Headteachers say that children are going to school malnourished. How can we reconcile this with the fact that official figures show that inequality has been, in Chris Giles' words, "remarkably stable" since the early 90s?

Very simply. The evidence that inequality has been stable comes from the fact that the Gini coefficient hasn't changed much. However, the Gini is a measure of the average of income inequalities. And of course, an average can be unchanged if some components rise whilst others fall. Which is just what's happened.

My chart gives a gist of what I mean. It shows some income ratios, based on real incomes after housing costs taken from the latest HBAI data.

What this shows is that we've seen a big increase in one inequality since the late 90s - that between the third decile and the bottom decile.

In fact, the bottom decile has done really badly. Its real incomes after housing costs haven't grown at all since 2000 whilst median incomes have risen over 27%. And since 2011-12, the bottom decile's real incomes have fallen 6.3%. This is consistent with reports of an increased intensity in poverty such as increased use of foodbanks and child malnourishment.

Those around the third decile, however, have seen their incomes rise - in part because they have benefited from a rising minimum wage. Their incomes have risen relative to those just above median incomes - those in the 6^th to 8^th deciles, thus reducing some inequalities. This is in part because of job polarization - the loss of job opportunities for those in once-decent to middling jobs such as secretaries and tradesmen.

And since 2010, we've also seen a narrowing of inequality between the top decile and middling incomes. (These data, however, tell us nothing about really top incomes which are hard to measure but which probably also fell since the crisis).

What we've seen, then, is an increase in one form of inequality: the very poorest have fallen behind everybody else. But this has been offset by falls in other inequalities, such as that between high incomes and those on slightly-below median incomes. A stable Gini coefficient is therefore consistent with an increased intensity of poverty and for that matter with very different types of society.

In this context, I disagree with those centrists and rightists who point to a flat Gini coefficient as if this sufficed to show that the left is wrong to worry about inequality. For one thing, flat overall inequality is consistent with increased poverty. For another, even if the Gini coefficient has levelled off (possibly (pdf) only temporarily) it might have done so at too high a level; we must avoid a form of anchoring effect whereby our perceptions of what's fair are overly coloured by actual inequality. And thirdly, just because inequality has stopped rising does not mean that the damage done by its earlier rise will go away. If a man has been hit by a bus you not restore him to health merely by stopping the bus. And of course, income inequality is only one of many inequalities: inequalities of power are also important (maybe more so).

We should ask: what sort of economy and society we want? A single statistic such as the Gini coefficient is not much help in answering this.