Purpose – To estimate gender disparity in high school dropouts when households experience financial shocks, this study aims to extend the simple decision-making process of school dropouts to allow for the multiplicity of dropout choices with respect to gender in households having exactly one boy and one girl in high school. We also account for heterogeneity in the households choosing the baseline “no dropout” category and resulting bias in the estimated effects of variables by comparing Baseline Inflated Multinomial Logit (BIMNL) models with Multinomial Logit (MNL) models. Design/methodology/approach – We filter households from India Human Development Survey (IHDS) dataset having exactly one boy and one girl in high school. We estimate Multinomial Logit (MNL) model and to account for the heterogeneity of the no dropout case, we estimate BIMNL models developed by Bagozzi and Marchetti (2017) and compare average partial effects (APE) and econometric performance of MNL and BIMNL models.
Findings – We find that Households facing shocks due to marriage or crop failure are more likely to drop out the girl. BIMNL only could capture the association of crop failure with dropout of girl. Econometrically BIMNL models performed better than MNL.
Practical implications – It would be pertinent to examine whether girls are especially at higher risk of high school dropout with these more sophisticated econometric tools to draw effective policy interventions in the backdrop of financial shocks due to Covid-19 / geopolitical crisis across the world.
Soumik Biswas, Indian Institute of Management Lucknow–Noida Campus, India
Kaushik Bhattacharya, Indian Institute of Management Lucknow–Noida Campus, India
Stream: Education and Social Welfare
This paper is part of the ACSS2023 Conference Proceedings (View)
View / Download the full paper in a new tab/window
To cite this article:
Biswas S., & Bhattacharya K. (2023) Choice of Gender in High School Dropout: A Comparison of MNL and BIMNL Models ISSN: 2186-2303 – The Asian Conference on the Social Sciences 2023: Official Conference Proceedings https://doi.org/10.22492/issn.2186-2303.2023.26
To link to this article: https://doi.org/10.22492/issn.2186-2303.2023.26
Comments & FeedbackPlace a comment using your LinkedIn profile
Share this Research