Fixed-term employment in Norway and Sweden: A pathway to labour market marginalization?

Abstract

We compare the effects of relatively liberal regulations on the use of temporary employment in Sweden and more restrictive rules in Norway. We find not only that temporary work may be a stepping stone out of unemployment but also that fixed-term employees are exposed to significant risks of long-term marginalization. Moreover, fixed-term employees in Sweden face greater risks of long-run unemployment and low earnings compared to those in Norway.

Keywords

Employment protection legislation fixed-term employment marginalization Norway segmentation Sweden

Introduction

This article studies the relationship between fixed-term employment and the long-term risk of labour market marginalization (defined as unemployment, low wages or dependence on social assistance after 5 years) in Norway and Sweden. Both countries can be considered inclusive employment regimes, where organized labour has strong influence and where policies are designed to achieve high employment levels and strong employment rights (Gallie, 2007). They are also prime examples of the Nordic welfare models, providing generous support in case of unemployment, sickness and old age (Esping-Andersen, 1990). Like most open economies, they have fostered mobility and adjustment through an interplay between market competition, solidaristic wage policies and active labour market policies.

However, in terms of employment protection legislation (EPL), the two countries have followed different paths. While both countries have strong protection for those on permanent contracts, Sweden has relaxed the rules on the use of temporary contracts several times in recent decades, and hence the share of such jobs is markedly higher than in Norway, where fixed-term employment remains strictly regulated. We therefore ask whether the risks associated with a temporary contract differ between the two countries.

Research has shown that temporary employment can function either as a stepping stone into a stable employment career, or as a trap leading into recurring insecure temporary jobs. The different trajectories result from employers’ and employees’ micro-level strategies. However, the choices and outcomes of these strategies are influenced by macro-level institutions, with EPL being one of the most significant (Blanchard and Landier, 2002; Polavieja, 2003). Strong protection for permanent workers entails high dismissal costs for employers, reducing their willingness to hire new employees on such contracts (Bassanini and Duval, 2006; Kahn, 2010) and increasing their incentives to use fixed-term contracts. If there is lax regulation of temporary contracts, jobseekers are therefore less likely to be offered permanent contracts, thus increasing temporary employment; but stricter regulation, however, creates barriers to their use. In such labour markets, employers may use fixed-term contracts primarily to increase numerical flexibility, which also secures the permanent core workers (Atkinson, 1984). High fixed-term employment shares may therefore lead to company-level segmentation between core and periphery (Olsen and Kalleberg, 2004) and broader labour market dualization.

Hence, our central research question is how far the effects of temporary employment on employees’ long-term risks of labour market marginalization vary between these countries. Deregulation of fixed-term employment in Sweden might lead to processes of dualization not expected in Norway. However, it is also possible that when used within an inclusive, high mobility employment regime, temporary employment functions more as a stepping stone than a trap, with limited long-term disadvantages for temporary employees. This would indicate that the Nordic model cushions the adverse effects of fixed-term employment deregulation.

We focus on the significance of temporary employment for the prime-age workforce (aged 25–54 years). Many young individuals enter the labour market from education with a temporary job, or use it to combine work and education, without seeking permanent or stable employment. Excluding those whose labour supply is affected by institutions outside the scope of this article (the education, pension and early retirement systems), the consequences of EPL come more to the foreground. Furthermore, as the phases of transition in and out of the labour market have become longer and more erratic than before, most fixed-term employees in Norway and Sweden are in other stages of their careers than labour market entry (Dieckhoff and Steiber, 2012: 22; Nergaard, 2016; Sackmann, 2001). While the proportion of temporary employees is highest among youth, most employees in the labour market are not in the youngest cohorts. Thus, most fixed-term employees in these countries have in fact passed the age of school-to-work transition phase.

Using a similar-country design, comparing two countries within an inclusive employment regime but with dissimilar regulation of temporary contracts, we study the long-term effects of temporary employment on individual risks of marginalization. Barbieri and Cutuli (2016: 510) find that EPL reforms at the margin hinder access to permanent employment and do not lead to improvements in overall employment rates. However, their findings also suggest that the fluid Nordic labour markets with low unemployment benefit more than other European countries from such reforms. Comparing these two countries, not together with several other countries by way of the crude OECD EPL index, but by a more informed study of national similarities and differences, provides a deeper understanding of the long-term consequences of fixed-term employment.

In both countries, EPL and the regulation of temporary employment are subject to heated political debate, as proponents portray partial deregulation as a way to include more vulnerable individuals into the labour market, while opponents claim it leads to increased labour market insecurity and dualization. However, there are few comparative studies of temporary employment within the prime-age workforce in these inclusive employment regimes, and none studying whether the long-term disadvantages of fixed-term employment are larger in Sweden than in Norway. A major strength of our study is that we combine the Labour Force Surveys (LFS) for the period 2000–2008 with detailed register data about employment and income components 2 years before and 5 years after LFS participation. Hence, we have a much longer panel for each cohort than usual in such studies, and more information about the respondents’ prior labour market connection than in ordinary LFS data, enabling us to evaluate the effects of temporary employment as elements of different trajectories for labour market integration.

Theory and previous research

As noted above, the effect of strong employment protection on aggregate employment and unemployment levels is ambiguous. Employers have limited incentives to use temporary employment in countries with lax regulation of permanent contracts, and fewer possibilities where the barriers to use of temporary contracts are high (Polavieja, 2003). A large gap between (strict) regulation of permanent contracts and (lax) regulations of fixed-term contracts should increase employers’ use of the latter, leading to labour market segmentation (Barbieri and Cutuli, 2016: 503; Passaretta and Wolbers, 2016; Polavieja, 2003). This would create a primary labour market with secure, permanent and stable employment, and a secondary market where there are insecure jobs, with temporary contracts, lower pay, worse working conditions and limited opportunities for mobility into the former (Doeringer and Piore, 1971).

A central issue in studies of fixed-term contracts is their impact on individual labour market careers. On one hand, temporary employment has been shown to work as a bridge from marginal labour market positions to more stable positions, thereby integrating fixed-term workers into the labour market. On the other hand, it can also work as a trap that leads to uncertain labour market positions with little prospect of transit into permanent employment, thus adding to the labour market segmentation.

Temporary employment might be expected to increase individuals’ probability of employment over time. It is difficult for employers to gain full insight into job applicants’ motivation and learning capabilities. Many staffing decisions are made under time strain, and thorough assessment of a number of applicants requires resources. Hence, fixed-term employment may be used to screen job candidates, serving as a form of probation (Houseman, 2001; Wang and Weiss, 1998).

Moreover, when job security for permanent employees is high, employers will tend to hire those who, through formal education or other traits, signal high productivity (Giesecke and Groß, 2003: 162). Those holding a temporary job rather than being unemployed preserve or increase their human capital (Becker, 1975); hence, a fixed-term contract can act as a bridge into more secure labour market attachment. In addition, a temporary employee can acquire job references, facilitating subsequent applications. Work experience in temporary employment can also build networks which provide access to unlisted jobs (McVicar et al., 2016). Hence, temporary jobs might serve as a stepping stone to more secure and better paid jobs.

While one study (Centeno and Novo, 2012) suggests that partial deregulation of fixed-term contracts leads mainly to job creation in the secondary labour market segment, other studies suggest that it lead to a substitution of permanent jobs by temporary employment (Blanchard and Landier, 2002; Gebel and Giesecke, 2011; Kahn, 2010). Previous research indicates that marginalized groups, such as youth and those with low education, are less likely to be employed in countries with strong EPL (Autor et al., 2006; Bassanini and Duval, 2006), but later studies indicate that this is not necessarily the case (Dieckhoff and Steiber, 2012; Kahn, 2010; Noelke, 2016). Thus, whether temporary positions integrate groups who have lower employment rates than others, such as youth, low educated and immigrants, by lowering the threshold for entry into the labour market, is still contested.

Some Nordic research supports the view that individuals on temporary contracts increase their chances of integration. In Norway, Nergaard (2004, 2016) finds that most fixed-term employees move to a permanent contract within 2 years, while Engebretsen et al. (2012) found that workers with a temporary position (especially those with low education and those older than 30 years) were more likely than those who were unemployed to obtain a permanent job within 1 year. Håkansson (2001) and Levin (1998), studying Sweden in the 1990s, found that fixed-term employees had a lower risk of ending up unemployed than those initially unemployed. In West Germany, Hagen (2003) found that temporary employment usually led to permanent work, while Booth et al. (2002), studying Britain, found that the wages of women who started in temporary employment and moved to permanent jobs caught up with those who started in permanent jobs. Finally, Picchio (2008) found that a temporary job, compared to unemployment, significantly increases the probability of having a permanent job 2 years later in Italy.

Becker (1975) asserts that the knowledge (human capital) used by a worker can be divided into two main types: general knowledge acquired through various forms of education, which can be used in many different companies, and specific knowledge developed in particular tasks in specific jobs which is often company-bound. Thus, those with valuable skills may be considered core workers, provided with permanent contracts, while those with limited skills are part of the periphery, only offered fixed-term contracts (Atkinson, 1984). Consequently, with workers on whom employers are less dependent, they may try to maximize numerical flexibility, and offer few training opportunities that facilitate promotion and wage increases. If legislation allows this strategy, peripheral temporary workers can function as a buffer, securing the jobs of insiders, creating labour segmentation at firm and labour market level.

Furthermore, temporary jobs can have a scarring (stigma) effect, as employers may wonder why a jobseeker did not previously hold a permanent job (Korpi and Levin, 2001; Yu, 2012); if so, we would expect lower earnings growth for temporary than for permanent workers, with recurring spells of unemployment (Mertens and McGinnity, 2005) and hence unstable careers. Research in a range of West European countries (Giesecke and Groß, 2003; Håkansson, 2001; Levin, 1998; Scherer, 2004) indicates that temporary employment and segmentation are interrelated and that temporary workers are more likely than permanent employees to become unemployed.

Regulatory and labour market differences between Norway and Sweden

As discussed above, employers’ motivations to use of fixed-term employment may vary with their institutional environments (Olsen and Kalleberg, 2004). According to the OECD EPL index (ranging from 0 to 6 with increasing strictness), dismissal of permanent employees was regulated with 2.33 strictness in Norway in 2010 and 2.66 in Sweden; but Swedish regulation of fixed-term contracts was only 0.81, while in Norway it was 3.0. The legislation on fixed-term contracts was liberalized in Sweden in the 1990s, making it among the most liberal in the EU. In Norway, the legislation has not undergone any substantial changes since 1996. In 2015, there were also significant changes in regulations, but outside the scope of this study.

Since 2007, employers in Sweden were not needed to specify any particular reason for using ‘general’ temporary contracts; previously, they were limited to a maximum of five ‘agreed’ fixed-term contracts. Fixed-term contracts can also be used for substitute workers, trainees, seasonal work and when the employee has reached the age of 67 years. In Norway, before 2015, temporary contracts could only be used for a temporary replacement or when ‘warranted by the nature of the work and the work differs from that which is ordinarily performed in the undertaking’; or for trainees, persons on labour market schemes, and for certain types of jobs within organized sports.

The rules limit the period a person may be temporarily employed in the same job. In Sweden, until 2007, a person employed on an agreed or a probationary fixed-term contract for 14 months within a 5 year period had a right to an open-ended contract. For substitutes, the same applied after 3 years. Between 2007 and 2016, an employee on a general fixed-term contract or working as a substitute had the right to a permanent contract if during a 5-year period they had been employed for more than 2 years. In Norway, until 2015, a person could be hired in the same temporary position or the same tasks for a maximum of 4 years before being considered permanently employed.

Reflecting the differences in regulation, over 11 percent of all employees aged 25–54 years (17% of all employees) were temporarily employed in Sweden in 2014, but under 6 percent in Norway (8% of all employees) (Eurostat, 2017a). The share of temporary employment in Norway has been around 6–7 percent in the age group 25–54 years since 2000, and there are no indications that the somewhat stricter regulations since 2006 influence the proportion. In Sweden, the share in the age group 25–54 years increased slightly from 2005, then decreased around 2008 before increasing again in 2014. Thus, there has been no sudden increase in the share of temporary employees following regulative liberalization in Sweden in 2003 and 2007. In both countries, the share of temporary employees is somewhat higher in care services and education, commonly public companies, and in the service sector.

Our comparison is based on a ‘most similar’ design (Ragin, 1987: 48), where the cases are similar in all but some respects. Still, there are some other differences between the two countries that might influence the long-term consequences of fixed-term employment. Not all between-country differences in labour market outcomes can be attributed to EPL differences. Changes in gross domestic product (GDP) may influence labour demand, but the trend throughout this period is rather similar, apart from a deeper economic recession and stronger recovery in Sweden in 2008–2010 (Eurostat, 2017b). The Swedish economy fared extremely well in the aftermath of the economic crisis of 2008. In general, full employment models with high labour demand should display less segmentation (DiPrete et al., 2001), and there were limited difference in employment levels between Norway and Sweden in the period 2000–2013, both around 83–86 percent in the age group 25–54 years (Eurostat, 2017c).

Labour market mobility and job openings are important for individuals’ possibilities of avoiding unemployment. Studies of mobility in Nordic labour markets in 2000–2006 show that the mobility rates from unemployment to employment within a 1-year period were higher in Norway than in Sweden (Berglund and Furåker, 2011). Conversely, mobility from employment to unemployment was higher in Sweden than in Norway. Furthermore, the likelihood of moving from temporary to permanent employment within 1 year was higher in Norway than in Sweden, while the probability of moving from a temporary job to unemployment was higher in Sweden (Svalund, 2013).

A study of the partial deregulation in Sweden in the 1990s found that most of the increased share in temporary employment in that period could be explained by the increase in unemployment at the beginning of the 1990s, and employers’ higher use of fixed-term employment in the subsequent period of job growth: depressed labour market conditions created incentives for firms to use, and exert pressure on employees to accept, temporary contracts (Holmlund and Storrie, 2002: 263).

The unemployment rate in the age group 25–54 years was higher in Sweden (between 3.7% and 6.3%) than in Norway (between 1.9% and 4.0%) in the 2000s, and the difference increased after the economic crisis of 2008. Summing up the institutional and labour market differences between the countries, we can conclude that Norway has stricter regulation of fixed-term contracts, higher shares of permanently employed and generally a better labour market situation than Sweden. Relating these facts, especially the differences regarding EPL, to previous studies (Olsen and Kalleberg, 2004; Passaretta and Wolbers, 2016; Polavieja, 2003), it seem reasonable to expect that Sweden may suffer labour market dualization (cf. Thelen, 2014: 188–190).

Hypotheses, data and methods

Our study focuses on the long-term consequences of temporary employment. Long-term is here defined as 5 years after the LFS measurement point (t). We examine the risks of marginalization that a temporary job may entail, taking unemployment, low earnings and receiving social assistance as indicators. While unemployment and low earnings signal a precarious labour market situation, dependence on social assistance can represent marginalization in a somewhat wider context, reflecting economic difficulties at household level. We anticipate the following:

The risk for marginalization at t + 5 is less for those with fixed-term employment at t than for unemployed at t (stepping stone)

There are larger long-term effects of fixed-term compared to permanent employment at t that increase the risk of marginalization at t + 5 (segmentation)

Dualization may entail a higher risk that temporary employees remain in the secondary segment of the labour market. Consequently, the probability of marginalization in the long-term should be greater in dualized labour markets. We therefore expect that

3. The difference in risk of marginalization between temporary and permanent employees is larger in Sweden than in Norway, and

4. The probability that temporary employees will be in a marginalized position is also larger in Sweden

To analyse the impact of temporary employment on future marginalization, we use the LFS from the years 2000–2008 (data for Sweden are missing for the year 2007) to identify labour market status: whether the individual had a permanent contract, a fixed-term contract or was unemployed. The LFS data for the last quarter of year t are combined with register data, providing information about individuals’ relation to the labour market (unemployment, low earnings and use of social assistance) 2 years before their participation in the LFS, and 5 years after. Thus, the data set, including the register information, comprises information spanning the years 1998–2013. The analysis is limited to individuals aged between 25 and 54 years; self-employed and respondents not in the labour force at time t are excluded.

Dependent variables

The three focal-dependent variables measure the situation of the individuals 5 years after participating in the LFS (t + 5). The basis for the variables is the FD-Trygd register in Norway and the Longitudinal integration database for health insurance and labour market studies (LISA) register in Sweden. These are administrated by the national statistical offices and collect administrative data such as incomes, income sources and payments from the unemployment or sickness funds.

The first variable indicates whether the respondent has received any unemployment insurance allowances during the year. This variable is narrower than the unemployment variable used in the LFS, only covering those who are entitled to and have applied for benefits. Furthermore, the unemployment insurance systems in Norway and Sweden differ, which can affect the coverage of the variable. Unemployment insurance is voluntary in Sweden but mandatory in Norway, where every employee whose working time is reduced by at least 50 percent, and who has met the minimum income requirement during the last 12 months or during the last three calendar years, is entitled to unemployment assistance. In Sweden, there is a two-tier system (Bengtsson and Berglund, 2012). The basic insurance scheme covers all who meet a monthly threshold of working hours, which was increased from 70 to 80 in 2007. The second system is based on membership of an unemployment insurance fund and is income related; entitlement requires contributions for the previous 12 months. A recent study (Lorentzen et al., 2014: 46–47) shows that during the period 2000–2010, coverage of unemployed persons aged 25 or older was higher in Norway (40%) than in Sweden. About 25 percent were uncompensated in Sweden between 2000 and 2005, falling sharply to approximately 50 towards 2010.

The second variable measures the risk of receiving low earnings, defined as annual pre-tax pay below 50 percent of median earnings. The figures are based on employers’ declaration to the tax authorities. Persons without any earnings across the year are excluded from the analyses.

The last dependent variable relates to reception of social assistance during the fifth year. This indicates whether the person belongs to a family that received such assistance during the year. In both countries, social assistance is paid by the municipality, but with more local discretion in Norway than in Sweden (Kuivalainen and Nelson, 2012). Hence, the circumstances in which a family is entitled to assistance may differ between the countries, and to some degree between municipalities. Generally, the social assistance system in Norway was somewhat more generous than the Swedish system during the period in focus (Kuivalainen and Nelson, 2012: 75–79).

Independent variables

We aim to analyse the risk of marginalization among temporary employees compared to permanent employees and those unemployed at t. The share of temporary employees and unemployed is larger in Sweden (10.8% and 3.9%, respectively) than in Norway (7.6% and 2.1%, respectively) in the survey data used in this article. These figures depart somewhat from official statistics, probably because only LFS data from the last quarter each year are used, excluding seasonal changes across the year.

In our regression analyses, three demographic variables (age at t, gender and country of origin), as well as educational attainment at t (primary, secondary and tertiary), are also used. There may be selection bias with regard to the dependent variables of marginalization. Hence, whether the respondents had been unemployed, were low paid or received social assistance 2 years before they participated in the LFS (t − 2) were controlled for by using register data. The share in each category is higher in Sweden (12.9%, 20.4% and 3.7%) than in Norway (7.3%, 15.6% and 2.3%).

To control for the demand and supply of labour during the period we include unemployment level at country level by t + 5, and to capture the significance of changes in labour demand/supply, we include changes in country-level unemployment from t + 4 to t + 5. We believe that the demand for labour the year before we measure the outcome variables is more an important control than at 5 years before (t).

Analysis

Our analysis starts by describing distributions of the dependent variables at t + 5 (Table 1), before conducting logistic regressions to estimate effects of these variables in Table 2. Because of strict regulation of administrative data by the national authorities, we are not allowed to pool the two data sets, and therefore cannot control for compositional differences between the labour markets (age, education, company size, as well as differences in unemployment levels). Consequently, our aim is to investigate whether the long-term outcomes resemble the (theoretical) expectations, without claiming that all between-country variation is caused by differences in EPL and the level of labour market dualization. The within-country selection of individuals into different labour market statuses (fixed-term employment, unemployment, permanent employment) are to some degree accounted for by using information at t − 2. In addition, we have made a separate analysis comparing unemployed and temporary employees who were all unemployed at t − 2, and the results, not shown here, were in line with the results in Table 2.

Table 1.

Social and labour market situation at t + 5 (%).

		Norway	Sweden
All	Unemployed	4.7	7.1
	Low earning	11.7	11.9
	Social assistance	1.4	1.9
Fixed-term employed at t	Unemployed	8.8	18.1
	Low earning	21.4	22.9
	Social assistance	3.7	3.5
Permanent employed at t	Unemployed	4.1	5.0
	Low earning	10.6	9.7
	Social assistance	0.9	1.2
Unemployed at t	Unemployed	15.9	22.7
	Low earning	32.6	34.8
	Social assistance	13.7	10.8

Table 2.

Risk of marginalization after 5 years (t + 5 years).

	Unemployment payments		Low earnings (50% of median)		Social assistance
	Norway	Sweden	Norway	Sweden	Norway	Sweden
Labour force status (Ref.: open-ended)
Fixed-term	0.578***	1.055***	0.218***	0.411***	0.783***	0.512***
Unemployed	0.851***	1.191***	0.660***	1.031***	1.425***	1.295***
Gender (Ref.: males)	−0.163***	0.028	0.669***	0.856***	−0.281***	0.041
Age, years (Ref.: 50–54)
25–29	0.412***	0.114*	−0.427***	0.383***	0.189	−0.372***
30–34	0.390***	0.031	−0.362***	0.117**	0.218	−0.126
35–39	0.191*	0.088	−0.508***	−0.310***	0.333*	0.211*
40–44	0.099	−0.016	−0.224***	−0.374***	−0.043	0.358***
45–49	0.030	−0.057	−0.292***	−0.264***	0.048	0.237***
Education (Ref.: tertiary)
Primary	0.628***	0.742***	1.178***	0.912***	1.763***	1.291***
Secondary	0.453***	0.657***	0.767***	0.545***	0.911***	0.882***
Country of origin (Ref.: domestic)
Nordic	−0.183	0.165*	0.000	0.085	0.322	0.312*
Europe/North America	0.453***	0.302***	0.183**	0.212***	0.334*	0.045
Outside Europe or North America	0.628***	0.263***	0.291***	0.325***	0.548***	0.436***
Unemployment t − 2 (Ref.: no payments)	0.968***	0.948***	0.126**	0.170***	0.334***	0.074
Low earning t − 2 (Ref: 50% or more of median)	0.249***	0.146***	1.561***	0.951***	0.971***	0.651***
Social assistance t − 2 (Ref.: No assistance)	0.338***	0.282***	0.655***	0.353***	2.242***	1.954***
Unemployment level (t + 5)	0.291***	−0.020	0.099***	−0.028	0.136	0.104**
Unemployment change ((t + 5) – (t + 4))	0.149***	0.091***	–0.099***	0.032*	−0.072	−0.002
Intercept	−4.820***	−3.496***	−3.526***	−3.102***	−6.737***	−6.321***
Nagelkerke’s R²	0.074	0.121	0.186	0.150	0.272	0.185
N	46,712	93,411	44,009	88,549	46,174	93,411

Coefficients that differ significantly (p < 0.05) between Norway and Sweden are shown in bold.

Levels of significance: *p < 0.05; **p < 0.01; ***p < 0.001.

Results

The distribution of the main indicators of marginalization after 5 years (t + 5) is shown in Table 1. In general, the highest long-term risk in both countries is becoming low paid (about 11%), while the risk of receiving social assistance is very small (1%–2%). The risk of being unemployed is considerably lower in Norway than in Sweden, while the differences are much smaller concerning the two other indicators.

Comparing the main categories of labour market status, those permanently employed have the lowest risks of marginalization 5 years later, regardless of type, while the unemployed have the highest, with temporary employees in between. The proportion of permanent employees who become unemployed or receive social assistance is marginally higher in Sweden than in Norway. Among fixed-term employees, far more are unemployed in Sweden than in Norway, but slightly fewer receive social assistance. The share of unemployed at t receiving unemployment benefits 5 years later is much higher in Sweden than in Norway, and a higher proportion receive low earnings. Finally, the share of those unemployed at t who receives social assistance is higher in Norway than in Sweden.

In Table 2, logistic regression is used to estimate effects of the independent variables on the risk of being marginalized after 5 years. We first turn our attention to being unemployed, and while several important controls are included, labour market status at t is still an important predictor. The risk of receiving unemployment payments 5 years later is significantly higher for those holding a fixed-term contract at t compared to permanent employees, though the risk is much higher in Sweden than in Norway. Confirming our initial findings in Table 1, those who were unemployed at t have an even higher risk of unemployment 5 years later; but in Sweden the difference between the coefficients for fixed-term workers and unemployed is not particularly large. For both categories the differences compared to permanent employees are larger in Sweden than in Norway.

Focusing on the control variables, males have a higher risk of being unemployed 5 years later in Norway. In both countries, the risk of unemployment decreases with age, but the effect is stronger in Norway than in Sweden. Furthermore, the risk of unemployment is higher for those without tertiary education. These differences are significantly higher in Sweden than in Norway. In both countries, the risk of unemployment is higher for respondents with origins outside Europe and North America; the effect is significantly stronger in Norway. In both countries, the variables controlling for whether the respondents were unemployed, had low earnings or received social assistance at t − 2 all predict the risk of unemployment 7 years later.

The risk of ending up with low earnings is also analysed in Table 2, and is higher for those temporarily employed at t than for permanent employees. However, the risk is particularly high for those unemployed at t. Being temporarily employed or unemployed at t has a stronger negative effect in Sweden than in Norway.

Focusing on the controls, the risk of low earnings is higher for females than for males, but significantly more so in Sweden than in Norway. In Sweden, the risk of low earnings is highest for the youngest age groups compared to the reference category (50–54 years old), but lower for the prime-age categories; but the risk of low earnings increases slowly with age in Norway. While the Norwegian and Swedish results are significantly different from each other for the youngest age groups, they are more similar in the age groups 40 years and above.

The risk of low earnings decreases with educational level, more strongly in Sweden than in Norway. In both countries, there is a somewhat higher risk of low earnings for individuals of non-domestic origin. Individuals in any of the marginal positions 2 years prior to t had a higher risk of low earnings 7 years later in both countries. However, in Norway, the risk related to social assistance receivers and low earnings at t − 2 is higher than in Sweden. Finally, unemployment levels and changes in unemployment levels influence the risk of low earnings.

Finally, in both countries, the risk of receiving social assistance after 5 years is much higher for temporary employees compared to permanent employees; but the coefficient for those unemployed at t is even larger, indicating a very high risk of receiving social assistance at t + 5. This risk is higher for males than females. Those aged 35–39 years have a higher risk than those aged 50–54 years in Norway, but the difference is rather small. In Sweden, the youngest group, aged 25–29 years, have a lower risk of receiving social assistance than those aged 50–54 years. The risk of receiving social assistance is higher among primary than tertiary educated; the difference is larger in Sweden than in Norway. Furthermore, those without a domestic or Nordic country background have a higher risk of receiving social assistance in both countries. In both countries, those who received social assistance or had low earnings 2 years before t have a much higher risk of receiving social assistance 7 years later. However, only in Norway have those who were unemployed 2 years before t a higher risk of receiving social assistance 7 years later.

Predicted probabilities

As logistic regression coefficients are not easy to interpret, we also present predicted probabilities. First, we used a model of a ‘typical’ person, an individual of domestic origin, 35–39 years, with secondary education, and without prior (t − 2) unemployment, low earnings or social assistance. The country unemployment level at t + 5 was decided to 3.9 percent in both countries, and the unemployment change to zero. The predicted probabilities are calculated for a permanent employee, a temporary employee and someone unemployed at t, and separately for both genders.

Second, theory predicts that those at the margin of the labour market benefit from less strict regulation of fixed-term employment, at least where permanent contracts are strictly regulated; we therefore made two predictions. First, models for each sex consisting of a temporary employee, 35–39 years with only primary education and originating from outside Europe or North America. Second, a young temporary employee at t, with domestic origin, 25–29 years and with only primary education.

Table 3 show that fixed-term employed ‘typical’ males have almost twice as high a risk of unemployment at t + 5 in Sweden compared to Norway. Comparing the probabilities for each category shows that temporary employees in Sweden are ‘closer’ to the unemployed than to the permanently employed; Norway has a similar pattern but smaller differences. Concerning the other two outcomes for the ‘typical’ male model, the difference between the countries is not great but somewhat higher in Sweden than in Norway. Comparing the differences between the categories, temporary employees are in this respect closer to permanent employees than to the unemployed.

Table 3.

Predicted probabilities: Five types of labour market situations at t + 5.

		Unemployment		Low earnings		Social assistance
		Male	Female	Male	Female	Male	Female
Permanent employee	Norway	4.6	3.9	5.3	9.9	0.4	0.3
Permanent employee	Sweden	5.6	5.7	4.9	10.8	0.8	0.8
Fixed-term employee	Norway	7.9	6.8	6.5	12.0	0.9	0.7
Fixed-term employee	Sweden	14.5	14.9	7.2	15.4	1.3	1.4
Unemployed	Norway	10.1	8.7	9.8	17.5	1.7	1.3
Unemployed	Sweden	16.3	16.7	12.6	25.3	2.9	3.0
Fixed-term employee, only primary education, origin outside Europe or North America	Norway	17.0	14.8	12.3	21.6	3.5	2.7
	Sweden	19.4	19.8	13.4	26.6	3.0	3.2
Fixed-term employee, 25–29 years, only primary education	Norway	12.0	10.4	10.3	18.2	1.8	1.4
Fixed-term employee, 25–29 years, only primary education	Sweden	15.9	16.3	18.2	34.4	1.1	1.2

In both countries, fixed-term employees with low education and origins from outside Europe or North America have an increased risk of all the three outcomes. However, in Norway, the risks compared to the ‘typical’ model increase considerably. For unemployment and low earnings, the risk roughly doubles, and the risk of receiving social assistance is almost nine times as large. In Sweden, the differences are more modest. For women in the same ‘marginal’ position in both countries, the risk of low pay strongly increases compared to the typical model of fixed-term employee.

The last model focuses on the outcome for young people with only primary education in temporary employment at t. In Norway, the risk of unemployment strongly increases, but is lower than for temporary employees from countries outside Europe and North America. It is the same pattern in Sweden, but the increase compared to typical temporary employees (who have a higher risk than in Norway) is very modest. For low earnings, the pattern is similar in Norway, but the risk increases much more in Sweden. The risk of social assistance is somewhat higher in Norway than in Sweden. Gender differences especially apply to one outcome in Sweden: young women with only primary education have a very high risk of low earnings at t + 5.

Discussion and conclusion

Our study analyses the long-term impact of fixed-term employment among adults in Norway and Sweden on the risks of ending up in a marginal labour market situation. From a theoretical point of view, fixed-term employment has been regarded both as a stepping stone to labour market integration and as a cause of labour market segmentation, leaving some categories of workers in risky positions.

First, if the long-term risk of marginalization is smaller for temporarily employed than for unemployed, the stepping stone hypothesis receives support. We can show that this is the case in our two Nordic countries, often described as inclusive employment regimes. In both countries, a temporary job reduces the long-term risks of marginalization compared to being unemployed. However, the risks involved in temporary employment are always higher than for those permanent employed. These results are in line with our second expectation: if differences persist in the long run, it is a sign of segmentation. Therefore, both stepping stone and segmentation theories are relevant for understanding the risks involved in temporary employment. Fixed-term employment may help some groups become more established in the labour market, but much less so for the most vulnerable.

Third, we anticipated greater long-term risks for fixed-term employees in Sweden than in Norway. However, the results do not unequivocally support this expectation. There is indeed strong evidence that the risks of unemployment are larger for temporary employees in Sweden than in Norway, but there is little difference in the risk of low earnings, and in relation to social assistance the risk was larger in Norway than Sweden. Fourth, we expected fixed-term employees to be marginalized to a larger extent in Sweden; but while this is indeed the case, the difference between Norway and Sweden in predicted probability of marginalization is smaller for temporary employees with low education, and from countries outside Europe and North America than more typical temporary employees. Thus, fixed-term employment does not help groups with low employment rates, even where the share of fixed-term employment is low and regulation is strict, as in Norway.

Why these inconsistencies in the differences between Norway and Sweden? There are at least four possible explanations. First, there is a much smaller share of temporary contracts in Norway than in Sweden, and one effect may be that the individuals ending up with such contracts are more likely to possess weak labour market attachment. This may explain the higher risk of dependence on social assistance in Norway. But the composition of the fixed-term employed and unemployed in Norway and Sweden at t − 2 is similar, with the exception of a higher proportion of unemployed and receivers of low earnings in Sweden. This may indicate that the Swedish labour market is more segmented, with relatively more people in disadvantaged labour market positions. We may not have controlled for all relevant risk factors in our analysis.

Second, the somewhat higher risk of social assistance in Norway may indicate differences in social security practices. Comparative analysis show that the Norwegian social assistance system is more generous than the Swedish (Kuivalainen and Nelson, 2012), probably making it easier to receive such assistance. Consequently, we could expect the difference in the share receiving social assistance to be smallest among the groups more prone to marginalization.

Third, we cannot measure pay per hour. Thus, the modest earnings differences found may be related not just to lower hourly pay, but also to differences in hours worked. With higher shares of under-employed and unemployed in Sweden than in Norway during this period, some of the difference in the rate of low earnings might reflect differences in working hours. That the differences in low earnings shares between Norway and Sweden were not higher may be because part-time shares and wage inequalities in the lower half of the distribution are higher in Norway than in Sweden.

Fourth, the general labour market situation may affect the results. During the period studied, unemployment was lower in Norway than in Sweden. Mobility into and out of the Swedish labour market as well as into and out of permanent jobs was also lower (Berglund and Furåker, 2011). This may explain the (very) high risk for temporary employees to be unemployed 5 years later, reflecting smaller chances to establish themselves in the labour market than their Norwegian peers. Temporary employment seems to a larger extent serve as a flexibility buffer for Swedish employers while shifting the risks to the employees, in line with the segmentation hypothesis. Thus, the more loosely regulated Swedish labour market showed stronger signs of dualistic tendencies (Obinger et al., 2012; Thelen, 2014: 188–190). With limited unemployment and high labour demand in Norway, employees were much better positioned, perhaps forcing employers to offer open-ended contracts. Consequently, we believe that the institutional setting (EPL) and the labour market situation (unemployment) interact in influencing the risks confronting fixed-term employees.

However, it is important to emphasize two findings that demand more research. First, unemployment is the greatest risk factor for marginalization. However, the rather high risk of unemployment for temporary employees, especially in Sweden, may indicate that temporary employment involves risks that amplify for even longer than 5 years. Fixed-term employees seem to cling to the labour market with recurrent spells of unemployment. Each of these episodes may carry a high risk of ending up in an even worse labour market situation. Second, in Norway, the additional risks that seem to be attached to temporary employment for those with origins outside Europe or North America, or who are young with low education, indicate a strong divide between labour market groups: some are integrated by way of fixed-term employment, others are marginalized. In Sweden, temporary employment seems to have been ‘normalized’ because of the high figures of such contracts (Berglund et al., 2017), while in Norway, specific categories may be more or less stigmatized in these positions. But the Norwegian figures for the young and immigrants, even though higher than for a ‘typical’ person in Norway, are still lower than for a comparable person in Sweden, with the exception for the share receiving social assistance.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was funded by the Norges forskningsråd (Research Council of Norway) programme on welfare, working life and migration (VAM) and by Forte (the Swedish Research Council for Health, Working Life and Welfare) (2013-0436).

Author biographies

Jørgen Svalund is a Researcher at Fafo, Oslo, Norway.

Tomas Berglund is a Professor of Sociology at the University of Gothenburg, Sweden.

References

Atkinson

(1984) The flexible firm and the shape of jobs to come. Labour market issues. no. 5. Oxford: Ruskin College Oxford, pp. 1–19.

Autor

Donahue

Schwab

(2006) The cost of wrongful-discharge laws. Review of Economics and Statistics 88: 211–231.

Barbieri

Cutuli

(2016) Employment protection legislation, labour market dualism, and inequality in Europe. European Sociological Review 32: 501–516.

Bassanini

Duval

(2006) Employment patterns in OECD countries: Reassessing the role of policies and institutions. In: OECD (ed.) OECD Economics Department Working Papers. Paris: OECD. Available at: https://econpapers.repec.org/paper/oececoaaa/486-en.htm

Becker

(1975) Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. New York: NBER.

Bengtsson

Berglund

(2012) Labour market policies in transition: From social engineering to standby-ability. In: Larson

Letell

Thörn

(eds) Transformations of the Swedish Welfare State: From Social Engineering to Governance? Basingstoke: Palgrave Macmillan, pp. 86–103.

Berglund

Furåker

(2011) Flexicurity institutions and labour market mobility. International Journal of Comparative Labour Law and Industrial Relations 27: 111–127.

Berglund

Håkansson

Isidorsson

Alfonsson

(2017) Temporary employment and the future labor market status. Nordic Journal of Working Life Studies 7: 27–48.

Blanchard

Landier

(2002) The perverse effects of partial labour market reform: Fixed-term contracts in France. Economic Journal 112: F214–F244.

10.

Booth

Francesconi

Frank

(2002) Temporary jobs: Stepping stones or dead ends? Economic Journal 112: F189–F213.

11.

Centeno

Novo

ÁA

(2012) Excess worker turnover and fixed-term contracts: Causal evidence in a two-tier system. Labour Economics 19: 320–328.

12.

Dieckhoff

Steiber

(2012) Institutional reforms and age-graded labour market inequalities in Europe. International Journal of Comparative Sociology 53: 97–119.

13.

DiPrete

Goux

Maurin

Tåhlin

(2001) Institutional determinants of employment chances: The structure of unemployment in France and Sweden. European Sociological Review 17: 233–254.

14.

Doeringer

Piore

(1971) Internal Labor Markets and Manpower Analysis. Lexington, MA: Heath Lexington Books.

15.

Engebretsen

Salvanes

Vassenge

(2012) Midlertidige stillinger – et springbrett til permanente stillinger? In: Norway

(ed.) Økonomiske Analyser. Oslo: Statistics Norway, pp. 33–37.

16.

Esping-Andersen

(1990) The Three Worlds of Welfare Capitalism. Cambridge: Polity Press.

17.

Eurostat (2017a) Part-time employment and temporary contracts – Annual data [lfsi_pt_a]. Available at: http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=lfsi_pt_a&lang=en

18.

Eurostat (2017b) GDP and main components (output, expenditure and income). Available at: http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=nama_10_gdp&lang=en

19.

Eurostat (2017c) Employment and activity by sex and age – Annual data [lfsi_emp_a]. Available at: http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=lfsi_emp_a&lang=eng

20.

Gallie

(2007) Employment Regimes and the Quality of Work. Oxford and New York: Oxford University Press.

21.

Gebel

Giesecke

(2011) Labor market flexibility and inequality: The changing skill-based temporary employment and unemployment risks in Europe. Social Forces 90: 17–39.

22.

Giesecke

Groß

(2003) Temporary employment: Chance or risk? European Sociological Review 19: 161–177.

23.

Hagen

(2003) Do fixed-term contracts increase the long-term employment opportunities of the unemployed? Available at: https://ub-madoc.bib.uni-mannheim.de/155/

24.

Håkansson

(2001) Språngbräda eller segmentering? Et longitudinell studie av tidsbegränsat anställda. Uppsala: IFAU.

25.

Holmlund

Storrie

(2002) Temporary work in turbulent times: The Swedish experience. Economic Journal 112: F245–F269.

26.

Houseman

(2001) Why employers use flexible staffing arrangements: Evidence from an establishment survey. Industrial and Labor Relations Review 55: 149–170.

27.

Kahn

(2010) Employment protection reforms, employment and the incidence of temporary jobs in Europe: 1996–2001. Labour Economics 17: 1–15.

28.

Korpi

Levin

(2001) Precarious footing: Temporary employment as a stepping stone out of unemployment in Sweden. Work, Employment and Society 15: 127–148.

29.

Kuivalainen

Nelson

(2012) Eroding minimum income protection in the Nordic countries? Reassessing the Nordic model of social assistance. In: Kvist

Fritzell

Hvinden

Kangas

(eds) Changing Social Equality : The Nordic Welfare Model in the 21st Century. Bristol: Policy Press, pp. 69–88.

30.

Levin

(1998) Arbetslöshet, tidsbegränsade anställningar och rörlighet på arbetsmarknaden i 1990-talets Sverige. Stockholm: Institutet För Social Forskning.

31.

Lorentzen

Angelin

Dahl

Kauppinen

Moisio

Salonen

(2014) Unemployment and economic security for young adults in Finland, Norway and Sweden: From unemployment protection to poverty relief. International Journal of Social Welfare 23: 41–51.

32.

McVicar

Wooden

Leung

(2016) Work-related training and the probability of transitioning from non-permanent to permanent employment. British Journal of Industrial Relations 54: 623–646.

33.

Mertens

McGinnity

(2005) A two-tier labour market for fixed-term jobs? Evaluating evidence from West Germany using quantile regression. Schmollers Jahrbuch: Journal of Applied Social Science Studies 125: 75–85.

34.

Nergaard

(2004) Atypisk Arbeid: midlertidige Ansettelser Og Deltidsarbeid I Norge, vol. 430. Oslo: Fafo-rapport.

35.

Nergaard

(2016) Tilknytningsformer I Norsk Arbeidsliv: Nullpunktsanalyse. Oslo: Fafo-rapport.

36.

Noelke

(2016) Employment protection legislation and the youth labour market. European Sociological Review 32: 471–485.

37.

Obinger

Starke

Kaasch

(2012) Responses to labour market divides in small states since the 1990s. In: Emmenegger

Häusermann

Palier

Seeleib-Kaiser

(eds) The Age of Dualization: The Changing Face of Inequality in Deindustrializing Societies. Oxford: Oxford University Press, pp. 176–200.

38.

Olsen

Kalleberg

(2004) Non-standard work in two different employment regimes. Work, Employment and Society 18: 321–348.

39.

Passaretta

Wolbers

(2016) Temporary employment at labour market entry in Europe: Labour market dualism, transitions to secure employment and upward mobility. Economic and Industrial Democracy. Epub ahead of print 8 June. DOI: 10.1177/0143831X16652946.

40.

Picchio

(2008) Temporary contracts and transitions to stable jobs in Italy. Labour 22: 147–174.

41.

Polavieja

(2003) Temporary contracts and labour market segmentation in Spain: An employment-rent approach. European Sociological Review 19: 501–517.

42.

Ragin

(1987) The Comparative Method: Moving beyond Qualitative and Quantitative Strategies. Berkeley, CA: University of California Press.

43.

Sackmann

(2001) Age and labour-market chances in international comparison. European Sociological Review 17: 373–387.

44.

Scherer

(2004) Stepping-stones or traps? Work, Employment and Society 18: 369–394.

45.

Svalund

(2013) Labour market institutions, mobility and dualization in the Nordic countries. Nordic Journal of Working Life Studies 3: 123–144.

46.

Thelen

(2014) Varieties of Liberalization and the New Politics of Social Solidarity. New York: Cambridge University Press.

47.

Wang

Weiss

(1998) Probation, layoffs, and wage-tenure profiles: A sorting explanation. Labour Economics 5: 359–383.

48.

(2012) Better off jobless? Scarring effects of contingent employment in Japan. Social Forces 90: 735–768.