When the new school league tables are published later this morning, there will be lots of parents and teachers who are shocked by the latest ranking of their school. More worryingly, the league table is going to pull head teachers in two directions and discourage them from using the pupil premium to help the poorest pupils.

The new table will measure how many pupils get good GCSEs in just five subjects – English, maths, science, a language and a humanities subject – rather than results across the board. The new measure, labelled an English Bac, is being introduced to combat what Michael Gove sees as a rise in "soft subjects". So many schools that rose up the league table last year because their pupils improved in other subjects are likely to see a sharp fall in their league table position today. Indeed, the government estimates that only 15 per cent of pupils will achieve the Bac.

The fact that the new league table has been introduced retrospectively has caused anger among teachers. They are concerned that they will be ranked using exams that were taken before the new measure was even in place. Up until this year, they were being actively encouraged to offer a broad curriculum, including information technology, diplomas and citizenship – none of which will be included in the new Bac. It is unfair, they claim, to penalise them for following the previous system.

Though these criticisms are justified, they are unlikely to cause much concern to Michael Gove. The very point of pushing the reforms so quickly is to shine a spotlight on the problem he has identified. The lower the number of schools passing the English Bac, the more evidence there is to back his claim that pupils are failing conventional GCSEs.

But what should give him a bigger headache is the impact the new league tables will have on the poorest pupils. Since becoming Education Secretary, Gove has made narrowing the attainment gap between rich and poor pupils his cause célèbre. Yet today's reforms to the league tables are likely to encourage schools to focus their resources on more affluent pupils.

IPPR analysis shows that, in 2009, only 26.6 per cent of pupils eligible for free school meals achieved five or more A*-C grade GCSEs or equivalent including English and maths, compared to 54.2 per cent of pupils not eligible for free school meals – an attainment gap of 27.6 percentage points. And only 10,000 children on free school meals got grades A*-C in a modern language – just one in 50 of that year's cohort of pupils.

The harsh reality is that the pupils most likely to achieve the Bac are those from more affluent backgrounds.

By placing the English Bac at the heart of the new accountability framework, the government is providing an incentive for schools to focus resources on those middle-class children likely to do well in a narrow range of academic subjects. Even the pupil premium, which is money intended for poorer children, is likely to be diverted to help boost a school's position in the league tables.

While Gove's rhetoric on narrowing the attainment gap between rich and poor pupils is laudable, he is sending a contradictory message with the new rankings. Until this contradiction is resolved, it will be the league tables that dictate where schools focus their effort.

Resolving this contradiction will require two things. First, additional support should be targeted towards the pupils for whom it is intended. The pupil premium, for example, could be an entitlement for every child on free school meals to activities such as extra catch-up tuition, small-group tuition or one-to-one teaching to stretch the most able. This would prevent the funds being diverted towards pupils more likely to boost a school's position in the league tables.

Second, the accountability system should be adjusted to give weight to a wider range of measures than attainment in "hard" GCSEs. New York City, for example, has introduced a school report card that includes measures such as the progress of children from low-income households. This would be a better way to hold schools to account for their performance against a wide range of criteria.

Jonathan Clifton is a research fellow at IPPR.