Labour market outcomes and their impact on tertiary decisions in Germany: class and gender differences

Abstract

In order to explain why students choose to enter or abstain from university education, economic theories of educational choice assume that the income returns to a degree are critical in motivating students’ educational decisions. However, important group differences between men and women as well as students from different class backgrounds that might influence the relationship between expected returns and educational choice have largely been ignored in this line of research. This article explores how and to what extent changes in income and unemployment returns for graduates influence the educational decisions of students that are eligible to enrol in tertiary education in Germany. A unique dataset from the German Higher Education Information System Institute consisting of large-scale surveys of university qualified students is analysed. Contrary to many previous findings, the results of the article suggest that variations in relative income returns do not seem to affect the educational decisions of school leavers while variation in unemployment seems to have significant effects on postsecondary decisions for women only. Furthermore, female students from lower-class backgrounds appear to be more responsive to changes in unemployment rates than their higher-class peers.

Keywords:

• higher education
• labour market outcomes
• educational decisions
• educational inequality
• gender differences

Acknowledgements

The author would like to thank Carlo Barone, Mads Meier Jæger and Kristian Karlson for helpful comments on previous versions of this paper. Michael Gebel's help with preparing the income data is thankfully acknowledged.

Notes

1. Gambetta (1987, 134) also criticizes the parking theory because of its short-sighted view that students’ choice of further education should almost solely be attributed to a lack of employment opportunities.

2. Control analyses with the second wave of the data yielded largely identical results (available on request).

3. This was done to avoid problems due to multicollinearity in the models because parental education and parental class are highly correlated.

4. The distribution across those categories is the following: service class with Abitur 15%/without Abitur 28%, intermediate class with Abitur 4%/without Abitur 32%, working class 21%.

5. A restricted Abitur (Fachhochschulreife) only permits entry to the universities of applied sciences but not traditional universities.

6. A restriction to younger graduates was not possible due to insufficient case numbers in that age group in the GSOEP.

7. Given that the students in the HIS survey were surveyed in December 1983, matching the 1984 income data does not seem very problematic.

8. Microcensus scientific use files from the following years were used: 1982 (matched to the 1983 cohort), 1989 (matched to the 1990 cohort), 1993 (matched to the 1994 cohort), 1999 (matched to the 1999 cohort), 2002 and 2004 (the average between 2002 and 2004 was matched to the 2004 cohort).

9. These can be broken down into five observation periods for each of the seven former West state-regions, two observation periods for West Berlin before reunification and three observation periods for five Eastern states after reunification (West Berlin was merged into a Brandenburg/Berlin cluster from 1994 onwards).

10. An alternative method to deal with the clustering at the state-level is the estimation of so called HLM – or Multilevel Models (Snijders and Bosker 1999). Given the small number of macro level state-units (13), the assumption that the regions are drawn from a random sample of states does not hold (e.g. Snijders and Boskers 1999, 43), which is why a logit model with year- and state fixed-effects was preferred.