The GPG is a statistic which is hazardous to interpret and variations in it need to be analysed with extreme caution. Different versions of the pay gap (all employees, FT, PT; hourly, weekly, annual earnings; mean, median) can give different answers and trends. It can change for reasons which are obscure: for instance one year the pay gap rose in both the public and private sectors, but the aggregate pay gap fell as the numbers employed in the private sector dropped while those in the public sector rose (the GPG generally being higher in the private than in the public sector).
Comparisons are very misleading. For example, those between countries: according to Eurostat, the unadjusted GPG for the UK in 2013 was 19.7%, in Italy only 7.3%. Nobody who knows anything about the Italian labour market would suggest that this is the result of more gender-blind behaviour by Italian employers. Only about 46% of Italian women aged 15-64 are employed, as against 67% in the UK; because of the high cost of employing low-skilled workers in Italy only relatively highly-skilled, and better-paid, women are in work. Thus Italy has a lower GPG. An extreme version of this effect is the case of Bahrein, which a decade ago had a negative pay gap – women earned more than men – of about 40%; only a handful of women worked in the formal sector, and these were largely highly-educated women.
The same issue arises within the UK, where Northern Ireland has a negative pay gap, probably the consequence of lower female participation and a concentration of employment in the public sector in the province.
Despite the gender pay gap being smaller on average in the public sector, there are in fact huge variations between government departments reflecting factors such as the type of employee and management structure rather than any difference in recruitment or HR practices.
Serious econometric work, rather than anecdotes, suggests that only a small element of the GPG can plausibly be attributed to discrimination after controlling for a range of observable factors which are known to influence pay, for example differences in educational and other qualifications, average age, experience, and hours worked.
Gender pay differentials between young people are now close to zero or negative – reflecting in part the better educational qualifications of girls and young women. It is only over the age of 40 that a gap really emerges. The obvious inference is that having children, caring responsibilities and other ‘lifestyle’ factors make for the difference in older age groups. Support for this comes from evidence that older single women seem to do as well as single men, while sexual orientation also has an impact (several studies suggest that lesbian women are paid more than straight women).
Furthermore, concentration on the over-40 pay gap in any case ignores ‘compensating differentials’ which offset lower pay to a considerable extent. There is evidence that other non-pay aspects of work done by women are more attractive than those for men. For instance women spend less time commuting, are less likely to work unsocial hours, less likely to face physical danger at work (the industrial accident rate for men is far higher than for women) and less likely to work outside or in isolated conditions. Wellbeing surveys suggest that they are on average happier at work.
There are many other ‘pay gaps’, some of which may arguably raise more important issues for public policy. There are known to be large pay gaps associated with (for example) ethnicity, religion and disability. In some of these examples discrimination may be a more obvious issue than is the case with the gender pay gap. For instance the Prime Minister has recently drawn attention to evidence that people with Asian/Muslim-sounding names are significantly less likely to be interviewed for jobs, evidence which is not found to the same extent for women.
One pay gap for which substantial evidence is beginning to emerge is that associated with looks: There is a ‘beauty premium’. After controlling for economically relevant factors, women rated attractive earn more than their plainer sisters. But interestingly the same effect also applies to men: handsome men do better in the labour market – and indeed the effect is stronger amongst men than amongst women.
All of this suggests that trying to impose policies to reduce the gender pay gap is difficult, as there are lots of other dimensions to pay inequality and the way in which pay develops over working lives is still incompletely understood. The danger in requiring organisations to publish GPG figures is that it will lead to some quite blameless organisations being castigated as discriminatory and sexist, while other organisations (which may actually treat women employees quite badly) will escape scrutiny as characteristics of their employment structure and recruits may produce a low or negative GPG.
This danger may in turn lead to employers trying to ‘game’ the situation by, for example, outsourcing unskilled work mainly done by females, or by turning contracted workers into freelancers, and thus reducing the measure by which they are judged. Such developments may not be in the interests of those whom policymakers wish to help.
Prof Len Shackleton is a Visiting Fellow at the IEA, and professor of economics at the University of Buckingham.