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

ABSTRACT

In this paper we analyse the health effects of fixed-term contract status for men and women in West-Germany and Spain using panel data. This paper asks whether changes in the employment relationship, as a result of the liberalisation of employment law, have altered the positive health effects associated with employment (Jahoda 1982; Goldsmith et al. 1996). Using information on switches between unemployment and employment by contract type we analyze whether transitions to different contracts have different health effects. We find that unemployed workers show positive health effects at job acquisition, and also find the positive effect to be smaller for workers who obtain a fixed-term job. We also establish surprising differences by gender and country, with women less likely to report positive health effects at job acquisition. For West-Germany, this was found to be a function of the dual-burden of paid and unpaid care within the home.

Keywords:

• temporary employment
• unemployment
• health

Notes

¹Own estimations of ECHP data, cross-sectional analysis of the labour force status of all fixed-term contract workers in 1995 by their labour force status in 1996, weighted data.

²Bohle et al. (2001) gives a very good overview of nearly 70 studies looking at health and safety effects of job insecurity conducted since 1966.

³We cannot identify agency workers at any point in both surveys. Agency workers may or may not classify themselves as on a fixed-term contract. While agency work has risen steadily in Germany in the last decade, it was still only 1.2 per cent of dependent employment in June 2000 (Bundesanstalt für Arbeit, 2001). Similarly Spain agency work accounts for approximately 0.8 per cent of total employment (Storrie 2002). Hence, we do not expect it to bias our results.

⁴Caution is required when using self-reported health measures in cross-cultural studies with different meanings attached to different response categories (Jürges 2005). For example, category 4 of our self-reported health status variable is ‘malo’ in the Spanish questionnaire and ‘weniger gut’ in the German questionnaire. However, ‘weniger gut’, which means a little less than good, has a more positive connotation than ‘malo’, which means bad. While these differences might affect a comparison of levels of health by country, our analysis removes this risk by looking at changes in health status.

⁵Both the ECHP and the GSOEP determine health status by asking respondents: ‘How is your health in general?’, with 1 = ‘Very good’ and 5 = ‘Very Bad’, resulting in a health change variable with nine different categories. We reverse code health status so that decreases in health status are represented by negative coefficients in the models. It should be noted that there is a strong correlation between subjectively defined health status and objective criterion (Table A1 in the appendix).

⁶ We decided to use random effect estimators after carrying out a Hausman test, which revealed that GLS was the most efficient estimator.

⁷By looking only at this selection of transitions from unemployment we lose 64 per cent of all observations in Spain and 58 per cent in Germany. The modal category of excluded respondents for both samples consist of those without complete information on labor force status for three years in a row. For those with complete information the modal categories for our German sample, accounting for 27 per cent of the sample, are those who leave unemployment for employment in t-1 and then re-enter unemployment in t and those who remain unemployed for two years and then enter employment in t. For Spain the corresponding figure is 26 per cent. To test whether our emphasis on three very specific transitions lead to biased results we introduce a dummy variable which includes all other possible transitions to the model. Our results were found to be consistent after introducing this dummy variable.

⁸Nested t-test established the size of the coefficients by contract type to be statistically different for Spain, at the 0.05 level, but not Germany.

⁹These figures are not directly comparable however, as the GSOEP data asks respondents how many hours they spent per working day whilst the ECHP asks the number of hours per week. This distinction is important as there tend to be considerable differences in the number of hours spent in care and household work on the weekend relative to the week.

¹⁰The non-significance of the Spanish female result could also be due to their disproportionate investment in housework. In Spain, both Ahn et al. (2003) and Alvarez and Miles (2003) find that women do the majority of housework, irrespective of employment status and job type. Nonetheless, in West Germany women also tend to do the majority of housework, working 35 hours a week compared to 17 hours a week for West German men (Rosenfeld et al. 2004: 119–/20). In Spain the corresponding figures are 35 hours a week for women and 4.5 hours for men (Ahn et al. 2003: 29). While Spanish women undoubtedly appear to have considerably less help from their partners than German women do, women in both countries engage insimilar amounts of unpaid work within the home.