The scandal of the gender pay gap

The gender pay gap is one of the most misunderstood areas of British public policy statistics. The only question is the extent to which this is accidental or deliberate obfuscation by pressure groups. The UK Statistics Authority needs to step in to do its part in getting better data, better explaining the existing data it publishes and correcting those who misuse it. It is a shame that the respected IFS has added to the deluge of confusion with its latest report published today.

So far as the statistics are concerned, the pay gap is the average amount of money paid to men in work versus the average paid to women. So far as legislation is concerned, the pay gap is the difference between the pay of an equally qualified and experienced man and woman doing exactly the same job. Sadly the rhetoric swings happily between the two helping no one. Every time this blows up I simply wish for better data so that we can really understand the issue and put ourselves in a position where we can develop policies that will put an end to discrimination. Instead we get (mostly) ill-thought out hot air. 

The latest research (“The Gender Wage Gap”, BN186, from the IFS) had wall to wall coverage in the media on a quiet news day in August. The skilfully served up headlines from the IFS are too easy for thinly staffed news desks to uncritically accept on a hot day.

Sky included: “Labour’s shadow women and equalities minister Angela Rayner called on Prime Minister Theresa May ….. to stamp out wage discrimination“. The Mirror went with the incredulous and militant merged, and probably incorrect: “Somehow the gender pay gap has got worse – this is what you can do about it”. Glamour Magazine had (note the capitalisation and punctuation) “Woman are paid HOW MUCH less than men?!

BBC could perhaps have done better too. The generally wonderful Nick Robinson fell for the story in presenting the Today programme on BBC Radio Four this morning, as did (unusually) Martha Kearney on World at One, despite the best efforts of Maria Miller to lead her to a more sensible place by talking about getting rid of the “jobs for girls and jobs for boys” culture. Perhaps this shows how deep the misunderstanding of the figures has gone.

Most of this is off mark judging by the comments of the UK Statistics Authority which pronounced on the gender pay gap in 2009. It said:

“Limitations of measures of the Gender Pay Gap

  1. The 2008 ASHE release states the following: Although median and mean hourly pay excluding overtime provide useful comparisons of men’s and women’s earnings, they do not reveal differences in rates of pay for comparable jobs. This is because such measures do not allow for the different employment characteristics of men and women, such as the proportion in different occupations and their length of time in jobs.”
  1. We agree with these important qualifying remarks which highlight some of the limitations of summary measures of the gender pay gap and the difficulties in providing a complete like-for-like comparison between men’s and women’s earnings. Although the estimates of the gender pay gap presented in this paper have value, they do not control for certain factors such as occupation and length of time in jobs which are likely to have a differential impact on the earnings of women and of men. These limitations need to be borne in mind when considering the extent to which measures of the gender pay gap provide evidence of inequality and discrimination in the labour market.”

The comments could be stronger and it needs to make sure that the ONS statistics are really given their true value (which is not much).

This ONS piece does at least include the comment: “These figures do not show differences in rates of pay for comparable jobs, and are affected by, for example, the proportion of men and women in different occupations.” But even the ONS cannot undermine its own data too much. As is shown in the annual ONS publication of gender pay gap (“Annual Survey of Hours and Earnings: 2015 Provisional Results”), despite the caveat, there are many charts and paragraphs of text giving the data more credibility than it deserves.

I will be forwarding this blog to UKSA and asking that they take action.

Back to today’s IFS report. There’s nothing wrong with what the IFS does, it’s what they don’t do that is the crime. You have to look hard for caveats about the data. The complete focus on aggregate data while ignoring composition changes in the labour market as people age is a major weakness in this work’s conclusions. The focus on average earnings not median is a risk – the average is distorted by a few very high earners, who tend to be men. The latest data used (from the BHPS) is for 2008 so there is every chance that the situation might have changed since then (as it has been doing steadily over time). There’s no recognition of the importance of job choice. No control, if you like, for different kinds of jobs and “on average” women making different choices from men – choosing to be nursery teachers not brick layers. The data is not fit to make the claims made and it’s a shame that IFS does not fully acknowledge that.

Consider what the report says:

  • “The figure shows that the gender wage gap widens dramatically in the years after the arrival of the first child.” What about the composition of those returning to work, or the jobs the returners do? Nothing is said about how that might be different from the pre-child profile. And why lower wage options might be preferred.
  • “It is worth noting that this story cannot explain the (much smaller) gender wage gap that already exists before the arrival of the first child. The reasons for that gap are certainly worthy of more investigation.” This seems a bit disingenuous. The pre-child gap is edging close to averaging 15% (reading off Figure 10) in the years prior to child birth. To minimise that by calling it “much smaller” is odd given the overall gap is only a bit higher at around 20%.
  • “It is possible that other factors could drive the divergence of wages as men and women age or after they have children, such as employers exercising market power or gender differences in choice of occupations.” They say “possible” and seem to limit it to when people age or have children. It is surely “certain” (given all we know) and it affects all ages. There’s no data to quantify it but also no data to refute it. But IFS should not downplay these factors so much.
  • The data samples are sometimes quite small. In Table 1 conclusions are drawn from just 700 cases.

The data weaknesses are known unknowns and IFS treats them like unknown unknowns. There are far too few caveats in the report. Like all data users, however, they get no prizes or re-commissions (and no headlines) for reports littered with caveats.

Large figures scare people and breed a sense of injustice. But is that the IFS aim? There is a pay gap and clearly something in the prescription IFS promotes but to ignore the choice of job is odd. It seems hard to imagine the gap closing until, to paraphrase, more women take on the jobs that pay more. Job choice is not covered by the equal pay legislation.

There are a few lone voices out there telling the story as it is, for example the Adam Smith Institute and the IEA’s “Employer discrimination not to blame for the gender pay gap”. Other groups like the Fawcett Society take a different line. The opening line of its briefing is “Despite the Equal Pay Act 45 years ago, women still earn less than men in Britain today.” This clearly conflates the discrimination and the labour market choices elements in an unhelpful way. The lack of data – and weak ground rules from UKSA – allows this confusion to reign.

The shame is that the much heralded measures proposed by government and lauded by those campaigning for a narrower gender pay gap will do nothing to help. A CIPD blog sets it out thus:

Public bodies subject to specific public sector equality duty requirements may already publish gender pay information, but the new regulations will oblige them to publish the same data as large private employers. They will have to report on the:

  • mean and median pay gap between men and women
  • mean and median bonus gap
  • proportion of men and women who received a bonus in the 12-month period to 30 April in the previous year
  • proportion of men and women by salary quartiles.

The timescales will also reflect the proposed private sector rules, with the first report required before April 2018 to cover the pay period up to April 2017, and annual reporting from April 2018 onwards.

This data will have just the same shortcomings as the ONS data that underlies the IFS study. A swift resolution to this problem seems far off unless UKSA decides to act.

Note added 24 August: Thanks to Ian Bright, I have been pointed to the ASA Sage publication containing “Overwork and the Slow Convergence in the Gender Gap in Wages“. It shows the same set of issues plaguing the US gender pay figures.

Note added 24 August: Delighted to see Prof Alison Wolf writing in today’s Times: “Theresa may is wrong about the gender pay gap“. (£) She says that “Women are not paid less than men because of discrimination; it’s because the broader economy is weighted against them”. So much in Alison Wolf’s 2013 book “XX factor” is worth reading. Here’s a Guardian write-up.


It is a shame that the IFS does not have the same standards relating to research release as is in place for the government data (on which this study is based). The press release was published on 23 August, hours after the press had, it seems, published the results. People have to consume this story via the media before the actual research is published.


One thought on “The scandal of the gender pay gap

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s