probabilities of reducing working hours for women using and not using flexitime, respectively. Women who were not using flexitime in either wave have a likelihood of about 53% to reduce their working hours after childbirth. The use of flexitime in both waves more than halved this likelihood to about 23%.

Figure 2

Predicted probabilities access to telework.

Predicted probabilities based on logistic regression of the likelihood of reducing working hours on access to telework.

Figure 3

Predicted probabilities use flexitime.

Predicted probabilities based on logistic regression of the likelihood of reducing working hours on flexitime use.

First-time mothers

We expect the influence of flexible working to be most pronounced for first-time mothers, as individuals who have had subsequent births during our observation period may have adapted their employment patterns prior to the observed birth. We do observe a more pronounced relationship between the access to flexitime (b = 1.06, p = .086) and telework (b = 1.64, p = .150), and the use of flexitime (b = 1.84, p = .110) on the likelihood of women being employed after the first childbirth (see Appendix Tables A2–5 and A2–6, available online as a data supplement). Although not statistically significant at the conventional .05 level, this may be owing to the small sample size (n = 131), and we believe that we should be cautious to reject this relationship purely based on significance levels.

The other results for this sample were similar to those found for the total sample. Access to teleworking significantly lowered the likelihood of mothers reducing their working hours, but for first-time mothers, only the access in the 4th wave was significant. Further, when combined with the access to flexitime in the same model, this effect became insignificant. Unlike in the general sample, for first-time mothers, access to flexitime seemed to lower the likelihood of reducing one’s working hours, especially when this was available in both waves (Table 3). Like the general sample, use of flexitime in both waves significantly lowered the likelihood of mothers reducing their working hours after childbirth, whereas use of telework was not significant (Table 4). Note that we encountered some problems owing to small sample sizes, so not all effects were estimated (see also note beneath table).

Table 3.

Logistic regression of the likelihood of reducing working hours (by more than 4 h/w) on access to flexible work – only first child.

	Flexible work		Flexitime		Telework		Flexitime & Telework
	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value
Constant	6.42	.122	9.25	.046	4.88	.240	9.48	.061
Main variables
No access to flexible work/time in either wave (ref)
Access to flexible work/time in wave 4 only	−.79	.488	−1.22	.312			−1.21	.326
Access to flexible work/time in wave 2 only	−.03	.973	.18	.847			.37	.703
Access to flexible work/time in both waves	−1.87	.017	−1.96	.006			−1.83	.027
No access to telework in either wave (ref)
Access to telework in wave 4 only					−1.88	.033	−.79	.434
Access to telework in wave 2 only					−.49	.682	−.35	.779
Access to telework in both waves					−.72	.359	.35	.705
Pseudo R2	.31		.32		.28		.33

Source: Understanding Society waves 2–4.n = 117. There are no women who work in a small part-time job in wave 2 in this sample. Same controls as Table 1 except for number of children. Detailed results are in Appendix 2, available online as a data supplement.

Table 4.

Logistic regression of the likelihood of reducing working hours (by more than 4 h/w) on use of flexible work – only first child.

	Flexible work		Flexitime		Flexible place		Flexitime & Telework
	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value	Coefficient	p-value
Constant	6.55	.116	7.35	.077	6.25	.129	7.97	.069
Main variables
No use of flexible work/time in either wave (ref)
Use flexible work/time in wave 4 only	−3.53	.022	–	–			–	–
Use flexible work/time in wave 2 only	.10	.926	−.07	.943			−.29	.792
Use flexible work/time in both waves	−1.60	.025	−2.09	.009			−2.44	.010
No use of telework in either wave (ref)
Use telework in wave 4 only					−1.75	.245	−1.17	.516
Use telework in wave 2 only					–	–	–	–
Use telework in both waves					−.34	.790	.70	.631
Pseudo R2	.33		.30		.27		.32

Source: Understanding Society waves 2–4.n = 117 for flexible work; 115 for flexitime as only two observations exist for flexitime in wave 4 only; 114 for telework as only three observation exists for telework in wave 2 only, 112 for flexible time and place. There are no women who work in a small part-time job in wave 1 in this sample. Same controls as Table 1 except for number of children. Detailed results are in Appendix 2, available online as a data supplement.

We again turn to predicted probabilities to have an idea of effect sizes. Examining working time reduction, the first thing to notice is that, for first-time mothers, the likelihood of reducing working hours is much higher than for the full sample (see Figure 4). Women who did not have access to flexitime in either wave or had it only in wave 2 had a likelihood of over 81–85% to reduce their working hours after the birth of a child. Access to flexitime in both waves reduced this likelihood to 56%. Turning to use of flexitime, women who did not use flexitime in either wave had a likelihood of about 79% to reduce their working hours, and the likelihood almost halved for women who used flexitime in both waves, to about 41% (Figure 5).

Figure 4.

Predicted probabilities access to flexitime – one child only.

Predicted probabilities based on logistic regression of the likelihood of reducing working hours on access to flexitime – one child only.

Figure 5.

Predicted probabilities use flexitime – one child only.

Predicted probabilities based on logistic regression of the likelihood of reducing working hours on use of flexitime – one child only.

Robustness checks

We performed a series of robustness checks to see how sensitive our results were to various specifications of our model (see Appendix 3, available online as a data supplement). First, a Heckman selection model was done to see whether the equations for being employed and reduction of hours were independent and, if not, whether treating them as such led to different conclusions. Second, multiple imputations were made and the analyses were repeated on 100 imputed datasets to know how much missing data affected our results. Third, we examined robust standard errors rather than the normal standard errors to see how sensitive the results were to misspecifications of the model. Fourth, the most influential cases in our model were identified, and we ran our model without these cases to see whether our results were solely owing to a few cases. Fifth, we included job change as a variable affecting the reduction of working hours (maybe someone changed jobs, hence changing the working hours and the access to flexible work arrangements). Sixth, we assess whether it mattered for our results that there was a cutoff of 70 hours/week for working hours, running the analysis without a cutoff. Seventh, we assessed whether only having access to flexible work/flexitime/telework but not using it as the reference category (rather than having in the reference category no access and no use) mattered for our results. Finally, we assessed various definitions of reduction of working hours (any reduction, more than 8 hours/week, and more than 10% per week). In short, these various robustness checks indicate that the effect of flexitime use on decreasing the likelihood of working time reduction is robust over various specifications of the model. There appears to be less evidence that the reduction of working hours was related to access to flexitime or telework, although for the latter the significance level does not drop drastically (see Appendix, available online as a data supplement).

Finally, we examined reduction in contractual working hours rather than actual working hours to separate the reduction that may be happening via overtime hours. We find that use of flexitime again remains significant when we exclude overtime hours from the analysis. In other words, women who use flexitime are more likely to keep (or increase) their contractual hours after childbirth compared with women who do not use flexitime. However, access to telework did not significantly decrease women’s likelihood of reducing contractual hours after childbirth, especially when access to flexitime was included in the model. This suggests that the lower likelihood to reduce working hours that was found previously may have been driven by an increase in overtime hours owing to teleworking (see also Bathini and Kandathil, 2017; Glass and Noonan, 2016), and not by maintaining contractual hours. Further, unlike in the analysis of actual working hours, perceived access to flexitime in both waves also results in a significant coefficient, decreasing the likelihood of mothers to reduce their contractual hours, but not when access to teleworking was also included in the model.

Controls

In addition to the impact of flexible work arrangements, we would like to explore some other factors that affected mothers’ likelihood of reducing working hours after childbirth. Unsurprisingly, women who worked shorter hours prior to childbirth were less likely to reduce their hours after childbirth; those who worked very long hours (48 or more) before childbirth were more likely to reduce it. First births were more likely to lead to a reduction in hours, most likely because women who had had children before might have adjusted their employment patterns before this observed birth. Interestingly, the working hour adjustment did not seem to happen straight after childbirth but perhaps a few months/years later; mothers whose youngest child was two years of age (thus those who gave birth in wave 2) were more likely to have reduced their hours compared with women who had just come out of maternity leave. Women in skilled work were more likely to keep their hours compared with women in managerial/professional occupations. This contradicts our expectation that women in higher occupational groups would be more likely to maintain their careers. Finally, living with a partner decreased the likelihood of reducing one’s hours, but only when the partner was unemployed or inactive. This is most likely owing to the breadwinning roles these mothers have (financial necessity), and because their partners can take up most of the childcare duties in this case (care capacity). See Appendix 2, available online as a data supplement, for more detail.

Conclusions and discussion

Despite the increased number of women in the labour market in the UK, many significantly reduce their working hours or leave the labour market altogether after childbirth. Previous studies of mothers’ employment patterns post childbirth largely focused on the family situation, the individual and their partner’s gender attitude and human capital, and/or on national family policy configurations. The studies that examined flexible working strategies women use to address work–life balance issues focused predominantly on part-time work. However, part-time work has limitations in being a desirable option for women given the low quality of part-time jobs, particularly in the UK. Women moving into part-time jobs or significantly reducing their working hours after childbirth may experience serious career consequences, which may reinforce gender inequality structures in the labour market and within the family. This article contributes to the literature by investigating a largely neglected area, namely whether perceived access to and use of two distinct types of flexible working – flexitime and teleworking – can influence mothers’ employment patterns post childbirth. Based on border and boundary management theories, we expected that women would be more likely to stay in employment and maintain their working hours after childbirth if they were able to control when and where they worked.

Results of this study provide evidence that flexitime and teleworking can help women sustain their employment status after childbirth. Although not significant at the conventional level, we find some suggestive evidence that access to and use of flexitime and teleworking may increase the likelihood of staying in employment for first-time mothers. In the UK, women are more likely to drop out of the labour market after the birth of their first child and only return back into the labour market after their second or final child is born (Vlasblom and Schippers, 2006). This explains why we observe more of an association between these two types of flexible working and employment for first-time mothers than for others. On the other hand, the association between flexitime/teleworking and mothers’ likelihood of reducing their working hours was found for all mothers. Women who had the ability to work from home when needed and those who used flexitime were much less likely to significantly reduce their working hours after childbirth. In the case of use of flexitime, the relationship was stable regardless of whether job changes were taken into account or when contractual hours rather than actual hours were examined.

It is important to note that it is the use of flexitime rather than perceived access to it that mattered most. For mothers with new-borns, perceived access to flexitime in itself may not be enough to tackle the work–life balance demands that they are faced with. The flexibility needs to be enacted to really make a difference. It is also important to note that it was for those who were already using flexitime prior to childbirth that flexitime seemed to matter most. Women who have been using flexitime prior to childbirth may be more likely to use flexitime again to adapt their work schedules to the new additional family demand. Further, women who have used flexitime prior to childbirth may be more aware about the consequences of using flexitime and may be more willing to use this arrangement to fit work around their family demands without the fear of ‘flexibility stigma’ – that is the stigma towards workers who use flexible work arrangements (Williams et al., 2013). The fact that these women were using flexitime prior to the birth of their child may also indicate that they did not experience negative consequences of using flexitime, or that the benefits outweighed the costs. Selection effects may be another reason why it is important to have used flexitime prior to childbirth. Even though we control for various job characteristics, it may be that there are certain aspects of the work environment and the occupational position of the worker that are not captured and that may be conducive for women to maintain their working hours. This could include things such as being more valued in the workplace, working in a more family-friendly workplace and/or being in a position with more vested interest to maintain their labour market status. We expect, however, that this is a less likely explanation as we also found a relationship between use of flexitime and reduction of working hours when comparing women who have access and use flexitime in both waves to women who have access but are not using flexitime in either wave (see Appendix for the results, available online as a data supplement).

This study has some limitations. Given that the data on flexible working were only available for two waves, we were restricted to a relatively small sample, with restrictive possibilities for analyses. More research on a larger sample will be necessary to see whether the effect of flexible working is more marked for certain groups of workers, for example, additional analysis of first-time mothers versus all mothers, and separate analysis for workers in different occupational groups, sectors and working hours. There may be a selection effect of flexible work arrangements in that higher-status workers may be more likely to have access to (see Appendix 4, available online as a data supplement) and further benefit from it. Therefore, additional analysis should examine whether the influence of flexitime and telework remains for women in lower occupational groups as well. It would also be useful to explore the career consequences of mothers who have used flexitime and telework to maintain their working hours. Although use of flexitime and telework have been linked to positive career premiums (Glass and Noonan, 2016; Lott and Chung, 2016), it can also come with ‘flexibility stigma’ (Williams et al., 2013) and can have negative consequences for one’s career, especially when the arrangements are used to meet family demands (Leslie et al., 2012). Future waves of these data will allow us to track longer-term consequences of these arrangements.

More waves of data will also allow us to investigate the influence of context in more depth. The data we used cover a rather specific point in time, that is, soon after the economic crisis of 2008/2009. This may have had an influence on the employment patterns of women as well as their access and use of flexible work arrangements. The economic conditions of the country, the region, and consequently economic cycles felt by the company may also have had an influence on the way in which these relationships played out. Whether the relationship found for the UK case holds in other countries with different institutional contexts should also be considered. Thus, future studies examining women’s employment patterns after childbirth should take a multilevel approach to explore these relationships and contextualize the results more. Finally, we have only examined the influence of flexitime and telework on women’s employment patterns post childbirth. Future studies should examine how fathers’ employment patterns as well as the amount of time they spend on childcare and domestic work after childbirth are influenced by these arrangements. This will provide us with a better understanding of the influence of flexible working on both men’s and women’s careers over the life course and how it may help reduce or increase the gender inequalities in the labour market.

Despite these limitations, this study has contributed to the existing literature and the border/boundary management theories in two ways. Firstly, it provides empirical evidence to show how flexibility and control over the temporal and physical boundaries of work can enhance the work capacity of workers in times of increased family demands. More specifically, when mothers have control over where but more importantly when they work, this allows them to sustain their careers after childbirth by helping them maintain their employment status and their working hours. Secondly, the results of this study also explain why previous studies found that flexitime and teleworking were not very effective in reducing work–family conflict. Unlike working part-time, flexitime/teleworking helps mothers to maintain their working hours and employment intensity, which may not necessarily help in reducing work–family conflict. However, by allowing women to maintain their career after childbirth, flexible working may help women’s role expansion (being able to carry out both work and family roles), which may be more effective in increasing life satisfaction (Grönlund and Öun, 2010).

Offering and actively promoting the use of such flexible work arrangements is crucial in supporting mothers who wish to remain active in the labour market in periods of high family demands. Flexible working may help alleviate some of the negative consequences of the motherhood penalty, by allowing mothers to remain in human-capital-intensive jobs, which can help diminish the gender wage gap (see also, Weeden, 2005). Further, allowing mothers to maintain their employment status will have major implications for retaining human capital for companies and society as a whole. Thus, it is pertinent to think of various ways to encourage employers to provide workers genuine access to flexible working, especially flexitime, and to encourage work cultures where flexible working is the norm rather than the exception.