Summary
The aim of this dissertation is to provide a better understanding of some of the factors causing sickness absence and gender differences in sickness absence. Such factors are complex and related to many different aspects of life which cannot all be covered. The scope of the dissertation is limited to aspects concerning individuals’ work and occupations, and more specifically to gendered characteristics of paid work. It attempts to address the following larger question: How does gender segregation in the labour market affect sickness absence and gender differences in sickness absence?
Gender segregation in the labour market implies that women and men are disproportionately allocated to occupations, as well as workplaces. I distinguish between possible effects of workplace and occupational gender composition both in the theoretical framework and empirical findings. My definition of sickness absence implies an understanding that it is influenced both by health, by motivation and the manner in which people interpret health problems and respond to them. The dissertation consists of four empirical studies. The first study is based on data from the European Labour Force Survey, and the remaining ones are based on Norwegian register data.
The first study sheds light on the “unhealthy/healthiness” of men’s and women’s work by addressing a claim that high sickness absence among women is caused by poor working conditions in female dominated occupations. This claim implies that the gender difference in sickness absence should be smaller when comparing men and women working in the same occupation. According to the results, this is not the case. On the contrary, detailed control for occupational categories leads to significantly larger gender differences in sickness absence. The findings indicate that women on average are in more healthy occupations than men.
Empirical studies of the occupational gender composition and sickness absence reveal a U-shaped pattern with high absence rates in both male- and female-dominated occupation. Theories concerning this issue mainly argue that characteristics associated with female and/or male-dominated occupations give unhealthy working conditions. However, the observed association may be confounded by selection based on stable individual characteristics. The second study examines the importance of such selection effects by the means of individual fixed effect models. When such models are employed, a negative and linear association between the proportion of women and sickness absence emerges. The findings indicate that the observed U-shaped pattern is primarily caused by overrepresentation of absence prone men in particular, but also of absence prone women, in female-dominated occupations.
When it comes to workplace gender composition, previous studies have found a weak positive association between the proportion of females and sickness absence. These findings have been taken as support for the idea that female dominated workplaces develop more lenient norms towards sickness absence. The last two studies in this dissertation give some further support for this claim using alternative measures of women’s influence at the workplace.
The third study concerns the association between the gender of managers at the workplace and sickness absence rates.The results indicate a weak positive relationship between female managers and sickness absence rates. The results are positive and significant also when gender, age, education, female proportion at workplace, working hours, number of children and occupational category is taken into account.
The fourth study explores the association between workplace gender composition and sickness absence among pregnant workers. The results show a weak positive association between the female proportion at workplaces and pregnancy-related sickness absence. Individual fixed effect models limited to variation within women with more than one pregnancy, give equally strong effects in the same direction. In other words, selection based on stable individual characteristics does not seem to explain the observed association.