The economic value of education : three statistical approaches

• Main Author: Requena Casado, Laura
• Other Authors: Risa Hole, Arne ; McIntosh, Steven
• Published: University of Sheffield 2017
• Subjects: 330
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.739833

This thesis deals with the topic of the economic value of education from three different substantive and statistical perspectives. In Chapter 2 the effects on educational subject choice of the increase in tuition fees after the 2012 higher education reform in the UK are analysed at individual level. A multinomial logit model with a difference-in-differences approach is estimated with data from HESA, choosing Scotland, where the 2012 reform had no effect, as the control group. Several models are presented taking into account students’ socioeconomic background and gender. Main results show that after the reform students are less likely to opt for Arts & Humanities and more likely to choose Health & Life Sciences. Chapter 3 estimates the returns to education in Spain using an instrumental variables approach. With data from the 2011 Living Conditions Survey, a sample selectivity model is estimated in order to avoid possible selection biases derived from only considering wage-earners. Family background variables are used as instruments for schooling. Beside showing that, in agreement with pertinent literature, ordinary least squares coefficients are downward biased, the analysis confirms that more educated individuals have the largest returns, although in Spain educational returns are higher for women than for men. In the context of increasing global migrations, recent research has shown relevant differences in education and skills between natives and immigrants in developed societies. With data from PIAAC, chapter 4 presents extended Mincerian equations to gauge differences regarding returns to skills and levels of education between natives and immigrants arrived to Denmark, France, Spain, and United Kingdom from OECD and non-OECD countries. The ordinary least squares analysis is complemented with a Bayesian estimation. Results suggest that there are significant differences between native and immigrant returns to education in some of the European countries considered.