Shorter working days, work engagement and productivity
By
Maarten Nieuwkamp
University of Groningen
Faculty of Economics and Business
BSc Bedrijfskunde
Supervisor: prof. dr. B.A. Nijstad
Abstract
The aim of this paper was to test whether shorter working days increase productivity, through work engagement. It is expected that when working hours are shorter, it is easier for
employees to put in effort, be persistent, be concentrated and that the challenge and happiness associated with shorter working hours would lead to an increase in work engagement. Then, the increased energy-levels, mental resilience, effort, persistence, well-being and generally working harder as a result of higher work engagement, could lead to higher productivity. Hypotheses were tested using a diary-study, within-person design in which one general and five daily surveys were taken in a big organization from the Dutch public sector. Data were analysed using multilevel analysis. While the main analyses show no evidence for a negative relationship between amount of hours worked and work engagement, explorative analyse indicate that this relation might exist, but only for people with a lower education.
Furthermore, the results show that there is support for the positive relationship between work engagement and productivity.
Keywords:
1. INTRODUCTION
‘’Employers in Sweden introduce six-hour work day’’ (Independent, 2016). These and similar headlines popped up in the news during the last months of 2016 and first months of 2017. The articles were referring to a study by the Swedish researcher Bengt Lorentzon (2016). In his study, nurses from the Svartedalen retirement home in Gothenburg switched from an 8-hour to a 6-hour work day for 23 months. To compensate for the loss of total working hours, 17 extra nurses were hired. The idea behind the experiment was that a shorter work day increases well-being of personnel and increases productivity through, for example, a higher work motivation. The results showed that nurses became healthier, more alert and were more calm than nurses working eight hours. Furthermore, sick leave among nurses decreased and
productivity, measured in terms of activities that they organized for residents, rose with 80%. Despite all these positive results, the experiment was terminated because the budget for the study was used up (Lorentzon, 2016).
The introduction of the 6-hour workday in modern society would be revolutionary, because the work-life balance of workers would change drastically. This, in turn, may be expected to increase the quality of life. As in the Svartedalen experiment, workers across the country may become more satisfied with their work and life overall. They potentially will have more leisure time, be healthier, experience less stress and be absent less often. In short, a shorter working day potentially has many positive effects on workers.
The question, thus, is whether productivity actually increases as a result of shorter working hours. And if this holds true, why exactly do workers become more productive if they know they have fewer hours to do the job? In this paper, the concept of work engagement is introduced as a mediator of the relationship between reduced hours of work and increases in productivity. Engaged employees are found to work hard, be dedicated and create happiness for themselves when working (Bakker, Schaufeli, Leiter & Taris, 2008), which is expected to increase productivity. Work engagement is different from similar concepts such as
workaholism or job involvement, and is defined as a positive, fulfilling, affective motivational state of work-related well-being that is characterized by vigour, dedication, and absorption (Bakker et al., 2008). Bakker et al. (2008) conclude that this relatively new concept has been shown to increase productivity as well as employee well-being. These same effects are also predicted by proponents of the six-hour work day and found in the Svartedalen experiment. The aim of this study is, therefore, to find out whether – besides stimulating employee well-being – shorter working days can actually increase employee productivity, and whether this effect is mediated by work engagement. To accomplish this aim, I will review the literature on working hours and its effects, and use this to create a theoretical model. Using a diary-study within-subjects design – a field experiment is beyond the scale of this research – the
hypotheses will be tested.
2. LITERATURE REVIEW
History of reduced working hours
Reduction of working hours is something that began in the late 19th century and first half of the 20th century (Dembe, 2011). In many countries, the workweek was shortened and became five days long instead of six days. Gradually, people changed from working days that were longer than twelve hours to the now so common 40-hour work week. As a result, the health and well-being of workers increased dramatically, as was intended and expected by
Working hours and employee well-being
Longer working hours and employee well-being
From a work-life balance perspective and from the perspective of employee health, there is extensive previous literature concerning longer working hours and its negative effects. Firstly, longer working hours are related to worse health. In two meta-analyses, Sparks, Cooper, Fried and Shirom (1997) and Spurgeon, Harrington and Cooper (1997) show that working longer hours increases overall, physiological and psychological health symptoms. There is even a strong link between long working hours and coronary heart disease (Virtanen, Heikkilä, Jokela, Ferrie, Batty, Vahtera & Kivimäki, 2012). Secondly, according to Spurgeon, Harrington and Cooper (1997), besides health problems longer hours also result in safety problems. This finding is supported by Rogers, Hwang, Scott, Aiken and Dinges (2004), who find that the risks of making an error increased significantly when shifts were longer than 12 hours, when working overtime or when making more than 40 hours in one week. Additonaly, Kirkcaldy, Trimpop and Cooper (1997) showed that longer working hours even led to more driving accidents. Thirdly, fatigue is a logical consequence of working long hours. Jungsun, Yangho, Chung and Hisanaga (2001) showed this in a subjective fatigue symptoms study. Besides pure physical consequences of working longer hours, longer hours also impact employee’s lives in other ways. For example, a higher number of working hours leads to a higher amount of stress (Cooper, Davidson & Robinson, 1982). Naturally, a reduction of working hours would thus reduce the amount of stress perceived by workers. Increased leisure time gives more room for relaxation and recreation, and especially physical activity during leisure time is shown to decrease stress levels (Aldana, Sutton, Jacobsen & Quirk, 1996). Moreover, related to the previous effects, is sick absence. Long total working hours are
related to job satisfaction. In general, according to the literature, employee well-being thus decreases with longer working hours.
Working part-time and employee well-being
Previous research on part-time employment, defined as working less than full-time hours in a continuous employment contract with a guaranteed number of hours or a somewhat fixed workschedule (Zeytinoglu, Lillevik, Seaton & Moruz, 2004), versus full-time employment, showed that part-timers had better overall health (Benavides, Benach, Diez-Roux & Roman, 2000), and slightly higher job satisfaction (Logan, O'Reilly & Roberts, 1973). A more recent study by Booth and Van Ours (2008), however, finds that women do have higher job
satisfaction when working part-time, whereas for men there is no effect on job satisfaction of working part-time compared to full-time. Although part-time employees work fewer hours and should thus have lower stress levels, according to the findings from the previous
paragraph, Zeytinoglu et al. (2004) found that part-timers could have higher stress as a result of inherent part-time-job characteristics such as lower payment than co-workers working full-time or, especially for women, still having the housewife role besides their part-full-time job. Finally, Higgins, Duxbury & Johnson (2000) showed that part-time work reduced work-to-family interference among women, and that work-to-family-to-work interference was lower for women that just wanted to earn money and higher for career women. Although dependent on gender, gender-roles and career aspirations, evidence in general tends to show better well-being for part-timers than full-timers.
Shorter working days and employee well-being
sick leave after 18 months was 2.8 times higher among nurses in the rest of the city of
Gothenburg than among Svartedalen nurses (Lorentzon, 2016). Furthermore, researchers have found that a 6-hour workday improves the work-family interaction (Anttila, Nätti & Väisänen, 2005). They showed that conflicts arising from work that affected family decreased as a result of reduced working hours, and that the results were the strongest for a 6-hour workday. Barnet and Gareis (2000) further showed that shorter working days only increased quality of life if the trade-off between work and non-work is not more stressful for the employees than the extra hours of work. Additionally, MacInnes (2005) found that there is extensive empirical support for reduced working hours, but that this support declines when combined with lower wages.
The second advantage of shorter working days mentioned by White (1987), lower
unemployment, is supported by Dembe (2011), and Booth and Ravallion (1993). The latter authors found that a cut in hours decreased unemployment in the UK, and that in Australia the unemployment results were ambiguous. Booth and Schiantarelli (1987) showed in general an ambiguous effect on unemployment. Something that both Booth and Schiantarelli (1987) and Mangan and Steinke (1988) mention is that for a positive effect of shorter working days on unemployment, productivity should rise, sales should increase or labour costs should be lower. White (1987) even finds that in practice, a reduction in working hours does not reduce unemployment. The cause he gives for this will be the topic of the next paragraph.
Shorter working days and productivity
Where the previous paragraph mentioned a great variety of effects of shorter working days, one effect is not yet discussed: productivity. Productivity is the relation of output and input (Tangen, 2002). Thus, as Tangen (2002) states, productivity is concerned with resources on the one hand, and with producing value on the other hand. According to the author, it is important to distinguish between production and productivity. Production is something that can be measured, for example in units produced, but productivity is a relative measure, that can only be compared with others or compared over time.
contrary to expectations of the philanthropic employers who wanted to increase employment in the 1890’s, reduction in working hours did not decrease unemployment. It appeared that the higher productivity after the instigation of shorter working days, increased the production to an extent that no extra workers had to be hired to produce the same output, or sometimes even more. In contradiction to this, Brunello (1989) found that unemployment does decrease as a result of shorter working days. Also, in a meta-analysis, Golden (2012) confirms the finding of higher productivity per hour when working days are shortened. Consequently, Latham and Locke (1975) find higher productivity when time restrictions are placed on workers’ working days. Lorentzon (2016) provides further evidence for increased productivity. During the Svartedalen experiment, the total amount of hours worked did not change, because extra nurses were hired. However, the production – measured in number of activities organized for tenants – rose with 80% in just under two years. This is a good example of the relationship between a shorter working day and productivity being tested in this paper. Additionally, Pencavel (2015) suggests that restrictions on working hours should not be seen as restraints for management, but should be seen as a way of improving efficiency.
There is thus empirical evidence that there is a relation between shorter working hours and increases in productivity, but this is not conclusive. A reason for this is that it is very expensive and difficult to conduct experiments such as the one in Sweden and find
compelling evidence. Therefore, a within-person design is used instead of a between-person design. This means that productivity between longer and shorter hour days is examined. Additionally, one thing that is not clear yet, is the mechanism through which this occurs. In the next section, I propose that the mediator in the relationship between shorter working days and increased productivity is work engagement.
Work engagement
characterized by vigour, dedication, and absorption. According to the same authors, vigour is characterized by high levels of energy and mental resilience while working, the willingness to invest effort in one’s work, and persistence even in the face of difficulties. They explain that dedication refers to being strongly involved in one’s work, and experiencing a sense of significance, enthusiasm, inspiration, pride, and challenge. Lastly, absorption is characterized by being fully concentrated and happily engrossed in one’s work, whereby time passes quickly and one has difficulties with detaching oneself from work. Bakker et al. (2008) summarize the literature and mention that job resources, such as autonomy, supervisory coaching and performance feedback, and personal resources, such as optimism, self-efficacy and self-esteem, are the most important factors increasing work engagement.
Golden (2012) mentions several factors that lead to improvements in productivity as a result of shorter working hours, such as less physical and mental fatigue, or creative improvements in time utilization, for example more flexible work schedules that reduced time wasted on the job. Besides the physiological aspects, White (1987) mentions increased motivation after reduced working hours.
In this paper, it is proposed that a shorter working day will lead to a higher engagement with the task to be performed. For example, a reduction of the negative feelings associated with having a long day of work ahead could result in a more positive view of the task. It is expected that it is easier for people to be mentally resilient, invest effort and be persistent when they have to work fewer hours in total, because there are fewer hours of work ahead and they can divide their already limited effort and concentration available over fewer hours. They are challenged to do things in less time and as a result their working pace can increase. When facing a shorter working day, it is easier to stay concentrated throughout the whole day, and people are probably more happy to go to work – as it takes less of their time. As a result, they could feel that time passes quickly and it may be difficult for them to detach from work. The first hypothesis is, thus, that lower day-level working hours will lead to higher day-level work engagement.
doing work and towards their job, their productivity is expected to increase. A high level of concentration and being completely engrossed in their work could greatly increase their output per hour. The second hypothesis is, thus, that higher day-level work engagement will lead to higher productivity. It is, therefore, proposed that work engagement mediates the relation between working hours and productivity.
3. METHOD
In this study a day-level diary-study design was used, because a (quasi-)experiment was beyond the scope and scale of this research. In using this within-person method, a good substitute of an experiment is found, as the amount of hours also varies within persons. This allowed me to explore whether people felt more engaged and productive on days with fewer than average working hours.
Sample and procedure
The data showed that there were only two parts where people stopped the survey prematurely. The first point was when they had to agree to the informed consent form and the second point was where they had to indicate whether they took part in the organization’s program where senior workers are able to work less hours just before retiring. An explanation for this could be that some people wrongly assumed that the survey wasn’t for them. Data for people that stopped in those two occasions were deleted. After deletion the sample consisted of 49 people that filled in the general questionnaire and 110 observations from daily observations.
The amount of daily surveys people eventually filled in varied from zero to five (M=2.22, SD=1.93). The average age of participants was 44.04 years (SD=11.34), and the distribution for gender was predominantly female (15 males, 34 females). Of participants, 26.5% had followed an education that was classified as being lower education (high school or
intermediate vocational education) and the other 73.5% had a higher education. On average, participants worked 7.32 hours per day (SD=2.53).
An assumption for the proposed relation was that people worked (relatively) independently and without external pacing of their work. For the survey, it was thus important that
The participants started with a one-time survey to gather personal information, such as gender, age and education; a general baseline measure of length of working day, work engagement and productivity; stress, time-pressure, job satisfaction, tiredness and autonomy scales; and information on how engaged they felt in their work and how productive they felt they were in general. Moreover, they received a survey to be filled in at the end of each working day, asking questions about the hours worked that day, the level of work
engagement, their productivity, shortened versions or single items of the scales mentioned for the general survey, and again how engaged and productive they felt. At the end, on the last day, there was an option for participants to give comments on the surveys.
All questionnaires were administered using the Qualtrics online survey platform. The link to the general survey was send to all employees using the intranet and internal email system of the organization, and the links for the daily surveys were send to participants over email after they had opted in using the general survey.
Data and measures
General survey
In the general survey personal information and general-level data on the theoretical constructs were measured. Participants agreed to participate in this research in an informed consent form. The surveys were taken in Dutch. An official translation of the UWES was used (Schaufeli & Bakker, 2003). The remaining scales that were originally in English –
performance self-report, job satisfaction, time-pressure, stress and tiredness– were translated using an (online) English-Dutch dictionary. The autonomy scale was originally in Dutch and was thus directly used in the survey. In this paper, the items for this scale were translated using a Dutch-English dictionary and shown in English.
Firstly, participants filled in their gender, age and highest education. Secondly,
General-level work engagement was measured using the Utrecht Work Engagement Scale (UWES) (Schaufeli, Bakker & Salanova, 2006). This scale includes 17 items (α=.92) divided over three dimensions: vigour, dedication and absorption. Examples of items are ‘At my work, I feel bursting with energy’, ‘I find the work that I do full of meaning and purpose’ and ‘Time flies when I am working’ (Schaufeli, Bakker & Salanova, 2006: 714). All of these items were measured on a seven-point scale: 0 Never (Never); 1 Almost never (A few times a year or less); 2 Rarely (Once a month or less); 3 Sometimes (A few times a month); 4 Often (Once a week), 5 Very often (A few times a week), 6 Always (Every day).
Performance was assessed using a supervisor performance scale rewritten for self-report (Henderson & Lee, 1992) consisting of 8 items (α=.84), and two items asking directly for participants’ productivity – the amount of work done per hour – and production – the total amount of work done per day. Examples of the performance self-report scale are ‘My
adherence to schedules’ and ‘My ability to meet set goals’, and these were measured on a seven-point scale ranging from ‘Extremely low’ to ‘Extremely high’. For the other two measures, the items were ‘How productive are you (compared to other employees with the same job), during the average workday when looking at the amount of work you produce per hour?’ and ‘How productive are you (compared to other employees with the same job), during the average workday when looking at the amount of work you produce per day (independent of the amount of hours worked)?’, respectively.
Participants also filled out a stress scale based on Evans & Johnson (2000) consisting of 7 items (α=.86), a time pressure scale (Furda, 1995) consisting of 4 items (α=.83), a shortened job satisfaction scale (Brayfield & Rothe, 1951) consisting of 5 items (α=.85), a tiredness scale (Pirinen, Kolho, Simola, Ashorn & Aronen, 2010) consisting of 3 items
Day-level survey
Scales for the daily surveys were rewritten to measure daily values of the variables. Firstly, the participants filled in the participant number they received during the general survey. Subsequently, they gave the number of hours that they had worked for that day and how many hours they planned to work the next day. Missing values for the number of hours worked were supplemented with values from the number of hours planned when
participants failed to fill in a number for number of hours worked for a given day. The
assumption here is that participants worked exactly the amount of hours they planned to work. A crosstab of the number of hours worked and the number of hours planned showed that this was the case as in the majority (75.78%) of cases participants worked the amount of hours they planned to work, and when they did not work their planned hours, the difference with planned hours was relatively small (M=0.11, SD=1.74).
Secondly, participants filled in the shortened version of the Utrecht Work Engagement Scale (Schaufeli, Bakker & Salanova, 2006) consisting of 9 items (α=.95), with the items adjusted in such a way that daily values for work engagement were measured. Examples of items are ‘At my work today, I felt bursting with energy’, ‘I find the work that I did today full of meaning and purpose’ and ‘Time flew when I was working today’.
Thirdly, daily performance was measured using 3 selected items (α=.82) from the performance self-report scale, and by using the productivity and production items discussed in the general survey section above. These were rewritten to measure daily values, for example ‘The quality of my work today’ for performance, ‘How productive were you, compared to for example your colleagues, during today's workday in terms of the amount of tasks you got done per hour?’ for productivity and ‘How productive were you, compared to for example your colleagues, during today's working day in terms of the total amount of tasks you got done? (Regardless of the amount of hours you worked)’ for production.
used. An example of one of the four items is ‘Today I had the possibility to choose my own way of working’.
The same scales as in the general survey were used. One person started a daily survey, but no questions were filled in, so data were deleted. Also, in one instance someone filled in 64787 as a participant number instead of 64878, so the former was replaced with the latter (the number of 64787 was not present in any of the other surveys, so it was assumed the participant made a type while filling in their participant number).
4. RESULTS
Descriptive analyses
Table 1: Descriptive statistics of general variables
Variable M SD
Correlations
2 3 4 5 6 7
1. Average Hours Worked 8.071 0.671 0.294* 0.195 0.016 0.052 -0.011 0.283* 2. General Work Engagement Score 5.075 0.735 1 0.288* 0.226 0.187 0.164 0.106 3. General Performance Score 5.135 0.611 1 0.450** 0.493** 0.051 -0.177 4. General Productivity Score 4.860 1.000 1 0.915** -0.056 -0.200 5. General Production Score 4.920 1.057 1 -0.055 -0.140
6. General Autonomy Score 3.722 0.755 1 0.317*
7. Education Level 3.960 0.865 1
*=p<0.05; **=p<0.01
Table 1 shows the descriptive statistics and bivariate correlations for the key variables
measured in the general survey (n=49). A mean of average working hours of 8.07 corresponds well with the (Dutch) standard of working eight hours per day and a standard deviation of 0.671 shows that most people are close to this standard. Generally, participants score high on work engagement (score 5.075 on a 7-point scale). Furthermore, there is a significant, but positive, correlation between the average amount of hours per workday and work
first hypothesis. As expected, the performance, productivity and production measures are significantly related to each other. Education is positively related to autonomy, which could potentially be explained by the fact that higher educated people have jobs with more decision-making freedom and thus more feelings of autonomy. The positive relation between education and average amount of working hours is less obvious, but it might be that people in jobs with more responsibility do more overwork, or like to work more because of higher job autonomy. Table 2: Descriptive statistics of daily variables
Variable M SD ICC Correlations 2 3 4 5 6 7 8 1. Actual Daily Hours Worked 7.321 2.534 0.115 0.674** 0.825** 0.251* 0.167 0.148 0.228* -0.136 2. Normal Daily Hours 7.447 2.151 0.007 1 0.732** 0.008 -0.096 -0.015 -0.003 -0.127 3. Planned Daily Hours 7.143 2.604 0.114 1 0.147 0.072 0.109 0.134 -0.109 4. Daily Work Engagement Score 4.748 1.189 0.447 1 0.602** 0.545** 0.597** 0.184 5. Daily Performance Score 4.782 0.886 0.183 1 0.671** 0.676** 0.107 6. Daily Productivity Score 4.220 1.254 0.078 1 0.874** 0.086 7. Daily Production Score 4.170 1.250 0.069 1 0.076 8. Daily Autonomy Score 3.888 0.869 0.415 1 *=p<0.05; **=p<0.01
across days during the week in which the study took place. Additionally, the mean and standard deviation of productivity and production are (just as in table 1) very similar, and it might be that participants find it difficult to distinguish between the two concepts, or that they simply are highly related. The latter is supported by their bivariate correlation. As would be logically expected, actual, normal and planned hours and performance, productivity and production are highly related to each other. Again, (actual) hours of work are significantly related to work engagement, and work engagement is positively related to performance, productivity and production (as opposed to only performance in table 1).
It should be noted that, besides autonomy, a number of other control variables were also measured (stress, time pressure, job satisfaction and tiredness). According to a correlational analysis, only the scores for autonomy, job satisfaction and tiredness were related to work engagement. However, the latter two were not included as control variables in the statistical models, because of the possible causal relationship between hours of work and job
satisfaction, and hours of work and tiredness. In other words, these variables may not be exogenous to the model, and therefore may not be suitable as control variables.
Main analyses
The first hypothesis of this paper was that a lower amount of working hours would lead to a higher work engagement among participants, which then would lead to higher productivity according to the second hypothesis. To test these hypotheses, a multilevel analysis was used in two steps. Firstly, the effect of the amount of hours on work engagement, and secondly, the effect of work engagement on performance, productivity and production was analysed. For every analysis, first an empty model was estimated, the control variable were then added to the model and thirdly the independent variable(s) was (were) added (one by one).
Hours on work engagement
Table 3: Multilevel analysis of daily work engagement score on hours worked
Model 1 (Empty) Model 2 (+ Control) Model 3 (+ Hours)
Daily autonomy score 0.108 0.162
Mean hours worked 0.351**
Deviation from mean hours worked 0.102
Variance (within) 0.757 0.769 0.778
Variance (between) 0.612 0.587 0.449
ICC (between/total) 0.447 0.433 0.366
Log Likelihood 298.875 300.444 293.929
+=p<0.10; *=p<0.05; **=p<0.01; ***p<0.001
The results for model 2 show that adding the control variable resulted in a slightly higher log likelihood of the model, which could have been the result of a changing sample size following the addition of the control variable due to missing data. The estimate of the coefficient for the daily autonomy score was not significant meaning that there was no relation between people’s autonomy on a given day and their work engagement on that day.
The results for model 3 show that adding mean hours worked and deviation from mean hours worked resulted in a substantially lower log likelihood of the model and mainly less
remaining unexplained between person variance. The estimate of the coefficient for the mean hours worked was positive and significant, meaning that people who generally work more hours have a higher work engagement. The estimate of the coefficient for the deviation from mean hours worked was insignificant, meaning that there was no relation between the amount of hours people worked on a given day and their work engagement on that day.
The first hypothesis was that a lower amount of working hours would lead to a higher work engagement among participants, which was not supported by the analysis above. Actually, an opposite (positive) relation was found at the between persons level and no relation was found within persons.
Work engagement on performance, productivity and production
Table 4: Multilevel analysis of daily performance, productivity and production on daily work engagement
Performance Productivity Production
Model 1 (Empty) Model 2 (+ Control) Model 3 (+ Hours) Model 4 (+ Work Engagement Model 1 (Empty) Model 2 (+ Control) Model 3 (+ Hours) Model 4 (+ Work Engagement Model 1 (Empty) Model 2 (+ Control) Model 3 (+ Hours) Model 4 (+ Work Engagement Daily autonomy score 0.084 0.118 0.052 0.110 0.140 0.029 0.098 0.138 0.022
Mean hours worked 0.160+ 0.042 0.173 - 0.004 0.269* 0.071
Deviation from mean hours worked
0.087 0.037 0.120 0.048 0.174+ 0.095
Mean daily work engagement score
0.299* 0.461** 0.502**
The results for model 2 show that adding the control variable resulted in no substantial change in the log likelihood of the model for performance, productivity and production. For all three measures the estimate of the coefficient for the daily autonomy score was insignificant, meaning that there was no relation between people’s autonomy on a given day and their performance, productivity and production on that day.
The results for model 3 show that adding mean hours worked and deviation from mean hours worked resulted in no substantial change in the log likelihood of the model for performance, but a lower log likelihood of the model for productivity and production. The estimate of the coefficient for the mean hours worked was positive and marginally significant for
performance and positive and significant for production, meaning it could be that people that work more hours generally have a better performance and production. For productivity this estimate was insignificant, meaning that there was no general relation between people’s work engagement and their productivity. The estimate of the coefficient for the deviation from mean hours worked was positive and marginally significant, meaning that it could be that people that work more hours on a given day also have a higher production on that day. For performance and productivity this estimate was insignificant.
The results for model 4 show that adding mean daily work engagement score and deviation from mean daily work engagement score resulted in a substantially lower log likelihood of the model for performance, productivity and production. There was less remaining between person variance for all three measures, meaning more between person variance was explained after adding work engagement to the model. The estimate of the coefficient for mean daily work engagement score was positive and significant for all three measures, meaning that people with higher work engagement generally have a better performance, productivity and production. The estimate of the coefficient for deviation from mean daily work engagement score was positive and highly significant for all three measures, meaning that people that have a higher work engagement on a given day also have a better performance, productivity and production on that day. The estimate of the coefficient for mean hours worked also became insignificant for performance and production and the estimate of the coefficient for deviation from mean hours worked also became insignificant for production.
The second hypothesis was that a higher work engagement in participants would lead to higher productivity, which was supported by the analysis above. There was a positive
Explorative analyses
To explore if the relation between the amount of hours and work engagement changes under different circumstances, some additional analyses were done.
Planned hours on work engagement
In a multilevel analysis similar to the ones above, the effect of hours planned for a day and the deviation from planned hours on work engagement was tested. The variables mean hours worked and deviation from mean hours worked from the original analysis were replaced by the variables hours planned (the amount of hours people planned to work on a given day) and deviation from hours planned (hours worked – hours planned, the difference between the actual and planned amount of hours). There was weak evidence for a positive relation between planned hours and daily work engagement score (b=.15, p=.094). It could be that people that generally plan more hours generally have a higher work engagement. Also, there was weak evidence for a positive relation between deviation from hours planned and daily work engagement score (b=.19, p=.103). It could be that people that work more than they have planned on a given day have a higher work engagement on that day.
Normal hours on work engagement
In the same way as for planned hours, a multilevel analysis was executed to test the effect of the amount of hours people would normally work on a given weekday and the difference between actual and normal hours on work engagement. The variables hours worked and deviation from hours worked from the original analysis were replaced by the variables normal hours (the amount of hours people would normally work on a given weekday) and deviation from normal hours (hours worked – normal hours, the difference between the actual and normal amount of hours). There was no evidence for a relation between normal hours and daily work engagement score (b=.14, p=.272). Also, there was weak evidence for a positive relation between the deviation from normal hours and daily work engagement score (b=.18, p=.065). It could be that people that work more than they would do normally on a given day have a higher work engagement on that day.
Education as a possible moderator
and keep them more engaged throughout the day. The relation between hours worked and work engagement could thus differ for different levels of education.
To find if education is a potential moderator in the relation between hours and work
Table 7: Multilevel analysis of daily work engagement on hours worked and education Model 3 (+ Hours) Model 4 (+ Education Dummy) Model 5a (+ Interaction Hours Mean and Edu) Model 5b (+ Interaction Hours Diff and Edu) Model 5c (+ Both interactions) Daily autonomy score 0.162 0.180 0.170 0.194 0.184
Mean hours worked 0.559** 0.576* 1.350* 0.517* 0.966 Deviation from
mean hours worked
0.200 0.171 0.172 - 1.177+ - 0.824
Education dummy 0.215 0.225 - 0.121 - 0.028
Interaction of mean hours worked and education dummy
- 0.912 - 0.511
Interaction of deviation from mean hours worked and education dummy 1.411+ 1.041 Variance (within) 0.778 0.790 0.792 0.810 0.805 Variance (between) 0.449 0.461 0.413 0.351 0.369 ICC (between/total) 0.366 0.369 0.343 0.302 0.314 Log Likelihood 293.929 288.346 285.952 284.963 284.332
+=p<0.10; *=p<0.05; **=p<0.01; ***p<0.001. Values for mean hours worked and deviation from mean hours worked were standardized.
Model 1, 2, 3 and their interpretations are exactly the same as in the main analyses.
The results for model 4 show that adding education dummy resulted in a lower log likelihood of the model. The estimate of the education dummy coefficient was insignificant, meaning that there is no relation between education and work engagement.
The results for model 5a show that adding the interaction of mean hours worked and
between mean hours worked and work engagement is stronger for people with a low education (b=1.35; p < .05) than for people with a high education (b=0.44; p < .10). The results for model 5b show that adding the interaction of deviation from mean hours worked and education dummy resulted in a lower log likelihood of the model, and that this interaction was marginally significant (b = 1.41, p < .10). Follow-up analyses showed that the relation between deviation from hours worked was negative and marginally significant for employees with a lower education (b = -1.18, p < .10), whereas it was positive and not significant for employees with a higher education (b = .23, p = .17). Thus, there is a possibility that there is a negative relation between hours worked on a given day and work engagement on that day for people with low education, but that this relation is positive for people with high education.
Summary of results
From the analyses above it can be concluded that there is no evidence that a lower amount of working hours leads to a higher work engagement in general. From the main analyses, it seems that people that generally work more hours generally have a higher work engagement and that there is no relation between the amount of hours people work on a given day and their work engagement on that day. The explorative analyses, however, show that there might be a relation when education is used as a moderator (see next paragraph). Furthermore, there is evidence that higher work engagement leads to higher performance, productivity and production. It seems that people with higher work engagement generally have a better performance, productivity and production and that people that have a higher work
engagement on a given day also have a better performance, productivity and production on that day. Additionally, there is weak evidence that people that work a higher amount of hours generally have a better performance, no evidence that they have a higher productivity, but strong evidence that they have a higher production. Also, there is some weak evidence that people that work more hours on a given day also have a higher production, while no such relation was found for performance and productivity.
The explorative analyses on planned and normal hours are in line with the findings of the main analyses in the sense that more hours generally go together with a higher work
was no evidence for a relation between the normal amount of hours and work engagement, but there is some evidence that people that work more than they would do normally on a given day have a higher work engagement on that day. Furthermore, both people with a low and a high education that generally work more hours seem to have a higher general work engagement. However, the relation is stronger for people with a low education (b=1.35) than for people with a high education (b=0.44). Remarkably, people with a low education that work more on a given day seem to have a lower work engagement on that day, while there is an indication that people with a high education that work more on a given day also have a higher work engagement on that day. Thus, there is a possibility that the relation between daily amount of hours and daily work engagement is negative for people with low education, and positive for people with high education, but the evidence for the latter relation is rather insignificant.
5. DISCUSSION
The aim of this paper was to test whether shorter working days lead to higher productivity, through work engagement. Data were collected using a diary study design in a big
organization from the Dutch public sector and analysed using multilevel analysis. While the main analyses show no evidence for a negative relationship between amount of hours worked and work engagement, explorative analyse indicate that this relation might exist, but only for people with a lower education. Furthermore, the results show that there is support for the positive relationship between work engagement and productivity.
Results showed that people that generally work more hours, plan more hours of work or work more than they would normally do seem to have a higher work engagement. From a logical standpoint, this could be explained by reversing causality: people that are generally more engaged would also be willing to work more. Also, this result was present for both people with a high and a low education, although the relation was stronger for people with a low education versus people with a high education. Intuitively, one would think that the relation would be stronger for highly educated people, because they would probably have more interesting jobs that would stimulate engagement more when working longer, but in this case it is the other way around. Why is still unclear.
during that day (through, for example, tiredness), as was argued in this paper, or a higher work engagement could lead people to work longer that day, but for both no support was found. It might be that these forces both work at the same time, in opposite directions, or that they prevail under different conditions, such as type of job. Maybe people with boring, monotonic jobs or with jobs they do not like, lose engagement as they have to work longer, while people with interesting jobs with a lot of variation or with jobs they love, tend to work more because they are engaged or get more engaged as they work more. To test this, in an explorative analysis, people’s level of education was used to proximate interestingness of jobs and added to the model as a moderator. Although evidence was weak, it was found that people with a low education that work more on a given day seem to have a lower work engagement on that day, while there is an indication that people with a high education that work more on a given day also have a higher work engagement on that day. This indicates that the relation between hours worked and work engagement could indeed be positive or negative, depending on the circumstances. In this case, it could be that people with a low education have, for example, a boring, monotonic job and their engagement with the work decreases as the working day is longer, possibly through repetitiveness. This is consistent with the findings of the Svartedalen experiment among nurses discussed in the introduction and literature review sections of this paper. Conversely, it could be that people with a high
education have, for example, a more interesting job with more variation and their engagement with the work increases as the working day is longer, possibly through a flow state.
For the relation between work engagement and productivity, the evidence is more clear cut. Higher work engagement is found to go together with higher performance, productivity and production. Firstly, people with higher work engagement generally seem to have a better performance, productivity and production. Intuitively, people who are more engaged with their work, will be more motivated to be effective, produce work of higher quality, produce more work per unit of time and consequently more work in total. However, the results in this paper provide no evidence for causality, so there is also a chance that people who are
effective, produce work of high quality, produce more work per unit of time and more work in total, feel more engaged with their work because of this.
could also be that after someone has had a day of work with above average performance, productivity and production, feelings of engagement are elevated in that person. Additionally, from these analyses it could be concluded that there was a positive relationship between hours worked and production, while there was no such relation between hours worked and
productivity (i.e., amount of work per time unit). As productivity is the amount of work people do per hour and production is the amount of work people do in total during a day, these findings could be explained by the fact that people that work more hours have more time to complete a total amount of tasks, but this does not mean that they also get more done per hour.
Implications
Theoretical implications are twofold. Firstly, there is evidence that indicates that the relation between work engagement and performance measures does not only exist at a general, between-person level, but that it also exists at a daily, within-person level. This opens up the possibility for research on day-level antecedents of work engagement and this may lead to very practical implications for, for example, employers. When employers know how to stimulate daily work engagement, they could indirectly increase daily performance, productivity and production.
Secondly, there is evidence that indicates that there is a negative relation between hours worked and work engagement for people with a lower education, and that this relation might be positive for people with a higher education. These findings could lead to further research on the existence of this relation, on other moderator variables in the work engagement – productivity relationship and possibly to extensive implications for the labour market in the future. If shorter working days lead to higher work engagement and subsequently a higher productivity among people with a lower education, it might be beneficial for both employers and employees to shorten working days for these people in the future.
As evidence is just hinting at the possible existence of these relationships, at this moment there are no direct practical implications resulting from this paper.
Limitations and future directions
variables. Especially for performance indicators, using a more objective measure would probably lead to more valid scores. Thirdly, the direction of causality of the found relations is unclear, as all variables were measured at the end of the working day. It would have been better to measure work engagement at the beginning of the day and indicators for
performance at the end of the day, to be able to determine the direction of causality for this relation. Fourthly, the study was executed within only one organization on a voluntary basis, so more research in other contexts and with a more randomized selection procedure would be needed for generalizability. Fifthly, the daily surveys were taken from Monday to Friday, but Saturday and Sunday are the days that people will often work just a few hours. It would be a good place to measure the effects of short working days. It might have been better to also send daily surveys on Saturday and Sunday, although the number of participants that
persevered until Friday already dropped significantly. Sixthly, the possibility for working zero hours on a given day was not taken into account when taking the surveys, such that people that did not work on a given day did not fill in the survey of that day, or if they did, they would get all kinds of irrelevant questions. It would have been better to consider this. Future research should focus on a number of things. To begin with, more research should be done on the relation between work engagement and performance on the within-person (daily) level to confirm the existence of this relationship and to determine the strength of this
relationship. Especially as it might open avenues for very practical research, such as how to increase employee performance on a daily level through stimulating daily work engagement. Moreover, future research should seek to provide further empirical support for the moderating role of education in the relationship between hours worked and work engagement, and for the sign and strength of this relationship under different levels of education. Also, research should look into why this relationship exists, as education could be just a proxy for another
moderator variable, such as certain job- or task characteristics (e.g., skill variety or job complexity). Furthermore, future research should aim to use a larger sample and more objective performance indicators. Subsequently, it would be wise to measure work
engagement at the beginning of the day and indicators for performance at the end of the day, to be able to determine a direction of causality between the two. Executing future research in varying domains and using randomized selection procedures would ensure a higher
Conclusion
6. REFERENCES
Ala-Mursula, L., Vahtera, J., Kouvonen, A., Väänänen, A., Linna, A., Pentti, J., & Kivimäki, M. (2006). Long hours in paid and domestic work and subsequent sickness absence: does control over daily working hours matter?.Occupational and environmental medicine,63(9), 608-616.
Aldana, S. G., Sutton, L. D., Jacobson, B. H., & Quirk, M. G. (1996). Relationships between leisure time physical activity and perceived stress.Perceptual and Motor skills,82(1), 315-321.
Anttila, T., Nätti, J., & Väisänen, M. (2005). The experiments of reduced working hours in finland: Impact on work–family interaction and the importance of the sociocultural
setting.Community, Work and Family,8(2), 187-209.
Bakker, A. B., Schaufeli, W. B., Leiter, M. P., & Taris, T. W. (2008). Work engagement: An emerging concept in occupational health psychology.Work & Stress,22(3), 187-200.
Barnett, R. C., & Gareis, K. C. (2000). Reduced-hours employment the relationship between difficulty of trade-offs and quality of life.Work and occupations,27(2), 168-187.
Benavides, F. G., Benach, J., Diez-Roux, A. V., & Roman, C. (2000). How do types of employment relate to health indicators? Findings from the Second European Survey on Working Conditions.Journal of Epidemiology and Community Health,54(7), 494-501. Booth, A. L., & Van Ours, J. C. (2008). Job satisfaction and family happiness: the part‐time work puzzle.The Economic Journal,118(526), F77-F99.
Booth, A., & Ravallion, M. (1993). Employment and length of the working week in a
unionized economy in which hours of work influence productivity. Economic Record, 69(4), 428-436.
Booth, A., & Schiantarelli, F. (1987). The employment effects of a shorter working week. Economica, 237-248.
Brayfield, A. H., & Rothe, H. F. (1951). An index of job satisfaction. Journal of applied psychology, 35(5), 307.
Cooper, C. L., Davidson, M. J., & Robinson, P. (1982). Stress in the police service.Journal of Occupational and Environmental Medicine,24(1), 30-36.
De Jonge, J., Landeweerd, J. A., & Van Breukelen, G. J. P. (1994). De Maastrichtse Autonomielijst: achtergrond, constructie en validering'[The Maastricht Autonomy
Questionnaire: background, construction and validation]. Gedrag en Organisatie, 7(1), 27-41.
Dembe, A. E. (2011). Factors shaping the development of working time regulation in the United States and Europe. International Labour Review, 150(3‐4), 419-429.
Evans, G. W., & Johnson, D. (2000). Stress and open-office noise. Journal of applied psychology, 85(5), 779.
Furda, 1995. Proefschrift, Universiteit Utrecht, Faculteit Sociale Wetenschappen. ??????
Golden, L. (2012). The Effects of Working Time on Productivity and Firm Performance, Research Synthesis Paper. International Labor Organization (ILO) Conditions of Work and Employment Series No. 33, Conditions of Work and Employment Branch, 2012.
Halbesleben, J. R., & Wheeler, A. R. (2008). The relative roles of engagement and
embeddedness in predicting job performance and intention to leave. Work & Stress, 22(3), 242-256.
Hallberg, U. E., & Schaufeli, W. B. (2006). “Same same” but different? Can work engagement be discriminated from job involvement and organizational
commitment?. European psychologist, 11(2), 119-127.
Hallberg, U. E., Johansson, G., & Schaufeli, W. B. (2007). Type A behavior and work situation: Associations with burnout and work engagement. Scandinavian Journal of
Psychology, 48(2), 135-142.
Henderson, J. C., & Lee, S. (1992). Managing I/S design teams: a control theories perspective. Management science, 38(6), 757-777.
Higgins, C., Duxbury, L., & Johnson, K. L. (2000). Part-time work for women: does it really help balance work and family?. Human Resource Management, 39(1), 17-32.
Independent (2016). Employers in Sweden introduce six-hour work day. Retrieved from:
Jungsun, P. A. R. K., Yangho, K., Chung, H. K., & Hisanaga, N. (2001). Long working hours and subjective fatigue symptoms.Industrial health,39(3), 250-254.
Kirkcaldy, B. D., Trimpop, R., & Cooper, C. L. (1997). Working hours, job stress, work satisfaction, and accident rates among medical practitioners and allied
personnel.International Journal of Stress Management,4(2), 79-87.
Latham, G. P., & Locke, E. A. (1975). Increasing productivity and decreasing time limits: A field replication of Parkinson's law. Journal of Applied Psychology, 60(4), 524.
Logan, N., O'Reilly, C. A., & Roberts, K. H. (1973). Job satisfaction among part-time and full-time employees.Journal of vocational behavior,3(1), 33-41.
Lorentzon, B. (2016). 23 månader med 6 timmar. Pacta Guideline, 1-74.
MacInnes, J. (2005). Work–life balance and the demand for reduction in working hours: evidence from the British Social Attitudes Survey 2002.British Journal of Industrial Relations,43(2), 273-295.
Mangan, J., & Steinke, J. (1988). Working‐time reductions: a survey of the Australian experience. Industrial Relations Journal, 19(4), 322-327.
Ohashi, I. (2005). Wages, hours of work and job satisfaction of retirement‐age workers.Japanese Economic Review,56(2), 188-209.
Pencavel, J. (2015). The productivity of working hours. The Economic Journal, 125(589), 2052-2076.
Pirinen, T., Kolho, K. L., Simola, P., Ashorn, M., & Aronen, E. T. (2010). Parent and self-report of sleep-problems and daytime tiredness among adolescents with inflammatory bowel disease and their population-based controls. Sleep, 33(11), 1487-1493.
Review of: White, M. (1987).Working hours: Assessing the potential for reduction. International Labour Office.
Rogers, A. E., Hwang, W. T., Scott, L. D., Aiken, L. H., & Dinges, D. F. (2004). The working hours of hospital staff nurses and patient safety.Health affairs,23(4), 202-212.
Schaufeli, W. B., Bakker, A. B., & Salanova, M. (2006). The measurement of work engagement with a short questionnaire: A cross-national study. Educational and psychological measurement, 66(4), 701-716.
Schaufeli, W. B., Shimazu, A., & Taris, T. W. (2009). Being driven to work excessively hard: The evaluation of a two-factor measure of workaholism in the Netherlands and Japan.
Cross-Cultural Research.
Schaufeli, W. B., Taris, T. W., Le Blanc, P., Peeters, M., Bakker, A. B., & De Jonge, J. (2001). Maakt arbeid gezond. Op zoek naar de bevlogen werknemer [Does work make happy, 422-428.
Sparks, K., Cooper, C., Fried, Y., & Shirom, A. (1997). The effects of hours of work on health: a meta‐analytic review.Journal of occupational and organizational psychology,70(4), 391-408.
Spector, P. E., Cooper, C. L., Poelmans, S., Allen, T. D., O'Driscoll, M. I. C. H. A. E. L., Sanchez, J. I., & Lu, L. (2004). A cross‐national comparative study of work‐family stressors, working hours, and well‐being: China and Latin America versus the Anglo world.Personnel Psychology,57(1), 119-142.
Spurgeon, A., Harrington, J. M., & Cooper, C. L. (1997). Health and safety problems associated with long working hours: a review of the current position.Occupational and environmental medicine,54(6), 367-375.
Tangen, S. (2002). Understanding the concept of productivity. In Proceedings of the 7th
Asia-Pacific Industrial Engineering and Management Systems Conference, Taipei (pp. 18-20).
Virtanen, M., Heikkilä, K., Jokela, M., Ferrie, J. E., Batty, G. D., Vahtera, J., & Kivimäki, M. (2012). Long working hours and coronary heart disease: a systematic review and
meta-analysis.American journal of epidemiology,176(7), 586-596.
White, M., Hill, S., McGovern, P., Mills, C., & Smeaton, D. (2003). ‘High‐
performance’Management Practices, Working Hours and Work–Life Balance.British Journal of Industrial Relations,41(2), 175-195.
Zeytinoglu, I. U., Lillevik, W., Seaton, M. B., & Moruz, J. (2004). Part-time and casual work in retail trade: stress and other factors affecting the workplace. Relations
