It’s April 2029 and the government is trumpeting the fact that no employer in the UK has a gender pay gap or ethnicity pay gap. All employers say that their median woman’s hourly pay is the same as the median man’s hourly pay and their median white employee’s hourly pay is the same as their median non-white employee’s hourly pay. Would you be joining in the celebrations?
If you are not a statistician, I suspect you would be. The party goes on for 3 months and then the Office of National Statistics (ONS) announces that there is a 10% gender pay gap i.e. the median woman earns 90p for every £1 earned by the median man. They also announce that the median non-white employee earns £1.10 for every £1 earned by the median white employee. Confusion abounds and before long statisticians are in the firing line for being party poopers.
This isn’t a fanciful scenario, it is a fact today that the ONS’ measures of the gender & ethnicity pay gap are in no way comparable with the pay gaps reported by employers. It is also fact today that the ONS gender pay gap is more negative for women and the ONS ethnicity pay gap is more favourable for ethnic minorities compared to data reported by employers. The best way to demonstrate these two facts is to imagine that every single employer has no gender pay gap and no ethnicity pay gap and ask what the national pay gaps as measured by the ONS would be in this scenario.
Why ONS Gender Pay Gap is more negative for women
Take a look at this graphic with 6 fictional employers, each with 12 employees. In each employer, the median woman earns the same as the median man and the average woman earns the same as the average man (slight difference for Office Cleaning & Co). 3 employers have male dominated workforces and 3 employers have female dominated workforces and there are 36 men & 36 women in total. Assume these 6 employers represent the entire economy that the ONS reports on, how much do the median man and the median woman earn?
The answer is given by the table on the right which lists the frequency distributions for men & women. The median man/woman are the averages of the 18th & 19th employees of those genders when they stand in a line in order of their hourly pay. The median man earns £30 and the median woman earns £25 per hour and thus for every £1 earned by the median man, the median woman earns 83p, a pay gap of 17p in the pound. A similar pay gap exists when the average man & woman are compared.
Why does this happen when none of the 6 employers had a pay gap? The answer should be obvious from the graphic since the 6 employers were displayed in order of average pay. Men are more likely to be found in high paying employers and women are more likely to be found in low paying employers. So it is no surprise that the national median woman will be earning less than the national median man.
The same effect is already apparent in the ONS pay gap figures. If we look at the 2020 gender pay gap, ONS estimate that the median women earns 84.5p for every £1 earned by the median man. Among full time employees, the median woman earns 92.6p for every £1 earned by the median man and among part time employees, the median woman earns £1.03 for every £1 earned by the median man. Note how the pay gaps for full time and part time employees are much smaller than the pay gap for all employees. Again the reason is obvious and follows the same effect as above. Women are more likely to work part time and part time roles earn less per hour on average than full time roles. Consequently, the median women’s pay is dragged down and the median man’s pay is dragged up when full & part time roles are combined.
The bottom line is that any employer that seeks to compare their pay gap with the ONS pay gap is making a false comparison. ONS pay gaps can never be comparable with pay gaps from GPG reporting and such efforts should be discontinued. Should we ever get to a situation where no employer has a gender pay gap, I would still be expecting the ONS to be reporting a pay gap against women of around 5-10p in the pound.
Why ONS Ethnicity Pay Gap is more favourable for ethnic minorities
In 2019, ONS reported the median ethnic minority employee earns 97.7p for every £1 earned by the median white British employee. I have seen some scepticism of this figure which appears to be a narrower pay gap than those reported by a number of employers. Again it turns out there an obvious reason why the ONS will always report a more favourable pay gap for ethnic minorities and once again, no employer should ever be comparing their ethnicity pay gap with the ONS figure.
Unlike gender, ethnicity is not uniform across the UK. Some areas are very white, other areas are ethnically very diverse. However, none of this would matter if the average pay was the same across the UK. As we all know though, pay in London is much higher with the ONS currently estimating the London premium to be at least 30%. At the same time, London is much more diverse than the rest of the UK. Consequently it is more likely that a non-white employee will be living in London and earning a premium compared to a white employee who is more likely to be living outside of London and not earning a premium. Thus it is not surprising that the national ethnicity pay gap should be so small.
To demonstrate this, I will again assume that no employer in the UK has an ethnicity pay gap. Rather than creating fictional employers, this time I have imagined each of the 632 parliamentary constituencies in Great Britain are employers (with approximately 100k employees including children!) and within each constituency, the median white employee earns the same as the median non-white employee. From the 2011 census, we know what % of the population in each constituency is white British and the ONS also report the median full time weekly wage by constituency. Whilst later data is available, my point still stands using the latest data I have in my possession which is for 2017.
The chart below plots weekly full time wage by % white British for all 632 seats. I’ve split England into 3 parts and plotted Scotland & Wales separately.
The chart may look messy but it is the open blue circles representing London that you need to concentrate on. On average, the 78 seats in London are 49:51 white:non-white whereas the rest of England is 87:13 white:non-white and Scotland/Wales are 93:7 white:non-white. At the same time, the average weekly wage in London £662 is 32% higher than that seen in Wales. The majority of seats in London have average weekly wages above £600 whereas the majority of seats outside of London have averages less than £600.
If we make the assumption that within each seat, there is no ethnicity pay gap, then one can take a weighted average of the weekly wages using the %white british as the weight to estimate that the average weekly wage for White British would be £545 and for non-white British, it would be £585. Thus for every £1 earned by a white British employee, a non-white British employee would be earning £1.073. Given that the ONS currently estimate this to be 97.7p instead, that implies that within each seat on average, the median non-white person is earning 90.4p for every £1 earned by a white person.
Indeed the ONS data shows that by region (see figure 8 of this link), non-white employees earn on average 7.5% less than white employees once the region is taken into account but this varies considerably between regions. In London, the median non-white person earns 73p for every £1 earned by the median white employee but note that non-white employees in London still earn more than non-white employees in any other region (bar the East). It just goes to show why an employer cannot compare their ethnicity pay gap with the ONS which is measuring something different.
Which pay gaps are more meaningful – ONS or Employer’s?
Over my 30 years as a statistician I have lost count of how many times I have said “Objectives! Objectives! Objectives!“. The question posed here is incomplete, what you need to ask instead is “what am I trying to do?” and “which measure is more meaningful given my objective”.
For an employer, the ONS figure is irrelevant. Indeed I argue far too much time is spent comparing one employer with another when employers should be comparing themselves over time to see if they are improving. I’ve said before that pay gap analytics is all about Continuous Improvement and is not about your place in a league table. Employer reported data is the best data to use if one is interested in trends over time as it allows a lot of true like for like comparisons as I did recently when estimating the trend for 2020.
On the other hand, if I want to see if the earning power of women or a specific ethnicity is improving, then the ONS figure is more meaningful. The fact that ethnic minorities are more likely to be in London does indeed boost their earning power and the ONS figure reflects that. Of course, it should be noted that ONS pay calculations exclude overtime, most bonuses, pensions and non-monetary rewards so they should be included if we truly want to know what the earning power of a group is. Indeed I would also argue that hourly pay which puts full time & part time employees on the same basis is misleading and the fact that women are more likely to work part time reduces their total earnings. So for me, earning power should in fact be total annual remuneration regardless of hours.
I am sure there are other objectives that would lead to different decisions but I hope I’ve made my point. What is reported by the ONS is sound and what is reported by employers is also sound. They are just not measuring the same thing and nor should they ever be expected to measure the same thing.
— Would you like to comment on this article? —-
— Subscribe to my newsletter to receive more articles like this one! —-
If you would like to receive notifications from me of news, articles and offers relating to diversity & pay gaps, please click here to go to my Newsletter Subscription page and tick the Diversity category and other categories that may be of interest to you. You will be able to unsubscribe at anytime.
— Want to know more about pay gaps? —
You will find a full list of my pay gap & diversity related articles here which are grouped by theme.
— Need help with understanding your pay gap? —
I offer the following services. Please click on the headings for more details.
- Analysis – I can dig deep into your data to identify the key drivers of your pay gaps. I can build a model using a large number of variables such as pay band, seniority, job function, location, etc and use this to identify the priority areas for closing your gaps.
- Training – I run training courses in basic statistics which are designed for non-statisticians such as people working in HR. The courses will show you how to perform the relevant calculations in Microsoft Excel, how to interpret what they mean for you and how to incorporate these in an action plan to close your gaps.
- Expert Witness – Has your gender pay gap data uncovered an issue resulting in legal action? Need an expert independent statistician who can testify whether the data supports or contradicts a claim of discrimination? I have experience of acting as an expert witness for either plaintiff or defendant and I know how to testify and explain complex data in simple language that can be easily understood by non-statisticians.
If you would like to have a no-obligation discussion about how I can help you, please do contact me.