VoxEU Column Education Labour Markets

Selective school systems: German evidence for the UK’s grammar schools debate

Selective schooling (or ‘tracking’) seeks to improve efficiency in education by tailoring curricula to students’ needs, but opponents argue that it can lead to segregation and lifelong disadvantages. This column draws on evidence from Germany to argue for the importance of flexibilities in any new system of selective education, as with the UK’s proposed reintroduction of grammar schools. Substantial disadvantages in terms of teaching curriculum and peer exposure do not need to have long-term consequences, as long as the education system allows revising initial choices at later stages.

Selective schooling (or educational tracking, as it is known in many countries) seeks to improve efficiency in education by tailoring curricula to students’ needs. Proponents of selective school systems stress that allocating students to learning environments more appropriate for their abilities can lead to more efficient educational production (empirical results in Duflo et al. 2011, for example, support this opinion).

UK Prime Minister Theresa May has suggested lifting the ban on new grammar schools, an institution that still exists in Germany. She stresses that the current education system effectively leads to selection by house price and hopes that grammar schools may make the system more meritocratic again. Similarly, the French Republican politician Bruno Le Maire has suggested a more ‘diversified education’ by abolishing the current comprehensive middle school system of the collège unique.

On the other side of the debate, opponents argue that tracking may lead to segregation and thus exacerbate initial disadvantages, because of the danger that mis-tracking due to a lack of information on academic abilities may translate into large differences in adult outcomes through exposure to very different learning environments (Hanushek and Wössmann’s 2006 empirical results suggest so).

But initial mis-tracking need not lead to lifelong disadvantage, as long as the tracking system has built-in flexibilities that allow correction of initial choices at a later point, and when more information about individuals’ innate abilities is available. This important aspect in assessing tracking systems is hardly considered in the academic literature, and has only recently entered the public debate when prime minister May suggested ‘points of entry’ in to grammar school at different ‘key stages’.

Measuring the long-term effects of tracking

To assess whether early track allocation does indeed have lifelong consequences, and whether a tracking system allows successfully for correcting initial mis-allocations, requires a comparison of long-term educational and labour market outcomes of individuals that were at the margin between two tracks, and who were randomly allocated into a higher or a lower track.

Our new study conducts such analysis. Of course, individuals who attend a lower track will on average be less talented than individuals who attend a higher track – a sorting that is the key purpose of a tracking system. To assess whether initial mis-tracking has long-term consequences, it needs to be possible to compare individuals who are equally talented when the tracking decision is taken, but who have been allocated to different tracks.

We draw on a natural experiment that allocates such ‘marginal’ individuals to different tracks and investigate the effects of being tracked into a higher versus a lower track on long-term labour market and educational outcomes. We explore this for Germany, a country with a particularly rigorous tracking system that very early on (at age 10) allocates students into three different school tracks based on ability and locks them into the assigned track throughout middle school (from grades 5 to 9, up to age 15/16).

Research design

Our research design exploits quasi-random shifts between tracks induced by date of birth. A previous school entry age rule requires students to enter school at age 7 when born in July and at age 6 when born in June.

At age 10, when the tracking choice is made, it is thus likely that children who entered school earlier may be slightly weaker academically than children who enter school later, even if they are equally talented. Thus, we should expect children born in June to be more likely to attend a lower track school than children born in July, even if they are equally talented. This is precisely what we find in our data, as indicated in Figure 1.

Figure 1 Higher (grammar) school track at age 14 by birth month

We then compare the long-term labour market outcomes as well as completed education between children who are allocated to different tracks solely because of their birth date, but who are likely to be very similar in terms of their ability. Note that since Germany does not have a compulsory school leaving age but rather a compulsory grade requirement, the age of school entry cannot affect the direction of schooling.


Despite the striking differences in learning environments between tracks, attending a more advanced track in middle school, induced through date of birth, has no effect on wages, employment or occupation choice at later ages (see Figure 2).

Figure 2 Wages (in logarithms and corrected for differences in work experience) by birth month

We attribute this surprising finding to the up- and downgrading of students between tracks that takes place at a later stage of their educational career. Indeed, we find a substantial amount of movement from the low or medium to the high track at the end of middle school (age 15/16) for marginal students, an upgrading facilitated by the school system that allows for multiple entry points to the highest track.

We further demonstrate a substantial amount of downgrading at the age of 18/19 through non-enrolment in college or university after graduating from the high track.

Due to this up- and downgrading, attending a more advanced track has, for marginal students, hardly any impact on the probability of graduating from a medium or high track, completing an apprenticeship or graduating from college or university. Of course, these findings of zero effects on long-term outcomes for marginal students by no means imply that there is no difference in observed outcomes of students who sort into different tracks in middle school.


Two important conclusions emerge from our research. First, when evaluating selective (tracking) school systems, it is important to consider not only the potential misallocation of students across tracks – even more important are the flexibilities in the school system that allow students to remedy initial track choices at a later stage when more information about their academic abilities is available. These flexibilities are an important, though often overlooked, feature of the German education system.

Second, our research shows more generally that substantial disadvantages in terms of teaching curriculum and peer exposure do not need to have long-term consequences, as long as the education system allows revising initial choices at later stages.

This second implication is not only relevant for tracking systems, but for education systems more generally. Introducing such ‘second chances’ through additional points of entry into higher track schools at different stages is certainly a substantial improvement that leads to more meritocratic outcomes.

This is also much in line with Prime Minister May’s vision of a more meritocratic school system in the UK in a speech delivered on 9 September 2016:

"A modern, meritocratic education system needs to be much more flexible and agile to respond to the needs of every child. So we will demand that new grammars make the most of their freedom to be flexible over how students move between schools, encouraging this to happen at different ages such as 14 and 16 as well as 11."

Such additional flexibility may be particularly beneficial for those from disadvantaged backgrounds, and for whom academic talent may need longer to develop.


Dustmann, C, P A Puhani and U Schönberg (2017), “The Long-term Effects of Early Track Choice”, forthcoming in the Economic Journal.

Duflo, E, P Dupas and M Kremer (2011), “Peer Effects, Teacher Incentives, and the Impact of Tracking: Evidence from a Randomized Evaluation in Kenya”, American Economic Review 105(5): 1739-74.

Hanushek, E A and L Wössmann (2006), “Does Educational Tracking A¤ect Perfor- mance and Inequality? Differences-in-Differences Evidence Across Countries”, Economic Journal 116(510): C63-C76.