ARE FIXED-TERM JOBS BAD FOR YOUR HEALTH?: A COMPARISON OF WEST-GERMANY AND SPAIN

There is considerable evidence which suggests that loosing one's job has negative psychological and physical consequences for the unemployed (Goldsmith et al. 1996; Hamilton et al. 1997; Theodossiou 1998; Romeu Gordo 2006). These studies underlined the social dynamics within employment which have positive implications for workers’ sense of self and well-being. The employment relationship has changed considerably, however, with growing numbers of fixed-term contracts following the liberalisation of employment law in the mid-1980s. This paper examines whether the reduced job quality (Kalleberg et al. 2000; Houseman 2001; McGovern, Smeaton and Hill 2004; Gash 2004a) and reduced job security (Gash 2004a; Gash and McGinnity 2005) associated with fixed-term contracts, decreases the positive health effects of employment for the unemployed.

The past two decades have seen an increase in fixed-term employment in the majority of OECD countries. This increase has been most dramatic in Spain where one-third of all employment contracts are fixed-term while in Germany fixed-term employment rates are below the EU average of 12–13 per cent of employment contracts (Eurostat 2002: 173). In Germany fixed-term contracts account for roughly 8 per cent of total employment (Rudolph 2000). Accordingly, there has been much discussion about the consequences of working in fixed-term jobs with mixed evidence for different countries. While in Spain fixed-term workers often seem to be trapped in a cycle of repeat spells of fixed-term employment and unemployment (Amuedo-Dorantes 2000; Polavieja 2003), there is less supporting evidence for Germany (Bookman and Hagen 2005; McGinnity et al.2005). We also know that fixed-term workers receive lower wages, (see e.g., Mertens and McGinnity 2004 for Germany and Jimeno and Toharia 1993 for Spain) and in many cases have fewer work related benefits (OECD 2002). Does this combination of job insecurity and poor working conditions have negative consequences on health? Might workers’ experience of fixed-term employment vary in countries with different risks and opportunities for fixed-term workers?

In this paper we use information on switches between unemployment and employment by contract type to analyze these questions. We expect evidence of an improvement in the health status of unemployed workers who obtain employment. If it is true that fixed-term contracts negatively affect health we should observe less of an improvement if the contract is fixed-term rather than permanent. One of the strengths of this research design is its provision of a comparatively homogeneous group of workers, given their shared prior experience of unemployment. This should control for some of the differences which exist between workers in different types of contract. We use data from the European Community Household Panel (ECHP) and the German Socio-Economic Panel (GSOEP) to test the impact of fixed-term contracts on health status and reveal whether differences by gender, country and duration of employment contract exist.

Fixed-term contract workers lose their jobs more frequently than those on permanent contracts simply because their contracts run out within very short periods of usually one to two years. This job loss often results in unemployment (after one year around 13 per cent of fixed-term workers become unemployed in Germany and 20 per cent in Spain¹), which usually causes a deterioration of general health indicators and self-reported health status (see e.g., Schwefel 1986; Kasl and Jones 1998; Murphy and Athanasou 1999). Why do we observe this close link between unemployment and health? In addition to the financial difficulties the unemployed face, which in themselves have been associated with extreme psychological strain (Pearlin, 1989; Whelan 1992), the unemployed loose many of the latent functions employment provides which are important to individual well-being (Jahoda 1982). Employment is seen to provide a structure to one's day, regular contact with others, as well as a sense of self-worth. Not only does unemployment deprive people of these functions, unemployment also implies skill attrition and a loss of or decrease in one's social status (Warr 1987).

We also expect fixed-term contract workers to be disproportionately affected by job insecurity, which is also thought to affect health status. One much cited study by Ferrie et al. (1995) shows that self-reported health status deteriorates when employees expect privatization and the accompanying job change or job loss. Crucially, this relative decline in health status was shown not to be linked with changes in health related behaviour. Burchell (1994, 1999) as well as Bohle et al. (2001) argue that job insecurity has obvious negative effects on physical and psychological well-being.² The estimated reduction in psychological well-being is very similar in magnitude to that caused by unemployment. One explanation for this could be the inability of fixed-term contract workers to plan and control their lives given the short-term nature of their jobs (Burchell 1994). Therefore, although a fixed-term job fulfils many of the conditions that Jahoda (1982) and Warr (1987) associated with employment, it might still have negative health effects as a result of the twin pressures of job insecurity and unemployment risk.

In addition to these twin pressures of job insecurity and unemployment risk there may also be stresses linked to the comparatively poor job quality of some fixed-term jobs. Fixed-term jobs are sometimes (not always) connected with relatively low pay (Mertens and McGinnity 2004; Gash and McGinnityforthcoming; Mertens et al. 2005), poor working conditions (Gash 2004) and reduced access to benefits (Houseman 2001; McGovern, Smeaton and Hill 2004). We anticipate this combination of relatively high job insecurity and lower job status to have implications for psychological health and health status.

The paper presents a comparative analysis of the health outcomes of contract type for men and women as we expect differences by gender and country. We expect women to be more likely to self-actualise in unpaid care work than men and expect this to protect them from some of the negative consequences of unemployment (Gallie and Russell 1998; Romeu Gordo 2006).This is especially likely in the two countries under consideration with relatively low female participation rates of just over 60 per cent in Germany and roughly 50 per cent in Spain.

Finally, we expect country differences as a result of the following: the different levels of unemployment, the different size of the fixed-term employment sector, the different opportunity structures fixed-term workers have and differing cultural attitudes to unemployment.

While both countries suffer from high unemployment rates, Spanish unemployment rates have been considerably higher and also differ considerably in their composition, with Spanish women experiencing very high unemployment rates relative to Spanish men. It could be argued that higher levels of unemployment will decrease the stigma associated with unemployment, with stigma less likely if higher proportions of ones social network face a similar situation (Clark 2003). There is also the suggestion that when the possibility of obtaining employment is very low, i.e., in situations of high and long-term unemployment, people may decide to ‘reject’ the role paid employment has in ones life (Gallie and Russell 1998). If this is the case we can expect Spanish women to be the least likely to exhibit positive changes in health status on entry to paid employment, if paid employment is no longer central to how they evaluate themselves.

There are differences between both countries in the proportion of workers on fixed-term contracts. In Spain, where one third of the workforce is a fixed-term contract worker, acceptance of these jobs might be higher than in Germany where just under 10 per cent of workers have fixed-term contracts. Because fixed-term contracts are more ‘abnormal’ in Germany we could expect a greater stigma associated with being a fixed-term contract worker in Germany and this might make them marginally more stressful. However, the different opportunity structures each country provides for fixed-term contract workers are also likely to influence the stresses workers have on the job. Spanish research suggests that fixed-term work is of poor quality (Jimeno and Toharia 1993; Jimeno and Toharia 2003; Amuedo-Dorantes 2000) and that fixed-term contract workers find it difficult to obtain permanent positions (Polavieja 2003). In Germany, the evidence is certainly more mixed. While there is also a wage penalty associated with fixed-term employment in Germany, it is not as large as the Spanish wage penalty and is also to some extent compensated for by higher wage growth (Mertens et al. 2005). There is also less evidence of fixed-term employment being a ‘trap’ in Germany (Boockmann and Hagen 2005), with the long-term repercussions of fixed-term employment found to decrease with time (McGinnity et al.2005).

Finally, we could also expect differences in the German and the Spanish work ethic to influence their response to both employment and unemployment. This assumption is based on research which suggested that the Spanish unemployed suffer less from unemployment as a result of their lower work ethic (Marsh and Alvaro 1990).

Before we analyse these questions empirically we provide a review of the literature in the next section.

There are few studies which analyse the direct health effects of fixed-term contracts and those that do have used a variety of health measures. For our purposes we will summarize the results into psychological and physical health effects.

We have argued that psychological well-being will be negatively affected by fixed-term employment, with these contracts considered stressful. Klein, Hesselink and van Vuuren (1999), using Dutch data, confirm this assumption reporting that roughly 44 per cent of fixed-term workers are worried about job insecurity, while only 15.5 per cent of permanent contract workers worry about insecurity. Similarly, Lasfargues et al. (1999) find evidence of lower psychological well-being amongst temporary workers in France. Nonetheless, fixed-term contract work is unlikely to have the same impact on all workers, as Bauer and Truxillo (2000) argue. Individual characteristics like tolerance for ambiguity and self-monitoring influence responses to stress and the selection process into permanent employment. Not all studies looking at psychological factors confirm our assumptions however. Benavides et al. (2000), in a cross-sectional study of 15 European countries, show that non-permanent employees tend to report lower work stress. Similarly, Sverke et al.'s (2000) report that fixed-term work has no effect on psychological well-being and, likewise, Artazcoz et al. (2005), in a Spanish health survey, find no association between fixed-term contracts and poor mental health.

The literature on physical health is even more ambiguous. While Benavides et al. (2000) find fixed-term workers to have worse physical health than permanent workers and Klein, Hesselink and Van Vuuren (1999) report slightly higher percentages of fixed-term workers with physical health complaints, other researchers have found fixed-term contract workers to have better health. Virtanen et al. (2002 and 2003) show that non-permanent employees in Finland report better health. Similarly, Sverke et al. (2000) find fixed-term contract workers to have better physical health in comparison with permanent workers. Fixed-term workers are sometimes reported to have lower rates of absenteeism than permanent workers (Benavides 2000; Virtanen 2003). Rates of sickness absence even tend to increase when employees move from fixed-term to permanent jobs. It is very unlikely, however, that this is caused by a real deterioration in health. It rather shows that fixed-term workers try to reduce absenteeism due to fears of job loss. Once in a permanent position this fear is reduced.

Finally, there are two papers on the topic which come closest to our research design. Strandh (2000) looks at the impact of different exit routes from unemployment on mental health in Sweden. He finds that mental health improves for unemployed workers who leave unemployment to either education or employment. He also finds that atypical workers, the self-employed and temporary workers, show less improvement in their mental health than permanent workers. Virtanen et al. (2003) also look at the impact of workers transitions on physical health and sickness absence in Finland. However, their study is only based on those working in two hospital districts and the number of switchers is very small. They find no change in health indicators when workers move from fixed-term to permanent jobs.

Our empirical study described in the following sections adds to the literature on the health consequences of fixed-term workers by examining the consequences of workers transitions from unemployment to employment on their health status.

For our study we use Spanish data from 1994 to 2001 from the European Community Household Panel (ECHP) and the German Socioeconomic Panel waves 1994 to 2004 for West Germany (SOEP Group, 2001). The ECHP provides full information on contract type from 1995 and the GSOEP from 1994 onwards.³ The panel component of both datasets allows us to analyse the impact of labour market transitions on changes in health status.⁴ By focusing on unemployed workers we reduce the problem of heterogeneity that comes with a comparison of fixed-term and permanent workers.

For our purposes we follow unemployed workers and compare their health status in at least two consecutive years, excluding respondents with less than two years complete information on our covariates of interest. Moreover, the sample is limited to those aged between 20 and 54 years of age, with those aged over 55 years more likely to have lower health status as a result of increasing morbidity.

To test the effect of labour market transitions on self-reported health status we estimate two models: one looking at the short-run immediate implications (Model 1) and another one looking at longer run consequences (Model 2).

Model 1: We first select individuals who are unemployed in t-1 and consider three possible labour transitions in the following year: (i) the individual remains unemployed; (ii) the individual gets a permanent contract; and (iii) the individual gets a fixed-term contract.

Model 2: In the first period (t-2) all individuals are unemployed. We then consider three different transitions in the labour market: (i) the individual remains without work till t; (ii) the individual finds a job with a fixed-term contract in the first period (between t-2 and t-1) and remains employed till t; (iii) the individual finds a job with a permanent contract in the first period (between t-2 and t-1) and remains employed till t. The objective in model 2 is to determine the long-term health implications of obtaining a fixed-term contract, with any positive effect of job gain expected to decrease if the worker becomes aware of reduced opportunities associated with their contract type.⁷

Model 1: Table 1 and 2 present the basic results for men and women. We find that for German men transitions from unemployment into employment have a significant and positive effect on health status for both fixed-term and permanent workers. Similarly, in Spain transitions from unemployment into employment have a significant and positive effect on health status for both types of contract but for Spain the effect of permanent employment is significantly larger than that of a fixed-term contract.⁸ So for Spanish men we can conclude that fixed-term contracts are significantly worse for your health.

For German women, on the other hand, we only find positive effects on health status if the contract is permanent. The transition from unemployment to a fixed-term contract has no significant effect on females’ health. Surprisingly, health status of Spanish women does not change in either case.

Now, it is possible that our results are biased because individuals with low levels of health are less likely to obtain a job, and/or to be less successful at obtaining a job with a permanent contract. Therefore, we estimate the same models homogenising the reported level of health in t-1. We select individuals whose reported health status in t-1 was at least 3, that is those with fair to very good health status. In Germany approximately 11 per cent of individuals report health status lower than 3, while in Spain the figure is 4 per cent. These individuals tend to have serious health problems, which are most likely to impact negatively on their labour market success. This is corroborated by comparing objective health measures for individuals who report health status lower than 3 with individuals whose reported health is at least 3 (see Table A1 in the appendix). We control for the possible health selection effect for Model 1 in Tables 3 and 4. While the results for German women do not change, they do change for Spanish women. After health selection, transitions into fixed-term jobs have a significant and negative effect on health status. The results do change for German men. Once we remove individuals with very low health status from our sample, we no longer find a positive relationship between health status and receipt of a fixed-term job. This result suggests that individuals with low health status are more likely to obtain fixed-term contracts, and that they derive a positive effect of obtaining employment. For Spanish men, the positive effect of transitions into fixed-term employment remains even after selecting on health and the difference between contract type remains significant. This may be due to the fact that fixed-term employment is more widespread in Spain so that health selectivity into fixed-term employment is less of an issue; with many individuals independently of their level of health likely to become fixed-term contract workers.

Summarising, we have found that job acquisition improves health status. We have also found contract type to play an important role in health status with workers who obtain a fixed-term contract exhibiting smaller increases in health status, though it is only statistically significant in Spain. We have also found women to exhibit different tendencies to men. For German and Spanish women fixed-term contracts have no effect on health status (with or without health selection). This result could be due to women's disproportionate investment in unpaid work within the home, making them less likely to exhibit positive health status change on receipt of paid employment. It might be, in fact, that the positive effect of job receipt is cancelled out by the stressful effects of the double-burden of paid and unpaid work. In order to analyse this question, we test whether the non-significance of the female result is driven by women who are engaged in intensive unpaid childcare duties within the home. Table A2 in the appendix presents the mean hours spent per day in unpaid childcare within the home and while we find full-time working women to have slightly lower levels of unpaid care the average hours spent are still very high, with German women spending four hours per day and Spanish women spending seven hours per day in childcare.⁹

We test the impact of childcare on health status by comparing the results of women who at t-1 were engaged in intensive childcare (more than four hours per day) and women whose’ child care load was lower (results not shown). In Germany women who were engaged in intensive childcare and obtain a job do not experience any significant effect on health (irrespective of contract type). However, women who were carrying out less hours of childcare experience a significant and positive effect of obtaining a permanent job. We carried out the same analysis for Spanish women; though found no significant differences between mothers with high and low child care hours.¹⁰

We also tried to identify fixed-term contract worker heterogeneity by looking at interaction terms of contract type by occupational skill level, education and age. These analyses were not found to be significant, however.

Model 2: From Model 1 we know that getting a job is good for your health, but that getting a permanent job is better. We would now like to know whether the positive effect on health remains once individuals have accommodated to their new status. For instance, we could expect fixed-term workers’ health to deteriorate if they believe that their contract will not be renewed nor converted into a permanent contract.

In the analysis that follows, the dependent variable measures changes in health status between t-2 and t. We analyse the effect on health of getting a job between t-2 and t-1 and remaining in this job till t. Again, we distinguish between becoming a fixed-term and a permanent worker but now analyse the effect on health when individuals remain in their job for more than one year. In Table 5 the estimation results for German and Spanish men are presented. We observe that for German men there is still a positive effect on health of obtaining a fixed-term job even if the individual remains in fixed-term employment for more than one year. However, for Spanish men, the positive effect of obtaining a fixed-term job disappears after one year in fixed-term employment. This may be a function of worker's stresses concerning their future unemployment risk given the short-term nature of their contracts. So while initially workers exhibit an increase in health status on job acquisition, this effect disappears with time, perhaps when they learn that their job is unlikely to lead to further employment. Table 6 presents the analysis for women where we find that the positive effect of permanent employment remains significant for the random effects estimation for German women.

Against the background of increasing numbers of workers in relatively insecure fixed-term contracts in Europe we investigated the relationship between health and contract type. We focused on unemployed workers exiting unemployment in Spain and Germany and compared the health consequences for different contract types. As we know, from the literature, that unemployed workers should on average experience a deterioration of their health we ask whether returning to work helps to restore their health.

As expected, job acquisition improves health status, with the exception of Spanish women and, for men, the positive effect is smaller if they obtain a fixed-term job. For women, however, the transition from unemployment to a fixed-term job has none of the positive health effects typically associated with paid employment. When we compared the longer-term effects of job acquisition, after a period of two years, we loose the positive health effects of job acquisition for Spanish men on fixed-term contracts. It is only German men who retain positive effects of fixed-term contract status over the longer-term which may support our expectation that German fixed-term contract employment provides greater opportunity structures than Spanish fixed-term employment.

Certainly, it is difficult to disentangle health and contract type effects, simply because the causality between contract type and health could run in both directions. On the one hand the insecurity associated with fixed-term contracts increases psychological pressure and might therefore also reduce physical health. On the other hand workers with serious health effects could run into problems finding a permanent job and therefore be condemned to switch alternately between unemployment and fixed-term contracts. In spite of this we have found our results to be relatively robust even when we control for health selection effects by focusing on those unemployed workers with fair to very good health. Even with selection on healthy workers we find that fixed-term jobs are not as good for your health as permanent jobs.

Summarizing, we believe that rising percentages of fixed-term contracts and the associated rise in insecure employment relationships has negative effects on health. This seems to be especially severe in countries like Spain where one third of all working contracts are fixed-term. Not only are individuals' lives affected by this but there may also be repercussions for the overall production possibilities of an economy.

Future research would do well to analyse the health effects of other forms of atypical employment, such as part-time work. Such an analysis could shed light on the health effects of employment for women who, in principle, are not suffering from the ‘double burden’ of paid and unpaid work.

Vanessa Gash Based at the Cathy Marsh Centre for Census and Survey Research at the University of Manchester. She is an economic Sociologist with an interest in comparative labour market research. She obtained her DPhil from Oxford University and spent two years at the Max Planck Institute for Human Development where the majority of the work for this paper was concluded.

Antje Mertens is Professor of Economics at the Berlin School of Economics. She received her PhD from the Humboldt University Berlin in 1998 and then worked as a researcher at the Max Planck Institute for Human Development, with which she is still associated. Her major research interests are labour mobility, wages and training.

Laura Romeu Gordo is a Research Fellow at the Institute for Employment Research (IAB) in Nürnberg. She is an economist and her research interests are labour economics, aging, and health economics. She is currently involved in a project about labour market policy evaluation.