Wednesday, 11 November 2009

The variability of the gender pay gap, and intersections

The Fawcett Society provides a breakdown by local government area of the gender pay gap. (It doesn't say in that document, but I assume it measures the difference in full-time mean annual pay)

It varies more than I'd expect, from East Renfrewshire, where women are paid 32.6% more than men on average, to West Somerset, where men's mean wage is 52.7% more than women's. The regional pay gap varies between 9.5% in Northern Ireland, to 25.4% in London (both in favour of men, of course). There are very few local government areas where women earn more, and it's usually by a small single-figure or even sub-percentage amount. There are far more where the pay gap is in the high twenties in favour of men.

County Durham, at 15.5%, has become a unitary authority, and so the variation within it will soon be hidden - from 2.8% in favour of women in Teesdale, to 26.8% in favour of men in Chester-le-Street, the variation reflects that across the country.

It's hard to tell exactly how many of these differences from the general trend are just statistical effects - the document doesn't give the error bars, but the sampling error for a district council will be much larger than the sampling error for a region. Nevertheless, the variation between regions is still large, and probably larger than statistical error would imply.

Part of the regional difference may be that the really high-paying jobs: senior managers at big corporations, bankers, lawyers, etc. are mainly available in the South East and London (both for where the jobs are and where people doing those jobs live: I'm not sure which this document measures), and mostly held by men. There seems to be a general trend that the pay gap is higher in the richer parts of the country.

The substantial differences at least show that the pay gap is not inevitable, but it's possible that the skewing effect of the highest paid jobs means that imbalances in the lower-paid jobs could be hidden. Quixotess looks at this for a different set of equal pay data that explicitly focuses on the payment to organisation heads (in this case, of US Jewish organisations).

While the headline figure for the mean pay gap may be interesting for showing that there's a problem, the breakdowns of the data might be more useful - how many men and women are in particular pay bands. Comparing only the full-time pay gap also hides class differences.