After two years of mandatory gender pay gap reporting, there is increasing pressure to bring in pay gap reporting for other protected characteristics. At the moment, ethnicity is receiving the greatest attention and a number of politicians are calling for the introduction of mandatory ethnicity pay gap reporting.
In this post, I will explain why I am opposed to an ethnicity pay gap reporting process which simply replicates the gender pay gap reporting process. In a future post, I will explore what an ethnicity pay gap reporting process should look like if parliament decides it wants to make this law.
In October 2018, the UK government launched a public consultation on whether employers should be required to publish ethnicity pay gaps. I was keen to submit a response to this as I had already flagged my concerns about this last year (see point #12 in the link). Unfortunately, I got the deadline wrong but also, I wanted to be constructive but struggled to come up with a suitable process at the time. In a future post, I will explain how I think ethnicity pay gap reporting should be done but I will say now that it will probably look like this ONS study of ethnicity pay gaps.
Last month, the Treasury Select Committee asked for my views on this and you can see my response here. This post expands on the points I made then but let me repeat something first of all that is very important to forestall potential objections.
“I completely support the idea that pay gaps for ethnicity and other protected characteristics should be measured. My concern is about how this is done since the gender pay gap reporting process is not fit for purpose when measuring minorities and an alternative process is needed.”
This article builds upon the points I made in my postand how measurements of the median gender pay gap become statistically unreliable when one gender is dominant in small employers. I produced this graphic which shows how the margin of error widens considerably in this instance and this statistical issue is the main reason I say the gender pay gap reporting process is not fit for the purpose of ethnicity pay gap reporting. In addition to Statistical concerns, I also have concerns on Data grounds and Ethical grounds.
Statistical Issues – the ethnicity profile of Britain is extremely variable
When it comes to ethnicity in Britain, what is not being appreciated is that the vast majority of employers will be dominated by white employees. To illustrate this point, let’s look at parliamentary constituencies since these are geographical areas of roughly equal size and therefore can be averaged across a number of constituencies to produce an estimate for a particular area. The 2011 census shows that of the 632 seats in Britain, 353 are over 90% white UK whilst only 53 are minority white UK of which 42 are in London as can be seen from the chart to the left. From the table below, it is clear that outside of London, minority white areas are mostly found in city centres.
So if you are an employer in the Lake District area say, you can expect 98% of your candidate pool to be white UK. An employer with 300 staff could then expect to have on average just 6 non-white employees. You don’t need to be a statistician to know that any comparison between the median of 294 employees with the median of 6 employees will be meaningless and is likely to change considerably just from adding or subtracting one non-white employee. If we are now talking about an employer of 2700 staff, the expected comparison would be between 54 non-white and 2646 white staff. If you are a professional statistician like me, then our training allows us to draw conclusions from such a small sample size for non-white staff but the skills required are way beyond what could be reasonably expected of an HR professional.
For an HR person to be able to measure the pay gap between any two groups with any degree of confidence, my professional opinion is that each group must have at least 100 employees. My opinion is backed by the Royal Statistical Society who recently published 10 recommendations for improving Gender Pay Gap Reporting. Recommendation number 8 calls for any employer with less than 100 men or women to be flagged as having unreliable data. Replace gender with an ethnicity category and the same recommendation applies.
Compared to gender though, the statistical constraints for ethnicity are bigger. That is because gender is binary whereas ethnicity is not. I doubt that employers and the public will find a White vs non-White comparison particularly informative which means that comparisons between multiple ethnicities will be called for. At a minimum, I would expect at least 5 categories (White, Black, Asian, Mixed, Other) which means that for an HR professional to have confidence in their figures, they are going to need at least 100 employees in each of these 5 categories. I repeat that a statistician is capable of drawing of conclusions with smaller sample sizes due to our training but I think you can see that realistically, only employers with 5,000 or more employees are likely to be in a position where an HR professional can carry out a meaningful ethnicity pay gap analysis.
Out of 10,500 employers who submitted gender pay gap data in 2018, there are only 500 (5%) employers with 5,000 or more employees. Half (5,163) employ less than 500 employees and will almost certainly end up with meaningless ethnicity pay gap figures. This means they end up having to perform a statutory administrative burden for no benefit.
One option I will explore in my next post on ways to measure ethnicity pay gaps is devolving the competence to do this to the London Assembly. As the chart above shows, London is extremely diverse ethnically and even employers with say 600 employees might have sufficient diversity to be able to calculate meaningful figures. The only thing to bear in mind is that the workforce of London relies heavily on commuters from outside of London and the commuter is much whiter than London itself.
Data Issues – will employees state their ethnicity?
Nearly every employer will have an ongoing record of their employee’s gender. The reasons for doing so are largely historical to do with tax, pensions and maternity leave. To calculate an ethnicity pay gap, the employer would have to get each employee to state their ethnicity. Unless parliament makes this mandatory, which I think is unlikely, declaring your ethnicity will be voluntary. My experience is that at least 20% of employees will not give their ethnicity. I list below 3 actual examples:
- A gender pay gap analysis for a financial sector firm where 40% did not declare their ethnicity.
- A staff satisfaction survey where 20% did not declare their ethnicity.
- Various market surveys where 5% do not declare their ethnicity.
This has the effect of creating an additional category of Unknown to whatever ethnicity categories are used and of course reduces the sample sizes of what remains. There is also the risk that the employees within the Unknown category are biased in some way. For example, if low paid non-white employees are suspicious of how their ethnicity data is going to be used, they may choose not to declare their ethnicity. That would have the effect of raising the median earnings for non-white employees elsewhere and could result in a smaller pay gap to what would have been calculated had all employees stated their ethnicity.
Another data issue is the categorisation used. It is very likely that the official census ethnicity categories will be mandated but this has the effect of shoehorning employees into a bureaucratically convenient identity rather than the identity that the employee sees his or her self as. I must say that in this day and age of personal identity, this does not sit well with me and as someone who is disabled, I have experience of this. Disability is a personal identity but what I often see in surveys is the definition used by the Disability Discrimination Act (DDA) and surveys often asked if you feel you come under that definition. The problem with the DDA definition is that it conflates disability with health issues. I may be disabled but I am not unhealthy and I object to being conflated with people with health issues. This conflation makes sense from an employer’s duty of care perspective as mandated by the DDA but if this gets rolled out for disability pay gap reporting, it will result in a misleading estimate of the disability pay gap. The reason why is that disabled people will often enter the labour market with their disability and have to overcome the issues it causes throughout their career. A healthy employee who then contracts cancer in their 40s will not have had these barriers to begin with in their career and therefore they will face different issues in overcoming their health issues. From a pay gap perspective, I see these as two very different categories.
Ethical Issues – do we want our personal characteristics permanently recorded?
In many ways, the ethical issues are my biggest concern. In order to carry out ethnicity pay gap reporting on an ongoing basis, it will be necessary for an employer to keep a permanent record of their employees’ ethnicities. I do not know of any society today or in the past where doing this has led to positive outcomes, rather the reverse tends to happen. You might counter that HR departments already record gender, marital status and disability and some times nationality, so what’s the big deal with recording race as well?
Let’s look at the historical reasons for recording gender. In the days of clear gender roles and expectations, it was obvious that a record of gender would be needed. This was especially the case with payroll, taxes and pensions that treated men and women differently. But that has gone now so is there any reason for HR departments to record gender today?
Obviously maternity leave is one reason but even then, the law now is that parental leave can be shared among parents rather than accruing solely to mothers. Finally with increasing debate about whether gender is binary or not, are we not now in a situation where there is little point in HR departments recording gender? We can’t be far off and yet on race, it is being proposed we go in the reverse direction?
Marital status is closely tied with gender as so much of what I say there also applies to marital status. This leaves disability. The requirement to record is mostly around the employer being aware of what specific needs the employee may have. Even then, this is more focused on the individual rather than the category. As someone who is deafblind, I will have needs but they will differ from another deafblind individual. In that sense, the recording of personal characteristics is more about making it easier for employers to fulfill their duty of care to their employees rather than it being part of a statistical exercise.
Finally, consider this. Once your characteristics have been recorded, you are then placing your trust in your employer’s IT security. We know that records can be hacked and I fear that it only needs two or three high profile hacks and employees will become more reluctant to have that data recorded.
So I hope you can see why I am uncomfortable with requiring employers to record their employees personal characteristics permanently. I finish by reminding you that I am in favour of measuring ethnicity pay gaps. What I can’t recommend at all is using the gender pay gap reporting process as a template for ethnicity pay gap reporting due to the statistical, data and ethical issues I have highlighted. In my next article, I will describe some alternatives that could be used instead.
– Need help with interpreting your pay gaps? –
I offer the following services for all kings of pay gaps.
- 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.
– Want to know more about the gender pay gap? –
I have written a number of articles about the gender pay gap covering these topics:-
- What gender pay gap data tells us, what it doesn’t tell us and how it can be misused
- Three distinct errors that have been made by at least 10% of all organisations when submitting their gender pay gap data
- How to distinguish between a true pay gap and a pay gap that arises naturally due to the laws of chance
- My 12 steps to improve public confidence in gender pay gap data
- My evidence to the Treasury Select Committee on how gender pay gap reporting could be improved
- Calculate your gender pay gap by downloading my free spreadsheet calculator!
- Did the gender pay gap narrow in 2018?
- How to identify unusual year on year changes in gender pay gaps
- Frequently Asked Questions about gender pay gaps.
Finally visit my Twitter thread to see my comments on gender pay gaps in the media. Some notable ones are here.