How do differing degrees of working-time autonomy and overtime affect worker well-being? A multilevel approach using data from the German Socio-Economic Panel (SOEP)

Abstract

Flextime, or Flexitime, leads to greater worker satisfaction and well-being, but evidence shows increased working-time autonomy also leads to a greater risk of burnout and overload. The aim of this study is to estimate the effects of working-time arrangements with differing levels of autonomy on job and leisure satisfaction as well as subjective health. It uses working excessive hours as the threshold moderator. Based on German data, hypotheses were tested using a balanced sample of 4019 individuals spanning 16,076 person-years. Changing to or remaining in autonomous working-time arrangements had a positive effect on job satisfaction. Advancing to self-managed working time (trust-based working time) had a negative effect on satisfaction with leisure time, although remaining in self-managed working time related positively to general health. This study shows that measures are needed to govern working-time autonomy in order to prevent employees excessively extending their working hours.

Keywords

Flextime general health leisure time multilevel Socio-Economic Panel trust-based working time within-person worktime control

Introduction

Flexible working-time schedules are widespread in organisations. According to data from the European Working Conditions Survey 2015, 20% of employees across Europe had flexible working-time regulations and 6% of employees were free to choose the beginning and end of their working day (European Union, 2015), with Germany showing a high proportion of flextime arrangements. The lack of working-time monitoring by employers is known as trust-based working hours (Singe and Croucher, 2003), which allows for the highest degree of working-time autonomy.

Studies on the effects of different flexible working-time arrangements on employees’ attitudes and well-being are inconclusive. Flexible working time, allowing for a certain degree of self-determination, positively relates to work–life balance, job satisfaction and well-being (cf. Baltes et al., 1999; Kelliher and Anderson, 2010; McNall et al., 2010; Nijp et al., 2012). Regarding trust-based working hours, which often accompany management objectives, mixed effects have been reported. Positive effects on motivation, work–life balance and lower levels of strain seem to go hand-in-hand with risks, such as overload and more overtime (e.g. Böhm et al., 2004; Glißmann and Peters, 2001; Kelly et al., 2011; Moen et al., 2011; Wingen et al., 2004).

This study contributes to disentangling these controversial results by examining within-person effects of differing degrees of working-time autonomy in relation to job satisfaction, satisfaction with leisure time, and general well-being. With respect to previous studies and within the limits of available variables in the data set, we chose job satisfaction, satisfaction with leisure time, and general health as outcomes to provide further insights into the effects of working-time autonomy on job-related well-being as well as general well-being and work–life balance. Although the consequences of flextime have received considerable attention in research (for an overview see Baltes et al., 1999; Nijp et al., 2012), trust-based working-time arrangements as an extreme form of working-time autonomy have received less empirical consideration (for exceptions, see e.g. Janke et al., 2014; Kelly et al., 2011; Moen et al., 2011; Singe and Croucher, 2003).

Using data from the German Socio-Economic Panel (SOEP), we analysed the effects of switching to different forms of working-time arrangements between measurement waves using a multilevel framework. We employed a within-person design by nesting measurement waves within persons. We therefore investigated intra-individual effects of working-time arrangements instead of comparing groups of employees. In addition, we propose that intra-individual effects within different degrees of autonomy in working time will be contingent upon the average amount of overtime. This offers new insights for career guidance as well as for organisational politics regarding working-time arrangements.

Working-time autonomy

In their meta-analysis, Baltes et al. (1999) reported positive correlations between flexible working time and productivity, job satisfaction and satisfaction with working time, as well as a negative relation to absenteeism. However, here the conceptualisation of flextime is gliding time with fixed-core time that has limited options for carry-over of overtime. These kinds of flextime arrangements include increased working-time autonomy when compared with fixed working-time schedules, but they are still rather restricted. In a more recent review by Nijp et al. (2012), flextime was operationalised by the free choice of start and end points of working time, which may include models with fixed core time but also more flexible models. In a summary of 53 studies, positive effects emerged with regard to better work–family balance, health, well-being and job satisfaction (Nijp et al., 2012). Accordingly, McNall et al. (2010) reported positive effects of flextime schedules on work-to-family enrichment.

However, if working-time autonomy is particularly high, as in the case of self-managed or trust-based working time, the work environment can also be challenging for employees. The concept of trust-based working time often also includes the expectation that employees should take on an entrepreneurial attitude towards their work, such as self-directed behaviour and increased responsibility for budgets and revenue. This is therefore an arrangement that entails more than the mere deregulation of working hours (Böhm et al., 2004). In this respect, the concept is similar to a results-only work environment (ROWE), as it also emphasises that employees’ performance should be measured by outcomes rather than by the amount of working time (Moen et al., 2011). In addition, there is sometimes an expectation that employees will prioritise the employer’s needs over their own (Pedersen and Jeppesen, 2012). Taken together, these challenges can turn into stressors and might result in self-harming behaviour, such as extending one’s working time beyond a reasonable level (Krause et al., 2015). Considering that different working-time arrangements offer different levels of autonomy, specifically working-time autonomy, the effects of autonomy will be discussed against the background of the job demands-resources (JD-R) model (Bakker and Demerouti, 2007, 2014; Demerouti et al., 2001) and action regulation theory (Frese and Zapf, 1994; Oesterreich, 1981; Volpert, 1987).

The JD-R model considers autonomy as a resource that has a positive effect on work engagement and acts as a buffer for negative effects of work demands (Schaufeli and Taris, 2014). Following this logic, working-time autonomy should have similar effects, and thus working-time arrangements, such as flextime or self-managed working time, should be beneficial for the well-being of employees. However, in highly autonomous working-time arrangements, such as self-managed working time, autonomy and specifically working-time autonomy is likely to be accompanied by increased demands, including complexity and responsibility. As mentioned earlier, an increased amount of self-directed behaviour is expected from employees (Böhm et al., 2004; Moen et al., 2011). Furthermore, complexity and responsibility are considered demands rather than resources (Schaufeli and Taris, 2014). Action regulation theory uses a concept similar to resources, namely ‘regulation resources’, that can be defined as ways to cope with the regulation requirements (i.e. demands of the task and the work environment), which mainly comprise ‘control’ or ‘autonomy’ over tasks (Frese and Zapf, 1994). This theory focuses on action or acting as the basis for understanding work by means of action sequences and goals (Volpert, 1987). Although autonomy is supposed to be a regulation resource, Volpert (1987) suggested that making use of autonomy is only possible if workers receive the necessary training to do so. This means that workers must be able to deal with the increased amount of complexity and responsibility they experience when facing an increase in autonomy. This is also supported by Roberts and Foti (1998), who found that workers with high levels of self-leadership had higher job satisfaction in highly autonomous work environments than workers with low levels of self-leadership. In addition, complexity is considered a regulation requirement (Frese and Zapf, 1994). In conclusion, autonomy can be a resource, but demands that can be difficult to cope with might also accompany such a resource. Theoretical models have not specifically included working-time autonomy, but they support the assumption that differentiation between working-time autonomy as a resource and autonomy-related demands, such as complexity and responsibility, is necessary. In highly autonomous working-time arrangements it is plausible to assume a high level of responsibility and task complexity in addition to a high level of autonomy. Therefore, a potential overload may be part of these arrangements.

We will distinguish between three forms of working-time arrangements according to the degree of autonomy for the self-management of working time they offer for employees. Fixed schedules are defined by clear start and end points of daily working time. Flexible schedules are defined by variable daily start and end times with some control over schedules by employees. Most often, employees are required to document their schedules in so-called time account systems. Self-managed working time is, unlike flexible schedules, completely self-scheduled with either no documentation or voluntary documentation. We assume the lowest level of working-time autonomy for the first option, and the highest level of working-time autonomy for the third option. To date, most studies have employed a correlational design, reflecting a between-person perspective. This perspective provides an insight into the differences between employees with different working-time arrangements. A within-person design, especially the control of changing arrangements, allows us to analyse effects within an employment career, and thus provides stronger evidence for the causality of proposed effects. The within-person approach actually controls for between-person differences, and thus we are able to get much closer to the ‘net-effect’ of a working-time arrangement. We believe that the concurrent analyses of within- as well as between-person effects is an asset of this study in terms of contrasting the two perspectives.

We will first look at lagged effects of being in one of the described time account systems on satisfaction with work and leisure as well as subjective health measured 2 years later. In a second step, we include information as to whether participants have changed the time account system in the last 2 years. Fixed working time is the reference category for the dummy variables. Changing the time account system comes along with gains or losses of working time autonomy. Such dynamics in the availability of resources are discussed within the conservation of resources (COR) theory framework. The general idea of COR is that ‘… people strive to retain, protect, and build resources and that what is threatening to them is the potential or actual loss of these valued resources’ (Hobfoll, 1989: 516). In other words, resource gains should relate to increased well-being. The within-person design relates the current working-time scheme (i.e. degree of working-time autonomy) to the outcomes in comparison with the individual average level of these outcomes across all waves. The proximity of changing of a working-time scheme is likely to influence the strength of effects. Therefore, we will not only relate the current working-time status to our outcome variables, but also look at changes between two measurement waves. Overall, it is possible to regard working-time autonomy as a resource that seems to have desirable effects on job satisfaction and well-being, leading to the following hypotheses (Hs):

H1: (i) Self-managed working-time arrangements or (ii) flextime with time accounts (in comparison to fixed schedules) positively relate to (a) job satisfaction, (b) satisfaction with leisure time and (c) general health.

H2: (i) Changing to self-managed working time arrangements or (ii) flextime with time accounts positively relates to (a) job satisfaction, (b) satisfaction with leisure time and (c) general health.

Overtime as a boundary condition

Flexible working-time arrangements allow varying degrees of self-managed scheduling of working time, and thus must balance various interests. This autonomy is associated with increased demands in self-regulation and planning and can be related to overload (Böhm et al., 2004; Glißmann and Peters, 2001; Wingen et al., 2004). Beckmann and Cornelissen (2014) reviewed data from the SOEP and reported that the difference between contractual and actual working time is especially great for employees in self-managed working-time arrangements. This also increases the risk of unpaid overtime (Lott and Chung, 2016; Matta, 2015). This expansion of worktime under working-time autonomy may be rooted in either higher work engagement (Beckmann and Cornelissen, 2014) or self-depletion due to the lack of control systems (Matta, 2015).

In trust-based working-time arrangements, the employer no longer controls working-time variability, which increases the expansion of working time and the blurring of boundaries between work and non-work domains (Bauer et al., 2004; Wingen et al., 2004). An important distinction seems to be whether worktime flexibility is an external demand or whether it offers opportunities to fulfil flexibility needs. Balancing organisational flexibility demands with employees’ needs is important for reducing risks for employees, such as overload and working overtime, and for increasing job satisfaction (Singe and Croucher, 2003). Costa et al. (2006) found that variability in working time demanded by employers was negatively related to well-being, whereas individual flexibility (i.e. working-time autonomy) had a positive effect on well-being. Working overtime may well distort the effects of flexibility needs fulfilled by employees. As reported in a systematic review by Bannai and Tamakoshi (2014), long working hours have significant adverse effects on a variety of health outcomes, including circulatory disease, psychological disorders and health-related behaviour. Therefore we propose that working-time autonomy as a resource may lose its positive potential within the constraint of overtime work.

In some studies within the framework of the JD-R model, overtime is considered a challenge demand and thus increases work engagement as well as work strain because it is stimulating and requires the depletion of energy (Van den Broeck et al., 2011). However, some studies only assume negative effects on well-being (Bakker et al., 2005; Boyd et al., 2011; Rodriguez-Muñoz et al., 2012). If complexity and responsibility act as demands in highly autonomous working-time arrangements, such as self-managed working time, overtime may increase their disadvantageous effects on well-being and satisfaction through increasing work strain. For all main effects hypotheses, we assume that working overtime moderates these relations by reducing the positive effects of flexible working-time arrangements on the outcomes:

H3: Working overtime moderates the positive effects of (i) self-managed working-time arrangements or (ii) flextime with time accounts on (a) job satisfaction, (b) satisfaction with leisure time and (c) general health. Under the condition of high levels of overtime, positive effects will diminish.

Methods

Sample

We make use of a large panel data set, the SOEP, which is a longitudinal panel survey of private households in Germany administered annually since 1984 that currently encompasses approximately 12,000 households with 21,000 persons. Between 2003 and 2011, questions regarding working-time arrangements were included in the SOEP every 2 years. According to Dormann and Zapf (2002), a 2-year lagged effect seems reasonable. They provide evidence for such exposition times of complex working conditions (in their case social stressors) to take full effect on well-being. Hence, for our analyses we used data from 2003, 2005, 2007, 2009 and 2011. We decided to use a balanced sample (i.e. a sample with data on all measurement waves; Andreß et al., 2013), as this reduced the sample size only marginally. The advantage of a balanced sample is that it reduces the noise in the data. In the within-person models used here, we want to measure effects of working-time arrangements on the well-being of individuals. The results should be more reliable if we have information for every individual at every wave. We excluded individuals who were self-employed or unemployed in any wave as well as those who were older than 65 years in 2011 (the official age of retirement in Germany).

These criteria resulted in a balanced sample of 4019 individuals, of whom 53.9% were men and 46.1% were women. The average age of respondents in 2003 was 42 years (SD = 8.63, min. = 17, max. = 57). As the operationalisation of working-time arrangement responses from one wave to the next were used, the analyses refer to dependent variables measured in the waves from 2005 to 2011. For the four remaining waves, this results in 16,076 person-years.

Measures

Working-time arrangements

Respondents were asked ‘Which of the following working-time arrangements best describes your employment situation?’. They could respond with one of the following response options: (a) ‘fixed beginning and fixed end of daily working time’; (b) ‘variable daily working times, appointed by the employer’; (c) ‘no formal scheme, I manage working time myself’; and (d) ‘gliding time with time account with a certain degree of autonomy regarding daily working time’. These schemes imply varying degrees of working-time autonomy. The highest autonomy is likely achieved under (c), and can be labelled as a general form of trust-based working time (Böhm et al., 2004; Haipeter et al., 2002; Hoff and Weidinger, 1999).

As we were also interested in effects associated with moving from or remaining in certain working-time regimes with varying degrees of working-time autonomy, we used dummy coding and merged the first two options, (a) and (b), into one category, which we labelled ‘fixed working time’ (‘fixed’), whereas option (d) was labelled ‘flextime with time account’ (‘time account’), and option (c) ‘self-managed working time’ (‘self-managed’). Table 1 displays the distribution of these working-time arrangements across time points. Over the years the proportion of fixed arrangements decreases slightly, whereas ‘time account’ and ‘self-managed’ increase in the overall study population by approximately 3%. Figure 1 depicts mobility between these working-time schemes. The bars shown in Figure 1 equal the dummy variables for subsequent multilevel analyses. Remaining in a fixed scheme was used as the reference category for the analyses. The dummy variables only consider changing to and remaining in the different arrangements. As far as the change is concerned, we did not differentiate the arrangement the person was in before. A person who changed from ‘fixed’ to ‘self-managed’ would thus be in the same category as a person who changed from ‘time account’ to ‘self-managed’.

Table 1.

Absolute and proportional amount of employees in different working-time arrangements across waves.

Working-time arrangements	2003	2005	2007	2009	2011	Total
Fixed	2,576	2,471	2,436	2,399	2,346	12,228
Fixed	64.10%	61.48%	60.61%	59.69%	58.37%	60.85%
Time accounts	927	1,003	1,013	1,030	1,057	5,030
Time accounts	23.07%	24.96%	25.21%	25.63%	26.30%	25.03%
Self-managed	480	502	514	528	567	2,591
Self-managed	11.94%	12.49%	12.79%	13.14%	14.11%	12.89%
Missing	36	43	56	62	49	246
Missing	0.90%	1.07%	1.39%	1.54%	1.22%	1.22%
Total	4,019	4,019	4,019	4,019	4,019	20,095

Figure 1.

Mobility between working-time arrangements 2003–2011.

We assume that changing from a more flexible to a less flexible working-time arrangement (downward change) does not have the same effects as changing from a less flexible to a more flexible arrangement (upward change). Therefore, we deemed it necessary to control for downward and upward changes. To incorporate this idea, we calculated a second set of models with two additional dummy variables that controlled for ‘upward change’ and ‘downward change’. As these control variables did not change the results, we did not include them in the final models.

Job satisfaction

Job satisfaction was measured by the item ‘How satisfied are you with your job?’ and rated on an 11-point Likert scale ranging from 0 (totally unhappy) to 10 (totally happy).

Satisfaction with leisure time

Satisfaction with leisure time was measured by the item ‘How satisfied are you with your leisure time?’ using an 11-point Likert scale ranging from 0 (totally unhappy) to 10 (totally happy).

General health

Subjective general well-being was assessed by asking the question ‘How would you rate your current health status?’. Answers were given on a 5-point Likert scale with 1 indicating ‘very good’ and 5 indicating ‘bad’. To be able to interpret the results using the positive connotation of ‘general health’, the variable was recoded in such a way that higher values reflected better subjective health. Single-item measures for satisfaction and general health are widely used internationally in household panel surveys and have been determined to be acceptably reliable and valid (Diener et al., 1999).

Control variables

On the person level, we controlled for gender (0 = male, 1 = female) and age (in the 2003 wave). We included weekly contractual working hours, the difference between contractual and actual average weekly working hours (i.e. overtime), net income, and years of schooling as time varying control variables because these variables are known to have an impact on satisfaction as well as general health (Boes and Winkelmann, 2010; Brunello et al., 2016; Powdthavee et al., 2015; Salinas-Jiménez et al., 2011). Table 2 presents correlations among study variables and descriptive statistics. Correlations among study variables are mostly of weak to moderate effect size. Net income and years of education positively relate to working-time autonomy as well as overtime. Employees working in time account or self-managed working-time systems show a higher risk of working overtime. Overall, the bivariate correlations provide no indication for multicollinearity.

Table 2.

Within-correlations of study variables.

	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1 Gender: female	0.47	0.50
2 Age (years) 2003	40.41	8.58	.02***
3 Satisfaction with work^a	6.86	2.11	−.02**	−.01*
4 Satisfaction withleisure^a	6.61	2.05	−.00	.02**	.26***
5 Subjective health^b	2.93	0.96	.01	.02**	−.00	−.02***
6 Years of education	12.81	2.71	−.01	.07***	.06***	−.03***	.10***
7 Actual working hours	39.30	11.18	−.46***	.00	−.02*	−.17***	.03**	.18***
8 Contractual workinghours	35.26	8.30	−.46***	−.05***	−.03**	−.08***	.04***	.06***	.86***
9 Net income	1,794.80	1,092.73	−.42***	.17***	.10***	−.06***	.07***	.43***	.55***	.46***
10 Time account^c	0.13	0.33	−.01*	.00	.02***	.01*	.22***	.18***	.07***	.11***	.17***
11 Self-managed^c	0.06	0.24	−.05***	.03***	.05***	−.02***	.16***	.20***	.14***	−.01	.26***	−.10***
12 Change to self-managed^d	0.05	0.22	−.03***	.01	.03**	−.03***	.00	.08***	.05***	−.01	.07***	−.14***	.60***
13 Change to timeaccount^d	0.06	0.24	−.03***	−.02	.03**	.01	.01	.04***	.03***	.04***	.05***	.44***	−.10***	−.06***
14 Change to fixed^d	0.06	0.25	−.00	.01	.01	−.03**	.01	.01	.00	−.00	−.01	−.16***	−.10***	−.06***	−.07***
15 Overtime	3.91	5.50	−.17***	.02**	−.03**	−.18***	−.01	.20***	.63***	.13***	.34***	−.02	.25***	.11***	.02	.02*

p < .05; **p < .01; ***p < .001.

Scale 0 ‘not at all satisfied’ to 10 ‘completely satisfied’;

scale 1 ‘bad’ to 5 ‘very good’;

dummy variables: 1 ‘Working time arrangement at current wave’, 0 ‘Rest’, reference category: ‘Fixed’;

dummy variables: 1 ‘Change from time 1 to time 2’, 0 ‘Rest’.

Results

In order to estimate the effects of changing of or remaining in a certain working-time arrangement on multiple dependent variables, we used multilevel modelling. Time-invariant variables (age and gender) were located at level 2 (person level), whereas variables that differed over measurement waves (contractual working hours, overtime net income and years of education) as well as the dummies for changing of and remaining in working-time arrangements constitute level 1 of the models (Raudenbush and Bryk, 2002). All level 1 variables (except for the dummies) were entered group-mean centred, and all level 2 variables were centred by the grand mean. Random intercept, fixed-slope models were used. In a first step, we calculated the intraclass correlation coefficients (ICCs) for the dependent variables in null models (satisfaction with work: 0.45, satisfaction with leisure time: 0.51, subjective health: 0.50). Next we tested H1 (Models 1–3) for the main effects of being in certain working-time arrangements, and H2 (Models 4–6, Table 3) to test the effects of changing working-time arrangements within a period of 2 years. The moderation hypothesis (H3) was tested by additionally including interaction terms between dummies and overtime (see Models 7–8, Table 4).

Table 3.

Multilevel models: Impact of working-time arrangement on several well-being indicators, including control variables for changes of working-time arrangements.

	(1)		(2)		(3)		(4)		(5)		(6)
	Satisfaction with work^f		Satisfaction with leisure^f		Subjective health^g		Satisfaction with work^f		Satisfaction with leisure^f		Subjective health^g
Level 1	B	SE	B	SE	B	SE	B	SE	B	SE	B	SE
Contractual working hours^a,h	.03^*	.01	−.10^***	.01	−.00	.01	.03^*	.01	−.10^***	.01	−.00	.01
Overtime^a,h	−.04^**	.01	−.13^***	.01	−.01	.00	−.04^**	.01	−.13^***	.01	−.01	.00
Net income^a,h	.04^*	.01	.01	.01	−.01	.01	.04^*	.01	.01	.01	−.01	.01
Years of education^a,h	−.01	.02	.00	.02	−.00	.01	−.01	.02	.00	.02	−.00	.01
Time account^b	.14^**	.04	.09^*	.04	.01	.02	.13^**	.05	.07	.05	.01	.02
Self-managed^b	.20^***	.06	−.15^**	.06	.03	.02	.28^***	.08	−.12	.08	.10^**	.03
Change to fixed^c							.15^*	.06	−.04	.06	.01	.02
Change to time account^c							.10	.06	.02	.06	.01	.03
Change to self-managed^c							−.07	.09	−.07	.09	−.11^**	.04
Level 2
Age 2003^d	−.01^***	.00	.00	.00	−.02^***	.00	−.01^***	.00	.00	.00	−.02^***	.00
Gender^e	.03	.05	−.06	.05	−.03	.02	.03	.05	−.06	.05	−.03	.02
Intercept	6.88^***	.04	6.64^***	.04	3.50^***	.02	6.85^***	.04	6.64^***	.04	3.50^***	.02
N	14484		14527		14546		14484		14527		14546
Var-L1	1.66		2.11		0.31		1.66		2.11		0.31
Var-R	1.99		1.97		0.33		1.99		1.97		0.33
ICC-L1	0.45		0.52		0.48		0.45		0.52		0.48
−2LL	−28267.54		−28632.03		−15472.55		−28262.74		−28631.35		−15467.65
df	8.0		8.0		8.0		11.0		11.0		11.0
AIC	56557.08		57286.06		30967.10		56553.49		5729.70		30963.30
BIC	5664.47		57369.49		3105.53		56659.62		57396.87		31069.49

p < .05; ^**p < .01; ^***p < .001.

Variables centred around person mean (Level 1).

dummy variables: 1 ‘Working-time arrangement at current wave’, 0 ‘Rest’.

dummy variables: 1 ‘Change from time 1 to time 2’, 0 ‘Rest’, reference category: ‘Fixed’.

variable centred around grand mean (Level 2).

dummy variable: 1 ‘Woman’, 0 ‘Man’.

scale 0 ‘not at all satisfied’ to 10 ‘completely satisfied’.

scale 1 ‘bad’ to 5 ‘very good’.

variables are z-standardised.

AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion.

Table 4.

Multilevel models: Impact of working-time arrangement on several well-being indicators, including control variables for changes of working-time arrangements and interactions of working-time arrangements and overtime.

	(7)		(8)		(9)
	Satisfaction with work^f		Satisfaction with leisure^f		Subjective health^g
Level 1	B	SE	B	SE	B	SE
Contractual working hours^a,h	.03^*	.01	−.09^***	.01	−.00	.01
Overtime^a,h	−.04^**	.01	−.10^***	.01	−.00	.01
Net income^a,h	.04^*	.01	.01	.01	−.01	.01
Years of education^a,h	−.01	.02	.01	.02	−.00	.01
Time account^b	.13^**	.05	.07	.05	.01	.02
Self-managed^b	.28^***	.08	−.10	.08	.10^**	.03
Change to fixed^c	.15^*	.06	−.04	.06	.01	.02
Change to time account^c	.10	.06	.02	.06	.00	.03
Change to self-managed^c	−.07	.09	−.08	.09	−.11^**	.04
Overtime* time account	.03	.03	−.07	.03	−.02	.01
Overtime* self-managed	−.01	.04	−.09^*	.04	−.02	.01
Level 2
Age 2003^d	−.01^***	.00	.00	.00	−.02^***	.00
Gender^e	.03	.05	−.06	.05	−.03	.02
_cons	6.85^***	.04	6.64^***	.04	3.50^***	.02
N	14484		14527		14546
Var-L1	1.66		2.11		0.31
Var-R	1.99		1.97		0.33
ICC-L1	0.45		0.52		0.48
−2LL	−28262.12		−28627.06		−15466.43
df	13.0		13.0		13.0
AIC	56556.24		57286.12		30964.87
BIC	56677.53		57407.46		31086.23

p < .05; ^**p < .01; ^***p < .001.

variables centred around person mean (Level 1).

dummy variables: 1 ‘Working time arrangement at current wave’, 0 ‘Rest’.

dummy variables: 1 ‘Change from time 1 to time 2’, 0 ‘Rest’, reference category: ‘Fixed’.

variable centred around grand mean (Level 2).

dummy variable: 1 ‘Woman’, 0 ‘Man’.

scale 0 ‘not at all satisfied’ to 10 ‘completely satisfied’.

scale 1 ‘bad’ to 5 ‘very good’.

variables are z-standardised.

AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion.

Effects of status of working-time schedule

In our first set of hypotheses, we proposed that being in a self-managed working-time or flextime arrangement (when compared with fixed schedules) would positively relate to job satisfaction and leisure time as well as general health. As can be seen from the estimates of the main effects in Model 1, self-managed working time is positively related to job satisfaction (ϒ = 0.20, p < .001); likewise, time accounts showed a significant positive relation with job satisfaction (ϒ = 0.14, p = .001), lending full support to H1a. Examining Model 2 in Table 2 shows that time account was positively related to satisfaction with leisure time (ϒ = 0.09, p = .037), whereas self-managed was negatively related to leisure time satisfaction (ϒ = −0.15, p = .009). These results support H1b (ii) but do not confirm H1b (i). For general health, a positive relation with self-managed working time was observed, but only in the models that controlled for change of working-time arrangements (ϒ = 0.10, p = .002). Hence, H1c (i) can be partly confirmed, whereas the results do not support H1c (ii), which relates to flextime with time accounts.

Main effects of changing working-time arrangements

In a second set of hypotheses, we proposed that changing to self-managed or flextime work arrangements (when compared with staying in fixed models) is positively related to job and leisure time satisfaction as well as general health. For job satisfaction, only the change to a fixed schedule showed a significant positive effect, which was not predicted. Thus, H2a must be rejected. Changing to the time account system within 2 years had no significant relation to leisure time satisfaction, lending no support to H2b. Changing to self-managed showed a significant negative effect on subjective health (ϒ = −0.11, p = .002), which is contrary to the predicted relation, and thus leads to the rejection of H2c.

Moderation by working overtime

Finally, we argued that working overtime would buffer or even reverse the proposed positive effects of working-time schemes with greater degrees of working-time autonomy. Overtime was found to be significantly negatively related to job satisfaction (ϒ = −0.04, p = .008) as well as satisfaction with leisure time (ϒ = −0.10, p < .001), but showed no significant relation to general health. Significant interaction effects of working-time arrangements with overtime were found for the prediction of satisfaction with leisure time. Simple slope analyses showed that under the condition of high overtime self-managed time account showed to have a significant negative effect on satisfaction with leisure time (reference category: ‘remaining in fixed’). The decline in satisfaction with leisure time was significantly steeper for ‘self-managed’ (b = −0.21, SE_b = 0.03, p < .001) than for the reference category ‘remaining in fixed’ (b = −0.12, SE_b = 0.01, p < .001). Thus, hypothesis H3b (i) was supported, whereas other combinations proposed under H3 were not.

Discussion

The aim of this study was to test for effects of different degrees of autonomy within working-time arrangements on job satisfaction, satisfaction with leisure time, and general health. In addition, working overtime was predicted to play a moderating role in these associations.

Our results show that highly flexible working-time arrangements, such as self-managed working time, positively relate to job satisfaction, but relate negatively to satisfaction with the availability of leisure time. Employees in self-managed working-time arrangements are less satisfied with their leisure time compared with employees with fixed schedules. This effect might be due to fragmentation of working time in the context of higher working autonomy, which may result in a permanent stand-by mode and reduction of recovery experiences during non-work times (Krause et al., 2015). The ongoing dissolution of working time and working place accompanied by management by objectives, which often emphasises goal attainment over work processes, may likely lead to a further intensification as well as expansion of work (Bauer et al., 2004; Wingen et al., 2004). Hence, it is important to sustain positive effects of increased working-time autonomy, allowing for a better life-domain balance, while also taking measures against negative effects on health and well-being.

Working overtime could be established as a boundary condition for effects of working-time arrangements on satisfaction with leisure time, but not for job satisfaction or general health. In self-managed working time, the effect of overtime on satisfaction with leisure time is much stronger than in fixed working time. This means that individuals working overtime experience a stronger decline in their satisfaction with leisure time when they schedule their work themselves than when the schedule is set by the employer.

Concerning the effects of changing the working-time arrangement within a period of 2 years, a negative effect of changing to self-managed schemes on subjective health was observed. As self-managed time accounts demand self-regulatory processes, especially when employees are used to other forms of time autonomy, this might become a stressor. The effects on general health should, however, be interpreted with caution, as the effects are very small and could be attributed in part to a change of working conditions or a change of employer.

Given the within-person design, it is important to consider that working overtime was centred by the person mean. Thus, these effects occur according to individual standards, rather than comparing different employees with differing degrees of overtime work. We may conclude that absolute working time in addition to variation in individual employment histories seems predictive of satisfaction and well-being.

We contribute to the research on the effects of working time as an important environmental factor in employment in at least two ways. Working-time autonomy should generally be promoted. However, there is also a risk of work expansion in the context of high working-time autonomy. This is even more crucial in the case of increased overtime. Under this condition, positive effects of high working-time autonomy cannot be expected. Working-time arrangements with a high level of autonomy cannot always be described as beneficial. When the accompanying working conditions, such as management by objectives and overtime, promote overload, these working-time arrangements could cause problems related to decreased well-being and work–life balance. This adds to previous findings that suggest a critical attitude towards arrangements such as trust-based working time (Böhm et al., 2014; Glißmann and Peters, 2001; Matta, 2015). Unlike other studies, we focused on transitions in working-time arrangements and studied the changing of and remaining in working-time arrangements instead of simply studying the changes (cf. Matta, 2015).

In theoretical terms, we could establish working-time autonomy as a resource when it comes to job satisfaction. We conclude that within the framework of the JD-R model and the action regulation theory, not only autonomy but also specifically working-time autonomy should be considered as motivational factors. Nevertheless, we also found negative effects of highly autonomous working-time arrangements on satisfaction with leisure. Hence, these arrangements seem to include demands that undermine the positive effects of working-time autonomy. Demands caused by working-time autonomy might relate to management by objectives (e.g. unrealistic goals) or an increased need of self-management. Such boundary conditions for the positive effects of resources have the potential for theoretical refinements of resource theories and require further investigation. When working in flextime arrangements, overtime seems to aggravate negative effects on satisfaction with leisure. This shows that, within the framework of the JD-R model, overtime is a hindrance instead of a challenge demand. To phrase it according to the COR theory (Hobfoll, 1989), in highly autonomous working-time arrangements, we see resource gains (with respect to working-time autonomy) that increase satisfaction with work. At the same time, there is the risk of resource losses related to demands caused by higher working-time autonomy, which decrease satisfaction with leisure time.

Organisations and individuals alike should be made aware of the risks of highly autonomous working-time arrangements. To avoid the problems associated with overload in self-managed working time, Matta (2015) suggested that employers should increasingly introduce regulated systems, including time accounts in order to detect problems associated with extensive overtime. Another option could be to strengthen a corporate culture in which employees feel free to admit overload and discuss it openly. If overload is experienced, there should be opportunities for intervention.

In addition, our study has shown that employees perceive working hours differently within different working-time schemes. As overtime has a stronger effect on leisure satisfaction in flextime compared with fixed working-time schemes, one may conclude that employees with flexible schedules tend to experience work as being more time-consuming and therefore more overwhelming. This could be because without a fixed end time of the working day, employees may feel there are no boundaries to work. As Matta (2015) suggested, more structured work environments could help employees have a better overview of their actual time spent working and thus escape the feeling of never quite finishing work by setting personal boundaries for work. Other options for organisations introducing flexible or highly flexible working-time schemes include allowing employees time to adjust to them, encouraging open discussion of potential risks and problems, and training managers to deal with employees’ schedules in a health-promoting way (Wieland and Scherrer, 2002). Highly flexible work schedules have a positive effect on work satisfaction, but they have a negative effect on satisfaction with leisure time. The influence of working overtime on satisfaction with leisure time is much stronger for highly flexible working-time schemes. When implementing highly flexible working-time schemes, organisations should take measures to protect employees from overload.

Limitations

It is important to acknowledge a number of limitations when interpreting the results of this study. First, we selected only those respondents for our analyses who were in gainful employment across all included measurement waves. We excluded respondents who reported spells of unemployment as this would have unnecessarily complicated statistical modelling. Second, we cannot infer from the data what happened between the waves. Despite the fact that a time lag of 2 years is reasonable (cf. Dormann and Zapf, 2002), the available data primarily set the design. Due to the very limited number of longitudinal studies linking working-time arrangements with well-being, we cannot draw any firm conclusions about optimal time lags for effects of working-time arrangements. However, this study can serve as a starting point motivating other scholars to collect further longitudinal evidence on optimal time lags. Third, as working-time arrangements can take various forms, the distinction made here is a rather crude categorisation justified by the large longitudinal sample. Fourth, we cannot rule out that confounding variables, such as change of employer and a change of working conditions other than the working-time arrangement, caused changes in the outcomes. Control for these was not possible within the study design and with the data set we used. Furthermore, the use of single-item measures may result in a low reliability of the measures.

These limitations are likely to contribute to an underestimation of the effects of working-time arrangements. Hence, future studies may opt to assess working-time arrangements in more detail and use shorter and longer time gaps between waves to cover changes in employment conditions fully across time, allowing the detection of more proximal as well as distal effects.

Conclusions

In this study, we used representative panel data to analyse effects of working-time arrangements on different outcomes. By differentiating between change and permanence of working-time schemes within a multilevel framework, we were able to extend the analysis of working-time scheme transitions beyond the scope of fixed-effects models.

Our results show that highly autonomous working-time arrangements can be beneficial in relation to job satisfaction, but at the same time they can reduce other aspects of well-being. If employees work overtime, these negative effects are increased. In future studies on the effects of working-time arrangements, it is important to consider the differential effects of working-time autonomy as a resource as well as complexity, responsibility, and overtime as demands. Indeed, distinct aspects of highly autonomous working-time arrangements can increase or decrease well-being.

Footnotes

Acknowledgements

This article is based on data collected and provided by the German Institute for Economic Research (DIW Deutsches Institut für Wirtschaftsforschung e.V.) and contains no analyses based on data collected by the authors. The Institute conducts its research in accordance with the highest ethical standards (see: ). Therefore, the authors did not deem it necessary to obtain ethics approval for this study. The authors declare that there has been no conflict of interest. This study has not received any funding by sources outside the university.

Access to the data can be obtained via a subscription. For details, see: . Syntax for the models used here can be obtained from the corresponding author.

References

Andreß

Golsch

Schmidt

(2013) Applied Panel Data Analysis for Economic and Social Surveys. Berlin: Springer.

Bakker

Demerouti

(2007) The job demands–resources model: State of the art. Journal of Managerial Psychology 22(3): 309–328.

Bakker

Demerouti

(2014) Job demands-resources theory. In: Cooper

Chen

(eds) Well-Being: A Complete Reference Guide. Chichester: Wiley-Blackwell, pp. 1–28.

Bakker

Demerouti

Schaufeli

(2005) The crossover of burnout and work engagement among working couples. Human Relations 58(5): 661–689.

Baltes

Briggs

Huff

et al . (1999) Flexible and compressed workweek schedules: A meta-analysis of their effects on work-related criteria. Journal of Applied Psychology 84(4): 496–513.

Bannai

Tamakoshi

(2014) The association between long working hours and health: A systematic review of epidemiological evidence. Scandinavian Journal of Work, Environment & Health 40(1): 5–18.

Bauer

Groß

Lehmann

Munz

(2004) Arbeitszeit 2003. Arbeitszeitgestaltung, Arbeitsorganisation und Tätigkeitsprofile. Berichte des ISO, 70. Köln.

Beckmann

Cornelissen

(2014) Self-managed working time and employee effort: Microeconometric evidence. SOEPpapers on Multidisciplinary Panel Data Research 636, DIW Berlin, The German Socio-Economic Panel (SOEP).

Boes

Winkelmann

(2010) The effect of income on general life satisfaction and dissatisfaction. Social Indicators Research 95(1): 111–128.

10.

Böhm

Herrmann

Trinczek

(2004) Herausforderung Vertrauensarbeitszeit. Zur Kultur und Praxis eines neuen Arbeitszeitmodells. Berlin: Edition Sigma.

11.

Boyd

Bakker

Pignata

et al . (2011) A longitudinal test of the Job Demands–Resources model among Australian university academics. Applied Psychology: An International Review 60(1): 112–140.

12.

Brunello

Fort

Schneeweis

Winter-Ebmer

(2016) The causal effect of education on health: What is the role of health behaviors? Health Economics 25(3): 314–336.

13.

Costa

Sartori

Akerstedt

(2006) Influence of flexibility and variability of working hours on health and well-being. Chronobiology International 23(6): 1125–1137.

14.

Demerouti

Bakker

Nachreiner

Schaufeli

(2001) The job demands–resources model of burnout. Journal of Applied Psychology 86(3): 499–512.

15.

Diener

Suh

Lucas

Smith

(1999) Subjective well-being: Three decades of progress. Psychological Bulletin 125(2): 276–302.

16.

Dormann

Zapf

(2002) Social stressors at work, irritation, and depressive symptoms: Accounting for unnmeasured third variables in a multi-wave study. Journal of Occupational and Organizational Psychology 75: 33–58.

17.

European Union (2015) First Findings: Sixth European Working Conditions Survey. Available at: http://www.eurofound.europa.eu/sites/default/files/ef_publication/field_ef_document/ef1568en.pdf (accessed 17 December 2015).

18.

Frese

Zapf

(1994) Action as the core of work psychology: A German approach. In: Triandis

Dunnette

Hough

(eds) Handbook of Industrial and Organizational Psychology. Palo Alto, CA: Consulting Psychologists Press, pp. 271–340.

19.

Glißmann

Peters

(2001) Mehr Druck durch mehr Freiheit. Die neue Autonomie der Arbeit und ihre paradoxen Folgen. Hamburg: VSA-Verlag.

20.

Haipeter

Lehndorgg

Schilling

(2002) Vertrauensarbeitszeit. Analyse eines neuen Rationalisierungskonzepts. Leviathan 30(3): 360–383.

21.

Hobfoll

(1989) Conservation of resources: A new attempt at conceptualizing stress. American Psychologist 44(3): 513–524.

22.

Hoff

Weidinger

(1999) Erfolgsfaktoren der Vertrauensarbeitszeit. Personal 8: 380–384.

23.

Janke

Stamov-Roßnagel

Scheibe

(2014) Verschwimmen die Grenzen? Auswirkungen von Vertrauensarbeitszeit auf die Schnittstelle von Arbeit und Privatleben. Zeitschrift für Arbeitswissenschaft 68(2): 97–105.

24.

Kelliher

Anderson

(2010) Doing more with less? Flexible working practices and the intensification of work. Human Relations 63(1): 83–106.

25.

Kelly

Moen

Tranby

(2011) Changing workplaces to reduce work–family conflict: Schedule control in a white-collar organization. American Sociological Review 76(2): 265–290.

26.

Krause

Baeriswyl

Berset

et al . (2015) Selbstgefährdung als Indikator für Mängel bei der Gestaltung mobil-flexibler Arbeit: Zur Entwicklung eines Erhebungsinstruments. Wirschaftspsychologie 1: 49–59.

27.

Lott

Chung

(2016) Gender discrepancies in the outcomes of schedule control on overtime and income in Germany. European Sociological Review 32(6): 752–765.

28.

McNall

Masuda

Nicklin

(2010) Flexible work arrangements, job satisfaction, and turnover intentions: The mediating role of work-to-family enrichment. The Journal of Psychology 144(1): 61–81.

29.

Matta

(2015) Führen selbstgesteuerte Arbeitszeiten zu einer Ausweitung der Arbeitsstunden? Eine Längsschnittanalyse auf der Basis des Sozio-oekonomischen Panels. Zeitschrift für Soziologie 44(4): 253–271.

30.

Moen

Kelly

Hill

(2011) Does enhancing work-time control and flexibility reduce turnover? A naturally occurring experiment. Social Problems 58(1): 69–98.

31.

Nijp

Beckers

DGJ

Geurts

SAE

et al . (2012) Systematic review on the association between employee worktime control and work-non-work balance, health and well-being, and job-related outcomes. Scandinavian Journal of Work Environment & Health 38(4): 299–313.

32.

Oesterrich

(1981) Handlungsregulation und Kontrolle. München: Urban & Schwarzenberg.

33.

Pedersen

Jeppesen

(2012) Contagious flexibility? A study on whether schedule flexibility facilitates work–life enrichment. Scandinavian Journal of Psychology 53(4): 347–359.

34.

Powdthavee

Lekfuangfu

Wooden

(2015) What’s the good of education on our overall quality of life? A simultaneous equation model of education and life satisfaction for Australia. Journal of Behavioral and Experimental Economics 54: 10–21.

35.

Raudenbush

Bryk

(2002) Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: SAGE.

36.

Roberts

Foti

(1998) Evaluating the interaction between self-leadership and work structure in predicting job satisfaction. Journal of Business and Psychology 12(3): 257–267.

37.

Rodriguez-Muñoz

Sanz-Vergel

Demerouti

Bakker

(2012) Reciprocal relationships between job demands, job resources, and recovery opportunities. Journal of Personnel Psychology 11(2): 86–94.

38.

Salines-Jiménez

MDM

Artés

Salines-Jiménez

(2011) Education as a positional good: A life satisfaction approach. Social Indicators Research 103(3): 409–426.

39.

Schaufeli

Taris

(2014) A critical review of the job demands–resources model: Implications for improving work and health. In: Bauer

Hämmig

(eds) Bridging Occupational, Organizational and Public Health: A Transdisciplinary Approach. Dordrecht: Springer, pp. 43–68.

40.

Singe

Croucher

(2003). The management of trust-based working time in Germany. Personnel Review 32(4): 492–509.

41.

Van den Broeck

Van Ruysseveldt

Smulders

De Witte

(2011) Does an intrinsic value orientation strengthen the impact of job resources? A perspective from the job demands–resources model. European Journal of Work and Organizational Psychology 20(5): 581–609.

42.

Volpert

(1987) Psychische Regulation von Arbeitstätigkeiten. In: Kleinbeck

Rutenfranz

(eds) Arbeitspsychologie. Göttingen: Hogrefe, pp. 1–42.

43.

Wieland

Scherrer

(2002) Vertrauensarbeitszeit – Merkmale einer gesunden und tragfähigen Praxis. Wuppertaler Psychologische Berichte 2: 1–16.

44.

Wingen

Hohemann

Bensch

Plum

(2004) Vertrauensarbeitszeit – Neue Entwicklung gesellschaftlicher Arbeitszeitstrukturen. Dortmund: Bundesanstalt für Arbeitsschutz und Arbeitsmedizin.