Change in Job Satisfaction for West-German
Workers: A Cause for Concern?
Rosan van Niekerk
Amsterdam, 15 May 2017
Student number: 11010797 Contact: rosanvann@gmail.com Supervisor: dr. Thomas Leopold
Second reader: dr. Stephanie Steinmetz University of Amsterdam
1
Change in Job Satisfaction for West-German Workers: A
Cause for Concern?
To date, it remains unresolved how job satisfaction levels change over time. Scholars who studied generational and age related differences in job satisfaction often relied on inadequate data and methodology. As a result, they could not distinguish between age and cohort effects. By making use of 31 yearly waves from the German Socio-Economic Panel Study covering the period 1984-2014, we were able to disentangle age and cohort effects on job satisfaction. Two research questions are answered by analyzing job satisfaction of West-German workers with a multilevel model. First, we explore how job satisfaction changes across the life course. Second, we explore how job satisfaction changes for successive birth cohorts. Contrary to existing knowledge, we find a negative age-job satisfaction relationship and a negative cohort effect. Not only do people become less satisfied with their job as they age, but recent birth cohorts also report considerable lower levels of job satisfaction compared to older birth cohorts.
Keywords: Job satisfaction, Age, Cohort, SOEP
Introduction
Job satisfaction is the most studied employee attitude in organizational fields (Spector, 1997; Judge et al., 2017). The majority of these studies aim to understand the causes and consequences of this work attitude. Studies addressing long-term developments in job satisfaction are more scarce and often lack proper data and methodology (Jürges, 2003; NG & Feldman, 2010; Riza et al., 2015). In this study, we investigate how job satisfaction changes across the life course and how these life course patterns differ for successive birth cohorts. Ageing and cohort succession are the processes that underlie shifts and trends in time (Riley, 1973). Researchers who tried to establish a relationship between age and job satisfaction mainly relied on cross-sectional data. However, if age differences in job satisfaction are measured at one point in time, the age effect could also be due to cohort differences. In this case, workers are more or less satisfied with their job because they are born in different times. Scholars who studied generational differences in work values and job satisfaction ran into a similar problem. The results of these type of analyses reflect a combination of generational differences and age differences. The key contribution of this study is our use of a
2
unique scope of panel data, which allows for more accurate conclusions on generational and age related differences in job satisfaction.
This paper is centered around two main research questions. First, we explore how job satisfaction changes over the individual life course. As people age, do they become more or less satisfied with their job? Second, we explore how job satisfaction changes for successive birth cohorts. Do recent birth cohorts experience similar levels of job satisfaction or can we point toward social change?
To answer these questions we will draw on the German Socio-Economic Panel Study (SOEP). This longitudinal dataset covers the population of Germany in the period 1984-2014. Job satisfaction is measured in each of the 31 waves. By using this type of data we can analyze change in job satisfaction as people age, covering major parts of the life course and a large range of birth cohorts. A multilevel regression analysis enables us to separately analyze age and cohort effects on job satisfaction. In addition, we can map job satisfaction levels at one particular age for several birth cohorts. This creates the possibility to explore social change in job satisfaction.
Satisfaction with the work you do is crucial for individual as well as organizational well-being and productivity (Spector, 1997). We add to the literature a precise description on how change in job satisfaction levels comes about. Accurately describing change in job satisfaction levels is an important condition for understanding and potentially explaining the phenomenon. Contrary to current knowledge, we demonstrate that job satisfaction declines as people age. In addition, we observe that recent birth cohorts are less satisfied with their job compared to older birth cohorts.
The paper is structured as follows. First, we show the variety in approaches to the concept of job satisfaction and discuss the drivers and consequences that have been studied. Second, we critically assess existing literature on the age-job satisfaction relationship and on generational differences in job satisfaction. Next, we formulate theoretical expectations on the age and the cohort effect and discuss the sample selection and the analytical approach. Finally, we present the results of the analysis and discuss the implications of our findings for the literature on generational and age related differences in job satisfaction.
3
Theoretical Background
The concept of job satisfaction
Job satisfaction has been of interest to scholars from a variety of disciplines like organizational psychology, social psychology, human resource management, sociology and economics. There is an overwhelming amount of research into the field of job satisfaction, and the concept requires a brief introduction.
In general, there are two definitions of job satisfaction: affective and cognitive job satisfaction. Affective satisfaction refers to the overall, global emotional response towards the job and is often measured with a single-item on a 5-,7- or 10-points scale. Cognitive satisfaction refers to the rational evaluation of different job conditions (Thompson & Phua, 2012). Cognitive satisfaction is sometimes referred to as the facet approach. Facets of the job are for instance pay, promotion opportunities, autonomy, job security, coworker support and so on (Spector 1997). The single-item measure is useful for measuring change in job satisfaction (Besen et al., 2013). However, there is disagreement on how global and facet job satisfaction relate to each other (Spector, 1997; Riza et al., 2015). Research shows that a global attitude is distinct from attitudes on aspects of the job (Jude et al., 2017). As a result, we do not know precisely how evaluations about individual job facets enter into the overall evaluation of the job (Weiss, 2002). Finally, many scholars cite Locke (1969) if they formulate a definition of the concept. According to Locke, job satisfaction results out of the subjective discrepancy between want you want in a job and what you have in a job. His definition is often used by economists (Judge et al., 2017).
What determines the level of satisfaction with your job? Scholars classify the literature into two perspectives. The situational or environmental approach dominated until the 1980s (Spector 1997). Within this approach researchers assume that job satisfaction is mainly caused by external factors like job characteristics. The most influential theory is the Job Characteristics Model proposed by Hackman & Oldham in 1980. This model explains how job characteristics, for example skill variety and autonomy in the job, create psychological states which in turn affect job satisfaction and motivation (Fried & Ferris, 1987).
Within the second approach, the dispositional approach, researchers assume that individuals differ in their tendency to be satisfied. Personality and individual traits are the factors that determine job satisfaction. Examples of individuals traits are negative affectivity and locus of control. The latter indicates the degree to which you believe you can control
4
events in your life (Bruk-Lee et al., 2009). To integrate the situational and the dispositional approach, researchers started to test models where job characteristics mediate the relationship between personality and job satisfaction (Judge et al., 2001).
Next to the drivers, the consequences of job satisfaction are extensively studied. Job satisfaction is related to overall life satisfaction and happiness but also to economic productivity, performance, job mobility and turnover or quit rates (Green, 2006; Rain et al., 1991; Judge et al., 2001; Judge et al., 2017; Sousa-Poza & Sousa-Poza, 2000). Economists entered the field of job satisfaction quite late, partly because of their distrust of self-reported survey evidence as it cannot be used to make interpersonal judgements about well-being (Blanchflower & Oswald, 1999; Green, 2006; Hamermesh, 2001; Sousa-Poza & Sousa-Poza, 2000). The main interest of economists is in observable phenomena, so developments in the ‘objective’ quality of work. Economic scholars thus examined how job satisfaction is related to other objective job conditions like wage inequality or take job satisfaction as a measure of job quality or utility (Bustillo Llorente & Macías, 2005; Hamermesh, 2001; Green, 2006). However, the latter is argued to be intrinsically problematic (Green, 2006). Bustillo Llorente and Macías (2005) find that variation in job characteristics do not explain variation in job satisfaction levels. They argue that this is not surprising because there is no reason to assume a lasting ‘lack of fit’ between work expectations and job characteristics, which should cause lower job satisfaction. First, persons look for a job which is in accordance with their expectations. Second, people adopt psychological coping mechanisms by adapting expectations and perceptions to their situation. Third, if the first two mechanism do not work, people will leave the job. They conclude: ‘If this were so, job satisfaction would be a variable of little interest to
sociology and economics, since it would not depend on social, exogenous, variables, but rather on subjective, endogenous processes and mechanisms’ (De Bustillo Llorente and Macías, 2005 p. 671).
Accepting the assumption that job satisfaction is mainly the result of psychological subjective mechanisms, renders explaining change in job satisfaction over time problematic. Before we turn to the literature on generational and age related differences in job satisfaction, we will first discuss important considerations if one wants to analyse change in job satisfaction levels.
Change in job satisfaction: Age-Period-Cohort effects
Studies that assess (cross-)national trends in job satisfaction levels show remarkable high levels of job satisfaction in many countries (Sousa-Poza & Sousa-Poza, 2000). Though there is relatively little variation in job satisfaction over time, some countries do show change in job
5
satisfaction levels. Blanchflower & Oswald (1999) find that job satisfaction is slowly declining in the United Sates from 1970 to 1990. In both Britain and Germany a similar decline is found (Green & Tsitsianis, 2005). A cross-national study of job satisfaction for 21 countries confirms these results and finds a decline in job satisfaction in Norway in the 1990s. Interestingly, this study also shows that job satisfaction is based on comparisons with the own immediate environment. For instance, the proportion of Russian workers that considered their jobs well-paid is higher than in Great Britain and Switzerland, while objective payment conditions would be considered as worse (Sousa-Poza & Sousa-Poza, 2000).
Studying attitudinal change over time can be fraud with interpretative and methodological fallacies as age, cohort and period processes occur next to each other (Riley, 1973). When studying long-term developments in job satisfaction, it is crucial to understand the differences between these three processes. Age effects are reflected in psychosocial and biological aging that can influence your work attitude (Rhodes, 1983). For instance, this can happen through a change in your expectations and needs when you grow older or the ability to have children. A cohort effect refers to a group of persons born around the same time who age together (Rhodes, 1983). In social sciences there have been many attempts to interpret cohort or generational effects in a meaningful way (Gilleard, 2004). Overall, birth cohorts are seen to have some shared traits, dispositions, behaviors and life experiences which are more or less stable over the life course and can potentially influence work attitudes (Rhodes, 1983; Gilleard, 2004). The general definition of the period effect is an event that occurs which affects everyone at the same point in time regardless of their age, for instance an environmental disaster.
Each of the empirical measures of age, birth and time is completely defined in terms of the other two (Riley, 1973). This ‘identification problem’ explains why it is not possible to model age, period and cohort effects simultaneously. If you have information on two aspects, you will always know the third. Age, birth and date of observation will thus be perfectly collinear. Therefore, one of the effects must be constrained on theoretical grounds. In the context of job satisfaction it can be argued that there is no external event that affects everyone at the same time independent of their age. The economic crisis, unemployment spells in 1990s or the dualization of the German labor market will certainly have had an effect on work conditions, but these effects will be different for different groups in society. Graduates entering the labor market will experience more difficulties during high unemployment spells compared to employees in their forties working on a permanent contract.
6
Previous research on trends in job satisfaction showed a decline in job satisfaction levels for Germany, especially in the 1990s (Green & Tsitsianis, 2005; Poza & Sousa-Poza, 2000). However, these trend studies cannot show the underlying dynamics of change across the life course and cohort succession.
We will first discuss existing literature on age and generational differences in job satisfaction. Informed by empirical and theoretical perspectives we end these sections by formulating hypotheses on the age and the cohort effects. Next, we present the methodological strategy that aims to describe change in job satisfaction through ageing and cohort succession.
The age-job satisfaction relationship
There is a considerable amount of studies that attempts to establish a relationship between age and job satisfaction (Besen et al., 2013). Studies that focus on providing empirical evidence for an age-job satisfaction relationship are abundant, while studies informed by comprehensive theoretical rationales are less widespread.
The results of these studies show an increase in job satisfaction over the life course with older workers reporting higher levels of satisfaction (Bernal et al., 1998; Besen et al., 2013; Chaudhuri et al., 2015; Rhodes, 1983; Riza et al., 2015; NG & Feldman, 2010). Furthermore, much attention has been paid to the form of the age-job satisfaction relationship, namely whether this relationship is linear or non-linear. Findings turned out to be inconsistent. Some studies found support for a linear relationship while others found support for a U- or J-shaped relationship. A U- or J-J-shaped relationship indicates that younger as well as older workers would report higher levels of job satisfaction (Clark et al., 1996; Hochwarter et al., 2001).
The majority of the research on the age-job satisfaction relationship adopts a hierarchical polynomial regression analysis on cross-sectional data. The use of cross-sectional data is highly problematic as it prevents scholars from disentangling age and cohort effects (Jürges, 2003). Even if the age-term has a large and significant effect on job satisfaction, this effect could still be biased by birth cohort effects. In order to examine true age-effects it is crucial to use longitudinal designs which track individuals over time (NG & Feldman, 2010).
Though there is extensive empirical research on age differences in work attitudes and job satisfaction, a systematic theoretical framework explaining the age-job satisfaction relationship is lacking (Kackmar & Ferris, 1989; Rhodes, 1983; NG & Feldman, 2010). As job
7
conditions partly depend on your career stage and your career stage is closely related to age, scholars made an attempt to link career stage theory to the age-job satisfaction relationship. First, as people proceed in their careers objective job conditions may change. In general, later stages in the career are associated with jobs with more status, power and prestige, increased promotion opportunities and better rewards in terms of pay and pension. Second, at later stages in the career you should expect a better match between the personal characteristics of the worker and the job as employees modify their skills, gain experience and have had the opportunity to search for the best-fitting job (Hochwarter et al., 2001; Kackmar & Ferris, 1989; Rhodes, 1983; Riza et al., 2015; White & Spector, 1987). A recent study into job satisfaction and its relation to time showed that job satisfaction increases with age but decreases as tenure in a given organization advances. But if workers enter a new organization, they experience a boost in job satisfaction (Riza et al., 2015).
A second approach tries to explain the age-job satisfaction relationship by relating personality developments across the life course to work attitudes (Besen et al., 2013; NG & Feldman, 2010). Socio-emotional selectivity theory states that as individuals age, they maximize their social and emotional gain. Conversely, for younger adults the primary goal is to gain knowledge. Older people are more likely to attend to and to recall positive information. They experience more positive emotions because they have a different perception of how long they will live. According to NG & Feldman (2010) this theory can explain their meta-analytic finding that age is positively related to job satisfaction. White & Spector (1987) argue that the personality trait of internal-external locus of control is related to age. As people age, their belief in personal control is enhanced which results in greater satisfaction. Both career stage theory and socio-emotional selectivity theory would thus imply an increase in job satisfaction across the life course.
A third approach integrates these two perspectives by examining how the relationship between job conditions, personality and job satisfaction may vary by age (Rhodes, 1983; Besen et al., 2013). Age may moderate the relationship between job conditions and job satisfaction as different work conditions can be valued in certain age stages. In addition, the relationship between personality and job satisfaction may become stronger across the life course. As people age, workers with higher levels of core self-evaluation (the fundamental evaluations that individuals hold about themselves) move into the better jobs (Judge & Bono, 2001). Evidence suggest that the strength of relationship between job characteristics and job satisfaction is indeed related to age. Besen et al. (2013) find that skill variety, autonomy and
8
friendship have a weaker effect on satisfaction as employee age increased. Though it is reasonable to expect that age acts as a moderator, the implications for the direction of the age-job satisfaction relationship are not clear.
Finally, it is argued that workers enter the job with high expectations, aspirations and idealistic work values. However, the everyday reality of the job can be disappointing and frustrating. As persons age and gain more work experience, they move to more realistic expectations and aspirations resulting in a rise in their job satisfaction levels. Rhodes (1983) refers to ‘cognitive distortion processes’ which account for the increase in job satisfaction as workers age. In this argument it is a subjective defense mechanism that causes higher job satisfaction levels instead of a change in objective job conditions, which is the case in career stage theory.
Clark et al. (1996) and Hochwarter et al. (2001) provide a theoretical account for a J- or U-shaped relationship between age and job satisfaction. Employees who enter their first jobs feel positive about their novel situation and have high expectations. At the same time, they also have a limited understanding of what makes a job a satisfactory job. While receiving salary may cause high job satisfaction in the early years, wrong expectations and increased boredom may result in a subsequent downward trend in job satisfaction. After a while, a ‘person comes to terms with his or her occupational role’ (Clark et al., 1996 p. 58) and perhaps moves into more rewarding positions, which results in an increase in job satisfaction. Furthermore, research shows that work-family conflict has a significant negative impact on job satisfaction. Around middle-age, parents with young children could therefore be less satisfied with their work (Grandey et al., 2005). This would contribute to a U-or J-shaped relationship between age and job satisfaction.
Taken together, empirical results, career stage and socio-emotional selectivity theory and the move towards realistic expectations of the job, all suggest that job satisfaction will be higher for older people. Furthermore, there are several arguments why workers should experience a drop in job satisfaction after their first years in the labor market, followed again by an increase into later life. Building on this literature, we hypothesize that job satisfaction levels are higher for older people.
Most studies we discussed make use of cross-sectional data. Higher job satisfaction for older workers can therefore not only reflect an increase in job satisfaction across the life course, but also lower job satisfaction for newly entering birth cohorts. Interestingly, if there is a decrease in job satisfaction across the life course combined with a decrease in job
9
satisfaction for the more recent birth cohorts, researchers working with cross-sectional data would conclude that there are no age effects at all. Again, this highlights the necessity of using panel data if one wants examine age-differences in work attitudes.
Job satisfaction across successive cohorts
While there is an extensive amount of research into age related differences in job satisfaction, studies that focus on cohort differences in particular are scarce. Cohort effects and corresponding explanations are briefly mentioned and often framed as a problematic factor in studying the age-job satisfaction relationship. However, in the field of human resource management generational differences are a widely debated topic. Baby Boomers (born between 1945-1965), Generation X (born between 1965-1980) and Generation Y (born between 1980 and 1990) are seen to have different work values and beliefs which challenges managers in effectively managing their workforce (Benson & Brown, 2011). However, these ideas mainly reside in more popular human resource management literature. Scientific articles providing empirical evidence for generational differences in work values are less widespread (Giancola, 2006). In addition, scientific studies that do try to test for generational differences adopt statistical analyses on cross-sectional data or short spells of longitudinal data (Benson & Brown, 2011; Krahn & Galambos, 2014). Consequently, they cannot properly distinguish between age and cohort effects.
In what follows, we divide theoretical arguments for cohort effects on job satisfaction between ‘cohort replacement’ and ‘diffusion’. Cohort replacement refers to a situation where job satisfaction levels change because new birth cohorts with specific characteristics replace older birth cohorts in the labor market. Diffusion happens when external circumstances like changes in job conditions affect a certain age groups or, similarly, persons belonging to a certain birth cohort.
Cohort replacement
White & Spector (1987) hypothesize that a contradiction between the needs of the system and the values of new workers can create dissatisfaction with the job. They support their argument with an example of General Motors that opened a new manufacturing facility and mainly hired workers in their twenties. This led to major worker unrest in this particular facility. In his work on global change in societal values over time, Ingelhart (1990) finds that older generations are more focused on material achievement and consumption while younger
10
generations are focused on postmodern values like life satisfaction and trust (NG & Feldman, 2010). Spector (1996) argues that older cohorts are more accepting of authority and expect less from their jobs. While these are plausible theories, the direct mechanisms through which different values of younger cohorts would affect job satisfaction levels remains unclear.
Research in the field of human resource management focuses in particular on generational differences in orientations, motivation and attitudes towards work. Some studies mainly find similarities across generations. Other studies find differences between generations with regard to motivation and preferred job characteristics (Appelbaum, 2005; Besen et al., 2013; Krahn & Galambos, 2014). Smola & Sutton (2002) state most firmly that generational experiences are of significant importance. Compared to the Baby Boomers, Generation X seems to move away from company loyalty and self-worth through one’s job. They have become less convinced that work should be an important part of one’s life and they seek more balance in their lives (Appelbaum, 2005; Smola & Sutton, 2002).
Even if generations have different work orientations, it is important to remember that only a discrepancy between work orientations and job conditions would result in lower job satisfaction (and not the changing values as such). However, labor market and job conditions change at the same time. This makes it a complex exercise to formulate theoretical expectations on generational differences in work orientations and its impact on job satisfaction. Studies that compared job satisfaction levels across generations find that Boomers were found to be significantly more satisfied with their job compared to Generation X or that results are inconclusive (Appelbaum, 2005; Benson & Brown, 2011).
Scholars studying these generational differences in work values use designs that cannot really provide an answer to their research question. Adopting a t-test on cross-sectional data to test for significant differences between birth cohorts does not rule out the influence of age. The study of Krahn & Galambos (2014) did make use of a longitudinal design, but tracked respondents only for seven years at the start of their career.
Another way in which cohort replacement can result in change in job satisfaction levels is through changes in cohort composition. The increased participation of women in the labor market and the expansion of higher education are two major societal transformations of the last decades. Consequently, the younger cohorts will have a higher share of female and high educated workers. Different job satisfaction levels of the more recent cohorts could partly be the result of this compositional effect. Indeed, gender and educational level are two demographic variables that seem to relate to job satisfaction. However, the exact relationship
11
is not straightforward. The higher educated have jobs with characteristics that are associated with higher job satisfaction, but the higher expectations of the better educated can ‘cancel out’ the effect of good job conditions or result in disappointment if they have to perform more routine tasks. By contrast, lower educated workers may have lower standards for what comprises a satisfactory job (NG & Feldman, 2010; Rhodes, 1983). With regard to gender, some studies mention the ‘gender paradox’. Women report high levels of job satisfaction, while they have less favorable work conditions and experience more obstacles. However, women’s disadvantaged position can result in lower expectations which in the end leads to higher satisfaction. Some scholars argue that because of the increased labor market participation of women, women in younger birth cohorts should have more equal expectations compared to men. This should make the gender paradox disappear over time (NG & Feldman, 2010; Souza-Poza & Sousa-Poza, 2003).
Finally, generations with high birth rates face stronger competition for the more desirable jobs. In Germany there is a drop in birth rates since the 1965. Cohorts born since 1965 entering the labor market around 1985 should thus experience less competition for available jobs, which could imply better self-selection into jobs they prefer, having a positive impact on their job satisfaction levels. However, in the 1980s and especially the 1990s unemployment levels in Germany were rising. The consequences of lower birth rates are therefore not evident.
In short, birth cohorts can differ in their job satisfaction level because older cohorts are replaced by younger cohorts who have higher and more idealistic expectations, different work values and orientations and a different composition in terms of gender and educational level. With regard to these cohort replacement arguments, the precise direction of the effect of belonging to one of the more recent cohorts is not easy to define. Therefore, we do not put forward any directional hypotheses on the difference in job satisfaction levels through cohort replacement.
Diffusion
Job satisfaction levels can also differ between birth cohorts because external circumstances affect certain age groups more than others. There are many ways in which job satisfaction could be affected through outside developments. In this section, we limit ourselves to the most important factors, namely the potential impact of economic conditions, labor market
12
conditions, school-to-work linkages and developments in the character and the quality of work.
Unemployment spells have a strong impact on the labor market, especially for new labor market entrants. Germany experienced a rise in unemployment since the 1980s which reached high levels in the 1990s (Sywottek, 2001; Dustmann et al., 2014). In 2005, the unemployment rate was 11%, from which it started to decline again (Krause & Uhlig, 2012). Consequently, birth cohorts who entered the labor market in the 1990s and around the new millennium experienced increased competition at the start of their career. According to Krahn & Galambos (2014), in a society with high unemployment rates and an expansion of high-educated workers who have a strong belief that they are entitled to a good job, there will be plenty of unmet occupational aspirations. This can even impact mental health conditions. In addition, the scholars find that long periods of unemployment made Canadian high school seniors downscale their employment ambitions by placing less value on intrinsic work values. In Germany in 2000, high educated labor market entrants had declining employment opportunities in top-paying and analytical jobs and entry wages had fallen (Reinhold & Thomson, 2016). The German Baby Boom generation entered the workforce during the 1960s and the 1970s in more standard and safe economic conditions. There was a widespread belief that, in contrast to the generation of their parents, their position was the result of choice instead of class-fate (Wierling, 2001). The older cohorts born between 1925 and 1945 lived through the second world war and shortly afterwards experienced cold and hunger. However, they also lived through increasing prosperity, especially in the 1970’s when pay increased, in particular for civil servants, and the living standards for families were rising (Sywottek, 2001). Taken together, it could be argued that for these older cohorts and the Baby Boom generation, economic and societal circumstances created favorable conditions for high satisfaction with the job. Conversely, for the more recent cohorts (born between 1970 and 1985) who entered the labor market around 1990 and onwards, high unemployment rates and increasing school-to-work mismatches could have negatively affected their job satisfaction level.
While these younger cohorts faced high unemployment when they entered the labor market, the characteristics of their jobs also changed compared to the work conditions of earlier birth cohorts. In general, jobs became more skilled and complex which created more opportunities and possibilities for learning (Mustosmäki et al., 2011). There is more room for workers initiative, influence and participation. In addition, the majority of the workers benefited from increased wages and work conditions became more safe and pleasant (Green,
13
2006). According to the Job Characteristics Model discussed in the beginning of this paper, people can be motivated by the intrinsic satisfaction they find in job tasks. More complex jobs are thus associated with high job satisfaction, motivation and performance for people who prefer challenge and interest in their work (Spector, 1997). This transformation of work tasks and conditions could result in higher levels of job satisfaction for these younger cohorts.
However, the literature on the quality of work that is describing a deterioration instead of an improvement of job conditions is more dominant. According to this literature, a deterioration of job conditions has taken place roughly since the 1980s into the 1990s as a consequence of macro-economic trends of globalization, technological change, liberalization of markets, declining unions and the rise of the service industry (Blanchflower & Oswald, 1999; Green, 2006; Mustosmäki et al., 2011; Smola & Sutton, 2002). These trends led to outsourcing, downsizing and the growth of non-standard work. It changed the nature of work in such a way that there are considerable negative consequences for the quality of jobs.
The new ‘knowledge economy’ consists of increasingly complex jobs which require high skills. While it is argued that increased complexity can create meaningful jobs, it is also argued that it makes jobs more demanding in terms of time and energy creating heavy work-life imbalances. Moreover, there is some evidence for skill polarization (Oesch and Menés, 2011). This not only implies a growth in high-skilled jobs, but also a growth in low-skilled (and unfulfilling) jobs. Related to the changing complexity of jobs is the intensification of work efforts and work-related stress. Stress is one of the most important and growing occupational health problem (Green, 2006). New forms of work organization often entail pressure from customers or patients and pay is increasingly linked to work-effort (Mustosmäki et al., 2011). In addition, a loss in task discretion, which is the autonomy to determine and choose your own activities, appears to be detrimental for job satisfaction (Mustosmäki et al., 2011). However, there is disagreement on the exact developments in workers discretion. On the one hand, workers are empowered as they are allocated more responsibilities and autonomy. On the other hand, there is stricter supervisory and technical control by the employer (Green, 2006; Mustosmäki et al., 2011). Furthermore, increased job insecurity is seen as one of the most important aspects that negatively affected the nature of work (Green & Tsitsianis, 2005; Mustosmäki et al., 2011; Olsen et al., 2010). According to Green, publishing his book in 2006, there has been no objective dramatic decline in the stability of jobs but perceptions of the risk of job loss increased, including in Germany. By now it seems evident that in many industrialized countries there is a growth of flexibility and non-standard work which leads to a growing
14
insecurity in work. Downsizing has become more regular and companies hire temporary employees or make use of subcontracts. Organizations increasingly treat employees as a means to an end which damages company loyalty (Smola & Sutton, 2011). In Germany, the growing availability of non-standard work increased pressure on workers to accept more flexibility (Eichhorst & Marx, 2011). Non-standard employment is associated with negative outcomes, including lower job satisfaction (Mustosmäki et al., 2011). Finally, the fairness of wages, especially between similar type of jobs, is argued to affect job quality and job satisfaction (Green, 2006; Spector, 1997). Hamermesh (2001) finds for the U.S. that recent shocks of increased income inequality negatively affected job satisfaction, but that the relationship between persistent inequality and job satisfaction is weak. In Germany income inequality increased in the 1990s and at the start of the new millennium, with declining wages especially at the bottom of the income distribution (Burda, 2016; Dustmann et al., 2014).
Olsen et al. (2011) compared the perceptions of workers on intrinsic and extrinsic job conditions in Germany at three points in time; 1989, 1995 and 2005. There appears to be an improvement in autonomy and interesting work, but a deterioration in perceived job insecurity, income level, advancement opportunities, heavy physical work and danger of work. How could the changing nature of work have influenced job satisfaction of workers? Probably, it did not affect workers of all ages in a similar way and to the same degree. Primarily vulnerable to the deterioration of job quality are the new labor market entrants. The cohorts who entered the labor market during the 1990s and in the new millennium have to cope with jobs with worse conditions like increased insecurity and an intensification of work efforts. Older workers from previous birth cohorts may already have secured their stable contract and position in the labor market. The younger cohorts could therefore experience lower levels of job satisfaction compared to cohorts who entered the labor market during times with more favorable job conditions.
In conclusion, unemployment spells, school-to-work mismatches and the deterioration of job quality could negatively affect job satisfaction levels. This would especially impact the younger cohorts born around 1970 and onwards. These more recent cohorts have less possibilities to self-select into jobs they like and have to cope with jobs with worse conditions. Therefore, we anticipate that these younger birth cohorts report lower job satisfaction than the older birth cohorts.
15
Synthesis
Several matters addressed in this paper complicate the assumption of job satisfaction reflecting changes in job quality. First, as discussed in the section on the concept of job satisfaction, some scholars argue that job satisfaction does not reflect job quality because it is mainly the result of psychological mechanisms rather than economic or societal change. Second, seeing changes in job satisfaction as reflecting trends in work quality relies on the assumption that work values against which satisfaction is measured do not change significantly (Green, 2006). However, there is some evidence for generational differences in work values and norms that could affect the evaluation of job conditions and thus job satisfaction. Finally, we do not know precisely how and when evaluations about individual job components enter into the overall evaluation of the job (Weiss, 2002). As a result, dissatisfaction with aspects of the job that changed over time, such as managerial supervision, may not be reflected in a single-item measurement of job satisfaction.
While being aware of these limitations when studying change in job satisfaction, we formulated the following expectations. First, we hypothesize that job satisfaction levels are higher for older people. Second, we expect that younger birth cohorts born around 1970 and onwards report lower levels of job satisfaction than older birth cohort.
To empirically test our expectations we have to decide on the right metrics of time. If we want to contribute to the empirical and theoretical literature on the relationship between age and job satisfaction, it logically follows that age is the best metric to measure every year increase. Yet, some theoretical arguments, especially with regards to the relation between age and career stages, would fit a metric of years of tenure since labor market entry. Recent findings also show that age and tenure in an organization have opposite relationships with job satisfaction. Therefore, we should be aware that we cannot automatically consider age as interchangeable with tenure in the labor market or in an organization.
To interpret change in job satisfaction in relation to the mechanisms of cohort replacement, birth cohorts would be the preferred measure to include in the analysis. For interpretation of the diffusion arguments, the construction of labor market entry cohorts could be useful. Using a variable of labor market entry cohorts could facilitate a comparison between respondents entering the German labor market during prosperous economic times or during unemployment spells. However, the interpretation of a model which combines age with a measure of labor market entry cohorts is overly complicated. Moreover, in the German Socio-Economic Panel Study information on labor market entry is not available for every
16
respondent, so half of the potential sample size would be omitted. Based on theoretical and empirical arguments we thus decide to adopt time metrics of age and birth cohorts in the analysis. In what follows, we will first discuss the sample selection and the model strategies. We then turn to the presentation of the results.
Method
Data and sample
To perform the analysis we used data from the German Socio-Economic Panel Study (SOEP, data for years 1984-2014, version 31; Wagner et al., 2007). The SOEP provides 31 annual waves of panel data for the period between 1984 and 2014, which makes it an ideal dataset to answer the research questions. It contains information on individuals and private households, covering a broad range of topics like employment, health and life satisfaction.
As the SOEP contains several subsamples, we first selected the relevant refreshment samples of 1998, 2000, 2006, 2011 and 2012 and excluded several oversamples (N = 48,221). Next, we decided to focus only on men and women who lived in West-Germany in 1989 and excluded individuals with a direct and indirect migration background. With this we aim to generate a coherent group which is more homogenous in terms of cultural values, historic experiences and socio-economic circumstances, which makes interpretation of the results more evident. Furthermore, we excluded people younger than 20 and older than 65 to focus on the general working population. Next, we selected respondents who are employed, being full-time, part-time or irregular employment. We also excluded respondents who were still in school or had missing values on the educational level. After these exclusions the sample consisted of 16,798 respondents and 134,413 observations (as with panel data respondents are observed multiple times). Next, missing values for the outcome variable job satisfaction were dropped. This results in a sample of 16,311 respondents and 129,940 observations. Finally, we created a categorical birth cohort variable, starting at the year 1925 and cut at every ten year. Respondents born before 1925 were dropped from the sample. This is crucial as every birth cohort category needs to have a sufficient amount of observations to generate reliable estimates. The final sample consists of 8553 men and 7622 women born between 1925 and 1989. The total amount of observations is 129623.
Job satisfaction is measured as a single-item construct. Respondents had to indicate their satisfaction with work on a scale from 0 to 10, 10 being the highest level of satisfaction.
17
Descriptive statistics indicate that the average job satisfaction level is relatively high. For this sample, job satisfaction has a mean of 7.2 with a standard deviation of 2.
Table 1. Sample selection
Starting Sample Total (N = 48,221)
Inclusion criteria Total N % (Total: 100)
Lived in West Germany in 1989 29258 60.7
No migration background 25995 53,9 Aged 20-65 21224 44 Employed 17048 35,4 Educational level 16798 34,8 Job satisfaction 16311 33,8 Birth cohorts 16175 33,5
Source: German Socio-Economic Panel Study (SOEP), 1984-2014, version 31.
Models
To analyze age and cohort effects on job satisfaction with panel data, a two-level multilevel model is applied. In these models, Level 1 observations are nested in Level 2 individuals. This means that every respondent has multiple observations, depending on how many years the respondent participated in the panel. We adopt a generalized least squares random effects estimator. This enables us to estimate change in job satisfaction as individuals age, while simultaneously estimating change in job satisfaction between individuals across cohorts. Job satisfaction is the dependent variable of the multilevel model and age and birth year are added as independent variables. Crucial in the construction of the model is to test for multiple functional forms of age. As discussed in the background section, a large part of the literature on the age-job satisfaction relationship is dedicated to research into the form of the age-job satisfaction relationship. Similarly, the effect of birth year on job satisfaction levels could differ in direction and strength across all birth years. We adopted a strategy where we go from simple to complex. First, a model with a linear age term and a linear birth year term was estimated. Step by step age was included as a polynomial from squared, cubic, quartic etc.
18
up to sceptic. The same procedure was followed for birth year. In addition, birth year was added to the models as a categorical birth cohort variable.
We selected the best fitting model based on the overall R-square, the significance of the age, birth year and birth cohort coefficients and the comparison of job satisfaction patterns by visual inspection of graphs. First, the descriptive life course patterns of job satisfaction were plotted for every successive cohort. By overlaying the descriptive pattern with the model estimated pattern, it is possible to visually inspect where the model deviates from the descriptive trend in terms of age and cohort patterns and the absolute level of job satisfaction. Finally, we added interactions between several polynomials of age and birth year to see if and in what way it would improve various models. Examining interactions between age and cohort effects is important as the life course pattern of job satisfaction can differ for successive birth cohorts. It turned out that none of the interactions led to a considerable improvement for any of the models.
We found evidence for a model with a quartic age effect and a categorical ten year birth cohort effect. This model was preferred over models with birth year as a continuous variable and models with several other polynomials of the age and birth year terms. Our final model includes a quartic age term which allows for three bends in the age-job satisfaction relationship. For the cohort variable, we defined the 1985-1989 cohort as the baseline category. Compared to the youngest birth cohort, successive older cohorts differ significantly in job satisfaction level.
Finally, we tested whether there were differences in age and cohort effects between males and females and between low, intermediate and higher educated respondents. As discussed in the background section, the participation of women in the labor market and the expansion of higher education are two major societal transformations that affect the composition of the cohorts in this study. We are interested in what happens to the age and cohort effects we observed initially, when we take into account gender and educational level in our model. To explore this, we first added gender and educational level separately to the model. Second, we simultaneously added gender and educational level to the model. Finally, we added interactions between gender and several polynomials of the age term and between gender and birth cohort, as the effect of gender on job satisfaction could differ for certain age stages or birth cohorts. The same strategy was adopted for educational level.
When studying change over time, it is crucial to consider the identification problem. Because each of the empirical measures of age, birth and time is completely defined in terms
19
of the other two, a constraint must be added to the model. In the background section we argued that it is theoretically justifiable to leave out survey year (time), assuming that there will be no period effects on job satisfaction. Probably, external conditions like unemployment spells will affect job satisfaction. However, this effect will not be completely independent of the year in which you are born. In addition, it should be emphasized that the main aim of this paper is to describe trends in job satisfaction, and not to explain why we find certain trends. Adding unemployment rates could be an important step in explaining trends in job satisfaction over the life course for successive birth cohorts. However, for this analysis we do not want to net out the effect of unemployment levels. On the contrary, we want to observe the ‘true’ descriptive trends in job satisfaction which, indeed, could be influenced by unemployment spells in the German labor market.
Results
The results of the analysis are presented in Figure 1 and Figure 2. Figure 1 shows the estimated trends in job satisfaction for the sample we selected in this study. The life course patterns in job satisfaction are plotted for seven successive cohorts, using ten year birth cohorts. The oldest birth cohort consists of respondents born between 1925 and 1934 and the youngest birth cohort consist of respondents born between 1985 and 1989. The length of the lines indicate the age range the respondents in this cohort are observed. The distance between the lines shows the extent to which job satisfaction levels differ between cohorts. The larger the distance, the larger the difference in job satisfaction levels. If the lines are close to each other or even overlap, job satisfaction levels have been quite similar between cohorts. Figure 1 is based on the model shown in Table 2. We used a multi-level model that includes a quartic age term and a categorical birth cohort variable. Results show that the linear age term is negative, followed by a positive, negative and finally a positive coefficient for the quartic age term. This allows the age-job satisfaction relationship to first decline, to slow down in this decline, to steepen again, and finally slowing down again or even increase in later life. The baseline category for the cohort variable is the 1985-1989 birth cohort. The coefficients demonstrate that compared to this most recent birth cohort, the older cohorts have significantly higher levels of job satisfaction, except for the cohorts 1965-1974 and 1975-1984. This is clearly visible in Figure 1. There is a large distance between the job satisfaction levels of the three oldest and the four youngest cohorts. The life course patterns of the more recent cohorts are very close, and for the two youngest cohorts the job satisfaction trends even overlap.
20
Figure 1. Change in job satisfaction, West-Germany 1984-2014
Note: N = 16175 individuals, N = 129623 observations. Y-axis shows job satisfaction scores (0
= completely dissatisfied, 10 = completely satisfied).
Source: German Socio-Economic Panel Study (SOEP), 1984-2014, version 31.
Figure 2 combines the descriptive data with the results obtained from the model. The dashed lines show the descriptive trends and the solid lines show the trends estimated by the model. The estimated model seems to be a rather good representation of the descriptive data as the solid lines match well with the dashed lines. The model deviates to some extent, especially for the cohorts 1935-1944 and 1945-1954. These cohorts experienced a considerable rise in their job satisfaction starting at age 56, which is only modestly modeled. For the cohorts 1955-1964 and 1965-1974 job satisfaction levels are slightly underestimated at the end of their observed age range. Overall, we see from the descriptive data that job satisfaction levels are quite high with scores between 6.8 and 8. Job satisfaction declined as people aged, but for some cohorts satisfaction levels started to increase again when they entered their final ten years of employment. People who are born earlier were more satisfied with their job than people born since 1955. Below, we will discuss the results in more detail in relation to the theoretical background and the formulated hypotheses.
21
Table 2. Random effects model for change in job satisfaction
*P-value <0.05; **P-value < 0.01; *** P-value < 0.001
Source: German Socio-Economic Panel Study (SOEP), 1984-2014, version 31.
B (SE)
Age linear -0.332*** (0.086)
Age squared 0.014*** (0.003)
Age cubic -0.000*** (0.000)
Age quartic 0.000*** (0.000)
Birth cohort (Ref: 1985-1989)
1925-1934 1.134*** (0.092) 1935-1944 0.776*** (0.081) 1945-1954 0.537*** (0.078) 1955-1964 0.236** (0.075) 1965-1974 0.105 (0.076) 1975-1984 0.004 (0.079) Constant 10.19*** (0.817) Number of individuals 16175 Number of observations 129623
22
Figure 2. Estimated and descriptive change in job satisfaction, West-Germany 1984-2014
Note: Solid lines represent the estimated trend, dashed lines represent the descriptive trend. N = 16175 individuals, N = 129623 observations. Y-axis shows job satisfaction scores (0 = completely dissatisfied, 10 = completely satisfied).
Source: German Socio-Economic Panel Study (SOEP), 1984-2014, version 31.
The age-job satisfaction relationship
We hypothesized that job satisfaction levels are higher for older people. It is immediately clear that this hypothesis is rejected. On the contrary, job satisfaction levels of West-German workers seem to decrease as they age. The birth cohorts born between 1935 and 1964, which are the red, green and the blue line experienced a drastic decline. For the three youngest cohorts, job satisfaction seems to decline at a slower rate. If the decrease we observed for the older cohorts would continue for these younger cohorts, a steeper downward trend could start at age 40. For the oldest cohort 1925-1934, the trend in the short observed age range is rather inconsistent. The descriptive data indicate a somewhat erratic pattern with a high peak and a sharp decline just before retirement. The rise in job satisfaction from age 56 until retirement for the three oldest cohorts is the only result that to some extent corresponds to
23
the formulated expectations; older people experience higher levels of job satisfaction. However, this is preceded by a sharp decline in job satisfaction over the life course, and only starts in the last ten or fifteen years of working life.
The results mainly contradict existing theoretical and empirical knowledge on the age-job satisfaction relationship. Previous empirical studies showed that age had a significant positive effect on job satisfaction which made researchers conclude that older workers are more satisfied with their jobs. Career stage theory, socio-emotional selectivity theory and the idea that workers move from idealistic to more realistic expectations all suggest that job satisfaction would increase as people age. However, our results suggest that this increase would only start around age 58. To theoretically support a J- or U-shaped relationship between age and job satisfaction it is argued that employees feel satisfied about their novel situation when they enter the labor market. But, wrong expectations may result in a downward trend in job satisfaction, which is reflected in the findings. Though it is suggested that after some years, this trend reverses again, our analysis suggests that this is not the case at all.
Job satisfaction across successive cohorts
Though the hypothesis on the age-job satisfaction relationship was quite evident, formulating expectations on the cohort effect appeared to be more difficult. Job satisfaction could change through the replacement of older cohorts by younger cohorts. However, we could not conclude on the precise direction of the cohort effect through the mechanism of replacement. We did anticipate that younger cohorts born around 1970 and onwards report lower level of job satisfaction due to changing external circumstances. We expected that changing labor market conditions more heavily affected job satisfaction levels of the more recent cohorts. In support of this hypothesis, we demonstrated that job satisfaction levels are significantly higher for the three oldest cohorts compared to the younger cohorts. Looking at the older cohorts, we see that there is a large distance between people born around 1930 and around 1940 (the black and the red line). The same applies between people born around 1950 and around 1960 (the green and the blue line). Inspection of the descriptive data shows that around age 58, the lines for these latter two cohorts start to connect, indicating that at later life they experienced similar levels of job satisfaction. For the four most recent cohorts, job satisfaction levels are very close to each other.
An advantage of the analysis is that it allows for a comparison of job satisfaction at a certain age for several birth cohorts. Taking age 40, we can compare job satisfaction levels for
24
five cohorts. We can clearly see the substantial difference between the older cohorts and the younger cohorts, which the coefficients in Table 2 also indicate. The data show that for the birth cohort 1935-1944 the level of job satisfaction at age 40 was 7.9. For the 1965-1974 birth cohort this was around 7.1, which is a difference of almost half a standard deviation.
The lower levels of job satisfaction for the three birth cohorts born since 1965 and onwards could be explained by the diffusion arguments discussed in the theoretical background. Increasing unemployment rates when these cohorts entered the labor market, school-to-work mismatches and the deterioration of job quality could be an explanation of their lower satisfaction with the job. Based on the generational literature, more idealistic expectations and different work values of the younger generations could also contribute to the lower satisfaction levels. The more recent cohorts seem to be more focused on life satisfaction and have become less convinced that work should be an important part of life. This may have created disappointment as well as a more critical attitude towards work. We have no theoretical explanation for the substantial difference between the four oldest cohorts. The oldest two cohorts lived through severe conditions after the Second World War but also experienced increased prosperity in the 1970s just before they entered the German Socio-Economic Panel Study. This could result in high job satisfaction at the end of their career. However, we did not expect the large differences in job satisfaction among the older cohorts.
If one wants to compare successive cohorts it is important to take into account the changes in cohort composition, for instance in terms of gender and educational level. From the literature, the exact relationship between gender, educational level and job satisfaction is not straightforward. Several arguments would be in favor of men and higher educated workers being more satisfied with their jobs, but arguments in favor of an opposite relationship also exist. Below, we briefly describe the outcomes of an analysis on the relationship between job satisfaction and gender and educational level.
Compositional change: women labor market participation and educational expansion
Increased participation of women in the labor market and the expansion of higher education impact the compositional structure of the more recent cohorts in this study. The descriptive statistics show that the proportion of men and women became more or less equal in the birth cohort 1955-1964 (see Appendix Table A). If women are considerably less satisfied with their job, this could be one of the reasons why we observe that more recent cohorts report lower
25
levels of job satisfaction. Higher education is gradually expanding for successive cohorts in our sample of workers. In the cohort 1975-1984 even 43% of the respondents is higher educated. In addition, we observe a downward trend in the share of lower educated workers (see Appendix Table B).
Interestingly, if gender is added to the final model it appears that there is no significant difference between males and females in their job satisfaction level. Regarding educational level, we observe that compared to the lower educated (the baseline category) intermediate and higher educated workers are more satisfied with their job. For both variables, the exploration of potential interaction effects did not have any far-reaching consequences for the results. What happens to the initial age and cohort findings if we add gender and educational level to the final model? The results show that the coefficients for the four age terms are rather similar compared to the model without gender and educational level, suggesting that life course patterns in job satisfaction are not considerably different for males and females or for lower, intermediate and higher educated workers (see Appendix Table C). We are most interested in understanding the extent to which the negative cohort trend is affected. The results show that the coefficients for the categorical birth cohort variable are rather similar compared to the initial findings. If we control for gender and educational level, the general negative cohort trend pertains. The positive effect of belonging to one of the older birth cohorts is even a bit larger in this second model, which is probably due to the positive effect of higher education on job satisfaction levels. As the more recent cohorts have a higher share of intermediate and higher educated workers, this positively impacts the job satisfaction levels of these cohorts. Thus, controlling for the effect of educational level makes the difference in job satisfaction between the younger and the older cohorts even larger. In short, our results suggest that the increased participation of women in the labor market and the expansion of higher education cannot explain the observed negative cohort effect.
Discussion
The purpose of this study was to gain insight into change in job satisfaction. Job satisfaction is the most studied employee attitude in organizational fields and is demonstrated to relate to overall life satisfaction and economic productivity and performance. However, existing research on long-term developments in job satisfaction often relied on inadequate data and methodology. By making use of 31 yearly waves of the German Socio-Economic Panel Study, we were able to accurately describe change in job satisfaction. The outcomes are rather
26
alarming. We demonstrated that for West-German workers, there is a negative cohort effect and a negative age-job satisfaction relationship. Not only do people become less satisfied with their jobs as they age, the younger generations also seem to report lower job satisfaction levels. These results cannot be explained by the rapid expansion of higher education or the fact that women started to increase their participation in the labor market.
The negative age-job satisfaction relationship is very remarkable in light of existing theoretical and empirical knowledge. Whereas other research has shown a positive relationship between age and job satisfaction, we demonstrated that job satisfaction declines as people age, followed by an increase in job satisfaction levels in the last ten years before retirement. How can we explain these contradictory findings? While most researchers relied on cross-sectional data, we were able to use a large scope of panel data which we analyzed with a two-level multi-level model. This allowed us to disentangle the effect of age and birth year on job satisfaction, while results from cross-sectional studies reflect a combination of the age and the cohort effect. The cross-sectional finding of age being positively related to job satisfaction could eventually be due to a negative cohort effect. People are not more satisfied because they are older, but because they are born earlier.
The findings of this study also direct attention towards the need for consistent theory-building on the age-job satisfaction relationship. At the moment, there is no systematic framework that could theoretically clarify the relationship between age and job satisfaction. Moreover, theories that do exist only provide arguments that support a positive relationship between age and job satisfaction. On the one hand, existing theoretical rationales like career stage theory and socio-emotional selectivity theory could simply be wrong. At least, the mechanisms these theories propose do not hold for the life course patterns of job satisfaction of West-German workers. Instead of a scenario where workers adapt their expectations, benefit from career opportunities and thus become more satisfied as they age, it may be that workers become and stay increasingly disappointed in their jobs and career prospects. On the other hand, external factors could have affected the age-job satisfaction relationship in such a way that it counteracted the mechanisms proposed by these various sociological and psychological theories. For instance, the deterioration of job quality starting in the 1990s and the new millennium, could impact the older cohorts around age 50, preventing them from benefiting from final, maybe even anticipated career advantages and opportunities. However, our results show that the decline in job satisfaction starts at an earlier age. In addition, it is likely that the deterioration of job quality will have the largest impact on the career and thus
27
the job satisfaction levels of the younger labor market entrants, instead of on the older workers who already secured their position in the labor market. Though this is not the place to formulate a coherent theoretical framework, our results certainly strengthen the need for more thorough theory building on the age-job satisfaction relationship.
Cohort effects have received little attention in the job satisfaction literature compared to the age-job satisfaction relationship. We anticipated that younger birth cohorts born around 1970 and onwards report lower levels of job satisfaction than older birth cohorts, which is confirmed by our findings. These younger cohorts are exposed to higher unemployment rates and worse labor market and job conditions. Consequently, they have limited opportunities to successfully self-select into jobs they prefer, which could make them less satisfied. The negative cohort effect could also be explained in terms of generational differences in attitudes and life expectations. Generation X and Y may adopt work values and expectations that are distinct from older birth cohorts. In addition, they may not adapt their expectations to the extent previous generations did. Disappointment and a more critical attitude towards working life among these more recent birth cohorts could contribute to the observed lower job satisfaction levels.
Importantly, both theoretical approaches cannot explain the clear negative cohort effect we found among the older cohorts. The high job satisfaction levels of the two oldest cohorts could partly be explained by the fact that they experienced the second world war and lived through severe conditions. Such a negative life-event may make you appreciate what you have in life even more, and this effect may persist as people age. Compared to the oldest cohort, the three successive cohorts report lower levels of job satisfaction. Currently, there is no conclusive theoretical rationale that could explain these results. The Baby Boom Generation grew up and entered the labor market in much more safe conditions as their parents. Apparently, did this not result in equally high job satisfaction levels.
There are two important limitations in studying change in job satisfaction levels. First, even with the design we adopted for this study it is not possible to observe pure age and cohort effects. Only panel data that covers the entire age range for every successive cohort would allow for this. With shorter spells of panel data there remains a possibility that the cohort effect is partly due to an age effect or the other way around. The younger generations could, for whatever reason, approach the higher job satisfaction levels of the older cohorts as they age. In addition, when the oldest cohorts were aged between 20 and 40, they could have been more dissatisfied with their working life, similarly to what we find for the two most
