12 April 2018

Pay Gap Flawed

Figures a dreadful mess.

By John Watson

The pay gap figures are now out and a dreadful muddle they are too.  That isn’t because of any failure by those reporting but rather because the formulae which generate the figures are so deeply flawed that they make the numbers they generate virtually meaningless.

The headline information in relation to each employer is the difference between the average rates of pay of male and female employees.  This is the figures which the government will focus on reducing.  It is also the figures which everyone in politics will cite when they wish to bash a public or private employer.  What a pity, then, that the regime is so badly designed that hardly any of those citing the figures will have a clue what they mean.

To understand why this is the case you need to descend into the topsy-turvy world of the regulations.  They are not a great read so let’s go direct to the formula which sets out how the main pay gap figure (the difference in average rates of pay as a percentage of average male pay) is actually calculated.  Subject to minor adjustments for special cases, it is simply:

[ (A-B) ÷ A ] x 100

where,

• A is the mean rate of remuneration for men and
• B is the mean rate of remuneration for women.

Short and sweet, eh?  Well, most sensible people distrust formulae so we’ll road-test it a bit.

Keeping in mind that “mean” is just mathematical slang for average, let’s take an employer who has the same number of men and women at every level of seniority.  As the men and women at each level are paid the same (these regulations are not about equal pay for equal work, which has long been protected by other laws), then A and B will be identical, so the pay gap is nil.  That is just as it should be.  Round one to the authorities.

Now let us suppose that this employer hires a number of secretaries.  They are predominantly young women and they are paid less than the highly specialist executives who make up most of the workforce.  That reduces the average salary paid to women, so B drops and the gender gap increases.  Is that because the status of women in the organisation is reduced?  No, it is because they decided that instead of outsourcing their secretarial side to an agency in the third world they should give local women the opportunity to work for them.  And yet they will be pilloried by the ignorant for increasing their gender gap.

Of course the gender gap would also increase if they hired more top level male executives because that would push up A, the mean salary for male employees.  So you get the same increase in the gender gap by hiring expensive males or hiring junior females.  Weird!

Actually, it is worse than weird.  It could lead to employers who are put under pressure about their pay gap (and what is the point of these figures but to put pressure on employers?) reacting by sacking low paid women and either contracting out their work or replacing them with low paid males.  Talk about the law of unintended consequences.  It may be that those who drafted the regulations secretly hoped to tilt the secretarial market in favour of men, but I rather doubt it.

Unfortunately it gets worse because employers are also required to produce figures by reference to median rather than mean remuneration.  That means that rather than a normal average they compare the middle woman with the middle man.  The results here are even more bizarre.  Try this:

Suppose there are 1002 employees in an organisation split equally between men and women at every level.  250 of each sex receives £100,000 a year and 251 of each sex are paid £20,000.  Both the median man and the median woman receive £20,000, so the difference is nil.  Then suppose that one man is promoted from the £20,000 level to the £100,000 level.  That will boost the median level for men to £100,000, five times that for women.  A pay gap of 80%.  If a woman had been promoted instead the gap would have moved to minus 80%.

Of course there is nothing wrong with using formulae to produce published information but it only works if the reader understands what the figure generated actually means.  The pay gap figures are already being debated in the media but without anyone commenting on their shortcomings as a form of information.  Could it be that none of the commentators have grasped how meaningless they actually are?

The annoying thing is that it is entirely unnecessary.  Company accounts have long shown directors’ remuneration etc by disclosing how many individuals fall into a particular salary band.  Why didn’t they do something similar here?  In fact the regulations go some distance in this direction by requiring employers to publish how many men and women fall into the highest paid quartile of employees and the three quartiles below that.  How easy it would have been to have done the job properly and to have used a band system.  But that would not have done, would it?  Those who designed this system wanted figures that could be used to exert pressure, a simple metric by which comparisons can be made.  Well, they have got them now and the figure are in their own image – entirely useless.

The regulations are signed by Justine Greening and require that she, as Secretary of State, review them from time to time.  She should start reviewing them now.