National-Level Family Policies and workers’ Access to Schedule Control in a European Comparative Perspective: Crowding Out or In, and for Whom?

Abstract This paper examines national-level family policies in a comparative perspective, to see whether they “crowd out” company-level family-friendly policies, namely schedule control. Further, it examines whether this relationship varies for different types of family policies, and for different groups of workers – i.e. distinguished by gender, parenthood status and skill divisions. The paper uses data from 27 European countries in 2010, and applies multilevel random slopes models with cross-level interaction terms. Results show that generous national-level family policies, in particular work-facilitating policies, “crowd in” company-level schedule control provisions, especially for high-skilled workers. However, very generous leave entitlements seem to crowd out schedule control provision.


Introduction
Many studies examine the cross-national variation of family policies or the extent to which policies support a dual-earner/carer system (e.g. Crompton 2006;Lewis et al. 2008;Korpi et al. 2013) with much of the focus on comparing national-level policies across different countries. Although national-level policies are important when examining the support available for parents, it is also crucial to examine the meso-and micro-levels (see also , Abrahamson 2007;Kvist and Greve 2011). On the one hand, companies may effectively The next section explains what is meant by schedule control, the varying access to it across different groups, and examines the theories on the relationship between nationallevel policies and company-level provisions. The following section examines the data, variables used, as well as the methodologies applied in the paper. The fourth section will present the analysis results, followed by some final concluding remarks and suggestions for future studies.

Theories, Definition
What Is Schedule Control?
The concept of schedule control builds on the job demands-control model developed by Karasek (1979), but focuses on the control over when work is done rather than how it is done (Kelly and Moen 2007). More specifically, it entails a worker's ability to change the timing of their work (that is, to alternate the starting and ending timesi.e. flexitime), and in some cases, the ability to change the number of hours worked per day or week, up to a worker having complete control over their working hours (working-time autonomy).
Work-family border theory (Clark 2000) and flexibility enactment theory (Kossek et al. 2005) suggest that having control over one's work schedule can help facilitate the integration of work and home roles. Schedule control provides workers with control over the time border between work and family, allowing workers to adapt the timing of work around family demands (Clark 2000). Especially given that normal fixed working hours (e.g. 9 a.m. to 5 p.m.) and family schedules/demands (e.g. school pick-up times at 3:30 p.m.) are not necessarily compatible, the control over when you work can help workers resolve some of the conflict arising from this incompatibility. There is evidence to show that having control over their work and more specifically having schedule control can help relieve workers' work-family conflict, that is the conflict between the demands of the work and home spheres (e.g. Chung 2011;Allen et al. 2013;Kelly et al. 2014).
However, schedule control is not only used to address workers' family demands, but also to enhance performance (Ortega 2009). Schedule control can be used as a part of a high-involvement system (Wood and de Menezes 2010) or high-performance strategy, which believes that allowing workers more discretion and influence over their work can help improve performance (Davis and Kalleberg 2006). In fact, studies have shown that workers increase their work intensity and working hours when using schedule control (Kelliher and Anderson 2010;Lott and Chung 2016), and accordingly it can lead to income premiums for workers (Leslie et al. 2012) and better performance outcomes for employers (de Menezes and Kelliher 2011).

Who Has Access to Schedule Control?
Who has access to schedule control will depend on the company's main purpose for introducing the policy. Swanberg et al. (2005) discuss three principles in which these can take form: principles of need, equity and equality. When employers are genuinely interested in addressing the work-family needs of workers, those with most family demands are likely to get access to the arrangements. In this case, we could expect parents/workers with care responsibilities to be more likely to have access to schedule control. This will particularly be the case for parents with young children, given the amount of care required Family Policies and Access to Schedule Control 27 in the early years of a child's life. Women still take, and are expected to take, the bulk of responsibility for household tasks and care (Eurofound 2013). Thus when schedule control is provided based on the principle of need, women should have more access.
On the contrary, when employers are more interested in the enhanced performance gained from introducing schedule control (principle of equity), those who are more likely to be able to increase work intensity/hours or contribute to the organisation will have more access (Swanberg et al. 2005). Furthermore, schedule control is more likely to be provided as a reward to those in higher status, supervisory roles (Schieman et al. 2013), or as a part of the increase in overall control these workers obtain over their work (Kelly and Kalev 2006). As employers hold discriminatory assumptions regarding women's use of schedule controli.e. for family-friendly purposes (Brescoll et al. 2013), men are expected to and in some cases actually do increase their work intensity more than women when using schedule control (Lott and Chung 2016). This, along with the fact that organisational hierarchies are gendered (Acker 1990), means that employers may be more likely to provide schedule control to men when the principle of equity is prioritised in deciding who gets access to schedule control. Lastly, when the equality principle takes precedence, access to schedule control will be provided to all workers equally regardless of their family demands or performance outcome projections (Swanberg et al. 2005).
Empirically, the principle of equity has been shown to be the strongest determinant of schedule control access. Those in high-skill positions, more highly educated workers and workers at higher occupational levels are most likely to have access to schedule control (Golden 2009;Ortega 2009;Brescoll et al. 2013). Workers in disadvantaged positionse.g. low wage, hourly paid, less well educatedare least likely to have such access (e.g. Swanberg et al. 2005;Golden 2009). In addition, studies have shown that men are more likely to have and be given access to schedule control (Golden 2009;Ortega 2009;Minnotte et al. 2010;Lyness et al. 2012;Brescoll et al. 2013), even if the reason for the access is for childcare purposes (Munsch 2016). Regarding parental status, there is evidence that being a parentespecially of young childrenincreases the availability of schedule control (Golden 2009;Minnotte et al. 2010), while others found parental status to be insignificant (Swanberg et al. 2005;Ortega 2009).

Impact of Institutions
Despite the number of studies examining access to schedule control, few examine the role national-level policies have in determining access to schedule control for workers in a comparative perspective. Furthermore, how national-level family policies influence the relative access different groups of workers have to schedule control is yet to be examined.
"Crowding out" theory argues that national-level policies will crowd out company-level engagements (Etzioni 1995;van Oorschot and Arts 2005), i.e. companies will not be willing to, or may not feel a need to, provide company-level family policies when there are generous policies at the national level. However, this will not necessarily stop companies from using schedule control for performance-enhancing purposes. Thus, following the "crowding out" assumption, we would expect the principle of equity to be prevalent in countries with generous national-level family policies, with more highly skilled workers, men and non-parents being more likely to have access to it.
The counter-argument to this comes from the "crowding in" theory (e.g. Künemund and Rein 1999), arguing for a positive, rather than negative, relationship between generous welfare policies and informal welfarein particular intergenerational welfare. Unfortunately, there is yet to be a "crowding in" theory that focuses on the relationship between national-level policies and corporate family policies. However, institutional theory provides us with some insights on these relationships. Institutional theory argues that institutions, bureaucratic systems, laws and policies put pressure on organisations to become similar to national institutions (DiMaggio and Powell 1983). Especially of interest to this paper are the normative isomorphic pressures, i.e. national-level policies changing the norm and subsequent public demand for companies to be more family-friendly (Den Dulk et al. 2013), and mimetic pressures, i.e. where companies imitate or mimic the practices of other organisations (Davis and Kalleberg 2006, pp. 199-202). Based on this theory, we can expect that when there are generous national-level family policies, this will raise the benchmark and change the culture companies operate in, making them more likely to provide company-level family-friendly policies. This normative change may also lead companies to follow the principle of need when providing schedule control, following what is done at the national level. In this case, we would expect that in countries with generous national-level family policies, employers will make schedule control more readily available for workers with the most family demands. This would result in women/parents having more access compared to men/non-parents. In addition, we would observe a reduced gap between workers with different skill levels, since schedule control would be given to workers regardless of their potential returns.
In sum, the hypotheses for this paper can be summed up as: H1a: In countries with generous family policies at the national level, workers are less likely to have access to schedule control ("crowding out"). H1b: In countries with generous family policies at the national level, workers are more likely to have access to schedule control ("crowding in"). H2a: In countries with generous family policies at the national level, men/non-parents and more highly skilled workers are more likely to have access to schedule control, and a larger skill level/educational gap will be found ("crowding out"). H2b: In countries with generous family policies at the national level, women/parents are more likely to have access to schedule control and a smaller skill level/educational gap will be found ("crowding in").
Empirically, the results are varied. Some qualitative studies provide evidence of "crowding out": that is, in countries without much statutory family policy provision, companies use family-friendly policies for staff retention or other strategic reasons (Den Dulk 2005; Ollier-Malaterre 2009). On the contrary, studies using quantitative data of companies have shown that company-level schedule control policies are more widespread in countries where there are generous family policies at the national level (Den Dulk et al. 2013;Chung 2014). Others using individual-level quantitative data argue that there are no clear relationships between statutory regulations and (extra) company-level provisions (Kassinis and Stavrou 2013;Präg and Mills 2014), and only when there is a very large involvement from the state can a "crowding out" impact be seen (Evans 2002). Only one individual-level data study shows that family policies at the national level have a positive impact on workers' access to schedule control (Lyness et al. 2012).
The different results may be due to the different types of data used, but also due to the different measurement of family policies at the national level -Evans (2002), Lyness et al.

Family Policies and Access to Schedule Control 29
(2012), and Präg and Mills (2014) focus on leave policies, Kassinis and Stavrou (2013) and Chung (2014) use family policy expenditure, while Den Dulk et al. (2013) use a single composite indicator encompassing both leave policies and childcare provision. However, efforts made by governments to support families are multi-dimensional and can have very different impacts on work-family outcomes (Misra et al. 2011), and do not necessarily correlate with one another (Korpi 2000). Following the distinction made by previous scholars (Korpi 2000;Misra et al. 2011;Korpi et al. 2013), this paper distinguishes family policies into three different types. Firstly, public expenditure on family policies as a percentage of GDP indicates a general level of effort made at the national level. Secondly, "work-reducing" measures (Misra et al. 2011) indicate the state's provision of the right to provide careallowing parents/mothers to stay at home during the early years of a child's life through parental leave. Lastly, "work-facilitating" measures (Misra et al. 2011) indicate the extent to which the state encourages women's labour market participation/dual-earner system, for example through public childcare provisions. Misra et al. (2011) note that while generous work-facilitating policies have a positive influence on employment outcomes for women, extensive work-reducing policies such as long leave entitlements may exacerbate the gender gap by enforcing "mommy tracks"easing mothers out of the markets. Similarly, I expect that work-facilitating measures will be especially closely linked to the provision of schedule control. Only when childcare is readily available can schedule control help parents/mothers back into work after childbirth (see also Chung and van der Horst forthcoming). In countries with long leave entitlements, having flexibility in one's schedule may not be as important since mothers/parents will be on leave during these care-intensive years.
Another limitation to previous studies is that most do not examine how the impact of institutions can vary for different groups in the labour market. To the author's knowledge, only two studies examine any variation. Den Dulk et al. (2013) show that generous family policies at the national level increase the gap between companies in their provision of schedule control; in countries with generous family policies, public and larger companies are even better at providing schedule control. Lyness et al. (2012), on the other hand, do not find a significant influence of generous paid in explaining the cross-national variance of the gender gap in access to schedule control. More investigation is needed to see how different types of family policies may influence the access gap across a wider range of workers.

Data
The European Working Conditions Survey collected by the European Foundation aims to provide information on a number of dimensions regarding the working conditions of workers across Europe. Individuals across 34 European countries were included in the survey: the EU27 plus candidate countries. Of the sample, I make use of the data for the EU27 countries. The survey was conducted in the first half of 2010, where individuals were interviewed using a face-to-face method at the respondent's home. A random stratified sampling procedure was used to gather a representative sample of those aged 15 or over and in employment at the time of the survey. The target sample for most countries was 1,000 respondents but this varied across countries. The average response rate was 44.2 per cent across all countries, ranging from 31 per cent (Spain) to 74 per cent (Latvia). Of the total sample, I restrict the analysis to those in dependent employment, and further exclude those in the armed forces, and in agriculture/fisheries due to the specific nature of these jobs. I also exclude those over 65 years of age. Further excluding cases with missing values in any one of our variables results in a total of 23,234 cases across 27 countries. See http://eurofound.europa.eu/european-working-conditions-surveys-ewcs for more on this survey.

Variables
The provision of schedule control has been measured through the following question "How are your working time arrangements set?", where the workers can answer: 1 -"They are set by the company/organisation with no possibility for changes"; 2 -"You can choose between several fixed working schedules determined by the company/ organisation"; 3 -"You can adapt your working hours within certain limits (e.g. flexitime)"; and 4 -"Your working hours are entirely determined by yourself". Those who have answered 3 or 4 to this question are considered to have schedule control. Note that we are examining access to schedule control, which is distinct from the use of it (McNamara et al. 2012).
The key independent variables are gender; parental status, i.e. whether or not the respondent lives with a child under 18 years of age, and whether the respondent lives with a preschool child under 6 years of age; the worker's skill level based on occupational category as a continuous scale (professionals/managers = 3, associate professionals/ managers = 2, generally skilled/vocational = 1, and low/unskilled = 0) (see appendix for detailed definitions); 2 and education level measured through two dummies (tertiary or above, upper and post-secondary (reference), primary and lower-secondary). Based on previous studies (e.g. Chung 2009;Lyness et al. 2012; Präg and Mills 2014) I also include age; whether the respondent lives with a partner; contract type; working hours; whether the worker holds a supervisory role; existence of an employee representative at the company; management support; gender of the direct manager; gender dominance of the post; and finally the size and sector (public vs. private, as well as the line of business, reference group: manufacturing) of the company the respondent works in.
Family policy is measured by three indicators. Firstly, general family-policy generosity is measured through public expenditure on family policies as a percentage of GDP for the year 2010 derived from EUROSTAT. Secondly, work-reducing family policy is measured through effective parental leave, i.e. duration of the parental leave (including maternity or paternity) multiplied by the income replacement rate throughout these months. This variable is derived from the Multilinks data set, and is for the year 2009 due to the lack of data for 2010. Thirdly, the proportion of children using formal childcare for age group 0-3 years for 2010 from EUROSTAT is used to indicate work-facilitating measures. I choose childcare coverage of 0-3 instead of 3-compulsory school age/6, for two reasons. Firstly, the 0-3 age group demands more intensive childcare and thus is much harder to provide, while it is most crucial in terms of allowing mothers to continue their labour market commitment after childbirth (Korpi 2000). Secondly, the compulsory school age varies widely across Europe (e.g. 4 for the UK, 7 for Finland) thus making the index harder to compare, and similarly, in many cases childcare for 3-6 year olds is often considered a part of the formal educational system for young children, rather than as Family Policies and Access to Schedule Control 31 policies to support mothers' employment (Misra et al. 2011). See appendix for more detailed information about the operationalisation of all variables.

Modelling Method
Two-level random-intercept/random-slope multilevel regression models are used for the purposes of this paper. Multilevel modelling methods are used when data is clustered, and contextual effects are taken into account (Hox 2002). Multilevel modelling assumes that the lower-level samplehere individualsis subject to the influences of groupings (Rasbash et al. 2009), such as countries in this paper's case. Firstly, before examining the predictors of schedule control, the interclass correlation (ICC) is examined. The ICC indicates the extent to which the variance of the dependent variable could be attributed to the country level. Secondly, I include individual-level variables to see the influence of individual (and company) characteristics that can explain workers' access to schedule control. Next, I include national-level variables in the model to explain the variance in the access to schedule control found between countries, after taking into account the composition effect, e.g. the fact that different countries have different compositions of workers. This allows us to test whether generous family policies at the national level influence workers' access to schedule control (H1). Afterwards, random slopes models are used to test the varying impact of gender; parental status; skill level; and educational level across different policy settings (H2). Through this approach, we are able to test whether the gender gap (differences between men and women); parental status gap (difference between parents and non-parents); skill gap (differences between higher and lower skilled workers); and the educational gap (differences between more and less highly educated workers) in the access to schedule control are equal across countries. A significant variance in the random slope entails that there are countries where these gaps are significantly different from the average gap found across Europe (coefficients found for the general model). For the variables with significant variance in the random slope, I include cross-level interaction terms with the family policy variables and the random slope variables, to see whether the different types of national-level family policies can explain some of this variation. Or to put it differently, we are testing whether the impact of family policies vary across different segments of the labour market (see also Chung and Mau 2014). This allows us not only to examine whether national-level family policies "crowd out" or "crowd in" access to company-level arrangements for the general population, but also to see whether certain populations are targeted in this process. STATA 14.0 meqrlogit is used for the analysis.

Descriptive Statistics
Looking at the national-level variation in the access to schedule control (Figure 1), on average about 22 per cent of all workers have access. At first glance, the positions of countries in their average access to schedule control seem to reflect the existing familypolicy regime typologies. In Northern European countries, along with the Netherlands, workers are more likely to have access to schedule control, with more than 40 per cent having access. In countries such as Austria, Germany, Luxemburg and Belgium, followed by Estonia, Slovenia, the UK and France, at least one in five workers have access to schedule control. In Southern and some Eastern European countries, workers do not have much access to schedule control, with less than 10 per cent having access in countries such as Bulgaria, Cyprus, Greece, Portugal and Romania.

Multivariate Analysis
The ICC in the empty model without any controls is 22 per cent, indicating that just above one-fifth of the variance in the access to schedule control across all workers in our survey is due to the country they live in. Table 1 provides the results of the multilevel analysis including individual/company-level characteristics to explain access to schedule control. Firstly, we examine the key variables of our interest, gender; parental status; skill; and educational level. The gender of the respondent is not significantly related to whether they have access to schedule control, and nor is parental status. Workers with preschool children are more likely to have access to schedule control, but the difference is rather small. On the other hand, there is a significant divide between workers of different skill levels in their access to schedule control. The higher the respondent's skill level and the higher the respondent's education level, the more likely that they will have access to schedule control. On average, one step change upwards in the skill hierarchy increases your likelihood of accessing schedule control by approximately 1.5 times. Similarly, those with tertiary education are 1.6 times more likely than upper secondary educated workers (the reference group), and more than twice as likely as primary/lower secondary educated workers to have access to schedule control. N level 1 = 23,234, N level 2 = 27 countries, *** p < 0.001, ** p < 0.01, * p < 0.05. N level 1 = 23,234, N level 2 = 27 countries, *** p < 0.001, ** p < 0.01, * p < 0.05.
Examining other control variables, older workersmost likely with more experiencethose in supervisory roles, and those with supportive managers are more likely to have access to schedule control. Those working in job posts that are predominately populated by men are less likely to have access to schedule control compared to posts where men and women are equally represented, yet working in posts where women are prevalent is far worse. Having a partner, working hours, having an open-ended contract, working in companies with employee representatives, and the gender of one's boss are all not significant in explaining workers' access to schedule control.
Working in a public company is negatively associated with access to schedule control, which may be because private companies are more likely to introduce new and innovative practices. Workers employed in large companies (250+) are more likely to have access to schedule control, followed by medium-sized (100-249) and micro companies (fewer than 10). Small companies (10-99) are the ones where workers are least likely to have access to schedule control. It is likely that medium-sized and larger companies may have more capacity to provide formal schedule control arrangements, while micro companies use informal arrangements. However, for small companies it may be difficult to do either. Looking at sectoral variations, electricity, financial services, real estate, public administration and other service sectors are those where workers are most likely to have access to schedule control. These sectors may be those where work is organised more flexibly, and there are fewer constraints. On the other hand, education, health and social services sectors, along with mining and quarrying are the worst in providing access. This is perhaps due to constraints coming from the nature of the work done or due to the lack of potential performance gain that employers expect from these sectors with the introduction of schedule control.

Country-Level Results
To address hypothesis 1, I examine the relationship between national-level family policies and workers' access to schedule control. The results in Table 2 show that countries with generous family policies are those where individuals are more likely to have access to Significance symbols represents the significant increase in log likelihood scores from the nested model.

Family Policies and Access to Schedule Control 35
schedule control, supporting hypothesis 1b. In countries with stronger work-facilitating policiesi.e. childcare provision, and where family policies are generous in generali.e. more spending on family policies, workers are more likely to have access to schedule control. Parental leave, representing work-reducing policies, has an inverted U-shape relationship to workers' access to schedule control, as shown by Evans (2002). In other words, in countries where effective parental leave is about the European average (about a year) workers' access to schedule control is at its highest. However, access to schedule control is restricted in countries where effective leave entitlements are very short or very long (see Figures 2 and 3). Family policy variables combined explain more than 50 per cent of the cross-national variation in the provision of schedule control (see model 2-4 in Table 2). Next, I examine whether national-level family policies can help explain the crossnational variation in the division between workers of different genders, parental status, skill and educational classes in their access to schedule control. First, we need to test whether there are significant cross-national variations in the slope of these variables using random slopes models. The results show 3 that for the gender gap (variance 0.036, p = 0.099), the gap between parents and non-parents (variance 0.000, p > 0.100), and the gap between upper-secondary educated workers vs. primary educated workers (variance 0.043, p > 0.100), there are no significant cross-national variations. However, there are significant cross-national variances in the gap in the access to schedule control between tertiary vs. upper-secondary educated workers (variance 0.113, p = 0.022) and that between high-skilled and low-skilled workers (variance 0.025, p = 0.025). Next, I examine the relevance of national-level family policies in explaining the crossnational variation in the gaps between workers with different skill/education levels. I find that the additional access that workers have to schedule control in countries with generous childcare coverage and in countries with about average levels of parental leave is mostly driven by the additional access gained by high-skilled and more highly educated workers (Table 3 and Figures 2 and 3). Put differently, the schedule control access gap between high-vs low-skilled/educated workers is largest in countries with extensive childcare coverage and about average level of parental leave. This is because high-skilled/educated workers are much more likely have access to schedule control in these countries compared to other countries, while the differences across countries in the access low-skilled/educated workers have are smaller.

Conclusion and Discussions
Despite the growing number of studies that deal with family policies, most cross-national comparative analyses focus either on national-level policies or company-level provisions, ignoring how the two levels interact. Those that do examine this relationship rarely explore whether it varies depending on the group of workers in question, as well as between different types of family policies, i.e. work-reducing vs. work-facilitating policies. This article tries to fill this gap and contribute to the literature by examining workers' access to one of the most prevalent forms of company-level family-friendly policy:  Significance symbols represents the significant increase in log likelihood scores from the nested model. schedule control. The paper first examines whether there are large gaps in the access to schedule control across workers of different gender, parental status and skill level. Then it goes on to examine how national family policy contexts can explain the cross-national variation in workers' access to schedule control using the theory of "crowding out" and "crowding in". It further examines whether family policies can explain the cross-national variation in the access gap between different groups of workers.
The results show that there are large variations across different groups of workers in their access to schedule control. Those who have greater potential to contribute to the company, and increase their performance/work intensity, seem to have better access to schedule control; i.e. more highly educated and skilled workers in supervisory roles are more likely to have access. There is little support for a needs-driven provision; although workers with preschool children are also slightly more likely to have access, there are generally no significant differences between parents vs. non-parents or men vs. women, confirming what has been shown in previous studies (e.g. Swanberg et al. 2005;Ortega 2009).
There is evidence that generous national-level family policies can both "crowd in" and "crowd out" company-level provisions, depending on the type of policy. Work-facilitating policies "crowd in" workers' access to schedule control, confirming earlier studies (Chung 2014;Lyness et al. 2012;Den Dulk et al. 2013). Work-reducing policies, however, "crowd in" schedule control only to a certain degree and then "crowding out" effects can be seen, similar to what was found for women's employment patterns (Misra et al. 2011). Family policies were also associated with gaps in the access workers of different skill and education levels had to schedule control. The additional schedule control access found in countries with generous work-facilitating policies and about average level of work-reducing policies was predominantly due to the additional access gained by highskilled/educated workers. Thus, we can argue that "crowding in" of occupational policies seems to be targeted towards the group of workers that employers have a stake in recruiting/maintaining, with greater potential to contribute to company performance outcomesi.e. the high-skilled workers. Yet employers may not feel a need to provide schedule control to workers, both high-and low-skilled, when very long leave entitlements are available, "crowding-out" occupational-level policies.
There are some points that need further investigation. Some of the relationships found here were rather exploratory, especially the context variables included to explain the cross-national variation in the gaps in access to schedule control between different groups of workers. Thus the results may be sensitive towards any confounders excluded from the analysis and future studies should examine other potential important contextual factorssuch as unemployment rates, labour market institutions, or work and gender norms. Additionally, due to the limitation of the cross-sectional data used here, we cannot be certain about the causality of the relationships founde.g. countries where organisations have a greater skill demand, managers may not only introduce generous family policies at the company level but also demand that generous family policies are developed at the national level (see also Fleckenstein and Seeleib-Kaiser 2011). Longitudinal data on the use of flexible work arrangements in a cross-national comparative setting will be of benefit to allow such analysis to take place. Finally, the results may be sensitive to the period under investigation. Major austerity cuts were implemented in 2009-2010, and similar behaviour may have occurred at the company level, which may distort our results. Further analysis using data from different years (pre-, post-crisis) would be of benefit.

Family Policies and Access to Schedule Control 39
Despite these limitations, this study has shown some useful insights on the underexplored issue of the relationship between national-and company-level policies, how this relationship may vary depending on the actual policy type examined, and across different groups of workers in the labour market. Firstly, an important result from this paper is that we cannot assume an equal influence of institutions across different groups of workers (see also Chung and Mau 2014;Chung 2016). This paper has shown the usefulness of random-slopes crosslevel interaction models to explore this issue with cross-national comparative data. Secondly, different results are found for different types of family policies, leading us to rethink how we conceptualise family policy, its role and its impact on work-family outcomes (see also Misra et al. 2011). This study provides further evidence that we need to be cautious about exactly what we are measuring when we examine family policy indicators, especially when using single or composite indicators. Finally, the results from this paper show that despite some cross-national variance, generally, disadvantaged labour market groups are least likely have access to company-level family-friendly arrangements such as schedule control (see also Swanberg et al. 2005;Golden 2009). This is especially problematic if we consider that these workers may be those in most need of family-friendly measures due to lack of other resources to address the conflicting demands of work and family life. Policy makers should thus explore ways to tackle this unequal access to company-level provisions of familyfriendly policies to ensure that everyone can make use of such benefits. Generous family policies at the national level may help in ensuring access for some workers, yet, as shown in this paper, this may not necessarily guarantee access for the lower ends of the labour market. A strong right to request flexible working, with greater protective measures, may be better suited to tackle this issue.

Notes
1. There are countries where workers have a right to request flexible working, such as in the UK. However this is a much softer right compared to the Dutch right to adjust working hours (WAA), where it is more difficult for employers to refuse the request. The Dutch, however, have passed a law on flexible working which will allow workers a stronger right to request flexible workinge.g. working from home or schedule controlas of 2016. 2. Skill level is included as a continuous variable to enable a more simplistic random slopes modelwhere several categorical variables would have resulted in a larger number of models to be tested. 3. Detailed results provided upon request.