HIGHLY EDUCATED BUT IN THE WRONG FIELD?

ABSTRACT

This paper investigates the impact of gender differences in tertiary education, i.e., field of study and level of tertiary degree, on two selected labour market risks: unemployment and low-status jobs. Using Labour Force Survey data from the year 2000, results of the logistic regression models and non-linear decomposition analyses generally confirm our expectation that the field of study explains a sizable portion of the gender gap in unemployment and low-status jobs in both countries. However, the level of tertiary degree earned explains only part of the female disadvantage behind holding a low-status job in Spain. The analyses also show that compared to men, women with a degree in a predominantly male field of study seem to be systematically disadvantaged in both Germany and Spain, particularly with respect to unemployment. Overall, the analyses reveal that gender differentiation in tertiary education leads to similar outcomes in two very different institutional contexts.

Keywords:

• gender
• gender segregation
• field of study
• education
• labour market

Notes

¹Calculations are based on the Mikrozensus and the EPA 2000.

²Conversely, there is also research that indicates that men receive a ‘bonus’, such as faster promotion opportunities, for entering traditionally female fields (Williams 1992).

³Foreign nationals were excluded because in many instances they obtained their tertiary degree abroad which limits comparability to the German/Spanish population.

⁴The ILO definition classifies persons as unemployed when they are not in employment at the time of the survey, are currently available and willing to take up paid work within 2 weeks, and were actively seeking work in the last 4 weeks.

⁵As the distribution of ISEI scores in our sample could not be considered metric, dichotomising the ISEI seemed appropriate. The cut-off at 50 points was chosen because average occupations with ISEI scores below 50 points – for example electronic equipment operators (48 points) or physical and engineering science technicians (49 points) – seem to be inadequate for somebody with a tertiary-level qualification. Nevertheless, some of the occupations with scores from 50 to 55 and from 45 to 49 are arguably hard to place in the dichotomy of low-status and adequate status.

⁶As graduates who were older than 35 years were excluded from the analyses, we chose to include the age variable with linear instead of polynomial functional form.

⁷The classification is based on gender composition of the previously analysed 12 fields of study. Typically male/female fields of study are those which have a share of at least 75% of male/female tertiary graduates. Apart from medical sciences (integrated in Germany, female in Spain) and natural sciences (male in Germany and integrated in Spain) all fields are classified the same way in Germany and Spain (see p. 8).

⁸We also computed differences in the predicted probabilities in unemployment for men and women with lower tertiary degrees. Because the results and are very similar to the results reported here, no additional figure is displayed.

⁹Differences in predicted probabilities to enter low-status employment for persons with lower tertiary degrees were also computed. The pattern of gender differences is similar to the results reported in Figure 2. In Spain however, the female advantage in typical female fields is considerably larger. In Germany, the female disadvantage is more pronounced across all fields.

¹⁰One could also write the decomposition (1.2) using the male coefficients as weights for the first term in the decomposition and the male distributions of the independent variables as weights for the second term (see Fairlie 2003: 3).

¹¹It follows that if there are more men in the sample, a subsample of women equal to the N of the male sample would be drawn.

¹²The decomposition was computed with the user written Stata program ‘fairlie.ado’ by Jann (2006).

¹³Using coefficient estimates from a female and a male sample, the results revealed that the magnitude of the individual contributions of the independent variables as well as the total contribution of all variables is lower when the female sample is used while the relative size of the contributions are about the same. The coefficient estimates from the male sample produces very similar results (see Reimer and Steinmetz 2007 for more detail).

¹⁴As in the decomposition of unemployment we used coefficient estimates from a male and female sample as alternative specifications (not reported). Again, the results revealed that the magnitude of the individual contributions of the independent variables as well as the contribution of all variables taken together is substantially lower when the female sample is used while results were largely the same with the male sample (see Reimer and Steinmetz 2007 for more detail).