Determinants of students’ salaries in the professional training year

ABSTRACT

This paper studies the main determinants of salaries for economics students in their year-long industrial placements. Using three different sources of data on three cohorts of economics placement students, including demographic characteristics, academic performance, programme of studies and employability-related characteristics, we find that academic performance, job location and industry type are the main determinants of placement salaries. We show not only that students’ academic performance can increase the returns of the placement year due to the possibility of high salaries, but such returns significantly increase at the top of the salary distribution. Students’ previous job experience also matters for high-paying placements. Conversely, demographic characteristics, such as age, nationality and ethnic background, do not appear to determine placement salaries. Finally, we find no evidence of gender differences in wages.

KEYWORDS:

• Academic performance
• placement
• salaries
• employability
• economics

Acknowledgments

We would like to thank Jo Blanden, Matthias Parey and João Santos Silva for their invaluable contributions, as well as participants at the School of Economics Seminar 2018, University of Surrey, and at The Developments in Economics Education Conference, University of Warwick 2019. The authors remain responsible for any possible errors.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 The response variable is in logarithmic form as this can show a constant percentage effect (Wooldridge 2014). That is, the effect of a change in average first-year mark is interpreted as a constant percentage change in real earnings throughout the mark distribution. For example, a mark increase from 55 to 60, or 65 to 70, would be associated with the same percentage change in real earnings. In addition, the log transformation can help correct the typical skewness in earnings distribution (see Figure 1).

2 An alternative to the dummy variable would have been to create a variable that counts each student’s accomplishments, but there is no scope for that in our data set because students report a very limited number of accomplishments, typically one or two.

3 We also run our model starting with a simple regression of the ln(salary) against two regressors (i.e. first-year mark and job experience) successively adding the rest of the control variables. For presentation purposes, we report only the results for the full set of explanatory variables.

4 A similar but less pronounced pattern is observed when considering the two highest quantiles, instead of just focusing solely on the top quantile. Also, the choice of three job experiences as the threshold value is in line with the average number of job experiences at this part of the salary distribution.