Young and at risk? Consequences of job insecurity for mental health and satisfaction among labor market entrants with different levels of education

Abstract

Young workers are often temporarily employed and thus likely to experience job insecurity. This study investigates associations of objective job insecurity (i.e., temporary employment) and subjectively perceived job insecurity with mental health, job satisfaction and life satisfaction among young workers, testing the moderating role of education. The longitudinal analysis based on 1522 labor market entrants from the German Socio-Economic Panel revealed that subjective job insecurity was associated with decreased mental health and lower satisfaction, whereas objective job insecurity was not. Three levels of education were differentiated: university degree, vocational training and low-qualified. There was weak evidence that those with vocational qualifications were more vulnerable to subjective job insecurity than either the low-qualified or university graduates. The results suggest that irrespective of education, detrimental consequences of subjective job insecurity emerge early in the career. Access to secure employment should be improved for young workers to prevent early dissatisfaction and impairment of their mental health.

Keywords

Education employee health job insecurity longitudinal analysis young workers

Introduction

To date, finding a secure job remains a priority for many adolescents in their early working life (Lowe and Krahn, 2000; Shell Deutschland, 2015), but it is not the reality they find in the labor market: employment stability among Europe’s younger workforce has declined (Cazes and Tonin, 2010). Temporary employment among new hires is increasing in many countries (OECD, 2014), and its prevalence among the under-25-year-olds has grown at twice the rate of older workers (ILO, 2012). Compared to 14% of the general working population, 42% of 15- to 24-year-old workers were in temporary jobs in 2012 (Eurofound, 2013). While temporary jobs are an important entry point into the labor market, they are associated with disadvantages such as lower wages and higher risks of unemployment. Especially low-educated workers have a higher risk of working in low-quality temporary jobs with little chances of upward mobility (Gebel, 2010; Gebel and Giesecke, 2009; OECD, 2002). Low stability and frequent job changes are to a certain extent normal in the process of settling into working life and thus nothing new (Quintini et al., 2007). However, since labor market entrants have been disproportionately affected by labor market deregulation, the uncertainty that characterizes this transition phase has intensified in recent decades (Buchholz, 2008; Mills et al., 2005).

The literature distinguishes between two types of uncertain employment: temporary jobs are objectively insecure, because they do not guarantee continuous employment (De Witte and Näswall, 2003; Klandermans et al., 2010), whereas subjective job insecurity concerns the perceived threat of job loss (Sverke and Hellgren, 2002). Evidence of negative associations with health and satisfaction is quite consistent for subjective job insecurity, but not for temporary employment (De Cuyper et al., 2008; Sverke et al., 2002). Moreover, few studies in this field have focused specifically on young workers (e.g., Fiori et al., 2016; Peiró et al., 2012).

The stratification of employment risks also calls for the exploration of social variations in health effects of job insecurity: higher education increases young adults’ chances to acquire quality jobs as well as their re-employment chances in case of job loss (Gebel, 2010; Mandemakers and Monden, 2013), and may thus buffer negative effects of job insecurity. However, highly-educated young workers may also react more negatively due to higher aspirations and threats of status inconsistency (Feldman and Turnley, 2004; Turner, 1995).

Young workers’ employment conditions are objectively less secure than those of middle-aged workers. As the entrance into the workforce signifies a formative phase of occupational socialization, experiences of job insecurity at this stage might have lasting consequences for the individual (cf. Peiró et al., 2012). This study investigated associations of job insecurity with mental health, job satisfaction and life satisfaction among labor market entrants in a longitudinal design, using the German Socio-Economic Panel Study (SOEP). Both objective and subjective indicators of job insecurity were included, testing the moderating effect of education. Indicators of mental health, job satisfaction and life satisfaction were chosen as outcomes for a balanced account of mental health. Both job and life satisfaction represent aspects of affective well-being as an integral part of mental health (Warr, 1994). All outcomes were thus understood as indicative of mental health in terms of a positive equilibrium between individuals and their environment (cf. Lehtinen et al., 2004).

This study contributes to the literature in several ways. First, it draws attention to the vulnerable population of young workers in the context of job insecurity. Second, investigating social variation in mental health reactions to job insecurity contributes to the understanding of working conditions as determinants of health inequality (Landsbergis et al., 2014; Marmot et al., 2011). Third, the longitudinal design allows for analyzing within-person variance in job insecurity, mental health and satisfaction, which strengthens causal inference by controlling for unobserved heterogeneity between persons.

Job insecurity among young workers

The term ‘young workers’ denotes various groups in the literature, such as an age group, students working part-time, or labor market entrants establishing themselves in the workforce (Loughlin and Barling, 2001). This study focuses on the last of these, using the terms ‘young workers’ and ‘labor market entrants’ interchangeably, while the transition from education to employment is construed as a process spanning several years rather than the single event of entering the first job (cf. Brzinsky-Fay, 2007).

Taking up employment after completing education is a critical part of the transition from adolescence to adulthood, and can be either a positive challenge or a stressful endeavor (cf. Schoon and Silbereisen, 2009). Entering the labor market from an outsider position and lacking experience, young workers face greater employment volatility than middle-aged workers (Gangl, 2002). Labor market deregulation, to differing extents in different countries, has increased uncertainty for young workers, because weaker employment protection and temporary jobs are typically allocated to newly hired outsiders (Cazes and Tonin, 2010; Gangl, 2002; Mills et al., 2005). Young workers today thus take longer to establish themselves in the labor market, which manifests itself for example in longer search periods to find the first job, or in increased prevalence of unemployment and temporary employment compared to older cohorts (Mills et al., 2005). Labor market entrants today are thus more prone to job insecurity than before.

The concept of job insecurity

Job insecurity can be defined in objective or subjective terms, but both definitions revolve around the threat of job loss. Objective job insecurity refers to structural conditions that threaten continuity in employment, such as downsizing or temporary employment (Büssing, 1999; Pearce, 1998). Especially the latter can be seen as an objective indicator of job insecurity, as it does by definition not guarantee continuous employment (De Witte and Näswall, 2003; Klandermans et al., 2010). The disproportional exposure of young workers to objective job insecurity is well documented (Allmendinger et al., 2013; Eurofound, 2013).

On the subjective side, psychological definitions of job insecurity refer to the perception that one’s job is at risk, that is, the anticipation of involuntary job loss and associated experiences of uncertainty and powerlessness (De Witte, 1999; Greenhalgh and Rosenblatt, 1984; Sverke and Hellgren, 2002). These individual perceptions tend to correspond to workers’ objective employment conditions, such that those exposed to downsizing or temporary employment also experience more job insecurity (Keim et al., 2014; Näswall and De Witte, 2003). This would suggest higher job insecurity among young workers given their lower levels of employment protection and high prevalence of temporary jobs, but the association of subjective job insecurity with age is less clear. It is plausible to assume that younger workers, who are more mobile and less likely to have family responsibilities, are less dependent on their jobs and thus perceive less job insecurity (cf. Keim et al., 2014; Pearce, 1998). This has been supported by Erlinghagen (2007), whereas Näswall and De Witte (2003) only found partial evidence. In a recent meta-analysis, contrary to what was expected, Keim et al. (2014) found a negative association between age and subjective job insecurity.

Consequences of job insecurity for mental health and satisfaction

Objective job insecurity

Objective job insecurity in terms of temporary contracts may affect young workers’ mental health and satisfaction through relative deprivation, which occurs when temporary workers prefer permanent employment and feel they are not receiving what they deserve compared to relevant others (Beard and Edwards, 1995; Feldman and Turnley, 2004). Labor market entrants with temporary jobs face wage penalties, as well as a higher risk of unemployment and cycles of repeated temporary contracts compared to their peers with permanent jobs, and it may take several years to recoup these initial disadvantages (Gebel, 2010). This may foster relative deprivation in the sense that young temporary workers may feel that they are not receiving adequate rewards for their work (i.e., income, security) compared to those in permanent jobs. Accordingly, an Italian study found reduced mental health among young temporary workers compared to those in permanent jobs (Fiori et al., 2016).

Empirical evidence from the general working population is mixed (De Cuyper et al., 2008). Some studies have found better mental health among permanent employees (e.g., Aronsson et al., 2002), whereas others have found better mental health and higher satisfaction among temporary workers (e.g., Mauno et al., 2005), and still others have found no differences between the two groups (e.g., Bardasi and Francesconi, 2004). These inconsistencies may be partly due to population heterogeneity: various work arrangements are subsumed under temporary employment. These may be associated with differing levels of job insecurity, which may also depend on country-specific employment protection standards (De Cuyper et al., 2008; Kim et al., 2012).

Workers’ motives for taking up temporary employment and their expectations regarding contract extensions are also heterogeneous: temporary employment may not lead to feelings of job insecurity or relative deprivation for everyone, especially when it is perceived as a voluntary arrangement (Bernhard-Oettel et al., 2013; Feldman and Turnley, 2004). Young workers can benefit from temporary contracts as a way to gain experience, since these often serve as entry jobs and probation periods for newcomers (Gebel, 2010; Nunez and Livanos, 2015). For many, temporary jobs serve as a stepping stone, whereas others become trapped in repeated temporary employment (see McLean Parks et al., 1998). Finally, temporary workers know their contract is limited and may not expect security from their employer, rendering experiences of job insecurity less severe (De Cuyper and De Witte, 2007; Klandermans et al., 2010).

Nonetheless, even though a contract is officially limited, some degree of uncertainty remains and many people enter temporary jobs hoping to secure a permanent contract (Bernhard-Oettel et al., 2013; Clinton et al., 2011). Employers may foster such expectations, but contracts are ultimately renewed or extended at the employer’s discretion, leaving temporary workers little control over the continuance of their job. Temporary employment among Europe’s younger workforce has been found to be in large part involuntary or motivated by instrumental reasons such as probationary employment (Eurofound, 2013; Nunez and Livanos, 2015). Notwithstanding empirical inconsistencies and population heterogeneity, temporary contracts should thus be taken into consideration as an objective indicator of job insecurity in this population.

Subjective job insecurity

The transactional stress model (Lazarus and Folkman, 1984) predicts workers will evaluate job insecurity as a threat and their coping resources as insufficient, because the situation is characterized by uncontrollability and unpredictability (De Witte, 1999; De Witte et al., 2016; Dekker and Schaufeli, 1995). Latent deprivation theory offers an underlying mechanism, as a potential job loss implies a threat to the manifest (income) and latent benefits of work, that is, activity, time structure, social contact, collective purpose and social status (Jahoda, 1982; Vander Elst et al., 2016). Since young workers are not a homogeneous group, they are likely to differ in the extent to which they evaluate job insecurity as threatening. According to transactional stress theory, these subjective appraisals are the decisive factor in eliciting strain (Lazarus and Folkman, 1984).

Correspondingly, the literature on subjective job insecurity is much more consistent. Its negative association with mental health and satisfaction has been supported by meta-analyses (Cheng and Chan, 2008; Sverke et al., 2002), quasi-experimental studies (Dekker and Schaufeli, 1995) and longitudinal studies (Ferrie et al., 2002; Heaney et al., 1994; Kinnunen et al., 2014). There is also evidence for causality, as job insecurity has been found to precede impaired mental health and reduced satisfaction rather than vice versa (De Witte et al., 2016; Hellgren and Sverke, 2003; Vander Elst et al., 2016). Consistent with another central notion of transactional stress theory that just anticipating an adverse event can induce stress (Lazarus and Folkman, 1984), job insecurity can be as detrimental as unemployment (Dekker and Schaufeli, 1995; Kim and Von dem Knesebeck, 2016). Focusing specifically on young workers, Peiró et al. (2012) found a negative association between job insecurity and work involvement among Spanish labor market entrants.

Job insecurity perceptions are likely influenced by young workers’ objective job security (Peiró et al., 2012), and objective job insecurity may be associated with reduced mental health as well (Fiori et al., 2016). However, based on theoretical considerations and existing evidence, subjective job insecurity can be expected to be decisive in predicting young workers’ mental health and satisfaction (see Bernhard-Oettel et al., 2005). The following was hypothesized:

Hypothesis 1: The negative association of job insecurity with mental health and satisfaction is stronger for subjective job insecurity than for objective job insecurity.

Educational differences in reactions to job insecurity

Both objective and subjective job insecurity may affect young workers’ mental health and satisfaction (cf. Fiori et al., 2016; Peiró et al., 2012), but the effects may not be the same for young workers with different levels of education (e.g., Domenighetti et al., 2000; Fiori et al., 2016). Education as a component of socio-economic status refers to an individual’s position in the social hierarchy, which involves differences in job prospects and working conditions. Working conditions contribute to health inequalities through differential exposure and differential vulnerability (Landsbergis et al., 2014; Siegrist and Marmot, 2004). Research on the general working population has shown that exposure to both objective and subjective job insecurity is higher among workers of lower status, but inquiries into differential vulnerability have been limited (Landsbergis et al., 2014). Education was thus investigated as a proxy for social status and potential moderator of the job insecurity–health relationship in this study. Three levels of education were distinguished: university degree, vocational training (i.e., an occupation-specific dual program of on-the-job training and a vocational school curriculum) and low-qualified, that is, not having any vocational qualifications. In the German labor market, vocational qualifications are decisive for employment chances (Gebel and Giesecke, 2009; Müller et al., 1995). Both low-qualified workers and university graduates have a higher prevalence of temporary employment, but the latter usually enter temporary jobs of higher quality and have more options on the labor market (cf. Gebel, 2010).

Higher education is associated with healthier working conditions and lower vulnerability to stressors (Marmot et al., 2011; Monden, 2005), higher employability in terms of an individual’s perceived chance to find a new job (Berntson et al., 2006), as well as re-employment chances in case of job loss (Mandemakers and Monden, 2013). This speaks of a buffering effect of education, because highly-educated young workers may be better equipped to cope with job insecurity. On the other hand, they likely have higher career aspirations than their low-qualified peers and may thus be more susceptible to relative deprivation (Feldman and Turnley, 2004). Consequently, they might be more vulnerable to job insecurity, because it poses a threat to their self-esteem through threats of status inconsistency or frustration of unmet expectations (De Witte, 1999; Domenighetti et al., 2000; Schaufeli, 1992; Turner, 1995).

Evidence that education moderates the job insecurity–health relationship is mixed. Fiori et al. (2016) found that temporary employment among younger workers was associated with poorer mental health among highly-educated men and women, but with better mental health among lower-educated men. With regard to subjective job insecurity, some studies of the general working population suggest that negative effects are weaker for higher-educated workers (Landsbergis et al., 2014). On the other hand, Domenighetti et al. (2000) found that higher education may aggravate negative effects of subjective job insecurity, while a large-scale study by László et al. (2010) found no moderating effect for education. However, as education is generally associated with access to resources, including better job prospects (Landsbergis et al., 2014), it seems more likely that education has a protective rather than aggravating effect for young workers exposed to job insecurity. The following was hypothesized:

Hypothesis 2: The negative association of objective job insecurity with mental health and satisfaction is weaker for higher-educated young workers.

Hypothesis 3: The negative association of subjective job insecurity with mental health and satisfaction is weaker for higher-educated young workers.

Method

Sample and procedure

Data were taken from the German Socio-Economic Panel (SOEP), an ongoing representative household panel study, which has been conducted annually since 1984 and includes over 20,000 persons in about 12,000 households in Germany. The same households are surveyed every year, while new samples have been added repeatedly to maintain representativeness. All household members aged 17 and older are surveyed individually, and children from participating households remain in the sample after they leave the parental household (Wagner et al., 2007). Initial household response rates ranged from 33% to 70%, with follow-up rates between 87% (2001) and 93% (2003) (Infratest Sozialforschung, 2011a, 2011b, 2012). Among young adults (18–30 years old) in participating households, 84% (2014) to 93% (2004) returned an individual questionnaire (own calculations). Analyses were restricted to the waves 2001–2014 due to data availability regarding mental health: the SF-12 health questionnaire (Ware et al., 1996) has been included in the SOEP biennially since 2002, so that all even-numbered years between 2002 and 2014 were included when analyzing mental health. Data on both job satisfaction and life satisfaction are available in all survey waves, so that all waves from 2001 to 2014 were included when analyzing satisfaction, in order to keep a comparable time frame and maximize the use of information for all dependent variables.

The study population consisted of young workers who entered the labor market between ages 18 and 30 within the study period. Labor market entry was defined as the first year in which respondents reported being employed after leaving education or vocational training, while working at least five hours a week with no return to education or training in the subsequent year. This ensured that all respondents were observed from roughly the same starting point of being part of the workforce for the first time, while at the same time excluding young adults on holiday jobs or taking gap years (see Gebel, 2010).

The longitudinal data were restructured according to time in the labor market: pooled across the waves 2001–2014, labor market entry was defined as Time 1 for all respondents, who were then followed for up to eight years. Common empirical cut-off points for the education-to-employment transition phase range from five to seven years after leaving education or entering the first job, respectively (Brzinsky-Fay, 2007; Gebel, 2010; Quintini et al., 2007). Due to the two-year time intervals between measurements of mental health this cut-off point was raised to eight years in this study to increase the number of measurements and within-person variance in mental health.

Only waves in which respondents were employed were retained in the panel, as job insecurity measures would not apply otherwise. Respondents were further dropped from the panel if they returned to education, entered into self-employment, into a workshop for people with disabilities or into retirement, that is, the subsequent waves were excluded, not the persons as such. The resulting panel was unbalanced (i.e., the number of observations per person varied from 2 to 8). To analyze within-person change, respondents identified as labor market entrants were included in the sample if they had at least two waves with complete data on both objective and subjective job insecurity and on the respective dependent variables while working, as well as valid observations of education. This resulted in varying sample sizes for the different dependent variables (ranging from N = 963 for mental health to N = 1522 for life satisfaction). The overall sample size was N = 1522.

A total of n = 407 were identified as labor market entrants but excluded from all analyses because they did not meet the criterion of at least two observations on the variables of interest. Compared with the rest of the sample when entering the labor market, these persons were on average younger (t (684.64) = −7.68; p = .000), worked fewer hours (t (499.18) = −7.39; p = .000), had a higher share of persons living with their parents (χ² (2) = 22.17; p = .000), a higher share of temporary contracts (χ² (1) = 11.54; p = .001), lower education (χ² (2) = 74.33; p = .000), and reported lower initial levels of job satisfaction (t (563.93) = −2.21; p = .028) and life satisfaction (t (587.00) = −3.12; p = .002).

Sample characteristics

Descriptive statistics refer to the year of labor market entry. Half of the respondents were women (50%) and the average age was 23.90 years (SD = 3.04). About half (54%) lived in their parents’ household, 30% lived with a partner and 16% lived in single households. Most respondents (75%) worked more than 38 hours per week (M = 40.12; SD = 8.85), average organizational tenure was 1.60 years (SD = 1.63) and 19% were employed in the public sector. Fewer than half entered the labor market through temporary employment (43%). A majority of 60% had completed a vocational training, 24% had a university degree and 16% had no vocational qualifications beyond elementary or upper-secondary school. The sample was fairly representative of younger German employees, except for a slight over-representation of public sector employment (see Federal Statistical Office, 2015; OECD, 2013).

Measures

Objective job insecurity

A dichotomous variable indicating permanent or temporary employment (coded 0 = permanent, 1 = temporary) served to measure objective job insecurity. Employment was considered permanent when respondents had an open-ended contract and were not employed by a temporary agency. Fixed-term contracts and employment with a temporary agency were subsumed under temporary employment, because the overall prevalence of temporary agency work was low (5%) with few transitions from or to permanent employment.

Subjective job insecurity

A single item asking respondents how much they worry about the security of their job (1 = very worried, 2 = somewhat worried, 3 = not worried at all; recoded such that higher scores reflected higher job insecurity) served to assess subjective job insecurity. Similar single-item measures have successfully been used in other studies on the health effects of job insecurity (e.g. László et al., 2010).

Mental health and satisfaction

The mental health component scale of the SF-12, a well-validated and widely used questionnaire of health-related quality of life (Ware et al., 1996), assessed respondents’ mental health in terms of affective well-being, social and emotional functioning. Six items (α = .75), covering both positive and negative aspects such as vitality, psychological distress and social role-functioning, asked respondents to rate the frequency of certain experiences during the past four weeks on a five-point scale from 1 = always to 5 = never (e.g., ‘How often did you feel energetic?’). The SOEP includes norm-based T-scores for the scale, ranging from 0 to 100 (M = 50.00, SD = 10.00) with higher scores indicating better mental health (Andersen et al., 2007).

Two single items each assessed job satisfaction (‘How satisfied are you with your job?’) and life satisfaction (‘How satisfied are you with your life, all things considered?’), both using an 11-point scale ranging from 0 = absolutely dissatisfied to 10 = absolutely satisfied. Similar single-item measures have been validated and widely used, and are considered reliable measures of the constructs for job satisfaction (Wanous et al., 1997) and life satisfaction (Diener et al., 2013), respectively.

Education

The CASMIN educational classification (Brauns and Steinmann, 1997) was collapsed into three categories to operationalize respondents’ educational level. The reference category ‘low-qualified’ (n = 235) ranged from having primary school education to the highest upper-secondary school degree (Abitur), with no further vocational qualifications. Respondents with an Abitur were included in this group, because vocational qualifications are considered the key to entering quality jobs. While the Abitur qualifies individuals to enter tertiary education, it has decreased in value on the labor market (Müller et al., 1995). The category ‘vocational education’ (n = 914) comprised all respondents who had completed a vocational training after secondary or upper-secondary school. All respondents with a tertiary degree from a university or a university of applied sciences (Fachochschule) were subsumed under the category ‘university degree’ (n = 373). Educational level at the time of labor market entry was conceptualized as a time-constant variable.

Control variables

Dummy variables for the survey waves served to control for period effects, such as fluctuations in economic climate. Since age, organizational tenure, as well as the household context as a source of financial and social support may influence the job insecurity–health relationship (Cheng and Chan, 2008; Lim, 1996), they were controlled for in the analysis. Household context included the categories living with parents, single and cohabitating with a partner. Further, working in the public versus the private sector was controlled for, as subjective job insecurity tends to be lower in the former (Erlinghagen, 2007). Because a change from temporary to permanent employment could result from the conversion of an existing contract or from a job change implying a change of work environments as well, a dichotomous indicator assessing whether respondents had changed their job since the previous year (coded 0 = no change, 1 = change and set to 0 for the year of labor market entry) was included as a control variable to reduce confounding.

Analyses

Panel regression models with fixed effects (FE) were calculated using Stata 13.0, with cluster-robust standard errors to correct for panel heteroskedasticity and autocorrelation within persons (Andreß et al., 2013; Rogers, 1993). In FE regression, within-person changes in the dependent variable are regressed on changes in the independent variables by analyzing deviations from person-specific means: all variables are mean-centered on the person level (Andreß et al., 2013). This offers the advantage of controlling for unobserved heterogeneity in terms of stable inter-individual differences in the sample, because only the variance within persons is included in the analysis and the time-constant person-specific error term is partialed out. Thus all time-constant variables (e.g., gender, innate abilities) are controlled for, whether they were observed or not (Allison, 2009; Andreß et al., 2013). This implies that the main effect of time-constant variables cannot be modeled, but their interaction with time-varying variables is possible (Andreß et al., 2013). The basic requirements for FE regression are that (a) the dependent variable is observed at least two times and (b) the independent variables change over time (Allison, 2009). The method is suitable for unbalanced panels (Andreß et al., 2013). Note that synchronous associations between predictors and outcomes averaged across waves of observation were estimated.

Nested models were calculated predicting mental health, job satisfaction and life satisfaction based on objective and subjective job insecurity (Step 1), then interactions between education and objective job insecurity (Step 2), and between education and subjective job insecurity (Step 3) were added, and finally both interaction terms were included in the same model (Step 4). Regression coefficients were not standardized, but outcome variables and subjective job insecurity were standardized within persons before the analysis. All other predictor variables were not standardized, because they were either dichotomous or measured on a ‘natural’ scale (e.g., age in years).

Results

Descriptive results

Zero-order correlations, as well as within-person and between-person variability in terms of standard deviations (continuous variables) or percentages (dichotomous variables) of the study variables are shown in Table 1. Between-person correlations (below the diagonal) were calculated based on the person-means of continuous variables pooled across all observation years and the modes of dichotomous variables, respectively. Within-person correlations were calculated after mean-centering the variables at the person level. The between-person percentages for dichotomous variables indicate the share of respondents falling into the given category, and the within-person percentages indicate the proportion of the time these persons fall into that category. For example, 55% of the respondents worked in temporary employment during the study period and they did so for an average of 53% of the time they were observed. Average levels of mental health (M = 51.21; SD_between = 6.04) were close to the norm scores reported for 18- to 24-year-olds in Andersen et al. (2007). Average job satisfaction and life satisfaction (M = 7.16; SD_between = 1.47 and M = 7.29; SD_between = 1.18, respectively) were comparable to the values reported for the whole SOEP sample between 2001 and 2013 (SOEP Group, 2015).

Table 1.

Descriptive statistics and between-person correlations (below diagonal) for all study variables, as well as within-person correlations (above diagonal) for time-varying variables, respectively, with listwise deletion of missing values.

Variable		M	s/%		1	2	3	4	5	6	7	8	9	10	11	12	13
Variable		M	btwn	wthn	1	2	3	4	5	6	7	8	9	10	11	12	13
1	Age	26.43	3.31	1.80	–	.04^*	.28^***	.03	.72^***	−.08^***			−.32^***	−.15^***	−.08^***	−.12^***	−.01
2	Single		32%	64%	.10^***	–	−.58^***	.06^***	.01	−.02			−.05^**	−.02	−.03	.01	−.06^**
3	Cohabitating		51%	75%	.34^***	−.36^***	–	−.01	.20^***	.01			−.05^**	−.06^**	−.04^*	−.06^**	.07^***
4	Work hours	41.01	7.36	5.14	.03	.11^***	−.12^***	–	.07^***	.00			−.06^**	−.08^***	−.07^***	.00	−.02
5	Tenure	3.04	1.95	1.55	.11^***	−.02	.02	.05	–	−.05^*			−.29^***	−.10^***	−.07^***	−.18^***	−.01
6	Public sector		26%	73%	.07^**	.05^*	.04	−.02	.07^**	–			.09^***	.04^*	.01	.07^***	.02
7	Vocational training		61%	99%	−.40^***	−.08^**	−.13^***	−.07^*	.09^***	−.10^***	–
8	University degree		26%	98%	.56^***	.14^***	.18^***	.14^***	−.07^**	.11^***	−.72^***	–
9	Objective JI		55%	53%	−.11^***	.05	−.04	−.03	−.29^***	.22^***	−.02	.01	–	.25^***	.02	.03	−.02
10	Subjective JI	1.71	0.56	0.48	−.12^***	−.07^**	−.05	−.05	−.09^**	−.05	.15^***	−.21^***	.17^***	–	−.10^***	−.08^***	−.10^***
11	Mental health	51.21	6.04	3.60	−.01	−.02	−.06^*	−.04	.01	−.04	−.08^**	.02	−.05	−.15^***	–	.22^***	.32^***
12	Job satisfaction	7.16	1.47	1.41	−.01	−.01	−.01	−.03	−.01	.08^**	−.08^***	.03	−.02	−.20^***	.37^***	–	.26^***
13	Life satisfaction	7.29	1.18	0.96	.02	−.02	−.07^**	.03	.04	.06^*	−.09^***	.12^***	−.04	−.25^***	.53^***	.43^***	–

Notes: N = 1428 persons for between-person correlations; N = 3038 person-years for within-person correlations. Pearson correlations based on person-specific means across observation years for time-varying continuous variables and person-specific modes for time-varying dichotomous variables, respectively. Within-person correlations based on group-mean centered variables. Dichotomous dummy variables represent the household categories, as well as the educational levels. btwn = between; wthn = within; JI = job insecurity. ^*p < .05, ^**p < .01, ^***p < .001.

With regard to the educational groups, low-qualified respondents were the youngest (F (2, 1514) = 415.03; p = .000), had a higher share of men (χ² (2) = 7.44; p = .024), lived more often with their parents (χ² (4) = 117.57; p = .000), worked fewer hours (F (2, 1496) = 9.66; p = .000), reported higher initial levels of subjective job insecurity (F (2, 1488) = 31.44; p = .000), but better mental health (F (2, 1437) = 5.65; p = .004) than their higher-educated peers. Respondents with a university degree lived more often with a partner, worked the most hours, and had the highest share of public sector employment (χ² (2) = 22.33; p = .000). Respondents with vocational training fell between the other two groups on all variables, with the exceptions that they had the longest tenure (F (2, 1514) = 41.06; p = .000), reported lower initial job satisfaction (F (2, 1489) = 5.42; p = .005) and lower initial life satisfaction (F (2, 1511) = 12.31; p = .000).

Fixed effects regression models

Table 2 shows the results of the FE regressions predicting mental health, job satisfaction and life satisfaction based on objective job insecurity, subjective job insecurity, and their respective interactions with education. The average number of observations per person was M = 2.7 years (range: 2–4) for mental health, and M = 4.1 years (range: 2–8) for both satisfaction variables. The coefficients of continuous variables represent changes from the person-specific means in mental health, job satisfaction and life satisfaction, respectively, as a function of changes from the person-specific means in the independent variables. Coefficients of categorical dummy variables represent the corresponding changes in the dependent variables as a function of being in the given dummy category compared to the reference category.

Table 2.

Fixed effects regressions of mental health, job satisfaction and life satisfaction on objective and subjective job insecurity, as well as their respective interactions with education.

Variable	Mental health				Job satisfaction				Life satisfaction
Variable	Step 1	Step 2	Step 3	Step 4	Step 1	Step 2	Step 3	Step 4	Step 1	Step 2	Step 3	Step 4
Control variables
Age	−.03^†	−.03^†	−.03^†	−.04^†	−.01	−.01	−.01	−.01	−.05^**	−.05^**	−.05^**	−.05^**
Household^a
Single	−.08	−.08	−.08	−.08	−.01	−.01	−.02	−.02	.05	.05	.05	.05
Cohabitating	−.11	−.11	−.11	−.11	−.04	−.04	−.05	−.04	.21^**	.21^**	.21^**	.21^**
Work hours per week	−.01^*	−.01^*	−.01^*	−.01^*	.00	.00	.00	.00	−.00	−.00	−.00	−.00
Tenure	.01	.01	.01	.01	−.10^***	−.10^***	−.10^***	−.10^***	.01	.01	.01	.01
Public sector	.06	.06	.06	.06	.29^***	.29^***	.29^***	.29^***	.13	.12	.13	.13
Job change	.11	.11	.11	.11	.16^***	.16^***	.16^***	.16^***	.09^*	.09^*	.09^*	.09^*
Job insecurity
Objective job insecurity^b	.01	.22	.01	.15	−.09^†	−.13	−.09^†	−.11	−.06	−.15	−.06	−.15
Subjective job insecurity	−.10^***	−.10^***	.06	.05	−.12^***	−.12^***	−.16^**	−.16^**	−.09^***	−.09^***	−.09^*	−.08^†
Education × objective JI
Vocational × objective JI		−.25		−.15		.11		.09		.11		.12
University × objective JI		−.25		−.19		−.10		−.14		.09		.08
Education × subjective JI
Vocational × subjective JI			−.21^*	−.20^*			.04	.03			−.01	−.02
University × subjective JI			−.14	−.12			.08	.08			.02	.02
Within-person R²	.02^*	.02^*	.02^**	.02^**	.06^***	.06^***	.06^***	.06^***	.02^***	.02^***	.02^***	.02^***

Notes: N = 963 for mental health; N = 1515 for job satisfaction; N = 1522 for life satisfaction. Period effects controlled for via dummy variables for survey years (not displayed in the table). JI = job insecurity. ^a Reference: living with parents. ^b Reference: permanent employment. ^†p < .1, ^*p < .05, ^**p < .01, ^***p < .001.

The results of Step 1 in each regression model showed that after controlling for period effects, age, household context, work hours, tenure, sector and job changes, increases in subjective job insecurity were associated with decreased mental health, lower job satisfaction and lower life satisfaction, whereas objective job insecurity was not, except for a tendency of p < .10 when predicting job satisfaction. That is, across observation years, when respondents reported subjective job insecurity above their personal average, they tended to report mental health and satisfaction scores below their personal average. Accordingly, the coefficient for objective job insecurity indicated that respondents tended to report job satisfaction scores below their personal average when they were in temporary employment compared to permanent employment.

Coefficients were larger for subjective job insecurity than for objective job insecurity on all outcomes, indicating that the former was overall more strongly associated with mental health and satisfaction than the latter. Wald tests for equality of parameters confirmed that effect sizes were not equal (b_obj. = .00; b_subj. = –.10, F (1, 962) = 3.99; p = .046 for mental health; b_obj. = –.05; b_subj. = –.12, F (1, 1514) = 4.34; p = .37 for job satisfaction; b_obj. = –.03; b_subj. = –.09, F (1, 1522) = 4.45; p = .035 for life satisfaction, respectively). Note that Table 2 reports coefficients for unstandardized objective job insecurity for the sake of interpretability, whereas Wald tests were calculated after standardizing the variable, so that objective and subjective job insecurity were compared to one another on the same scale. Hypothesis 1 was confirmed.

When interaction terms of education with objective and subjective job insecurity were introduced into the respective models, the coefficients for the two dummy variables ‘vocational’ and ‘university’ indicated the degree to which the association between the respective variables and the outcomes among respondents with vocational training and those with a university degree each differed from the effect in the reference group of low-qualified workers. The main effects of the independent variables indicated the effect size within the reference group. The main effect of education as a time-constant variable was controlled for as implied by the FE analysis.

Overall, interactions between education and objective job insecurity were not significant. Hypothesis 2 was rejected. Only one interaction was detected for subjective job insecurity: respondents with vocational training reported poorer mental health than the low-qualified when subjective job insecurity increased, and the pattern remained when the interaction between education and objective job insecurity was also accounted for (Step 4 of the model). Figure 1 illustrates that subjective job insecurity was associated with declines in mental health for higher-educated respondents, but not for the low-qualified. Simple slopes analyses revealed that the decline in mental health was significant among respondents with vocational training (b = –.15; p = .000), but not among the low-qualified (b = .05; p = .557), or university graduates (b = –.07; p = .143).

Figure 1.

Within-person changes in mental health as a function of within-person changes in subjective job insecurity from one standard deviation below, to one standard deviation above the personal mean, by educational level.

When interactions with education were introduced, subjective job insecurity remained associated with decreased job satisfaction among low-qualified respondents, and with reduced life satisfaction by a tendency of p < .10. Higher-educated respondents did not differ from the reference group. Support for Hypothesis 3 was mixed.

Discussion

The aims of this study are twofold: (a) to compare the relative importance of objective versus subjective job insecurity in predicting young workers’ mental health and satisfaction; and (b) to investigate educational differences in the respective associations of job insecurity with the outcomes. The study extended previous research on the relationship between objective and subjective job insecurity, which has predominantly relied on between-person comparisons (Bernhard-Oettel et al., 2005; De Cuyper et al., 2008; Klandermans et al., 2010), by investigating within-person changes in a specific risk population with high prevalence of temporary contracts. Moreover, testing for educational differences in reactions to job insecurity among young workers contributes to the literature on occupational health inequality (Landsbergis et al., 2014; Marmot et al., 2011), as well as to an unresolved puzzle in job insecurity research, namely whether higher-status workers are more or less vulnerable to its effects (De Witte, 1999).

The relative importance of objective and subjective indicators

The results supported Hypothesis 1 insofar as subjective job insecurity predicted mental health and satisfaction among young workers, whereas objective job insecurity did not. This was in line with theoretical expectations and existing evidence for the general working population (De Witte, 1999; Lazarus and Folkman, 1984). The finding that the same young workers were by and large not more or less healthy in temporary versus permanent employment stands in contrast to the findings of Fiori et al. (2016), and lends support to the notion that the formal contract per se is neither good nor bad for mental health (see Bernhard-Oettel et al., 2005). However, there was a tendency for young workers to be less satisfied with their jobs when in temporary employment, suggesting relative deprivation with regard to work-related outcomes (Beard and Edwards, 1995; Feldman and Turnley, 2004).

Some authors have questioned altogether whether temporary contracts are appropriate indicators of objective job insecurity (De Witte and Näswall, 2003), while permanent employment is not equivalent to security, either: especially young workers lacking seniority are often the last hired and first fired and not necessarily safe in ‘permanent’ jobs (see Gebel, 2010). On the bivariate level, young permanent employees in this study did feel more secure than their peers on temporary contracts, and the same persons also felt less secure when they worked with a temporary contract compared to times when they were permanently employed. Previous research has also shown an overlap between objective and subjective job insecurity (Keim et al., 2014; Näswall and De Witte, 2003). Yet the pathway from temporary employment to mental health and satisfaction is still unclear. Some authors have argued for interactions between objective and subjective job insecurity such that temporary workers are more exposed to, but less affected by subjective job insecurity, because they do not expect security from their employer (e.g., De Cuyper and De Witte, 2007). Such expectations, as well as volition, should be accounted for in future research. Further, the study design did not account for chronic vs. occasional job insecurity or the timing of different employment forms. Future research should take this into account using more dynamic methodologies.

The results concerning subjective job insecurity add to the evidence of its negative associations with mental health and satisfaction. Few studies have so far employed a within-person design (e.g., Vander Elst et al., 2016) to reduce endogeneity between subjective job insecurity and outcomes. This study also clearly showed that despite younger workers reacting less strongly to job insecurity than their older colleagues (see Cheng and Chan, 2008), they should not be overlooked. The effect sizes of subjective job insecurity were rather small, but this was not surprising given that within-person variability in the variables under study tends to be smaller than between-person differences (cf. Kinnunen et al., 2014), and that young workers are probably quite resilient against stressors. Nonetheless, in line with Peiró et al. (2012), the associations between subjective job insecurity and deteriorations in mental health and satisfaction are already substantial at the beginning of the career.

The role of education

There was some evidence of an interaction between educational level and subjective job insecurity when predicting mental health, which provided mixed support for Hypothesis 3: young workers with a university degree were better off than those with vocational training when experiencing job insecurity, suggesting a protective effect of tertiary education. However, both of these groups were worse off than the low-qualified, whose mental health was virtually unaffected by changes in subjective job insecurity. This could be attributable to lower expectations among the low-qualified, such that having a job at all is decisive for their mental health. They also enter employment at a younger age and are not trained for specific occupations, so they may be less attached to the labor market and have a less stable occupational identity. The association of job insecurity with mental health among university graduates was negative, but smaller than among those with vocational training and not significant. This may be attributable to the smaller subsample size, as well as better job prospects among university graduates. It is noteworthy that the middle group of workers with vocational training was the worst off in terms of mental health. They have already accumulated a few years of experience during their training and might therefore expect more security. Furthermore, as the vocational training system is designed to ensure a smooth transition from education to employment, young adults who value security more than others are likely to self-select into this stream of education.

However, the interaction between educational level and subjective job insecurity was neither replicated across measurements of job insecurity, nor across outcome variables. It is possible that the interaction that was detected with regard to mental health resulted from a selection bias: mental health is included only biennially in the SOEP, whereas job satisfaction and life satisfaction are available in every wave. This implied that to be included in the analyses, respondents had to be observed as working not just in any two years within the study period, but two even-numbered years. This resulted in a larger dropout when predicting mental health, while dropout was in turn related to lower education. Perhaps the associations between job insecurity and mental health were underestimated in the group of low-qualified young workers, because only the particularly resilient ones managed to maintain employment when confronted with job insecurity. Descriptive analyses (available upon request) revealed that the proportion of persons staying in employment after the first labor market entry was lowest in the group of low-qualified workers, while the respective shares of unemployed persons and those returning to education were the highest. Moreover, the healthy worker effect has been found to be particularly pronounced among persons with lower socio-economic status (Dahl, 1993).

Evidence of educational differences in health reactions to job insecurity among young workers was overall weak, although effect sizes were not always small. This may be attributable to the fact that FE regression tends to yield less efficient estimates with larger standard errors than other regression methods, because only within-person variance is used (Allison, 2009). Future studies may be able to detect differing vulnerabilities across educational groups with higher statistical power.

Finally, even if in reality there are no educational differences at all in reactions to job insecurity, that does not necessarily mean that education does not matter in this context: low-qualified young workers are in a particularly weak position in the labor market fostering disproportionate selection into insecure jobs (Gebel and Giesecke, 2009), and thus at greater risk of experiencing job insecurity early in their career.

Strengths and limitations

Some general strengths and limitations of this study should be kept in mind when interpreting the results. First, the sample was selective as dropout was related to some of the study variables, for example, temporary workers and low-qualified workers were more likely to drop out. The former may have biased the estimates for objective job insecurity, whereas the latter may have contributed to a healthy worker effect (Dahl, 1993). The generalizability of the findings is still high, as a large sample of young workers was investigated across regions, sectors and occupations.

Whether the findings can be replicated across different countries and labor market contexts is subject to future research. The operationalization of educational levels reflected the German peculiarity of a standardized vocational training system, and the importance of certified field-specific qualifications in the labor market (Mills and Blossfeld, 2005): vocational training was thus distinguished from other types of education, and respondents with upper-secondary school education but no further qualifications were classified as ‘low-qualified’, although they are certainly not low-educated compared with adolescents who have only compulsory school education. This approach was chosen to reflect respondents’ qualifications, because work-related aspects such as job prospects and quality were assumed as key mechanisms underlying the moderating effect of education. Other country contexts or a focus on other mechanisms would call for a different operationalization.

Context influences on the job insecurity–health relationship were not considered, such as the availability of secure jobs in a given industry or job prospects associated with occupation-specific types of qualifications, which might entail differences within educational groups. This was beyond the scope of this article, but suggests avenues for future research, for example comparing levels of job insecurity and its associations with mental health and satisfaction among labor market entrants across a theoretically-informed selection of sectors. Moreover, gender differences in the experience of, and vulnerability to job insecurity were not considered. Keim et al.’s meta-analysis (2014) found no substantial gender difference with regard to the experience of subjective job insecurity, but Buchholz (2008) observed that young women faced greater difficulties entering permanent employment than young men. Gender differences due to selection into certain types of sectors and work arrangements should be investigated in more detail in future research. Finally, the use of single-item measures may have reduced the reliability of the results and did not allow for differentiating between different facets of job insecurity (e.g., cognitive vs. affective, see Sverke and Hellgren, 2002). This was a matter of data availability and considered an acceptable drawback of secondary analysis in relation to the benefits of using large-scale longitudinal data.

Regarding causal inference, the directionality of effects was not tested, and a potential endogeneity bias due to self-selection into temporary employment or due to time-varying unobserved variables cannot be ruled out. However, the former has been done elsewhere (e.g., Hellgren and Sverke, 2003; Vander Elst et al., 2016). With regard to the latter, controlling for job changes at least ruled out the possibility that changes in outcomes were attributable to changes in work environments, which might cause both job insecurity to rise and mental health and satisfaction to decline.

Nevertheless, the longitudinal design observing young workers from a comparable starting point was a major strength with regard to causal inference, because changes within persons were analyzed. The majority of job insecurity research has focused on inter-individual differences. Keim et al.’s (2014) call for FE analysis to reduce endogeneity between subjective job insecurity and its predictors thus also holds for the analysis of job insecurity and its outcomes. The strength of this method is that unobserved heterogeneity is controlled for (Andreß et al., 2013). This is an advantage for stress research, because the reported association of subjective job insecurity with mental health and satisfaction cannot be explained as a function of stable dispositions.

Implications

This study confirms the role of perceived job insecurity as a stressor for young workers. The findings further support transactional stress theory (Lazarus and Folkman, 1984) in emphasizing the relative importance of subjective appraisals of the stressor in predicting mental health and satisfaction. As for objective job insecurity in terms of temporary employment, further theory development is needed to conceptualize the pathway from the contract type to individual outcomes, taking into account volition and preferences, but also such temporal aspects as time left before contract expiration, repeated temporary contracts, and expectations about future contract renewal or permanent employment (see Clinton et al., 2011). The role of time is also important to consider from a within-person perspective, as workers might not benefit from reductions in subjective job insecurity to the same extent as their mental health is reduced through increases (see Ferrie et al., 2002). Likewise, transitions from permanent to temporary employment versus transitions from temporary to permanent employment may not result in equivalent changes in mental health. With regard to educational differences, this study provided little support for differential vulnerability, suggesting that differential exposure might be the more important mechanism through which working conditions contribute to occupational health inequalities (see Landsbergis et al., 2014). Nevertheless, when investigating specific stressors such as job insecurity, future research may benefit from theorizing and directly testing different mechanisms such as employability or relative deprivation through which education may moderate their association with outcomes.

Finally, this study has some practical implications. Young workers are in a formative phase of occupational socialization during which persistent work attitudes are shaped (Peiró et al., 2012). Employers and policy makers should thus create sustainable career paths for newcomers to avoid early frustration and dissatisfaction through experiences of job insecurity. Whenever possible, temporary jobs should be designed as a bridge into more stable employment. Open communication about future prospects after contract expiration, as well as support in career development may help reduce job insecurity and protect young workers’ mental health from more severe impairment in the future (see Clinton et al., 2011; Keim et al., 2014).

Conclusion

This study is the first to investigate the association of subjective and objective job insecurity with mental health and satisfaction among young workers with three different levels of education (low-qualified, vocational training and university degree), thus highlighting an emerging population at risk. The findings underline the relative importance of subjective job insecurity compared to objective indicators. A consistent intra-individual decline in mental health, job satisfaction and life satisfaction when perceiving job insecurity was already visible at the beginning of young workers’ careers. Evidence of differential vulnerability to job insecurity across educational levels was weak and suggests nonlinear effects of education as a moderator. Future research is needed to solve both the puzzle of how temporary employment affects mental health and well-being, as well as how education might affect susceptibility to job insecurity.

Footnotes

Acknowledgements

I would like to thank Hilke Brockmann, Sonja Drobnič, Magnus Sverke and the two reviewers for their helpful comments on earlier versions of this article, as well as Ulrike Ehrlich, Nora Waitkus, Lara Minkus and Olaf Groh-Samberg for sharing their expertise about the SOEP data.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: BIGSSS (Grant / Award Number: ‘DFG grant GSC 263’).

Author biography

Katharina Klug is a fellow at the Bremen International Graduate School of Social Sciences at the University of Bremen and Jacobs University. Her research interests include work stress and employee well-being, job insecurity and occupational health inequalities.

References

Allison

(2009) Fixed-Effects Regression Models. Thousand Oaks, CA: Sage.

Allmendinger

Hipp

Stuth

(2013) Atypical Employment in Europe 1996–2011. WZB Discussion Paper 3, Berlin.

Andersen

Mühlbacher

Nübling

, et al. (2007) Computation of standard values for physical and mental health scale scores using the SOEP version of SF-12v2. Schmollers Jahrbuch 127: 171–182.

Andreß

H-J

Golsch

Schmidt

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

Aronsson

Gustafsson

Dallner

(2002) Work environment and health in different types of temporary jobs. European Journal of Work and Organizational Psychology 11(2): 151–175.

Bardasi

Francesconi

(2004) The impact of atypical employment on individual wellbeing: Evidence from a panel of British workers. Social Science and Medicine 58(9): 1671–1688.

Beard

Edwards

(1995) Employees at risk: Contingent work and the psychological experiences of contingent workers. In: Cooper

Rousseau

(eds) Trends in Organizational Behavior, Vol. 2. New York: John Wiley & Sons, pp. 109–126.

Bernhard-Oettel

Rigotti

Clinton

, et al. (2013) Job insecurity and well-being in the temporary workforce: Testing volition and contract expectations as boundary conditions. European Journal of Work and Organizational Psychology 22(2): 203–217.

Bernhard-Oettel

Sverke

De Witte

(2005) Comparing three alternative types of employment with permanent full-time work: How do employment contract and perceived job conditions relate to health complaints? Work and Stress 19(4): 301–318.

10.

Berntson

Sverke

Marklund

(2006) Predicting perceived employability: Human capital or labour market opportunities? Economic and Industrial Democracy 27(2): 223–244.

11.

Brauns

Steinmann

(1997) Educational Reform in France, West-Germany, the United Kingdom and Hungary: Updating the CASMIN Educational Classification. MZES Working Paper 21, Mannheim.

12.

Brzinsky-Fay

(2007) Lost in transition? Labour market entry sequences of school leavers in Europe. European Sociological Review 23(4): 409–422.

13.

Buchholz

(2008) Die Flexibilisierung des Erwerbsverlaufs. Eine Analyse von Einstiegs- und Ausstiegsprozessen in Ost- und Westdeutschland. Wiesbaden: VS Verlag für Sozialwissenschaften.

14.

Büssing

(1999) Can control at work and social support moderate psychological consequences of job insecurity? Results from a quasi-experimental study in the steel industry. European Journal of Work and Organizational Psychology 8(2): 219–242.

15.

Cazes

Tonin

(2010) Employment protection legislation and job stability: A European cross-country analysis. International Labour Review 149(3): 261–285.

16.

Cheng

GH-L

Chan

DK-S

(2008) Who suffers more from job insecurity? A meta-analytic review. Applied Psychology: An International Review 57(2): 272–303.

17.

Clinton

Bernhard-Oettel

Rigotti

, et al. (2011) Expanding the temporal context of research on non-permanent work: Previous experience, duration of and time remaining on contracts and employment continuity expectations. Career Development International 16(2): 114–139.

18.

Dahl

(1993) Social inequality in health: The role of the healthy worker effect. Social Science and Medicine 36(8): 1077–1086.

19.

De Cuyper

De Witte

(2007) Job insecurity in temporary versus permanent workers: Associations with attitudes, well-being, and behaviour. Work and Stress 21(1): 65–84.

20.

De Cuyper

De Jong

De Witte

, et al. (2008) Literature review of theory and research on the psychological impact of temporary employment: Towards a conceptual model. International Journal of Management Reviews 10(1): 25–51.

21.

De Witte

(1999) Job insecurity and psychological well-being: Review of the literature and exploration of some unresolved issues. European Journal of Work and Organizational Psychology 8(2): 155–177.

22.

De Witte

Näswall

(2003) ‘Objective’ vs ‘subjective’ job insecurity: Consequences of temporary work for job satisfaction and organizational commitment in four European countries. Economic and Industrial Democracy 24(2): 149–188.

23.

De Witte

Pienaar

De Cuyper

(2016) Review of 30 years of longitudinal studies on the association between job insecurity and health and well-being: Is there causal evidence? Australian Psychologist 51(1): 18–31.

24.

Dekker

SWA

Schaufeli

(1995) The effects of job insecurity on psychological health and withdrawal: A longitudinal study. Australian Psychologist 30(1): 57–63.

25.

Diener

Inglehart

Tay

(2013) Theory and validity of life satisfaction scales. Social Indicators Research 112(3): 497–527.

26.

Domenighetti

D’Avanzo

Bisig

(2000) Health effects of job insecurity among employees in the Swiss general population. International Journal of Health Services 30(3): 477–490.

27.

Erlinghagen

(2007) Self-perceived job insecurity and social context: A multi-level analysis of 17 European countries. European Sociological Review 24(2): 183–197.

28.

Eurofound (2013) Working Conditions of Young Entrants to the Labour Market. Dublin: Eurofound.

29.

Federal Statistical Office (2015) Stand und Entwicklung der Erwerbstätigkeit in Deutschland. Fachserie 1 Reihe 4.1.1. Wiesbaden: Destatis.

30.

Feldman

Turnley

(2004) Contingent employment in academic careers: Relative deprivation among adjunct faculty. Journal of Vocational Behavior 64(2): 284–307.

31.

Ferrie

Shipley

Stansfeld

, et al. (2002) Effects of chronic job insecurity and change in job security on self reported health, minor psychiatric morbidity, physiological measures, and health related behaviours in British civil servants: The Whitehall II study. Journal of Epidemiology and Community Health 56(6): 450–454.

32.

Fiori

Rinesi

Spizzichino

, et al. (2016) Employment insecurity and mental health during the economic recession: An analysis of the young adult labour force in Italy. Social Science and Medicine 153: 90–98.

33.

Gangl

(2002) Changing labour markets and early career outcomes: Labour market entry in Europe over the past decade. Work, Employment and Society 16(1): 67–90.

34.

Gebel

(2010) Early career consequences of temporary employment in Germany and the UK. Work, Employment and Society 24(4): 641–660.

35.

Gebel

Giesecke

(2009) Labour market flexibility and inequality: The changing risk patterns of temporary employment in West Germany. Zeitschrift für Arbeitsmarktforschung 42(3): 234–251.

36.

Greenhalgh

Rosenblatt

(1984) Job insecurity: Toward conceptual clarity. Academy of Management Review 9(3): 438–448.

37.

Heaney

Israel

House

(1994) Chronic job insecurity among automobile workers: Effects on job satisfaction and health. Social Science and Medicine 38(10): 1431–1437.

38.

Hellgren

Sverke

(2003) Does job insecurity lead to impaired well-being or vice versa? Estimation of cross-lagged effects using latent variable modelling. Journal of Organizational Behavior 24(2): 215–236.

39.

ILO (2012) Global Employment Trends for Youth. Geneva: ILO.

40.

Infratest Sozialforschung (2011a) SOEP 1990/91 – Methodenbericht Ostdeutschland zu den Befragungsjahren 1990/91 (Welle 1/2 – Ost) des Sozio-oekonomischen Panels. SOEP Survey Papers 14: Series B, Berlin.

41.

Infratest Sozialforschung (2011b) SOEP 2003 – Methodenbericht zum Befragungsjahr 2003 (Welle 20) des Sozio-oekonomischen Panels. SOEP Survey Papers 46: Series B, Berlin.

42.

Infratest Sozialforschung (2012) Soep 2011 – Methodenbericht Zum Befragungsjahr 2011 (Welle 28) Des Sozio-Oekonomischen Panels. SOEP Survey Papers 108: Series B, Berlin.

43.

Jahoda

(1982) Employment and Unemployment: A Social-Psychological Analysis. Cambridge: Cambridge University Press.

44.

Keim

Landis

Pierce

, et al. (2014) Why do employees worry about their jobs? A meta-analytic review of predictors of job insecurity. Journal of Occupational Health Psychology 19(3): 269–290.

45.

Kim

I-H

Muntaner

Vahid Shahidi

, et al. (2012) Welfare states, flexible employment, and health: A critical review. Health Policy 104(2): 99–127.

46.

Kim

Von dem Knesebeck

(2016) Perceived job insecurity, unemployment and depressive symptoms: A systematic review and meta-analysis of prospective observational studies. International Archives of Occupational and Environmental Health 89(4): 561–573.

47.

Kinnunen

Mäkikangas

Mauno

, et al. (2014) Development of perceived job insecurity across two years: Associations with antecedents and employee outcomes. Journal of Occupational Health Psychology 19(2): 243–258.

48.

Klandermans

Hesselink

Van Vuuren

(2010) Employment status and job insecurity: On the subjective appraisal of an objective status. Economic and Industrial Democracy 31(4): 557–577.

49.

Landsbergis

Grzywacz

LaMontagne

(2014) Work organization, job insecurity, and occupational health disparities. American Journal of Industrial Medicine 57(5): 495–515.

50.

László

Pikhart

Kopp

, et al. (2010) Job insecurity and health: A study of 16 European countries. Social Science and Medicine 70(6): 867–874.

51.

Lazarus

Folkman

(1984) Stress, Appraisal and Coping. New York: Springer.

52.

Lehtinen

Ozamiz

Underwood

, et al. (2004) The intrinsic value of mental health. In: Promoting Mental Health. Concepts – Emerging Evidence – Practice. Geneva: World Health Organization, pp. 46–58.

53.

Lim

VKG

(1996) Job insecurity and its outcomes: Moderating effects of work-based and nonwork-based social support. Human Relations 49(2): 171–194.

54.

Loughlin

Barling

(2001) Young workers’ work values, attitudes and behaviours. Journal of Occupational and Organizational Psychology 74(4): 543–558.

55.

Lowe

Krahn

(2000) Work aspirations and attitudes in an era of labour market restructuring: A comparison of two Canadian youth cohorts. Work, Employment and Society 14(1): 1–22.

56.

McLean Parks

Kidder

Gallagher

(1998) Fitting square pegs into round holes: Mapping the domain of contingent work arrangements onto the psychological contract. Journal of Organizational Behavior 19: 697–730.

57.

Mandemakers

Monden

(2013) Does the effect of job loss on psychological distress differ by educational level? Work, Employment and Society 27(1): 73–93.

58.

Marmot

Siegrist

Theorell

(2011) Health and the psychosocial environment at work. In: Marmot

Wilkinson

(eds) Social Determinants of Health, 2nd edn. Oxford: Oxford University Press, pp. 97–130.

59.

Mauno

Kinnunen

Mäkikangas

, et al. (2005) Psychological consequences of fixed-term employment and perceived job insecurity among health care staff. European Journal of Work and Organizational Psychology 14(3): 209–237.

60.

Mills

Blossfeld

H-P

(2005) Globalization, uncertainty and the early life course. A theoretical framework. In: Blossfeld

H-P

Klijzing

Mills

, et al. (eds) Globalization, Uncertainty and Youth in Society. London: Routledge, pp. 1–24.

61.

Mills

Blossfeld

H-P

Klijzing

(2005) Becoming an adult in uncertain times. A 14-country comparison of the losers of globalization. In: Blossfeld

H-P

Klijzing

Mills

, et al. (eds) Globalization, Uncertainty and Youth in Society. London: Routledge, pp. 423–441.

62.

Monden

CWS

(2005) Current and lifetime exposure to working conditions. Do they explain educational differences in subjective health? Social Science and Medicine 60(11): 2465–2476.

63.

Müller

Steinmann

Ell

(1995) Education and Labour Market Entry in Germany. MZES Working Papers 10, Mannheim.

64.

Näswall

De Witte

(2003) Who feels insecure in Europe? Predicting job insecurity from background variables. Economic and Industrial Democracy 24(2): 189–215.

65.

Nunez

Livanos

(2015) Temps ‘by choice’? An investigation of the reasons behind temporary employment among young workers in Europe. Journal of Labor Research 36(1): 44–66.

66.

OECD (2002) OECD Employment Outlook 2002. Paris: OECD Publishing.

67.

OECD (2013) Government at a Glance 2013. Paris: OECD Publishing.

68.

OECD (2014) OECD Employment Outlook 2014. Paris: OECD Publishing.

69.

Pearce

(1998) Job insecurity is important, but not for the reasons you might think: The example of contingent workers. In: Cooper

Rousseau

(eds) Trends in Organizational Behavior, Volume 5. Chichester: Wiley, pp. 31–46.

70.

Peiró

Sora

Caballer

(2012) Job insecurity in the younger Spanish workforce: Causes and consequences. Journal of Vocational Behavior 80(2): 444–453.

71.

Quintini

Martin

(2007) The Changing Nature of the School-to-Work Transition Process in OECD Countries. IZA Discussion Paper 2582, Bonn.

72.

Rogers

(1993) Regression standard errors in clustered samples. Stata Technical Bulletin 13: 19–23.

73.

Schaufeli

(1992) Unemployment and mental health in well and poorly educated school-leavers. In: Verhaar

Jansma

(eds) On the Mysteries of Unemployment: Causes, Consequences and Policies. Dordrecht: Kluwer Academic, pp. 253–271.

74.

Schoon

Silbereisen

(2009) Conceptualising school-to-work transitions in context. In: Schoon

Silbereisen

(eds) Transitions from School to Work. Cambridge: Cambridge University Press, pp. 3–29.

75.

Shell Deutschland (2015) 17. Shell Jugendstudie. Jugend 2015. Frankfurt am Main: Fischer.

76.

Siegrist

Marmot

(2004) Health inequalities and the psychosocial environment: Two scientific challenges. Social Science and Medicine 58(8): 1463–1473.

77.

SOEP Group (2015) SOEP 2013 – SOEPmonitor Individuals 1984–2013 (SOEP v30). SOEP Survey Papers 284: Series E, Berlin.

78.

Sverke

Hellgren

(2002) The nature of job insecurity: Understanding employment uncertainty on the brink of a new millennium. Applied Psychology: An International Review 51(1): 23–42.

79.

Sverke

Hellgren

Näswall

(2002) No security: A meta-analysis and review of job insecurity and its consequences. Journal of Occupational Health Psychology 7(3): 242–264.

80.

Turner

(1995) Family, economic context, and the health effects of unemployment. Journal of Health and Social Behavior 36(3): 213–229.

81.

Vander Elst

Näswall

Bernhard-Oettel

, et al. (2016) The effect of job insecurity on employee health complaints: A within-person analysis of the explanatory role of threats to the manifest and latent benefits of work. Journal of Occupational Health Psychology 21(1): 65–76.

82.

Wagner

Frick

Schupp

(2007) The German Socio-Economic Panel Study (SOEP) – Evolution, Scope and Enhancements. SOEPpapers on Multidisciplinary Panel Data Research 1, Berlin.

83.

Wanous

Reichers

Hudy

(1997) Overall job satisfaction: How good are single-item measures? Journal of Applied Psychology 82(2): 247–252.

84.

Ware

Kosinski

Keller

(1996) A 12-item short-form health survey: Construction of scales and preliminary tests of reliability and validity. Medical Care 34(3): 220–233.

85.

Warr

(1994) A conceptual framework for the study of work and mental health. Work and Stress 8(2): 84–97.