ISER Working Paper Series 2006-44
The long term impacts of compulsory schooling: evidence from a natural experiment in school leaving dates
24 Nov 2006
For a very long time compulsory schooling has been considered a cornerstone of education policy. However, in recent years, the emphasis on requiring students to complete a minimum number of years in school as a policy instrument has dwindled in importance, while focus has shifted to a stronger emphasis on choice and quality. Among economists, compulsory schooling laws have also attracted considerable attention in recent years as part of a widespread effort to find credible instruments to identify the causal effect of schooling on labour market outcomes. Starting with Angrist and Krueger's 1991 seminal paper, a strand of research has attempted to address the causality problem by using 'natural experiments' driven by the institutional structure of the education system and its unequal treatment of otherwise similar individuals on the basis of their date of birth.
Most of the institutional settings analysed to date, however, may confuse the 'true' schooling effect with other potentially important factors, such as the impact of relative age within a school cohort. This is because identification of schooling effects usually relies on a comparison between the oldest individuals in the class, who according to the rules in place are more likely to leave without completing the full grade, and their younger peers, who are compelled to stay on and attain that year's grade.
This paper approaches such problems by considering a unique set of institutional circumstances which were in place in England and Wales over much of the second half of the twentieth century. This structure was organized such that, depending on their date of birth, individuals within the same school cohort were allowed to leave school only after one of two specific dates upon reaching their final year of compulsory schooling. Since no other institutional differences, apart from school exit rules, apply in the period around the date of birth that marked this separation, we can safely attribute observed differences in performance to the impact of the school exit rule and build our estimation strategy around what is effectively a regression discontinuity design.
We show that this rule had a strong impact on school leaving behaviour, on qualifications achieved by age 16 as well as adult educational outcomes. We then address the question of whether differences in educational outcomes induced by the rule translate into significant differences in adult labour market performance. Our IV estimates suggest significant qualification effects on participation, employment and earnings, which are usually of a higher magnitude than those obtained through OLS. Consistently with other studies, we find that formal education is a more important driver of future economic success for women than it is for men.
Our research makes a contribution to the understanding of regulatory obligations as policy instruments. We show that well-considered interventions of this kind can actually have a positive impact on the subjects whose behaviour is constrained as a result of their enforcement. We argue that policies intended to improve educational outcomes should not fully dismiss the regulatory approach in favour of other financial incentives when both can be powerful complements. The right balance between these two approaches, however, still needs to be addressed.
This study also highlights the interaction between education, as measured by the time input devoted to schooling, and the credentials which are acquired as a result. The rules we consider here induce a fairly small difference in potential length of schooling - approximately three months. We demonstrate that the same school leaving rules imply stronger effects when the school leaving age is timed to compel individuals to complete a year in which they can be awarded nationally recognized qualifications. In other words, our results suggest that the effect of gaining a certification and not just merely length of schooling alone plays an important role in explaining future economic outcomes.
We thus conclude that policies aimed at increasing time in formal education should at least allow individuals to receive credit for the additional learning experienced. While there is no reason to assume that an investment model of schooling is the right framework for understanding teenage behaviour in relation to school, qualifications and skills, it is doubtful they will remain oblivious to the potential risk of not obtaining a labour market premium for the additional efforts.