Abstract
The so-called flexicurity approach suggests that security for employees can be successfully combined with flexibility for organizations and companies. This article studies if affective job insecurity (worry about losing one’s job) is compensated for by perceptions of employment security (possibilities of finding an equal or better job) and income security. Data derive from a survey carried out in 2010 among employees in Sweden. The main findings are that cognitive job insecurity (the perceived risk of job loss) increases affective job insecurity, whereas both employment and income security have the opposite effect. Moreover, cognitive job insecurity and employment security interact, implying that the effect of cognitive job insecurity on affective job insecurity is reduced in the presence of employment security but is reinforced in the absence of it. These results are discussed in relation to the flexicurity approach, concluding that flexicurity may be a risky venture for employees.
Introduction
In the discussion about ‘flexicurity’, it has been suggested that security for employees can be combined with flexibility for organizations and companies (e.g. Bredgaard et al., 2005; Jørgensen and Madsen, 2007; Wilthagen and Tros, 2004). Three forms of labour market security – job, employment and income security – have been particularly in focus (Gazier, 2007; Leschke et al., 2007; Wilthagen and Tros, 2004 1 ). The first has to do with possibilities of keeping a certain job, the second with the chances of being employed but not necessarily with the same employer and the third with financial compensation in case of joblessness. Searching for a flexible labour market and a competitive economy, advocates of flexicurity argue in favour of a specific combination of policies related to these forms of security. They assert that it is important to avoid strict job protection in order to grant firms numerical flexibility. A possible implication is that job-holders become insecure in their positions. The proponents however believe that this can be compensated for by a high degree of employment security; flexicurity should generate a dynamic economy with high levels of job creation and an institutional environment facilitating transitions to them. In addition, income security should be created by generous benefits for those who cannot find work immediately. Denmark and the Netherlands have often been held up as flexicurity model countries.
It has been questioned whether the flexicurity path leads to the goals its proponents wish to attain (e.g. Berglund and Furåker, 2011; Burchell, 2009; Calmfors, 2007). From the perspective of employees this path may be a risky endeavour; several studies show that the experience of job insecurity is an important stressor in working life (Cheng and Chan, 2008; Sverke et al., 2002). Such an experience may, for example, contribute to impaired health, job dissatisfaction, low organizational commitment and strong turnover intentions. The key question raised in this article is whether employment and income security really counteract or compensate for the stress of job insecurity. Do good possibilities of finding a new job ease people’s worries of losing their job? And is that kind of worry reduced by knowing that – in case of unemployment – one has some reasonable income protection? To throw light on these questions we analyse data from a survey carried out in 2010 among employees in Sweden. More than 2000 respondents are included in the dataset. Sweden must be considered an interesting case, because it is often described as a country in which people are highly protected from the risks associated with unemployment and income loss (e.g. Kenworthy, 2008; van den Berg et al., 1997).
The present article is organized in the following way. Next we provide a conceptual and theoretical discussion on different types of labour market security and insecurity. Then there is a section on the dataset and the methods to be used. This is followed by a presentation of the empirical analyses and the results. At the end we summarize the conclusions to be drawn and discuss the risks associated with flexicurity in a Swedish context.
Conceptual and theoretical issues
The flexicurity paradigm implies that employment protection legislation (EPL) is weak and this is expected to promote a volatile labour market with high rates of job creation and job destruction. Consequently employees may experience recurrent job losses and perhaps also recurrent periods of unemployment. In a Nordic study of labour market mobility, higher rates of transitions into unemployment were found in Denmark than in Norway and Sweden (Berglund and Furåker, 2011; Berglund et al., 2010). At any rate, flexicurity policies can be assumed to bring about a labour market with relatively high levels of job insecurity for employees.
Job insecurity
Job insecurity can be defined in different ways. A first distinction is usually made between an objective and a subjective dimension (De Witte, 2005; Sverke et al., 2002). Objective job insecurity refers to the actual risk – as assessed by some outside observer – of a job loss for a given category of workers. It is then, among other things, relevant to consider the higher precariousness of fixed-term contracts compared to open-ended contracts (Berglund et al., 2010). Subjective job insecurity, on the other hand, is a matter of individuals’ own perceptions of the continuance of their employment. In a meta-analysis by Sverke et al. (2002: 243) it is defined as ‘the subjectively experienced anticipation of a fundamental and involuntary event related to job loss’. Notably, two persons equally insecure in an objective sense – for example, by having identical fixed-term contracts – may interpret their situation quite differently.
Referring to the subjective dimension, De Witte (2005: 2) defines job insecurity as ‘the perceived threat of job loss and the worries related to that threat’. This definition includes both the assessment of the situation and the emotional reactions to it. We believe, however, that it is important to distinguish between the informational and emotional sides of the phenomenon. Other researchers have made a distinction between the cognitive and the affective side of job insecurity (Anderson and Pontusson, 2007; Borg and Elizur, 1992; Huang et al., 2010; Muños de Bustillo and de Pedraza, 2010). The cognitive aspect refers to individuals’ assessments of the probability of losing their job. Although these assessments may include errors of judgement, there is generally a positive correlation between individuals’ evaluation of risks and de facto job loss (Dickerson and Green, 2012; Klandermans et al., 2010).
The other aspect – affective job insecurity – is the emotional reaction to estimations of a job loss. This conceptualization implies that the cognitive side precedes the affective side (Anderson and Pontusson, 2007; Huang et al., 2010). However, affective job insecurity is not only a consequence of the information of a possible job loss but also of limited means to handle the situation or inability to cope with this threat (Greenhalgh and Rosenblatt, 1984).
The loss of a job could consequently mean very different things to different persons. For an employee who has just planned to start an education it may not be important at all, whereas it may be a serious blow to another employee who has no alternative plans or immediate possibilities of making a living and of coping with the threat. Worry of losing one’s job can be regarded as a type of stress reaction and a risk factor for deteriorated well-being (De Witte, 1999; Huang et al., 2010).
Factors influencing subjective (cognitive or affective) job insecurity range from individual characteristics, for example personality traits, age, sex and education, to macro-economic factors, for example unemployment levels or changes in the unemployment level (De Witte, 2005; Erlinghagen, 2008). The relative precariousness of the work situation is important as well. Näswall and De Witte (2003) found that, in particular, class position (white-collar/blue-collar workers) and form of employment contract (permanent/temporary) matter.
Employment security
In the flexicurity formula, employment security is assumed to be a significant factor to counteract the negative consequences that a job loss can imply and it refers to the possibilities of staying employed but not necessarily with the same employer (Bredgaard et al., 2005: 23; Leschke et al., 2007: 340). These possibilities are believed to be enhanced by life-long learning institutions and active labour market policies, besides a dynamic economy in general (see also Berglund and Furåker, 2011).
Another concept with a similar meaning is that of employability. It is used in different ways in the literature. A typical definition is the following: ‘Employability involves the ability of workers to remain attractive for the labour market in terms of their skills and qualifications, by reacting and anticipating changes in tasks and the work environment, facilitated by the human resources development opportunities offered to them’ (Eurofound, 2008: 14). This concept focuses on individuals’ ability to qualify for jobs rather than on the demand for labour. Its impact in the debate represents a normative move; it ‘illustrates a shift from a systemic view of the labour market to a focus on individuals and their qualities’ (Garsten and Jacobsson, 2004: 2).
There are nevertheless many who use the concept of employability but pay attention to both supply and demand in the labour market. Forrier and Sels (2003: 106) underline that employability must be looked upon as a function of both individual characteristics (motivation, skills, educational level) and contextual factors (e.g. the availability of jobs). In this understanding the concept comes close to that of employment security.
It may be added that even the concept of ‘employability security’ has been suggested (Gazier, 2007; Leschke et al., 2007: 340). In this article, however, we stick to the concept of employment security and we then focus on individuals’ perceived possibilities of finding an at least equally good job with another employer. We try to avoid the normative emphasis on individual abilities and responsibilities but just assume that people’s chances of finding another job may be due both to individual characteristics and to demand in the labour market.
In the flexicurity perspective, a shift from job security to employment security is supposed to take place without too severe consequences. Affective job insecurity is one negative effect that is likely to be counteracted by employment security. In studies of employability, it is believed to work as a coping mechanism; by providing choices in the labour market, the anxiety of a possible job loss is reduced (De Cuyper et al., 2008). There is some empirical evidence in support of this assumption. Anderson and Pontusson (2007) have shown that individuals’ perceptions of how easy or difficult it would be to find an acceptable job are linked to affective job insecurity. Furthermore, Berntson and Marklund (2007) found that employability is positively associated with health and mental well-being. In a study by De Cuyper et al. (2008) a relationship between employability and subjective job insecurity is established.
Research focusing on employment security (e.g. Anderson and Pontusson, 2007; Furåker, 2010a, 2010b) and employability (e.g. Berntson et al., 2006) has emphasized several important antecedents. One crucial factor is age. In general, youth seem to be more optimistic about employment opportunities than older individuals. Another important factor is education; people with higher education generally have better chances in the labour market and this is mirrored in their outlook. Furthermore, the general labour market situation appears to be vital for perceptions of employment security (Furåker, 2010a).
Income security
A fundamental problem with a job loss is the risk of losing the main source of income and the associated financial worries (Sinclair et al., 2010). In response to these risks, the flexicurity paradigm emphasizes unemployment insurance as a potential compensating institution to create income security. There is some evidence of the mitigating consequences of unemployment insurance. Anderson and Pontusson (2007) found a significant association between macro-level spending on unemployment insurance and people’s worries of losing their job. However, there is less research examining the effects of subjective income security on affective job insecurity. When it comes to antecedents of income security, some clues appear in a study by Burgoon and Dekker (2010). These authors use a question on whether respondents experience a risk of falling into poverty. Such a risk is more frequently felt by part-time workers, temporary workers, people with previous unemployment experience and people with low incomes. In the present study income security refers to employees’ assessment as to whether they can manage to live on unemployment benefits in case of unemployment.
Expected relationships
The starting point for the present analysis is the flexicurity idea that job protection should be substituted for by employment and income security and the assumption that this change does not have to be detrimental for employees. One possible negative consequence of decreased job protection would be an increase in employees’ perceived risk of losing their job, that is, cognitive job insecurity. This threat is also believed to result in greater anxiety among employees of losing their job, that is, affective job insecurity. However, following the theoretical understanding of job insecurity, the affective response is not only a function of the severity of the perceived risk, but also of the available means to cope with the situation. Employment and income security may work by mitigating the consequences of a job loss for the individual (cf. Anderson and Pontusson, 2007).
We may also think of possible interaction effects of cognitive job insecurity, employment security and income security on affective job insecurity. Previous research has shown interaction effects between subjective job insecurity and employability in relation to different outcomes, for example health (Sverke et al., 2002). Green (2011) has demonstrated that job insecurity among employees who perceive few external job opportunities in the labour market is much more detrimental to health than job insecurity in combination with good job prospects. Silla et al. (2009) found that employability moderates the effect of job insecurity on life satisfaction but not on psychological distress. These studies do not, however, make use of the distinction between cognitive and affective job insecurity. Still, the results make it reasonable to assume an interaction effect between cognitive job insecurity and employment security reducing the impact of cognitive job insecurity on affective job insecurity. High levels of employment security may be looked upon as buffering the anticipated risks of a job loss. A similar reduction can also be expected due to the interaction between cognitive job insecurity and income security. In addition, there is a possible interaction between employment security and income security, but this combination can be supposed to reinforce the negative impact of both variables on affective job insecurity. We have not found any previous studies in which the last two interaction effects have been tested.
Data and methods
The data to be used here derive from a postal survey conducted in Sweden during the autumn of 2010. As the ordinary labour force survey (LFS) made up the sampling frame, two advantages were achieved. First, specific categories in the labour market could easily be selected. In the present study, employees aged 16–64 years were targeted. Second, we got access to the LFS variables and in that way we have detailed information about the respondents’ labour market situation besides what was covered in our own questionnaire. With a response rate of 54%, the number of respondents in the survey amounts to 2023.
The significance of different factors on employees’ experience of security and insecurity is estimated by means of ordinal logistic regression. The main reason for choosing this estimation technique is that the dependent variables consist of measurements on an ordinal-scale level (Long, 1997: 114ff.). However, to get the estimations to converge in reliable models (with not too few cases in the categories and meeting the parallel line assumption) some mergers of the categories in the dependent and independent variables must be carried out. Below, the variables and their function in the analyses are described.
The key dependent variable is affective job insecurity, theoretically described as people’s anxiety about losing their work. It is measured by the following indicator: ‘To what extent do you worry about losing your present job?’ Four alternative answers (besides ‘Do not know’) were presented: ‘I worry a great deal’, ‘I worry to some extent’, ‘I worry a little bit’ and ‘I do not worry at all’. In the models below, however, the two first alternatives have been collapsed, transforming it into a three-step ordinal variable.
Three other variables are of central importance. Cognitive job insecurity, employment security and income security will first constitute dependent variables. We then examine how social and structural factors affect employees’ perceptions in these regards. In a second step these variables will work as independent variables in an analysis of their relationship to affective job insecurity.
Cognitive job insecurity has to do with individuals’ perceived possibilities of losing their job. It is measured by the following question: ‘How do you assess the risk that in the coming 12 months you will lose your job?’ Five alternative answers (besides ‘Do not know’) were given: ‘Very large’, ‘Quite large’, ‘Neither large nor small’, ‘Quite small’ and ‘Very small’. In the analyses the alternatives ‘Very large’ and ‘Quite large’ and ‘Quite small’ and ‘Very small’ have been collapsed respectively, whereas the alternative in the middle remains intact.
Employment security refers to people’s assessments of their chances in the labour market and it is gauged by the question ‘In general, what do you think of your possibilities for the present of finding another job which is equal to or better than your current job?’ In other words, the question does not just refer to any kind of work but to an option which is at least equivalent to the one people already have. Response alternatives are the same as for cognitive job insecurity and are merged in the same way.
The measurement of income security rests upon the following question: ‘How would you/your household manage financially if you became unemployed and had to rely on unemployment benefits for three months but less than six months?’ Actually we also asked about both longer and shorter periods than three but less than six months, but we have chosen the latter period as the most relevant indicator for our purpose. The respondents could choose between five answers ranging from ‘Very well’ to ‘Very poorly’ (in addition to ‘Do not know’). On this variable the alternatives have been collapsed in the same manner as above. This way of operationalizing income security links respondents’ economic situation to the institutional context or, more precisely, to unemployment insurance. There may be other ways of measuring income security, for example by information about money saved or supportive social networks. We believe, though, that it is reasonable to put emphasis on unemployment insurance, as it has a crucial role in the discussion of flexicurity (cf. Bredgaard et al., 2005; Jørgensen and Madsen, 2007; Wilthagen and Tros, 2004). Notably, the question does not exclude the possibility that people have other resources – besides unemployment benefits – to fend for themselves, for example some buffer money or a spouse with a stable income.
In the coming analyses we examine the significance of a quite wide range of independent variables – most of them expected by earlier research to be related to perceptions of insecurity and security in the labour market (see the theoretical section). For the purpose of finding effective and parsimonious models, however, only variables with statistically significant relationships to the dependent variables are kept in the final analyses. Gender is actually the only original variable that has no significant relationship to any of the dependent variables; it has been considered but is not included in any of the models to be presented.
Most of the remaining independent variables are derived from the regular LFS. Age is added to the LFS from Swedish register data. Educational level is operationalized by the SUN 2000 standard, which is very comparable to the international standard ISCED 1997. It has been transformed into three categories, where the primary level collapses categories 1 and 2 in ISCED 1997, the secondary level is equal to category 3 and the tertiary level comprises ISCED categories 4–6. There are also variables on people’s family situation. Marital status (married, cohabitant or single) and children living at home (yes or no) are variables from the LFS. Household income is derived from the additional questionnaire. A question with nine alternative answers ranging from an income below 100,000 SEK up to 801,000 SEK or more have been changed into a variable with three categories (400,000 SEK or below, 401,000–800,000 SEK or 801,000 SEK or above). The characteristics of the employees’ jobs (indicators from the LFS) are captured by socioeconomic category (the Swedish SEI index in a five-category version), working time (up to 34 hours or 35 weekly hours or more), type of employment contract (temporary or open-ended), sector of employment (public or private), size of the workplace (1–10, 11–19, 20–49, 50–99, 100–499 or 500+) and tenure (up to 2 years, 3–10 years or 11 years or more). Finally, we have a variable measuring the unemployment rate on the county level (Swedish Län), aimed at controlling for the effect of the general labour market situation. The figures originate from Statistics Sweden and refer to the annual unemployment average in 2010.
The analyses start with a description of the distributions of securities and insecurities in the labour market. Then we examine the relationships between various individual, family and work factors and cognitive job insecurity, employment security and income security. This will give us a picture of how these forms of insecurity/security are structured among social categories in Sweden. Thereafter we concentrate on affective job insecurity, which is analysed in a series of models. The first merely includes the independent variables referring to individual, family and work characteristics. A second model also entails the perceptions of cognitive job insecurity, employment security and income security, now appearing as independent variables. At last, three additional models will be run in which the different interactions between cognitive job insecurity, employment security and income security are brought in.
Results
Our first step is to show the distributions of the perceptions of securities and insecurities in the labour market and this is done in Table 1. A little more than 5% think the risk is very or quite large that they will lose their present employment in the coming 12 months; that is, they have a rather high degree of cognitive job insecurity. Affective job insecurity is more widespread among employees in Sweden. About 9% worry a great deal or to some extent about losing their employment and if we also include the ones who worry a little bit (17.3%) we see that more than one-quarter of the employees report some degree of anxiety. In other words, some individuals are bothered about losing their job, although they do not consider this likely to happen. With respect to employment security, 36% perceive very or quite good opportunities of finding a new job that is at least equally good as their present employment. Finally, about half of the employees believe that they would manage well financially if they had to live on unemployment benefits for between three and six months.
Perceptions of various forms of insecurity and security among employees in Sweden (in percentages).
In Table 2 we present the results from the multivariate analyses regarding cognitive job insecurity, employment security and income security. On the first dependent variable, type of employment contract and tenure turn out to have the strongest impact. Temporary employees are much more likely than permanent employees to report cognitive job insecurity. This also holds for people with shorter tenure compared to those with the longest. Both these results should be seen in relation to the Swedish EPL, which is rather strict by international comparison although liberalized in the last decades (OECD, 2004: Ch. 2; Venn, 2009). Without going into details, we can note that open-ended contracts and longer tenure usually mean stronger protection.
Factors impacting on cognitive job insecurity, employment security and income security (ordinal logistic regression; odds ratios).
Levels of significance: * p < .05; ** p < .01; *** p < .001.
NS means ‘not significant’ and the variable is therefore not included in the model presented.
Other important factors are socioeconomic category, size of workplace and working time. Non-skilled workers perceive the highest risk of losing their job, but also skilled workers and lower and middle-level white collars are more insecure than higher white collars. The pattern concerning size of workplace has no obvious explanation. A perception of risk also appears more often among part-time workers compared to full-timers.
The last two variables with significant relationships to cognitive job insecurity are age and marital status. A little surprisingly, younger respondents to a lesser degree than the oldest perceive some considerable risk of losing their job. This result is contrary to the findings by Anderson and Pontusson (2007), but the positive view among the young is revealed only after control for type of employment contract and tenure, variables not included in the study mentioned. The pessimism among those who are single is not in the same way related to their labour market situation.
Concerning employment security, several factors turn out to have a significant impact. Youth are obviously more optimistic about their job opportunities in the labour market than are older employees, a result fully in line with earlier research (e.g. Furåker, 2010b). Educational level is important as well, which has also been shown before (Anderson and Pontusson, 2007; Berntson et al., 2006; Furåker, 2010b). Employees with primary or secondary education tend to assess their job prospects more negatively than employees with tertiary education. A related variable is socioeconomic category: non-skilled workers and lower white-collar workers are apparently less secure in this sense than higher white collars. The impact of socioeconomic category when educational level is controlled for indicates that skills acquired in working life may have an independent effect on perceived labour market prospects. Other significant variables are household income, working time and sector of employment. Employees with the lowest income see fewer opportunities in the labour market. The same pattern is found for part-time workers compared to full-time workers and for public sector employees compared to private sector employees.
Moreover, there is a significant pattern for tenure: the longer people stay in the same workplace the less likely it is that they perceive great possibilities of finding an equal or better job in the labour market. This result is especially interesting in contrast to that concerning cognitive job insecurity, according to which short time at the workplace is associated with insecurity. Finally, unemployment on the county level also shows a negative effect, that is, the higher the unemployment rate the lower the perceived employment security.
The last column in Table 2 presents the analysis regarding income security – measured by the question on how one would manage financially if one became unemployed and had to live on unemployment benefits for three up to six months. The youngest obviously consider themselves most secure in this respect. To have children living at home is accompanied with a relatively low degree of income security. Furthermore, household income has a decisive influence: the lower the income, the less inclined are respondents to report that they would do well or rather well on unemployment benefits. Yet another important factor is socioeconomic category: especially blue-collar workers tend to feel less financially protected than higher white collars.
Next we focus on the affective dimension of job insecurity. Table 3 shows how different factors impact on this kind of feeling. Five different models are presented. The first model analyses the background variables without control for cognitive job insecurity, employment security and income security, but the latter three factors are all included in Model 2. The last three models demonstrate, separately, each of the possible interaction effects between the key variables.
Factors impacting on affective job insecurity (ordinal logistic regression coefficients; odds ratios).
Levels of significance: * p < .05; ** p < .01; *** p < .001.
CJI = cognitive job insecurity; ES = employment security; IS = income security.
Looking first at Model 1 in Table 3, we find that age is significantly related to affective job insecurity: youth are less likely to be worried about losing their job than the oldest age category. Having children at home also has a clear impact; it implies a higher risk of anxiety. Household income is another important factor: people with smaller earnings are more often worried than those with larger earnings. A further major result is that temporary employees are more likely to be bothered than permanent employees. Similarly, shorter time at the workplace means a relatively greater tendency of being anxious about suffering a job loss. We also find that public sector employees are to a lesser degree worried than private sector employees. The result regarding size of workplace is less easy to interpret.
In Model 2 cognitive job insecurity, employment security and income security are introduced as independent variables. All three have significant effects in the expected directions: cognitive job insecurity entails a much higher risk of being worried, whereas employment security and income security are both associated with a lower degree of anxiety. To some extent, the introduction of these variables reduces the impact of some of the variables in Model 1. There are no longer any statistically significant differences on age, employment contract and size of workplace.
The last part of the analysis focuses on three interaction effects and the outcomes are shown in Models 3–5. However, it is only in Model 3 that a statistically significant interaction effect appears. In substance it implies that the impact of cognitive job insecurity on affective job insecurity is conditioned by employment security. To illustrate the interaction effect, additive and multiplicative mean models are presented in Figures 1a and 1b. Figure 1a (the additive model) shows that employees who see substantial opportunities in the labour market have a rather low probability (about 30%) to worry a great deal or to some extent, even though they believe the risk is large that they will lose their job in the coming 12 months. If we also take into account the interaction effect (Figure 1b), the probability drops to less than 20%. However, if employees think they have small opportunities to find a new job in the labour market, the probability of worry when the risk of losing one’s job is considered high is about 56% in the additive model (Figure 1a). Paying attention also to the interaction effect means that the probability of being worried goes up to 67% (Figure 1b).
Effects of cognitive job insecurity (CJI) on predicted probabilities to worry a great deal or to some extent for different values of employment security (ES) without interaction effects between CJI and ES. Mean model.
Effects of cognitive job insecurity (CJI) on predicted probabilities to worry a great deal or to some extent for different values of employment security (ES) with interaction effects between CJI and ES. Mean model.
Perhaps somewhat surprisingly we do not find any significant effects of the other two interactions – especially not of that between cognitive job insecurity and income security. With respect to this interaction, Table 3 shows a rather low odds ratio for the same category (1*1) that is significant for cognitive job insecurity and employment security, but this result is not significant (p = .177). There are few observations in the category and consequently a rather high risk of a type-II error. This justifies future research on the hypothesis of an interaction effect between cognitive job insecurity and income security.
Conclusion and discussion
According to our study during the autumn of 2010 very few employees (5%) in Sweden felt some considerable risk that they would lose their job in the coming 12 months. The most important factors explaining cognitive job insecurity are type of employment contract and tenure. Concerning tenure there are certainly informal processes of becoming established in the workplace that can explain the pattern: a long period of service tends to make people more safeguarded. There is also the formal aspect of the Swedish EPL; with up to 10 years of continuous employment, individuals get increasingly stronger legislated protection.
High employment security – which means perceiving good possibilities of finding another job equal or better than the current – was experienced by nearly 36% of employees in Sweden. The main factors accounting for this optimism are age, education, socioeconomic category, tenure and unemployment rate. The effects of age and education have also been found in previous research (Anderson and Pontusson, 2007; Berntson et al., 2006; Furåker, 2010a, 2010b). It seems that youth have greater confidence in their abilities and are also more adaptable to labour market demands; for example they can relatively easily switch to another job or occupation even if it requires that they move geographically. The significance of education is certainly a consequence of the strong demand for highly educated persons in the labour market. We can also expect to find some co-variation between psychological characteristics – for example self-efficacy – and education (see Berntson, 2008). Socioeconomic category is yet another factor with an impact on perceived employment security. When education is controlled for, we discover a unique effect of socioeconomic category that is not directly related to the skills developed in the formal educational system. Human capital may also be transmitted from the work situation – through training and work experience (Berntson et al., 2006).
A main effect on employment security refers to tenure. The pattern is totally reversed compared to that regarding cognitive job insecurity. This may be a matter of the possibilities of an internal career. Staying longer with an employer increases an employee’s chances of climbing up the organizational hierarchies. Even when people remain in unskilled jobs, without making a formal career, tenure may lead to improvements in the work situation. All this implies increasing difficulties – or is experienced as increasing difficulties – of finding an equal or better job in the external labour market. In other words, there seems to be a ‘lock-in’ effect of tenure (cf. Furåker, 2010a). The longer people stay with an employer, the less risk they see of losing their job, but at the same time the drive to try one’s wings in the labour market falters, because the perceived possibilities of obtaining a new job, equal or better than the present, tend to decline.
Finally, the significance of unemployment for employment security should be emphasized. In counties with high unemployment rates there is less optimism about opportunities to find a new job than in counties with lower unemployment rates. A similar result is found by Anderson and Pontusson (2007), who show that changes in national unemployment levels have an impact on people’s assessment of how easy it would be to find an acceptable job. Both results underline the importance of the demand for labour for the estimation of employment security in the labour market.
Income security here refers to employees’ judgement of how they would manage financially if they became unemployed and had to rely on unemployment benefits for three to six months. Almost 50% thought they would manage very or quite well. Employees with children living at home showed a lower degree of income security than those without children at home. Higher income and higher socioeconomic position seem to make people relatively more optimistic in this regard. The income variable in our survey does not measure the individual wage but household income, that is, there may be earnings from other family members as well. Another possibility is that high income earners to a larger extent have access to other assets, for example savings.
The last step of the analysis was to study the effect of cognitive job insecurity and employment and income security on worry of losing one’s job. Such worry summarizes the emotional reactions to cognitive job insecurity and the possible compensatory impact of employment and income security. At the time of our survey about one in 10 employees in Sweden felt some considerable anxiety about losing their job. Cognitive job insecurity is strongly associated with being worried and employment security has the opposite effect. Income security has a much weaker but still statistically significant influence in the presumed direction.
It was also hypothesized that interaction effects might exist between the different kinds of insecurity/security on anxiety about losing one’s job. However, only one of them was statistically significant; it is the interaction between cognitive job insecurity and employment security that has an impact on the relationship to affective job insecurity. It means that an employee turns out to be less worried if cognitive job insecurity is combined with employment security than if we look at the effects additively. This interaction is in line with an understanding of employment security as an important coping mechanism for job insecurity. At the same time, the interaction implies an increased risk of worry if an employee perceiving a high level of cognitive job insecurity sees few options in the labour market.
Let us then return to the flexicurity paradigm and its suggestion of a shift from job protection to employment and income security. A key question is whether the latter two forms of security really counteract or compensate for the stress that can be expected from having less job protection. Our study shows that cognitive job insecurity has an impact on people’s worries. If cognitive job insecurity to some extent mirrors objective insecurity, it appears likely that, ceteris paribus, a liberalization of the EPL would increase fears of a job loss. These negative feelings can, however, be counteracted, especially by good opportunities in the labour market but also to some extent by income compensation in case of unemployment.
From the perspective of employees, at first sight the flexicurity paradigm might seem to offer a reasonable tradeoff concerning insecurity and security, even though the estimated effect of cognitive job insecurity on affective job insecurity is larger than the counteracting effects of employment and income security. However, the revealed interaction effect between cognitive job insecurity and employment security also has to be taken into consideration to understand the tradeoff. In essence, it implies that EPL liberalization may not be very harmful to employees in an upward economic trend, but in a downward trend – when few job options exist – fears may be more severe than without this kind of liberalization.
In a Swedish context, the key institutional adjustment in accordance with the flexicurity paradigm would be liberalization of EPL. Sweden has recently made some significant changes of that kind, although they only concern temporary employment; regular employees are still well protected by international comparison (Venn, 2009). A central compensating institution in case of increased insecurity among employees is unemployment insurance. In recent years Swedish unemployment benefits have been made less generous: stricter work requirements and longer waiting periods have been implemented, the demands on the unemployed individuals’ search activities have been strengthened and the replacement rate has been lowered (Bengtsson and Berglund, 2012). The compensatory effect of income security on affective job insecurity found in the present study may therefore be lower than it used to be when unemployment insurance was more generous.
Contrary to the development in Sweden, the flexicurity paradigm suggests a strengthening of the compensatory effect of unemployment insurance. If we also take into consideration the general labour market situation in Sweden with high unemployment levels and no relief in sight, the flexicurity paradigm, especially a liberalization of EPL, seems to be a risky venture for employees. All in all, it appears that the compensatory effects of employment and income security might fail to appear. Flexicurity may be a good institutional set-up under certain circumstances, but if these are not at hand it may create greater risks for employees compared to more regulated conditions in the labour market.
The conclusions in this study must be evaluated with an eye on its limitations. First, the data at hand are cross-sectional, which means that we have to be careful with causal inferences. There is always a risk of reversed causality with such a design. To illustrate this, people who tend to be anxious may also tend to be pessimistic about their chances in the labour market. Second, the survey has a rather low response rate. Although we have not discovered any great biases, the response rate is lower for males, youth, temporary employees and employees with primary education. These categories are in general relatively more exposed to job insecurity and we may thus slightly underestimate the prevalence of it in Sweden. Third, one may ask how generalizable our findings are to contexts other than the Swedish. Is, for example, the significance of tenure for cognitive job insecurity and employment security a peculiarity for countries with the same kind of EPL? And do the effects of employment and income security vary between countries with better or worse labour market situations and more or less generous unemployment insurance? These are possible future research questions raised by the present study.
Footnotes
Acknowledgements
We wish to thank two anonymous reviewers for fruitful comments and Maria Solevid for valuable help with the statistical analysis.
Funding
The research reported in this article was funded by VINNOVA (2009-02883).