I felt a twinge of disappointment on hearing about the government’s decision to shelve its grammar school plans.

I went to the sixth form of a grammar - William Ellis in North London, which is now non-selective - and before that I went to two middling comprehensives and failed all my O levels, apart from one. Had I not retaken my O levels and got into William Ellis, I might not have ended up at a good university.

I would not want to deny to other children the opportunity I had or one I might easily have chosen for my own children, had I not succeeded in setting up the West London Free School.

Most participants in the grammar school debate refer to “the evidence” as if it was unequivocally on the side of the opponents, but that is not the case.

As Theresa May has pointed out, children on free school meals perform better in selective schools, on average, than they do in non-selective schools, and that remains true if you control for prior attainment.

But the difficulty with increasing the number of selective school places as a solution to the problems identified in this week’s Social Mobility Commission report is that there are not that many of these children.

At present, just under half of England’s 163 grammar schools give preferential treatment in their admissions arrangements to applicants from disadvantaged backgrounds, yet less than 3 per cent of those who get in are on free school meals and very few are from the UK’s lowest achieving demographic - poor white British boys.

New selective schools could be encouraged to do outreach of various kinds to try and get these numbers up, and introduce quotas, but they would be unlikely to rise by much.

One recent report found that 2.4 per cent of the children being admitted to grammars schools were on free school meals, whereas pupils eligible for free school meals make up 6 per cent of high-attaining children at the end of key stage 2 (ie, children likely to pass the 11-plus).

At present, there are about 500 children eligible for free school in each year group at England’s 163 grammars out of a total of roughly 20,850. That would only increase to 1,250 per year group if the percentage of high-attaining children on free school meals at grammars reflected the national average.

The main beneficiaries of an increase in grammar school places would likely be the children of middle-class and lower-middle-class families, just as they were when grammar schools were expanded in the wake of the 1944 Education Act.

According to the Crowther Report (1959), around 36 per cent of the sixth-form pupils at grammar schools were classified as members of the “professional and managerial” class, 18 per cent as “clerical”, 36 per cent as “skilled manual”, 7 per cent as “semi-skilled manual” and 3 per cent as “unskilled manual”.

Limited impact on social mobility

The fact that new selective schools would be unlikely to do much to increase the participation in higher education of poor white boys is not, by itself, a reason to oppose them. They might have facilitated less dramatic forms of social mobility - from the third socio-economic quintile to the second, for instance, rather than bottom to top.

Some teachers at grammar schools praise the hothouse atmosphere of their classrooms, pointing out that it is possible to learn at a faster pace and explore subjects in greater depth, thereby preparing children better for elite universities.

Roger Scruton, himself the beneficiary of a grammar school education, believes selective schools play an important role in the preservation and expansion of knowledge, steering exceptionally bright children towards academic careers.

Many conservatives believe the expansion of grammar schools is justified from a parental choice point of view and argue that it is unfair that parents of highly able children cannot send them to selective schools if they do not live close enough to one of the 163 existing grammars and cannot afford to go private.

Opponents generally focus on the supposed harm that more grammar schools will do to children at neighbouring, non-selective schools - by skimming off the most able pupils and the best teachers. But the evidence for this is inconclusive.

In 2016, the Education Policy Institute published a report entitled “Grammar Schools and Social Mobility” that, among other things, compared the performance of children at non-selective schools in selective areas with that of children at comprehensives in non-selective areas.

It found that if you define “selective areas” quite broadly - ie, allow for the fact that some children travel large distances to attend grammar schools - and control for prior attainment, the two groups perform no differently.

That suggests that, in aggregate, the presence of a grammar school in an area does not have a harmful effect on children at non-selective schools.

However, the EPI did find that children on free school meals performed slightly worse in non-selective schools in selective areas than their counterparts at comprehensives in non-selective areas, but this difference was partly attributable to the fact that the most able children on free school meals in the selective areas were at grammar schools.

Across all state schools, 33.3 per cent of pupils eligible for free school meals got five A*-C GCSEs, including English and maths, in 2015 compared with 30.1 per cent in selective areas, suggesting that, as things stand, grammars have a small negative effect on the most disadvantaged.

The EPI also found that children in the top quartile of the ability range, as measured by prior attainment, performed no better in their best eight GCSEs at grammars than they did at good comprehensives, defined as the top 25 per cent as measured by value added - and there are five times as many of these high-quality comprehensives as there are grammars.

However, high-ability children did perform worse at the remaining 75 per cent of comprehensives.

The EPI concluded that more selective schools would be unlikely to raise or lower standards overall, which it thinks is a good argument against the policy. Why bother to create more selective school places if it is not going to bring about any system-wide improvement?

But it could just as easily be an argument in favour. After all, if creating more grammar schools will not have any significant negative effects, why not go ahead and do it since it would clearly be popular with many parents and increase school choice?

To my mind, the best argument against is that increasing the number of selective schools would lead to a more differentiated curriculum being used across the school system, with fewer children being taught academically challenging subjects and more being steered down vocational pathways in key stage 4.

I also worry that a great deal of political capital and energy would be needed to create more selective places - the reason it was dropped from the Queen’s Speech - and a government that did this might not have a great deal to show for it after five years.

With a fair wind, the total number of grammar schools could have increased to 200 between 2017 and 2022, assuming not many non-selective schools converted. That would still only be 200 schools among roughly 24,300 state schools in England - less than 1 per cent of the total.

My reluctant conclusion, because I’m a Conservative and believe passionately in school choice, is that the opportunity cost associated with creating new grammar schools would be too great.

I do not think it would have much impact on attainment across the school system, and might even have a small positive effect if planned carefully, but the effort-to-reward ratio means the game is not worth the candle.

All that energy and political capital would be better spent on other projects that would yield greater benefits, such as improving behaviour and helping all schools enter as many children as possible for the English Baccalaureate.

In other words, it would be better to focus on creating more “comprehensive grammars” rather than more grammars.

Toby Young is a free schools founder. He tweets @toadmeister

Want to keep up with the latest education news and opinion? Follow Tes on Twitter and like Tes on Facebook