Flextime/Flexspace for All in the Organization? A Study of the Availability, Use, and Consequences of Flexible Work Arrangements for Low and High SES Employees in Nine European Countries

: This article investigates the relationship between employee socioeconomic status (SES) and the availability, use, and consequences for employees of flexible work arrangements (working from home and flexible starting and finishing times). Multi-level analyses based on the European Sustainable Workforce Survey (11,011 employees nested in 869 teams at 259 organizations in nine European countries) reveal a negative relationship between low SES employees and the availability of working from home. Lower-status employees also perceive working from home and flexible work times as less available to them and use these arrangements less than higher-status employees. Findings suggest similar outcomes of use for both groups. We found almost no differences between lower and higher SES employees in how using flexible work arrangements affected performance, commitment, and work–life conflict.


Introduction
In the past 30 years, advances in information and communication technology and changes in the relationship between work and family domains have led more organizations to adopt flexible work arrangements, such as working from home and flexible work hours (Eurofound and International Labour Office 2017;Den Dulk et al. 2013;Hammer et al. 2013;Kelly et al. 2008;Chung and van der Lippe 2020).Organizations offer employees flexible work arrangements (FWA) to enhance their performance, commitment, and work-life balance (Kelliher and de Menezes 2019;Clark et al. 2008; Van der Lippe and Lippényi 2019).The COVID-19 pandemic has accelerated the introduction of FWA, increasing the scale and extent of working from home dramatically (De Haas et al. 2020;Savić 2020;Aczel et al. 2021) and showing consequences of key distinctions in working from home for wellbeing (Yang et al. 2023).In the current debate on new forms of work and their flexibility, it has become clear that there are fundamental changes in work and work-life dynamics (Ewers and Kangmennaang 2023).The question is whether all employees are benefitting in the same way from FWA.The main focus in such research is generally on men and women (Dermott and Gatrell 2018; Thébaud and Pedulla 2022) and we learn that differences between men and women therein are dependent on different types of FWA and whether the focus is on availability or use of FWA.There has recently been a call in the HR literature to examine the causes and consequences of HR differentiation for employees and organizations (Jackson et al. 2014;Schmidt et al. 2018) and how to address employees' social class in recruiting, selection, and development practices.High-class employees tend to receive more opportunities and resources compared to low-class employees (Guerci et al. 2023), including access to FWA like teleworking (Eurofound 2023).This growing attention is doing justice to the growing socio-economic divide in many countries in the Western world (OECD 2017).Some authors look now specifically at the consequences for higher-and lower-educated women of using flexible work arrangements (Fuller and Hirsh 2019), perceptions of the availability of lower-educated fathers (Cook et al. 2021), and how gender intersects with occupational class when studying the relationship between FWA and the division of unpaid work among working parents (Chung and Booker 2022).We aim to contribute to this field by studying differences between low and high socioeconomic status (SES) employees and pose the following research questions.Are flexible work arrangements available to low and high SES workers to the same extent?Are low and high SES workers able to make use of these arrangements to the same extent and does such use contribute to better performance, commitment, and less work-life conflict for both groups?
This study contributes to previous research in the following ways.First, existing studies on organizational FWA have shown that there is often a gap between availability and use.Workers in the same company appear to vary in terms of their take-up of FWA and other 'available' work-life supports in ways not fully accounted for by differences in need (cf.Blair-Loy and Wharton 2002).Moreover, there is empirical support for the idea that FWA is more readily available to higher-educated, high-skilled, higher-SES workers than their lower-educated, low-skilled, and low SES counterparts (Cook et al. 2021;Eurofound 2023;Golden 2009;Riva et al. 2018).Unless we study the consequences, however, we cannot say whether inequalities in the distribution of FWA availability are in fact problematic.So, we argue that only studying the difference between availability and use is not enough; consequences should be taken into account as well.Although multiple studies show that flextime policies are beneficial for workers, studies focusing on flexspace policies are less clear about positive or negative outcomes (Van der Lippe and Lippényi 2018).In this study, we build on previous research by examining formal FWA availability as provided by the organization, employees' perceived availability of FWA, their actual use of FWA, and the consequences for them of such use.We thereby pay specific attention to flextime and flexspace.
Second, our focus on comparing low and high SES employees is driven by the assumption that inequality shapes how organizations operate-externally in the marketplace and internally in the workplace.Firms seek to maximize performance (Van Jaarsveld et al. 2009), which might imply that even if possible, FWA will not be made available to all employees.Some forms of FWA might be limited available to low SES due to the nature of their work, such as working from home as a blue-collar worker (Cook et al. 2021) but for flextime policies, this argument is less clear.For the types of FWA which are useful also for low SES employees, organizations may profit less if they limit such policies to higher SES employees (Schmidt et al. 2018).We therefore need to understand how inequality shapes the choices of organizations and even more so when the outcomes of their decisions affect employees and their families.
By studying FWA, we are also turning our attention to gender inequalities and socioeconomic status.Although many previous studies related national work-life policies to the intersection of SES and gender (Esping-Andersen 2009; Korpi et al. 2013;Mandel and Semyonov 2006) and there are ongoing public policy debates on how to support low-income workers and reduce poverty and income inequality in the process (Lambert 2009), research has largely ignored the role of the organization in this area.SES inequalities have increased and national work-life policies differ in the extent to which they improve opportunities for lower SES (female) employees compared with their higher SES counterparts (Cooke 2011;Korpi et al. 2013).Our study asks whether it is also true for organizational FWA that they improve opportunities for lower SES (female) employees compared with their higher SES counterparts.We argue that this inequality has a direct impact on organizational performance and functioning because it influences the behavior of male and female employees by affecting their interactions in the workplace and by shaping an organization's institutional environment.
Third, unlike most research in this field, we go beyond studying a single organization or a working population sample without reliable organizational information by using multilevel data on employees nested in organizations.Our data are drawn from a large-scale survey of 259 organizations, 869 teams, and 11,011 employees across multiple economic sectors in Bulgaria, Finland, Germany, Hungary, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom (Van der Lippe et al. 2016).This multi-level dataset enables us to study multiple theoretical explanations for an SES difference in the availability, use, and consequences of flexible work arrangements in organizations while taking the characteristics of the organization and the country into account.This means that we do not include self-employed workers or the unemployed in our study.
In general, SES relates to an individual's ability to produce and consume resources and is frequently measured in terms of placement along a continuum of several attributes (income, educational level, and occupational status).Our operationalization of SES refers to a widely used socio-economic index of occupational status which scales occupations by the average level of education and average earnings of job holders (ISEI) (Ganzeboom et al. 1992).We are aware that SES may also function as a proxy for job roles so that high SES jobs are more likely to involve the use of information and communication technology (ICT) than low SES jobs, which might make it harder in some low SES jobs to work from home (Eurofound and International Labour Office 2017).In studying the impact of FWA on low and high SES employees, we focus on three consequences: perceived performance, commitment and work-life conflict.Performance is the ultimate goal for firms, since better employee performance leads to better organizational performance.As commitment and work-life conflict are often a prerequisite for better performance in addition to better work-life balance and health of individual employees, we also take these consequences into account (Beauregard and Henry 2009).
The COVID-19 crisis has made the questions raised in this study even more pertinent, with major differences in how various SES groups have coped with the impact of the pandemic on their work lives (Aczel et al. 2021).In this study, we investigate two types of FWA at the organizational level, namely (1) flextime policy or arrangements that allow employees to determine the hours at which they start and stop working and (2) flexplace arrangements, where employees work part of the work week from home (Blair-Loy and Wharton 2002).We study these indicators for FWA separately to identify any differences between low and high SES employees or to determine whether the mechanisms apply to all forms of FWA in the same way (Allen et al. 2013).Finally, whether organizational FWA is available, whether employees are able to make use of it, and the possible outcomes of such use may depend on the country in which the employee and organization are located.While it is beyond the scope of this article to study differences between individual countries, we test whether our findings are robust overall, across countries.

The Importance of SES for the Availability, Use, and Consequences of Flexible Work Arrangements
Our theoretical approach focuses on analyzing the availability of FWA in the workplace, the take-up by employees, any discrepancy between low and high SES employees in that regard, and the consequences of these arrangements for employee performance, commitment, and work-life conflict.Although our aim is to understand the importance of SES specifically in terms of the use and consequences of flextime and flexplace policies, we are aware that gender is also an important factor.Where possible and necessary, we address gender and the intersection of SES and gender as well (Korpi et al. 2013).
In this paper, our approach is that we draw on the dominant theoretical explanations for SES differences in availability and use of FWA and test these empirically by addressing the differences in FWA availability, use, and consequences by SES while accounting for the mechanisms described in the literature.The dominant theoretical explanations are the business case argumentation, organizational stratification theory, and institutional theory for availability; for use, theoretical arguments are based on norms and support from the environment.We thereby show the explanatory power of the different mechanisms for understanding SES differences in FWA availability, use, and usefulness and show whether a net difference remains and how substantial this difference is.

Availability of Flexible Work Arrangements
Organizational stratification theory is useful for understanding why organizations allow employees to work flexible hours or away from the workplace because it explains how employers distribute 'opportunities' such as FWA.Through the mechanisms of formal human resource policies and more informal day-to-day practices, firm-level labor markets define and distribute opportunities among job holders (Doeringer and Piore 1985), including FWA.Given their limited resources, employers distribute these opportunities in the manner that is most profitable for the organization (Sweet et al. 2014).They do so in line with business case argumentation, which states that policies, such as FWA, should serve to attract talent and increase employee commitment, satisfaction, and performance.
In line with both business case argumentation and organizational stratification theory, an organization is more likely to offer FWA to higher SES employees because it perceives them as more valuable and expensive to lose than lower SES employees and will thus invest more in them (Schmidt et al. 2018).These 'insiders' on the labor market enjoy higher wages, better working conditions, and more job stability, with employers investing more because they expect higher returns from them than from the 'outsiders,' who are easier to replace (Osterman 1994).One possible unintended consequence is that this unequal distribution of benefits and arrangements will result in the disadvantaged group showing less commitment and poorer performance (Lewis 1997).Making FWA policies available is also dependent on the type of FWA.
Institutional theory adds to this explanation by deliberately taking the context of the organization into account.The theory argues that organizations respond to institutional pressure in the context in which they operate.Pressure to make FWA available to all or only to specific groups may be the result of social pressures, legislation, labor market conditions, or professional norms (Riva et al. 2018).Numerous studies show that high SES employees are more likely to have access to FWA (Golden 2009;Chung 2019;Riva et al. 2018), whereas low-wage, low-skilled, and lower-educated workers are least likely to have such access (Chung 2019;Golden 2009;Swanberg et al. 2005).Flexspace arrangements are difficult for jobs that involve less use of ICT and work that is not portable (i.e., connected to a certain space), such as bus drivers or construction workers.Flextime, on the other hand, is less dependent on the use of ICT or the portability of work.However, flexspace and flextime are not always easy to disentangle since spatial flexibility is also likely to result in flexibility over time because employees are less subject to direct supervision when working away from the office (Gajendran and Harrison 2007).In the test of the expectations, we will take the difference between the two into account.
We have so far only discussed whether employers make FWA policies available but availability also depends on the employees and whether they see such policies as available to them.Employee awareness of FWA is likely to be contingent on their job level (Ollier-Malaterre and Andrade 2016).Employees in higher level jobs have a larger skillset and will be better able to exercise agency (Kalleberg 2009).Moreover, higher SES employees have more autonomy and exercise control in their jobs, making it more likely that they perceive FWA policies as being more available to them (Riva et al. 2018).We therefore expect that low SES employees see organizational policies as less available.
The foregoing leads to hypothesis 1: (a) organizations are less likely to make FWA available to low SES employees than to high SES employees and (b) lower SES employees are less likely than higher SES employees to perceive FWA as available.

Use of Flexible Work Arrangements
Although FWA might be available to employees, not all of them will make use of these arrangements to the same extent.Again, we expect individual agencies to play a role, in the sense that employees who have fewer resources at their disposal will be less capable of using FWA.Having the skills needed to negotiate the use of FWA with managers and to investigate which organizational policies are available to them will influence take-up by employees and the associated consequences (Hobson 2014;Cook et al. 2021).Lower SES employees have fewer resources than higher SES employees, for example, a suitable workspace at home or the ability to negotiate the use of FWA with their managers.They are therefore less likely to make use of these policies.
Whether employees make use of FWA also depends on their environment and specifically on the norms of and the support provided by the workplace (Blair-Loy and Wharton 2002; Kossek et al. 2010).Behavior is embedded in a social context that shapes people's willingness and capacity to act (Pfeffer 1997).For most people working in organizations, the most potent and relevant context is that of the group with which they engage.If making use of FWA is the norm in the workplace because many other employees also do so and if such use is supported by the managers, then all employees, regardless of SES, will feel encouraged to do the same (Den Dulk et al. 2018).Having co-workers who work from home or who vary their starting and finishing times may help to create norms that support the take-up of FWA and encourage greater use of them.
Low SES employees often work in teams with other low SES employees and it may be more likely that the culture in such teams is not to use FWA.Such employees have less job autonomy, for example, and the thinking maybe that low SES employees will be less productive if they work from home.Managers may therefore be reluctant to let them take advantage of available FWA. Employees first need to show that they can be productive working away from the office.Higher SES employees, on the other hand, may already have considerable job autonomy and will be given more freedom to decide when and where they work, although they may fear that using available FWA will affect their careers.At the most general level, we expect that group-level factors will influence employee policy use, independent of individual-level factors.
In addition to norms in the workplace, family support and family norms influence whether employees make use of policies; if the employee's partner supports their making use of FWA and other friends and family members also make use of these arrangements, thereby showing that it is the norm, the employee is more likely to make use of flexible work hours or work from home.Working from home also requires having a suitable workspace and not all employees have a spare room where they can focus on work.Low SES employees in particular may lack this resource.All in all, this leads us to Hypothesis 2: lower SES employees are less likely than higher SES employees to make use of organizational FWA.

Consequences of Using FWA for Performance, Commitment, and Work-Life Conflict
According to the self-interest utility theory (Lind and Tyler 1988), FWA will increase performance and commitment and decrease work-life conflict if employees find such arrangements personally useful and perceive the benefits (Butts et al. 2013).Employees who see benefits in especially flextime are expected to experience less work-life conflict and stress and be more productive at work (Beauregard and Henry 2009;Butts et al. 2013).Flextime allows employees to align their working times to their personal needs and responsibilities outside work, which decreases work-life conflict.Moreover, the freedom to decide when to work allows them to work at times that work best, to be more productive and effective, which enables them to perform better.For flexspace, the expectation is less clear: on the one hand, working from home saves commuting time and allows employees to work undisturbed and to better concentrate at work.On the other hand, working from home can lead to blurring boundaries and can make it harder to disconnect from work.However, based on the social exchange theory, scholars also argue that employees who are offered more opportunities by the organization, in this case FWA, will reciprocate by working harder and performing better (Gouldner 1960;Kelliher and Anderson 2010).This follows the implicit rules of exchanging favors with the firm, with the expectation of a 'quid pro quo'.
This begs the question as to whether we would expect the consequences of FWA to be similar for lower and higher SES employees.We contend that both low and high SES employees are expected to reciprocate the 'favor' that their employer has done for them and perform better.With respect to work-life conflict, lower SES employees may even benefit more because they have less autonomy and more time restrictions at work (see also (Fuller and Hirsh 2019) on the larger motherhood penalty associated with working from home for higher educated women).Being allowed to work from home or set one's own hours relaxes the time restrictions at work and thus leads to less work-life conflict.Given all this, we refrain here from hypothesizing about whether the outcomes of use differ for lower and higher SES employees.Although we expect that using FWA will influence performance, commitment, and work-life conflict, any such effects will not be differentiated by employee SES.

Sample
To test our hypotheses, we used the European Sustainable Workforce Survey (ESWS), a multi-actor survey of organizations conducted in 2016 in Bulgaria, Finland, Germany, Hungary, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom (Van der Lippe et al. 2016).We used national business lists of organizations in six chosen industries-manufacturing, health care, higher education, transportation, financial services, and telecommunication-as our sampling frame and chose establishments belonging to one of these industries, dividing them into three size categories: 20-99 employees, 100-250 employees, and 250+ employees.We then randomly selected one organization from each sampling cell to contact.If an organization within a particular industry and size group refused to participate in the study, we used a matching strategy to select a new organization within the same industry and size category.Both private and public sector organizations were included in our sample.After the organization (often the HR director) agreed to participate, employees and their department managers were contacted at work and asked to participate in an online or paper-and-pencil questionnaire.In 2016, a total of 11,011 employees nested in 259 organizations participated in the survey.The participation rate among organizations varied from 5% to 20% across countries.Given how difficult it is to gain access to organizations, non-response is a significant problem in the majority of studies that sample organizations.Once an organization joined our research, the response rates were good.The within-organization response rate was 61% for employees and almost 98% for HR managers.

Missing Values
We addressed missing values in the aggregate analyses at team and organizational levels by means of listwise deletion (organizational level: 18 missing cases or 7%; department level: 179 missing cases or 20%).In the employee-level analysis, all respondents with missing values for one of the outcome variables were removed from the analytical sample (n = 750, 6.6%).Employee-level missing values for the explanatory variables were imputed using chained equations in Stata 15.1 (30 imputations).All variables covered in the analysis as well as education and income were included in the imputation procedure.Missing values for organizational and departmental control variables were replaced by the reference category or the sample mean, respectively, and a dummy variable indicating missing organizational or departmental information was included in the analysis (between 8% and 14% of cases).

Measures
To measure the availability of organizational policies, we asked the HR manager and team manager whether employees were allowed to work from home during normal work hours (0 = No and 1 = Yes) or to maintain flexible starting and finishing times (0 = No and 1 = Yes).To measure availability as perceived by the employee, we asked employees whether their organization allowed them to work from home during normal work hours (0 = No and 1 = Yes) and to determine their own starting and finishing times (0 = No and 1 = Yes).To measure the use of FWA, we asked employees how often they had worked from home during normal work hours in the past 12 months (0 = never to 6 = 4 or 5 days a week), with responses being recorded to a dummy (0 = No and 1 = Yes if the employee worked from home at all) and whether they had determined their own starting and finishing times in the past 12 months (0 = No and 1 = Yes).
For consequences of using FWA, we measured contextual performance, commitment, and work-life conflict.We measured contextual performance based on the contextual performance scale of the Individual Work Performance Survey (IWPS) battery, which was developed to produce comparable measures of self-evaluated performance across different types of jobs and has been shown to have satisfactory content and construct validity (Koopmans et al. 2014).Contextual performance is measured by five items on a 5-point Likert scale.Examples are 'I took on challenging new tasks when they were available' and 'I actively participated in meeting and/or consultations.'The Cronbach's alpha of the contextual performance battery is 0.80.We constructed individual-level performance by averaging the scores on the five items for all respondents who had provided a valid answer to at least three items.Higher values indicate better performance.To measure commitment, we slightly modified the scale from the value commitment battery of Angle and Perry (1981) and used the following four items: (1) I am willing to go above and beyond the call of duty to help my organization be successful, (2) I tell my friends that my organization is a great place to work, (3) I am proud to tell others that I am part of this organization, and (4) I really care about the future of this organization, with answer categories ranging from 1 (strongly disagree) to 5 (strongly agree).The reliability of the scale is 0.84.Answers to the four commitment items were averaged for all respondents who had provided at least three valid answers.Higher values indicate more commitment.To measure work-life conflict, we used part of the SWING scale developed by Geurts et al. (2005) with the following three items: how often does it happen that (1) you do not have the energy to engage in leisure activities with your family or friends because of your job? (2) you have to work so hard that you do not have time for any of your hobbies, and (3) your work obligations make it difficult for you to feel relaxed at home, with answer categories ranging from 1 (never) to 5 (always).The reliability of the scale is 0.86 and answers were averaged for all respondents who provided at least two valid answers.Higher values indicate more work-life conflict.See the Table S1 for the results of an exploratory factor analysis of all items of these scales.
To determine the socio-economic status of employees, we used the international socioeconomic index of occupational status (ISEI), as its concept of status encompasses both educational credentials and income and offers a continuous measurement of socio-economic status as opposed to occupational classifications such as the International Standard Classification of Occupations (ISCO), which are categorical in nature.Conceptually, the ISEI scales occupations by the average level of education and average earnings of job holders (Ganzeboom et al. 1992).We derived the ISEI by using the ISCO-08code, based on the question: 'What is your occupation?Please give a full description of your occupation, for example nurse at the intensive care, cashier at the bakery counter.'We recoded the occupation status scores from the ISCO-08 code using the ISEI transformation scale (Ganzeboom 2010).The ISEI scale runs from 10 to 90, with the lowest value representing occupations such as agricultural workers and the highest value referring to medical professionals.We refer to this measure as SES in our text and in the results.
At the organizational and team levels, employee SES was measured by asking the HR manager or team manager what percentage of employees in the organization/team are in skilled or high-skilled positions.Answers were recorded on a 10-point scale ranging from 1 (none at all) to 10 (all).

Control Variables
We controlled for a number of variables that the literature identifies as influencing the availability of policies, their use, and consequences for performance, commitment, and work-family conflict.For employees, these include gender (0 = male and 1 = female), number of work hours, tenure (i.e., number of years working for the organization), permanent contract, autonomy in work tasks, and age.To control for family circumstances, we checked for the presence of a partner (married or cohabiting) and minor children in the household and the age of the youngest child.To control for organizational characteristics relevant to the provision of FWA, we added the log indicating the size of the organization, whether it is a public sector organization, and the share of women in the organization.We also controlled for industry (manufacturing, health care, higher education, transportation, financial services, or telecommunication) and the country in which the organization is located.Descriptive statistics of all variables can be found in Table 1, including the number and percentage of imputed cases for each variable.

Analytical Strategy
We performed four series of explanatory analyses.In the first, we analyzed the relationship between employee SES on the one hand and availability of FWA as indicated by the HR and team managers on the other, using aggregated analysis at the organizational and team level with the proportion of highly skilled employees at each level as the main predictor.Organizations (data reported by the HR manager) and teams (data reported by the team manager) served as the unit of analysis in these models and we added controls at both levels.In the second and third series, we analyzed the relationship between employee SES, availability of FWA as perceived by the employee, and use of FWA, using logistic regression three-level (organization-team-employee) random intercept models.In the fourth series, the relation between FWA use and performance, commitment, and work-life conflict was interacted with by the SES of employees using three-level OLS regression models.Figure 1 shows the model to explain this for different SES employees.Models predicting availability, use, and consequences of FWA at the employee level were estimated with SES interacted by gender to explore whether SES and gender intersect in their effects.Because we found no significant gender differences in the effects of SES on FWA availability, use, or the consequences of FWA use, results are not reported by gender.Results pertaining to the hypotheses are presented graphically as average marginal effects; in addition, estimates for all coefficients from the full regression models are presented in Tables.

Hypothesis Testing
With regard to Hypothesis 1a, Figure 2 shows the outcomes for FWA policies as reported by the HR manager and team manager.The graphs show average marginal effects in % points of a one-category change (roughly 10%) in the proportion of skilled workers on the two FWA.The results partially confirm our expectation: managers in organizations

Hypothesis Testing
With regard to Hypothesis 1a, Figure 2 shows the outcomes for FWA policies as reported by the HR manager and team manager.The graphs show average marginal effects in % points of a one-category change (roughly 10%) in the proportion of skilled workers on the two FWA.The results partially confirm our expectation: managers in organizations and teams that have a higher proportion of employees in high-skilled positions are more likely to allow their employees to work from home.This result is robust to the inclusion of various relevant control variables at the organizational level.The models predicting whether HR or team managers offer their employees flexible work times also show they are more likely to offer such an arrangement if a larger proportion of their employees is highly skilled but after including the control variables, this effect is no longer statistically significant.Hypothesis 1a, which states that organizations are less likely to make FWA available to lower SES employees, can therefore only be confirmed for the policy on working from home.Estimates from the full models with all control variables are presented in Table 2.
( With regard to Hypothesis 1b, the employee-level results presented in Figure 3 (lefthand panel) show that higher socio-economic status is related to a higher probability of employees reporting that they are allowed to work from home and work at flexible times.While the effects become substantially smaller for both dependent variables after adding the control variables, they remain statistically significant in both cases.As Figure 3 shows, an increase on the ISEI scale of roughly 10 points is related to an increase of approximately 4% points in the probability of employees perceiving flexible work times and working from home as being available to them.As the ISEI scale runs from roughly 10 to 90 points, the predicted difference between lower and higher SES workers in perceived availability of FWA is substantial.The finding that lower SES employees perceive policies on working from home and flexible starting and finishing times as less available to them confirms Hypothesis 1b.Estimates from the full models with all control are presented in Table 3.
We now consider employee use of the different types of FWA as reported in Figure 3 (right-hand panel).The results show that lower SES employees make less use of working from home and flexible work times than higher SES employees, confirming Hypothesis 2. These findings mirror the availability findings, with a higher SES being associated with a significantly higher probability of an employee actually making use of flextime and flexspace arrangements.The effect remains significant in the full models (see Table 3) and is substantial and comparable in size to the effect on availability.Bivariate correlation analyses indeed show that perceived availability and use are strongly correlated at the employee level (r = 0.76 for flexible times and r = 0.63 for working from home), a finding that we discuss in our conclusion.Estimates from the full models with all control variables are presented in Table 3  With regard to the consequences of using FWA, we did not formulate an expectation but rather took an exploratory approach.The models predicting contextual performance, commitment, and work-life conflict by the use of FWA interacted by employee SES are shown in Table 4.For contextual performance, the results show that higher SES and use of both FWA are associated with higher reported contextual performance but for working from home, the significant negative interaction indicates that among high SES employees, the benefits of working from home are smaller (see the left-hand panel, Figure 4).This could be driven by a selection effect of cause, by which employees in high-status jobs take work home to combat excessive workloads.Organizational commitment is not affected by SES or the use of flexible times in the models with all control variables but, similar to contextual performance, working from home appears to be weakly associated with lower commitment among higher SES employees (see right-hand panel Figure 4).Work-life conflict is higher for higher SES workers and for those working from home but, since the interaction term is not significant, these appear to be independent effects.The two significant FWA use-SES interactions are illustrated in Figure 4, which shows that the effects of the working from home-SES interaction on contextual performance (left-hand panel) and commitment (right-hand panel), while being statistically significant, have no substantial influence on the outcomes.As noted above, most of the effects in the fully controlled model are statistically insignificant and those that do reach statistical significance are small effects.Since we did not expect to see any differences between low and high SES employees in terms of performance, commitment, and work-life conflict outcomes of using FWA, our findings, in fact, more or less confirm our expectations.Moreover, FWA use had no clear beneficial effects on the outcomes we studied.4).

Robustness Checks
All models were also estimated using a categorical specification of the proportion of skilled workers and the ISEI measure, respectively, to for any non-linear re-Results confirmed that a linear specification was a good fit for the data in all models.
We also re-estimated the employee-level models using educational level and income as alternative operationalizations of SES.The results of these calculations did not differ substantially from the ones obtained by using ISEI.
As we have nine countries at our disposal, we performed the jackknife procedure to investigate the influence of single countries on our results.They remained largely the same after removing single countries from the analyses.The outcomes of the additional analyses are available from the authors upon request.

Conclusions and Discussion
First of all, we can conclude that the SES of employees makes a significant difference to FWA availability in the workplace after accounting for the autonomy in tasks, characteristics of the workers related to relative bargaining power such as tenure, permanent position, occupational status, and work hours, and the actual need for work-family support indicated by workers' sex and the presence of young children in the household.This is most clearly the case for employees' perceived availability.Lower SES employees report less availability than higher SES employees of arrangements for working from home or flexible starting and finishing times.Lower SES employees may therefore feel that their organization is investing less in them than in higher SES employees.Close examination of the data provided by HR and team managers, however, shows that this is less the case than these employees perceive.While working from home is indeed less available to them, that is not true of flexible starting and finishing times, according to HR and team managers.Thus, as expected in our theoretical section, it is maybe more the case that employees in higher level jobs have a larger skillset and will be better able to exercise agency (Kalleberg 2009).Moreover, it is often argued that working from home may be less available to low SES employees due to the nature of their jobs, as low-skilled work is often less 'portable' than higher-skilled work.In our analysis, however, we included control variables that are clearly related to job content.Our findings therefore suggest, in line with Cook et al. (2021) or Eurofound (2023), that it is not only the nature of the job that plays a role in whether employees are given the option of working from home but also a sense of entitlement or the level of trust.This conclusion is additionally bolstered by the fact that our result remains intact when, instead of the socio-economic status scale (which contains occupation and thereby bears a stronger relation to work characteristics), education is used as an indicator of SES.
With respect to flexible work hours, there appears to be a gap between the availability of flextime as reported by managers and the perceived availability of flextime among lower SES employees.Overall, our findings confirm theoretical ideas about organizational stratification and business case argumentation only to a certain extent.With respect to flextime, they suggest that organizations are indeed willing to invest in FWA for low SES employees, thereby considering the long-term future of the firm.The problem is that low SES employees do not perceive these policies as being available to them.In other words, they do not appear to be obtaining the signal (Spence 1973) that this specific policy is also available to them.One implication is that organizations should be communicating more effectively about the FWA available to their employees.This has become only more relevant given the current debate about new forms of work and the significant role of communication therein.Having communication control as an employee appears to be important for good functioning at the workplace (Ter Hoeven and Van Zoonen 2023).As Kirby and Krone (2002) suggest, it would be valuable to examine how such work-life policies are enacted through discourse and interactions, since the policy intent will not come to fruition until the policies are put into practice.Going a step further, this gap could also explain why lower SES employees feel more frustration in organizations: their perception is that they have fewer opportunities than their higher SES counterparts (Payne 2007).It might be interesting to look at how employees are informed about policies and how the diffusion of this information takes place.Diffusion may be faster in teams consisting solely of high SES employees because they have more agency.
Our second conclusion, then, is that lower SES employees make less use than higher SES employees of the FWA available to them.While take-up is limited because these employees do not perceive the FWA as available to them, their agency and skills may also be factors.Lower SES employees may not have the skills necessary to use the policies available to them (Robeyns 2017).If that is so, HR managers and team managers must develop a different strategy: one that goes beyond communicating about the policies and also teaches employees how to make use of them.During the pandemic, many employees, including low SES employees, were obliged to work from home to slow the spread of the coronavirus (Bick et al. 2020;De Haas et al. 2020;Savić 2020).As a result, low SES employees gained experience in working from home and setting their own work hours.Future research could examine whether this learning experience resulted in more low SES employees working from home and working flexible hours.
The third conclusion is that we found almost no differences between lower and higher SES employees in how using FWA affected performance, commitment, and work-life conflict.Where we did find differences, our findings indicate that lower SES employees benefit more than higher SES employees from using FWA but the effects are too small to be considered significant.Our results show that a higher SES and working from home both lead to more work-life conflict.This may be because boundaries become blurred when working from home, making it harder for employees to disconnect from work.However, there also might be reversed causality in this case: people with more work-life conflict work more from home (Allen et al. 2013).There may also be a ceiling effect for higher SES employees: since they already have considerable job autonomy, working from home does not impact their work-life conflict.It may also be that the low SES employees constitute a selective group.If lower SES employees in fact do derive the same benefits from FWA as higher SES employees, employers have an extra reason to invest more in lower SES employees.Our results show that, generally speaking, use of FWA does not appear to have many consequences for employee performance, commitment, and work-family conflict.This confirms previous research; for example, in their meta-analyses, Allen et al. (2013) found only a small significant effect and more for availability than for use.The conditions under which policies are used appear to be important.
Due to research limitations, our results leave several questions unanswered, which lead to several recommendations for future research.First and foremost, it is likely that organizations with ample FWA are selective in their hiring.If so, then it is not that they are making policies available only to certain employees but rather that they are choosing certain employees to match specific policies.We have tried to take this into account by controlling for family background characteristics but it would even be better in future research to have longitudinal data over a longer time frame.Second, we are aware that there are possible reverse causality issues.As far as performance (or commitment) is concerned, we could imagine that employees who perform better are more able to bargain to be offered FWA.Conversely, we could think that employees with lower commitment would try to obtain more FWA to 'escape' the organization.Finally, on work-life conflict, employees with high level of work-family conflict could be more likely to ask for FWA.Although endogeneity issues are partly ruled out due to the great set of variables included in the analysis, in future research, it would be good to investigate these causality issues further.Third, as already said, the data for this study are pre-pandemic.It could be the case that we underestimate SES differences, given that SES differences have increased since then.But this is also for future research to investigate.Fourth, quantitative research might not be able to explain why lower SES employees perceive FWA as less available to them.We encourage other researchers to examine this question by performing qualitative research and conducting interviews with lower SES employees.Fifth, these data from the company come from their HR managers, so it could be possible that HR managers are unlikely to admit having policy gaps or treating employees unfairly.This could be studied more in-depth in future research.Finally, the current measurement items were authored by highly educated scholars and validated using highly educated sample populations (Agars and French 2016).The items can furthermore vary in meaning and may be open to interpretation.If we are serious about conducting more research on lower SES employees, then it is high time to take this group into account when constructing new measurement items.

Institutional Review Board Statement:
The study was conducted in accordance with the European Commission, and approved by the Faculty's Committee on Ethical Issues of Utrecht University (O&O 13.21178, 9 October 2013).The study is a non-interventional study and a survey.

Informed Consent Statement:
With respect to informed consent for data collection, the employees and manager were informed among others about the expected duration of the survey, a description of the procedures followed, why the survey is conducted, a statement that participation is voluntary, information about who is organizing and financing the research, a statement describing the procedures adopted for ensuring data protection/confidentiality/privacy including storage of personal data, and a reference to whom to contact for answers to pertinent questions.

Figure 1 .
Figure 1.Model to explain the availability, use, and consequences of flexible work arrangements for lower and higher SES employees.

Figure 1 .
Figure 1.Model to explain the availability, use, and consequences of flexible work arrangements for lower and higher SES employees.

Figure
FigurePredicted contextual performance and commitment by employee SES and working from home (based on the full model, see Table4).
K.B.; Visualization, K.B.; Supervision, T.v.d.L.; Project Administration, T.v.d.L.; Funding Acquisition, T.v.d.L.All authors have read and agreed to the published version of the manuscript.

Funding:
The research leading to these results received funding from the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013) / ERC Grant Agreement n. 340045.This study is also funded by the Dutch Research Council (NWO) and the Dutch Ministry of Education, Culture and Science (OCW) in the context of its 2017 Gravitation Program (grant number 024.003.025)Sustainable Cooperation-Roadmaps to Resilient Societies (SCOOP).

Table 1 .
Descriptive statistics of all variables used at the employee level.

Table 1 .
Cont.Source: ESWS wave 1 (2016), n = 10,281 after listwise deletion on dependent variables; a Descriptions based on available cases per variable before imputation.Na = Not applicable.

Table 2 .
Estimates from the logistic regression model FWA at the team level and organizational level by the proportion of higher-skilled employees at the team/organizational level (odds ratios).
Figure 2. Effect of a higher proportion of workers in highly skilled positions on probability of FWA availability; data aggregated at organizational and department level.Note: Control variables are Figure2.Effect a higher proportion of workers in highly skilled positions on probability of availability; data aggregated at organizational and department level.Note: Control variables are organizational size (log-transformed), organization in public vs. private sector, % female employees in the organization, sector (six categories), and country (nine categories).See Table2for estimates.

Table 3 .
.Exponentiated coefficients; t statistics in parentheses; variance components not shown.+ p < 0.10, * p < 0.05, ** p < 0.01.Estimates from a three-level random intercept logistic regression model predicting FWA availability and use at the employee level by employee, organization, and department level characteristics (odds ratios).
Figure3.Effect of employee SES on perceived availability and use of FWA.Note: Control variables are age, presence of children <18 in the household, age of the youngest child, presence of a partner in the household, autonomy in work, work hours, tenure in the current organization, employment contract (temp vs. perm), organizational size (log-transformed), organization in public vs. private sector, % female employees in the organization, sector (six categories), country (nine categories), availability of each work-family policy at the organizational and team level, and proportion of workers in skilled positions at the organizational and department level.See Table3.Figure3.Effect of employee SES on perceived availability and use of FWA.Note: Control variables are age, presence of children <18 in the household, age of the youngest child, presence of a partner in the household, autonomy in work, work hours, tenure in the current organization, employment contract (temp vs. perm), organizational size (log-transformed), organization in public vs. private sector, % female employees in the organization, (six categories), country (nine categories), availability of each work-family policy at the organizational and team level, and proportion of workers in skilled positions at the organizational and department level.See Table3.

Table 4 .
Estimates from a three-level random intercept regression model predicting consequences of FWA use by SES at the employee level.

Table 4 .
Cont.Standard errors in parentheses; variance components are not shown.+p< 0.10, * p < 0.05, ** p < 0.01.Predicted contextual performance and commitment by employee SES and working from home (based on the full model, see Table4).