Tomorrow, thousands of 15 and 16 year olds will receive their GCSE results. In the future, these exams may be considerably different. It has been reported that the education secretary, Michael Gove, is planning to make radical changes to the structure of GCSEs in England. One potential change is to ensure there is only one exam board such that there is no competition between providers encouraging schools to choose the easiest option. The other is to replace the current system of GCSEs with a dual system of exams, one harder and one easier. Recent analysis by IFS researchers throws light on the desirability of these potential reforms and whether they would actually represent a substantial change compared with the current system. The evidence also raises a third and more fundamental question: what are GCSEs for?

Since the first group of 15 and 16 year olds sat GCSEs in 1988, grades have trended up year-on-year. Between 1996 and 2011, the proportion getting five GCSEs at A*-C (including English and Maths) increased from 35% to 58%. These increases could reflect improvements in student's knowledge. However, the sheer pace of this increase may reflect 'grade inflation', with exams getting easier. It is hard for outsiders to judge - itself a problem to the extent that it reflects a lack of confidence in the system. And the regularity of the year on year increases can reinforce that concern.

However, there are reforms that could reduce perverse incentives for schools in the current system, which is one motivation behind reforms to GCSEs proposed by the education secretary.

Choosing exam boards and qualifications

The qualifications system in England and Wales currently involves a number of exam boards competing for shares in a regulated market. One danger is that they might compete by offering qualifications that are easier to pass. A school choosing between two qualifications which will count the same in league tables, and perhaps to the outside world more generally, has a clear incentive to choose the one it thinks more of its students will pass - the easier one. It is hard to judge the degree to which this has actually contributed to rising GCSE results and multiple exam boards may be able to offer more diversity than just one. However, the incentives are clear. The theoretical case for eliminating such perverse incentives seems strong.

Rising GCSE results in the late 2000s were also buoyed by a proliferation of "GCSE equivalent" qualifications, which meant that by 2009-10 42% of pupils took 'Vocational-Related Qualifications', 30% took BTECs and 10% took GCSEs in vocational subjects. These 'GCSE-equivalent' qualifications tend to be taken by more disadvantaged pupils. Work by IFS researchers showed that schools that had moved most aggressively into these GCSE-equivalent qualifications were also the ones that improved their league table position by the most. This is suggestive evidence that schools use variation in qualification difficulty to game the league-tables. Correlation is not causation, but evidence also led the Wolf Review to similar conclusions. The government is currently implementing the recommendations of this review, including slimming down the number of vocational qualifications that count towards league tables.

A dual system

At the same time, Mr Gove is reported to be considering replacing GCSEs with a dual system of harder exams like the old O levels and easier ones like the old CSEs. His stated aim is to challenge more able students and to ensure that higher standards are maintained. Given that there are already different tiers of GCSEs and many other vocational qualifications, it is not yet clear how much difference any such change would make in practice. Nor is it clear how a new system of exams would affect the behaviour of schools and parents.

Of course there is much room for improvement in school standards. According to international rankings, the UK performs around the OECD average for reading and maths, slightly above average for science, but well behind high-performers such as Canada, Finland and South Korea. Perhaps Mr Gove's proposals could help deal with this?

The trouble is that if you dig under the average results from international comparisons what you find is that the performance of pupils in England is rather unequal with some strong associations between social class and performance. A worrying finding from the research cited above is that it has tended to be poorer pupils who have been directed away from traditional GCSEs in recent years. Recent research published in Fiscal Studies also suggests that by international standards there is a particularly big gap between social groups in the performance of the most able - with the most able pupils from lower social groups doing much worse than the most able from more privileged backgrounds. Seeking to stretch the most able pupils from all backgrounds may well help to deal with some of these problems. However, there may well be other factors holding back able pupils from disadvantaged backgrounds, such as differences in pupil attitudes or aspirations. Such a policy also seems unlikely to aid less able pupils in general.

The role of GCSEs

There is perhaps a third more fundamental question that we should be asking, and which certainly needs to be stated clearly before any reforms are introduced. And that is, what exactly are GCSEs for? England is actually rather unusual in having a high stakes school leaving exam at 16. Most countries focus on exams when most young people in fact leave school at 17 or 18. The system in England looks rather like a left over from a time when the majority of young people did expect to leave school at 16. Now that the vast majority stay on past 16 to do further qualifications there must be some question over the role of a set of exams which may signal to some that leaving at 16 is expected, particularly in the context of government policy to raise the "education participation age" to 18.

GCSEs do perform other roles as well. They are often used to hold schools to account for their performance and are one of the only external measures of attainment universities can see when making offers of places. However, other measures of school success could be used in league tables (such as a core of subjects or post-16 results) and may well be more desirable if schools have been using the present system to boost their league table position. It would also be odd to justify retaining GCSEs on the basis they are used for university admissions. Currently, the majority of children don't go to university and other reforms could improve the flow of information to admissions tutors, such as entrance exams or running the application process after A-Level results have been published. England is also extremely unusual in allowing those who do stay on past 16 to drop study of maths and English, as was pointed out in the Wolf Review.

Perhaps an even more radical rethink of the role of GCSEs and the structure of the public examination system is called for if we are to ensure that these exams serve a valid purpose and young people are best served for the future.