Overeducation wage penalty for university graduates: evidence from the MENA region using machine learning techniques

ABSTRACT

This study estimates the causal effect of overeducation on wages using cross-sectional survey data from three countries in the Middle East and North Africa region. The Labour Market Panel Survey data from Jordan in 2016, Tunisia in 2014, and Egypt in 2012 and 2018 are used. Overeducation occurs when an individual works in a job that requires someone with less education. Our analysis focuses on employees with high school or college education in paid-wage jobs. Overeducation in the data is self-reported. Between 10% and 50% of the samples in the various countries are overeducated, with overeducation being more pronounced among high school than college-educated people. The novel causal forest method as well as the nearest neighbour and propensity score matching methods are used to estimate the over-education wage penalty. The overeducation job mismatch wage penalty is significant in both total wage and basic wage. The size of the penalty, however, varies considerably between the three countries and is negatively associated with the prevalence of overeducation in the job market in each country. The penalty in total wage ranges from 17% in Egypt in 2012 to approximately 50% in Tunisia. High school graduates have smaller penalties. We recommend policies such as conducting more active labour market programmes to reduce the stock of mismatched workforce, enhancing job search services to improve the quality of matching in new vacancies, and following strategies that motivate entrepreneurship.

JEL CLASSIFICATION:

• C21
• C50
• J31
• J24
• I26

Acknowledgments

I am gratefully acknowledging the crucial support of the Deanship of Scientific Research at King Faisal University (KFU). The author's deep gratitude goes to Martyn Andrews at the University of Manchester for his help, I am grateful for your support. The author would like to thank Len Gill, Isam El laythy and Hisham Mansour Elhaggaz for their constructive comments and help. I am gratefully acknowledging the crucial support of the colleagues in the Economic Research Forum and for providing me access to the data for conducting this research. I would also want to thank my colleagues in the Department of Economics and the College of Business Administration at King Faisal University and my family, Zeinab, Duaa and Ammar.

Disclosure statement

No potential conflict of interest was reported by the author(s).