A row on Friday night between Anna Chen and Laurie Penny has reminded me to get round to writing a white male economists' perspective on the hot topic of intersectionality.

This, it seems to me, is a fancy word for an obvious idea - that people experience inequalities differently. The experience of black women - to take Kimberle Crenshaw's original example (pdf) - is different from that of white women; working class women suffer a different form of inequality than richer women; and so on.

This matters, because there can be trade-offs when we try to promote localized forms of equality. White feminists are sometimes accused of marginalizing black or trans women, for example, and demands for more women in boardrooms or the media leave class inequalities unchallenged. And so on.

I have four observations here:

1. Inequalities are not always additive. Some evidence on this comes from a paper (pdf) by Hilary Metcalf on wage inequalities*. She shows that, controlling for qualifications, black Caribbean and Indian women earn more that white ones. Yes, there's a gender pay gap and an ethnic pay gap, but they cut across each other. And there's also evidence that lesbians actually have a wage premium - although, paradoxically, they tend to have live in households (pdf) with lower earnings.

2. There's a curious omission in the inequalities which identity politics worry about - that between good-looking at ugly people. One UK study (pdf) - which is consistent with international evidence - found that boys who were considered unattractive by their teachers at the ages of 7 and 11 earned 14.9 per cent less than averagely attractive boys at the age of 33, even controlling for qualifications; ugly women suffered a 10.9% penalty. This pay gap is bigger than Metcalf's estimate of the adjusted pay gap between men and women or whites and blacks, and I suspect it's not offset by uglies getting a good deal in other spheres of life. The inequalities we hear about, then, are only a subset of those that actually exist.

3. The danger with identity politics is that it can degenerate into narcissism, or what Phil calls a "project of individual self-presentation." As Dr Crenshaw said (pdf): "the moment where a different barrier affects a subset of us, our solidarity often falls apart." I suspect this isn't wholly the fault of feminists, but is instead exacerbated by a form of projection: there's a tendency to presume that Laurie Penny or Caitlin Moran are "speaking for women" whereas nobody assumes that, say, Nick Cohen or James Delingpole are speaking for white men.

4. Although there's a tendency to regard the many inequalities as different, they have a common link - power. The powerful use their power to favour their group and disfavour out-groups. The fact that there's more ethnic diversity in the City than in the arts tells us that we'll not achieve equality merely by ensuring that the powerful have "liberal attitudes". This, though, poses the question: what would a society without power inequalities look like?

And herein lies a paradox. Perhaps the most widely studied blueprint we have for such a society is to be found not in leftist literature but in the economic textbooks. Under perfect competition, employers wouldn't have the power to discriminate. Of course, this alone wouldn't eliminate all inequalities, but it'd be a start.

* Of course, these are only a subset of inequality, but they give us a measure of how capitalism produces some systematic inequalities.