Take two schools - let's call them Mill Lane and Brick Street. Both are large comprehensives with sixth forms; both serve urban areas with high levels of deprivation outside London.
Mill Lane and Brick Street come up on the Government's statistical database of schools as almost identical. They have the same proportion of pupils on free school meals and the same number with special needs children. So why are Mill Lane's GCSE results better than Brick Street's? Are the standards of teaching better at Mill Lane? Not at all.
There is just one key reason for the difference: race. Mill Lane has a very high proportion of pupils from ethnic minorities while Brick Street does not. At Mill Lane, more than two-thirds of the pupils do not speak English at home, while at Brick Street just one in 20 comes from a non English-speaking household. Yet Mill Lane's pupils manage an average of nine B grades; Brick Street's between seven and eight.
Shouldn't that be the other way around? Haven't we spent years debating the reasons why pupils from certain ethnic groups do less well than their white counterparts? Shouldn't the high-performing school be the one with the low proportion of pupils from ethnic minority backgrounds?
This might once have been the case, but not any more. National figures on the GCSE results of different ethnic groups still show white British pupils outperforming those from Caribbean, African and Pakistani backgrounds. But not by much - and not if you take their parents' incomes into account. The truth is that the main reason most Pakistani, African or black Caribbean pupils do worse than their white classmates is because they are poorer.
It has long been recognised that family incomes are one of the most potent forces in determining whether or not pupils do well in exams. And ethnic minority families tend to have lower-than-average incomes. But factor this out and a very different picture emerges.
If you look at the exam results of pupils from the poorest backgrounds - those on free school meals - white pupils are now at the bottom of the heap, and falling further behind every year. Last year, white British pupils from poor families were outperformed by every other ethnic group - with the sole exception of the tiny number of pupils from traveller families.
Indeed, even among pupils with better-off parents, white pupils lagged behind those whose families came from India or China at GCSE, while those from black African and Bangladeshi families only trailed by a fraction.
Even among those groups that were still behind the white British pupils - mainly black Caribbeans and Pakistanis - the rate of improvement over the past few years was much faster than the average so we can expect them, too, to overtake their classmates in the near future.
So why, then, does the Government continue to spend Pounds 200 million each year on the Ethnic Minority Achievement Grant (Emag)? The big argument for this funding, of course, is that pupils arriving in this country with little or no English need intensive support to help them catch up - and so they do. But there is other money available for language support, and in fact the vast majority of those non-native speakers seem to suffer little disadvantage - just a couple of percentage points - at GCSE.
This funding puts schools with a high proportion of ethnic minority pupils at an unfair advantage against those with a large number of pupils from poorer white families. This year, Mill Lane received an extra Pounds 50 per pupil to spend on schemes designed to narrow "achievement gaps" between different ethnic groups.
So which gaps would those be? Would it be acceptable, for example, for Mill Lane to spend the money on raising the attainment of white pupils? It would seem not. Schools must spend this grant on black and Asian pupils, even though their underperformers are more likely to be from poor white families.
The Emag funding is allocated on the basis of numbers of pupils from "underperforming" ethnic groups, which include black Africans and Bangladeshis, who barely underperform at all. It also targets areas where lots of pupils have "English as an additional language", completely ignoring the fact that many speak Urdu or Gujarati at home, even though their parents were born here.
Pressure groups still lobby the Government for continued ring-fenced funding for "bilingual pupils" even though there is no evidence that being bilingual carries any educational disadvantage.
What is this money actually spent on? Some of it goes on language support, but large sums would be much better spent on all underachievers, regardless of where their grandparents or great grandparents were born. One school in London, for example, reported recently that it spent the money on "all of those pupils who belong outside of the majority community", presumably including those who outperformed their white classmates at every stage.
These pupils were treated to extra monitoring and support, and were given opportunities to "celebrate linguistic and cultural diversity" with special displays and dedicated learning areas. One could only hope their white classmates, who had probably had fewer opportunities in their lives to celebrate "cultural diversity", were allowed into those areas too.
Spending hundreds of millions of pounds each year on blunt-edged schemes aimed at "ethnic minorities" doesn't make sense practically because it doesn't address the most pressing issues of underachievement; and it doesn't make sense politically because it fuels the ill-conceived notion that white pupils are falling behind because their teachers don't care enough.
If the Government wants to do one thing to foster better race relations in schools, it should abolish this funding.
Fran Abrams, Investigative journalist and author of 'Seven Kings'.