![Sample our Education journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5932/TOC-Subject-Sample-2021-Education.jpg"})
Abstract
This article addresses the question of why fields of study differ in early labour market returns. It is argued that the higher the potential training costs of a field of study the more problematic the labour market integration of graduates. This is due to the fact that employers use the occupational specificity and selectivity of a study programme as a signal for the expected training costs. In addition, the article suggests that structural relations between fields and occupations act as mediators for the effect of field of study on non-pecuniary returns. Using the German HIS Graduate Panel 1997, the results indicate that a lack of occupational specificity is partly responsible for differences between fields of study. Selectivity measures do not contribute to an explanation. As expected, working in the public sector and the required expertise for a job strongly mediate field of study differences.
Notes
1. The data from 1997 characterise the German higher education system in the pre-Bologna Process. At present higher education is differentiated into a bachelor's and master's degree. Moreover, the so-called Exzellenzinitiative tends to establish stronger hierarchical levels between universities in terms of prestige. As the internal differentiation and the accentuation of single universities may complicate the analysis of field of study mechanisms, the ‘old’ data seem to be advantageous.
2. The sample is restricted because the hypotheses do not refer to self-employed. As to graduates with a further field of study the dataset lacks additional information.
3. I only have information on job characteristics for the first (not implicitly the first significant one according to the relevant question) and current job in the first and second wave. If the first significant job that is extracted from the employment history is not equivalent to the first job in the dataset and does not fall in the period of the first or second wave, I do not have further information on this job. Therefore, the sample size is smaller for the analyses on overeducation and job mismatch than for job search.
4. The first significant employment does not include stop-gap jobs or marginal employments (all kinds of minor work such as internships, summer jobs or other casual employment). The variable search time contains 103 (1.78%) right-censored cases.
5. For those graduates who did not find a job I assign the mean value of occupation-based dispersion from their field of study. Due to perfect collinearity I cannot estimate the effects of the fields of study and a measure of the dispersion of occupations in fields in the same model. Thus, I consider the occupation-based dispersion index as a proxy of a field's occupational specificity. In the original formula the range is between zero and one, where higher values indicate a more heterogeneous distribution. In order to ease interpretation the pattern is reversed.
6. The Abitur is the necessary requirement for eligibility for higher education in Germany. For some fields of study requiring Numerus Clausus average Abitur grades are important prerequisites for admission. In Germany grades range between 1 and 6 where 1 represents the best grade and 6 the worst. In order to ease interpretation average Abitur grades and their standard deviation are reversed.
7. In every analysis health and welfare is the reference category because bivariate analyses show that these graduates have the smoothest transition according to all indicators.
8. However, the effect has to be seen with caution as there might be endogeneity problems.
9. The standardization was calculated with the STATA-ado listcoeff (Long and Freese 2006). The y*-standardized coefficients can be interpreted as follows: for a unit increase in xk, y* increases or decreases by βk standard deviations, holding all other variables constant.