Educational effects of widening access to the academic track: a natural experiment

It is difficult to know whether widening access to schools which provide a more academically-orientated, general education makes a difference to average educational achievement. Although, there has been a shift in this direction in many OECD countries, reforms have been difficult to evaluate because they are often accompanied by other important changes to the educational system or because they have been introduced at the same time across regions. However, the consequences of such reform are deeply controversial and very much a current policy issue. In particular, contenders argue that an increase in access to the ‘academic’ track harms the quality of education for everyone (through contextual effects) without improving the prospects of those enabled to attend the more academic track. In this paper, we make use of a unique natural experiment to identify the net effect of widening access to schools which provide a more academically-orientated, general education on overall education outcomes. The reform consists of widening access to the more academic track within one specific province (Northern Ireland) at a particular point in time. This is the only differential change that happens across the regions considered – England and Northern Ireland – which otherwise have a similar curriculum with the same national examination for students at age 16 and 18. The reform enabled a very significant increase in the number of Northern Irish pupils who could attend the more academic track (‘grammar schools’) at the end of primary school, between the pre-reform (born in 1978) and post-reform cohort (1979). This natural experiment makes it possible to identify the effect of widening access to the more academic track on overall educational attainment, by comparing educational outcomes in Northern Ireland and England, before and after the reform. Using a ‘difference-in-differences’ analysis performed on administrative data, it can be seen that the ‘open enrolment’ reform of 1989 (affecting the 1979 birth cohort) has been accompanied by a clear impact in Northern Ireland relative to England. A 15 percentage point increase in the number of people enabled to attend grammar school (at age 11) was accompanied by shifts of a similar magnitude in the number achieving 5 or more GCSEs at A*-C and 1 or more A-level. This suggests a strong causal effect of expanding the more academic track on overall educational achievement. Just before the ‘open enrolment’ reform, there was a change affecting admissions in a qualitative way. Up to 1988, girls and boys were assessed in different categories such that the same percentage of entrants to the admission test would obtain a given grade (determining whether or not they could be admitted to grammar school). Following a high court ruling in June 1988, this practice was discontinued and from then on, girls and boys were assessed together (which affected grammar school intakes in 1989, i.e. the 1978 cohort). This change was to the advantage of girls since they outperformed boys on the ‘verbal reasoning’ tests that were the basis of selection at this time. The one year gap between this qualitative change to admissions and the ‘open enrolment’ reform has generated significant upward and downward shifts in the relative proportion of girls enabled to attend grammar school across cohorts 1977-1980. We show that these shifts have been followed by parallel shifts in girls’ subsequent relative outcomes at age 16 and 18. This confirms the considerable effect of grammar school entry on educational outcomes using a different source of identification to that described when comparing outcomes over time between England and Northern Ireland. We also replicate this latter analysis for boys and girls separately and affirm our previous results. Thus, whether we compare girls and boys within Northern Ireland or make comparisons by gender between Northern Ireland and England, it is clear that grammar school reforms have a strong impact on educational outcomes and that the design of the educational system (in this case, the mechanism of entry into grammar school) has consequences for gender differences in educational outcomes. As well as considering the overall effect of expanding the academic track on educational outcomes, we are able to use the same experiment to consider whether the selective system is a contributory factor to observed inequalities between socio-economic groups with regard to later educational outcomes. Specifically, we are able to analyse the effect of the reform according to whether children are eligible for free school meals (FSM), which roughly corresponds to families at the bottom quartile of income. Before and after the reform, there is a big difference between the probability of lower-income groups entering grammar school and achieving good educational outcomes at age 16 and 18. The reform has an equal impact on children with and without FSM in terms of entry to grammar school and educational achievement at age 16 and 18. Hence, we are able to conclude that grammar school attendance has no less effect on relatively disadvantaged pupils than it has on more advantaged pupils. Therefore, the barriers that make it difficult for FSM children to enter into grammar schools in the first place (e.g. lower test scores at age 11 because of lower parental resources) has a long-term effect on inequality through (in part) to the lower probability of FSM children entering into grammar school. Although this paper cannot be interpreted as evaluating the overall effects of a comprehensive or selective/ ‘tracked’ system of education, it is an example of where widening access to the more ‘academic track’ has generated positive net effects. . It illustrates the high price individuals pay from being excluded from the ‘academic’ track, even when they are someway down the ability distribution within the cohort. Also, this study provides clear quasi-experimental evidence that selection into the more academic track really has a causal impact – it is not simply an artefact of the selection process.