Female-typical Subjects and their Effect on Wage Inequalities among Higher Education Graduates in Germany

ABSTRACT

In the search for an explanation of wage inequalities between men and women, it is often argued that women's disadvantages on the labour market are the result of their lower investment in human capital. However, this argument loses for at a time when highly educated women are outperforming men in terms of educational attainment. But even within the group of higher education graduates, gender differences remain in the form of gender-typical enrolment in academic subjects. The present article analyses the wage effects of how graduation in typically female subjects and subsequent work in typically female occupations generate wage inequalities 1 and 5 years after graduation. For understanding these effects, it follows the hypothesis of a socio-cultural devaluation of female-dominated subjects and occupations. Based on the 1997 HIS Absolventenpanel, we estimate gross hourly wages of male and female higher education graduates as they enter the German labour market. Results show that, 1 year after graduation, men receive a higher wage penalty than women for graduating in a female-dominated subject. However, working in typically female occupations affects women's wages more strongly than it does men's immediately after graduation. In addition, women have fewer possibilities to compensate for this penalty in subsequent years.

Key words:

• horizontal segregation
• wage inequalities
• higher education
• gender
• Germany

Notes

1 Evaluative discrimination is thus related to processes of devaluation due to horizontal segregation of fields of study. When, however, women and men with the same human capital (meaning academic subjects) are hired for different hierarchical positions and consequently receive different wages, we talk about allocative discrimination. Thus, men are ascribed more competence in leadership positions, since the cultural background of gender status beliefs (Ridgeway and Correll 2006) leads decision-makers to perceive femininity and professional leadership as incongruent (Eagly and Karau 2002; Meng 2002).

2 These findings contradict another prominent view supporting the opposite causal direction, namely that wage rate influences sex composition (Strober 1984; Reskin and Roos 1990).

3 In order to account for the continuously different labour market situation, our models include a variable controlling for whether a person has a job in East or West Germany.

4 As a robustness check, we re-estimated the models with data from the 2001 graduate cohort. The only remarkable descriptive difference is the larger share of women in the sample. As to our research question, we find largely the same results with the exception that the interaction between female-dominated occupation and time is not significantly positive any more for women, i.e., neither men nor women can compensate the negative effect of female-dominated occupations between T1 and T2. This is in line with findings that the phenomenon of the gender wage gap has been stable since the mid-1990s (Statistisches Bundesamt 2012).

5 In the HIS survey, respondents were asked about their gross monthly labour earnings at T1 and T2, but not for all employment spells since graduation. The choice to calculate hourly wages instead of monthly earnings aims at examining gender differences net of working hours (Petersen 1989). The possibilities to calculate hourly wages with HIS-data are, however, somewhat limited since contractual working hours are only reported for part-time workers. Therefore, full-time workers are assigned 40 weekly working hours. The alternative option of estimating gross monthly earnings would have forced us to exclude part-time workers in order to control for the working hours effect. Yet, this would have increased the problem of selectivity – especially regarding female working biographies at reproductive ages –, which exists anyway since we focus on graduates in employment only. In addition, we excluded self-employed individuals and those working abroad, since their wages are likely to be determined by mechanisms different from those graduates in dependent employment in Germany.

6 In order to check for differences between the results obtained by RE and FE models, we estimated FE models for the time-varying covariates. These models show that the effects of occupational segregation are the same as in our RE model, they vary only somewhat in size.

7 Despite a more detailed coding of subject areas in the data set, we opted for this rough classification in order to allow for a sufficiently large number of respondents in each subject category.

8 We are aware that the analysis of sex segregation is sensitive to the number of fields of study entered in the analysis (Smyth 2005). Therefore, we used the more detailed field of study classification to define female-dominated subjects (see Appendix A for the relationship between the detailed and rough classification as well as the percentage of women and the gender of each subject category).

9 The percentage of women required for an occupation to be considered female-dominated has been discussed controversially. While Smyth (2005), for example, assumes a threshold of 60%, Anker (1998: 80–6) suggests a stricter definition of between 70 and 75% female or male. We followed the latter suggestion; however, models with different specifications yielded similar results.

10 This operationalisation may, however, be criticised for not being able to differentiate between evaluative discrimination and mere crowding effects. In order to test the alternative explanation that women crowd in a relatively small number of occupations which they consider family-friendly, we additionally control for the perceived family-friendliness of the workplace.

Additional information

Notes on contributors

Kathrin Leuze

Kathrin Leuze is a full professor of Education Sociology at the Institute of Sociology of the Leibniz University of Hannover. She received her PhD from the Graduate School of Social Sciences (GSSS) at the University of Bremen in 2007. She is interested in quantitative life course research with special emphasis on higher education, employment biographies, gender inequalities and cross-national comparisons.

Susanne Strauß

Susanne Strauß is an assistant professor of Sociology at the Institute of Sociology of the University of Tuebingen. She received her PhD from the Graduate School of Social Sciences (GSSS) at the University of Bremen in 2007. Her research interests are: quantitative life course research with special emphasis on employment biographies, old age security, gender inequality and civic engagement.