Month of birth and aspirational career choice: a gender perspective

ABSTRACT

This paper investigates the impact of students' month of birth (MOB) on their university career choices. Specifically, we analyze whether the oldest students in their academic cohorts show more aspirational preferences when expressing their first choice of university degree. Using administrative records for students in a large university district and applying a sharp regression discontinuity design, we find that the oldest female students are more likely to express a preference for high-earning, academically selective, and STEM degrees than their youngest peers. The results for STEM are concentrated in degree courses where women are a minority and in the technology and engineering fields. Conversely, we find much smaller or non-significant MOB effects in most cases for male students. Results further indicate that the MOB effect is reinforced in families with high socioeconomic status.

KEYWORDS:

• Month-of-birth
• academic achievement
• behavioral responses
• gender
• sharp regression discontinuity

JEL CLASSIFICATION::

• I21
• J01
• J16

Acknowledgments

We thank the Universitat de València and the Universitat Politècnica de València for providing the data. We would also like to thank Judit Vall as well as participants at the Sixth Lisbon Research Workshop on Economics, Statistics and Econometrics of Education 2022, the Simposio de la Asociación de Economía Española, SAEe-2022, and the Annual Conference of the Royal Economic Society 2023 for their useful comments.

Disclosure statement

None.

Notes

1 In many countries, the academic year starts in September and finishes in June, but cohorts are formed by children born from January to December. In other countries, such as the UK, the cohorts are formed by children born from September to August.

2 Month of birth has been also seen as a factor influencing a wide variety of other long-lasting effects coming from victimization in school (Mühlenweg 2010), ADHD diagnosis (Elder 2010), mental health (Black, Devereux, and Salvanes 2011) or success in politics, academia and professional career (Du, Gao, and Levi 2012; Fukunaga, Taguri, and Morita 2013; Muller and Page 2016). See Peña (2017) for a comprehensive review of the month-of-birth effects on different outcomes.

3 According to data for the OECD countries, excluding health-related sciences, only 13$\text{\%}$ of female students as compared to 39$\text{\%}$ of male students graduated in STEM in year 2019 (OECD 2021). In Spain, the proportion of university graduates is even lower than the OECD average, with only 11$\text{\%}$ of women as compared to 36$\text{\%}$ of men graduating in STEM in that year. In health sciences, the percentages are 23$\text{\%}$ for women and 10$\text{\%}$ for men.

4 Around 75% of a student cohort entering university in a given year in the region of Valencia choose a public university in the region (Spanish Ministry of Science and Innovation, MICIN, https://tinyurl.com/2egtxvzm, pp. 30–31, last accessed February 2023.)

5 Delaney and Devereux (2021) use the same variable as we do to analyze the different aspirations of men and women in career choice for Irish students. See also Saygin (2016) for Turkey.

6 In an Appendix at the end, we replicate the main results presented below using instead the degree in which students actually enroll. Those results indicate that the actual enrollment seems to be a much weaker factor for identifying the expected effects.

7 https://forbes.es/empresas/8820/las-10-profesiones-mejor-pagadas-en-espana/, last accessed May 2020; Informe Anual ESADE-Infojobs: https://nosotros.infojobs.net/prensa/informes; https://www.mentedidactica.com/carreras-universitarias/, last accessed May 2020).

8 Some of the average earnings for degrees in this group are: engineering project management 70,956; dentist 54,748; IT management 53,396; orthodontist 52,798; commercial management 52,406; etc. According to INE, in 2018, the average earnings for workers with tertiary education were 29,463 (Encuesta de Población Activa).

9 We believe the cutoff at the 90th percentile is a reasonable solution compared to the alternatives of setting a higher cutoff, which would mean very few degrees were classified as selective, or setting a lower cutoff, which would mean we might pick up degrees that are not really selective.

10 Students mark their options in the admission application form once they know their final entrance grade and the entry cutoff of each degree, which is also made public in advance. Thus, they might try to target the 'best they can', being aware that it does not make sense to apply for degrees too far above their entrance grades. Only in some specific circumstances can students enter a degree with an admission cutoff substantially above their entrance grades. These admissions tend to be the result of preferential policies of the university, targeting, for instance, elite sports students or students with some degree of disability. We leave these cases out of our estimation sample.

11 Clasificación Nacional de Ocupaciones, CNO-11, INE. https://www.ine.es/daco/daco42/clasificaciones/.

12 In a standard parametric estimation, accounting for the possible non-linearity of the effect at both sides of the cutoff is crucial to rule out the possibility that what looks like a jump might be simply an unaccounted for non-linearity in the relationship. The non-parametric local estimation avoids this problem. By default, we use second-order polynomials all through the main analysis that we present below. Nevertheless, we provide in the Appendix robustness checks where all the RD estimates have been replicated using first-order and third-order polynomials

13 As a result of the method's automatic choice of the bandwidth – MSE-optimal bandwidths – the number of days selected in our estimates below range from 60 days to the right to 75 days to the left, depending on the specification. Thus, roughly speaking, we will be comparing students born in February with those born in November–December.

14 Implemented using rdrobust package in Stata.

15 If it is true that students become more mature as they get older, it could be the case that, for instance, younger students are more likely to live at home with their parents than older ones, or that older students are more/less likely to divide their time between study and work.

16 This is a working paper with preliminary work on the MOB effects for university students at UV. The sample (more limited, including only students of economics and medicine at UV) and outcome variables (entrance grades and university GPA in students' freshman year) were different to the ones used in this paper.

17 More specifically, the points in the figure are IMSE optimal number of evenly spaced bins on the support of the running variable which is selected to balance squared bias and variance in order to approximate the underlying conditional expectation globally.

18 To classify each degree in the different fields we follow the classification made by the US Department of Homeland Security STEM Designated Degree Program List. https://tinyurl.com/2lsaamsb, last accessed February 2023.

19 This positive result of column 1 for males would correspond to the positive jump observed in the third plot of Figure 6.

Additional information

Funding

P. Beneito and Ó.Vicente-Chirivella acknowledge financial support from Grant PID2021-124266OB-I00 funded by MCIN/AEI/10.13039/501100011033 and by ‘ERDF A way of making Europe’. Javier Soria-Espín acknowledges funding by the Bank of Spain Graduate Fellowship and by the Agence Nationale de la Recherche (ANR) under the framework of the investissements d'avenir programme reference ANR-17-EURE-001 and Generalitat Valenciana [PROMETEO - 2019-95].