Gender differences in the labour market entry of STEM graduates

ABSTRACT

Many women do not work in science, technology, engineering, and mathematics (STEM) occupations even though they have degrees in these subjects. To shed light on this problem, we use information from the German Graduate Panel and show a significant gender gap among STEM graduates working in degree-related occupations after graduation. Therefore, we focus on university graduates’ transition into the labour market and include male and female non-STEM and STEM graduates. We find that male STEM graduates are more likely to work in a degree-related field than other men. A gender gap in degree-related work in STEM occupations shows that this is not the case for women. Separating STEM into engineering and computer science (EngComp) and mathematics and natural sciences (MatNat) shows that EngComp graduates are the main driver of the STEM effects. The estimations remain robust to a comprehensive set of individual background information. Moreover, bearing children before graduation or at the beginning of one’s professional career does not explain the lower entry behaviour of female EngComp graduates. Possible channels for why women with an EngComp degree are not as likely as men to start their professional life in an EngComp occupation are discussed.

KEYWORDS:

• University degree in engineering
• University-to-work transition
• Higher Education
• Gender-differences in STEM
• Education job mismatch

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The DZHW surveys graduates only every four years.

2 See appendix A1 for additional discussion on the construction of the dependent variable

3 Official statistics to assess the reliability of our measure are difficult to find. The OECD reports a field of study mismatch for Germany of 20%; however, data only exists for the years 2015 and 2016 (OECD 2017).

4 This includes the fields of mathematics, physics, chemistry, pharmacy, biology, geo-sciences geography, as well as computer science and all engineering fields.

5 Results are robust to not adding all other fields of study.

6 Estimations using random-effects or logit models lead only to minor changes after the third decimal and do not influence the interpretation of the coefficient or the significance level.

7 This information comes from the second wave of the survey, and we assume that desire is constant over both waves.

8 See additional discussion on regionality in appendix A4.

Additional information

Notes on contributors

Jakob Schwerter

Dr. Jakob Schwerter is a PostDoc TU Dortmund, Center for Research on Education and School Development, and an associated member of the LEAD graduate school and research network. For his Ph.D., he worked at the University of Tübingen, Chair for Econometrics, Statistics, and Quantitative Methods. The title of his dissertation is Econometric Analysis of the Effects of Educational Decisions on Labour Market Outcomes and the Influence of Self-Testing on Learning Outcomes. His main research focus lies in the field of education economics, educational effectiveness, and evaluation, as well as policy analysis.

Lena Ilg

Lena Ilg was a student at the University of Tübingen who wrote her Master’ Thesis about this topic. She is now working at Porsche in the department of Analyst Customer Intelligence.