When gender pay gap reporting was introduced by the government for the 2017 snapshot date for all employers with a headcount of 250 or more, it was made clear they would evaluate how the legislation had worked after 5 years. We are now in the 5th year of pay gap reporting and I would like to submit to the government 7 recommendations to improve the way employers’ data is reported and 5 recommendations to improve the data used in the calculations and to reduce various distortions.
My 7 recommendations for improved gender pay gap reporting
- GEO to put Gender Breakdowns by the 4 Pay Quarters (Fingerprints) front & centre of the GPG reporting regime & abolish the other statistics.
- Employers to submit data for 4 categories Men, Women, Other & N/A
- Employers to submit actual numbers per category per Pay Quarter
- Employers to submit average pay per Pay Quarter
- GEO to calculate & publish Gender Ratios, Swap Numbers & Pay Ratios for employers
- Employers to include a 2nd breakdown by gender in their pay gap report
- Employers to perform steps 2 to 6 above twice, including & excluding Variable Pay.
This article focuses on gender but all 7 recommendations also apply to any ethnicity pay gap reporting regime. The main difference is with recommendation #2 (what categories to report) for which I will be writing a separate blog post about and will link to here when published.
You can hear me talk about how I came to these 7 recommendations in this Gender Pay Gap Webinar on 23rd June 2021 (link to be made available soon) hosted by the Economics Statistics Working Group. My talk starts 45 mins in and goes on for 25 minutes.
My 5 recommendations for improved calculations & reduced distortions
- A – All calculations based on a 12 months/52 weeks period
- B – Reconsider current employee exclusions
- C – Use Total Remuneration Package as much as possible
- D – Change Bonus Pay to Variable Pay
- E – Conglomerates to report group level data
Recommendations A to E were first flagged in my article “10 quick and easy ways to close your gender pay gap.”
Why have I made these recommendations?
Regular readers of my pay gaps & diversity blogs will not be surprised by many of these recommendations as many have been flagged for a while. This article simply pulls all of these into one place but I will now go through each recommendation in turn. As much as possible I will link to my other articles in addition to the details I’ve given below.
Recommendation 1 – Make Gender Breakdowns by the 4 Pay Quarters (Fingerprints) front & centre
This is the one feature of the current GPG reporting regime that gives genuine insight about why an employer has a pay gap. I call the 4 pay quarter breakdowns Fingerprints and I explain why I find fingerprints so useful in these two articles.
- “Why Gender Pay Fingerprints are superior to Gender Pay Gaps“
- “Employers should report their Ethnicity Pay Fingerprints“
To help make these front and centre in the public’s mind, employers should no longer be required to report their mean and median gender pay gap and bonus pay gap. Employers can continue to report them in their narratives if they wish but I consider them poor statistics in that it is hard to translate these into actions to close a pay gap. Fingerprints, on the other hand, are much easier to translate into actions hence why they need to be front and centre.
Recommendation #5 which calls for fingerprints, swap numbers and pay ratios to be published provides for more meaningful statistics than the existing pay gap statistics. The swap number is intended to be the successor to the median gender pay gap. Should it be decided that pay ratios are not feasible or desirable then it may be worthwhile retaining the mean gender pay gap. I demonstrate why in recommendation #5
Recommendation #7 which calls for two sets of numbers from employers, first excluding variable pay and second including variable pay, is intended to be the successor to the existing bonus pay statistics.
Recommendation 2 – Employers to submit data for Men, Women, Other & N/A
If an employee does not identify as either male or female or does not declare their gender, the employer is allowed to omit that employee from their calculations (see tool 3 of my article “10 quick and easy ways to close your gender pay gap!“). I find this unsatisfactory in a world where gender is not always seen as binary and I believe that employers should report such data separately as an Other category. Additionally, there could be a large discrepancy between an employer’s headcount and the actual numbers of employees that end up in a pay gap calculation which will not be apparent to the public. By publishing N/A numbers separately, this discrepancy would be captured.
Since recommendation #1 places fingerprints front and centre of an employer’s report, should say 50% of employees at an employer not declare their gender, it will be immediately obvious in the category breakdowns when plotted. An example can be seen with employer F in the chart shown in my ethnicity fingerprint article where half of employees do not declare their ethnicity. N/A is much more likely to be an issue in ethnicity pay gap reporting but it is not unknown in GPG reporting.
However, an extremely important issue must be considered before employers are required to submit data for any category for any protected characteristic and that is anonymity. It is standard practice in a lot of market and social research to never make public data for any category where the sample size is less than a minimum number. I believe this is also a practice recommended by the Information Commissioner who oversees GDPR and Data Protection enforcement. In my experience, this minimum varies between 5 & 20 people and I tend to err on the high side of 20 when I report results of surveys that I have worked on.
The reason for this is that declaring such data could identify individuals. Suppose an employer has 3 employees who identify as non-binary. Suppose the gender fingerprint is published and all 3 are in the upper pay quarter. Someone who is familiar with the employer might be able to guess who those individuals are and how much they are paid which is not the point of pay gap legislation. By requiring a minimum sample size for anonymity, this risk can be mitigated.
My personal preference is to set this minimum at 20 but that may mean that very few employers will end up submitting gender pay date for the Other category say. With that in mind, I do not rule out the current system of allowing employers to exclude Others & N/As and not submit data for these categories but I think this point should be debated.
For ethnicity though, determining how many categories should be reported and how they are defined is a complicated topic in its own right but the same issue of anonymity still applies. I will be publishing a separate article on this topic soon and will provide a link here when published.
Recommendation 3 – Employers to submit actual numbers per category per Pay Quarter
Employers submit their pay gap statistics today via a portal operated by the Government Equalities Office (GEO). This is not user friendly and it is too easy for employers to submit incorrect data.
To mitigate this risk, I am calling for the GEO portal to be changed so that an employer’s actual numbers for each pay quarter is submitted rather the % breakdown. For example, an employer with 800 employees might say there are 150 men & 50 women in the upper pay quarter 120 men & 80 women in the upper middle pay quarter, etc, not 75%/25%, 60%/40%, etc.
One reason for doing it this way is that it will catch employers who are calculating pay quarters incorrectly. By definition an employer with 800 employees must have 200 in each of the 4 pay quarters yet I have seen some employers think that pay quarters require the pay scale to be divided into 4 equal chunks. For example, they might incorrectly think that their pay scale which runs from £10 an hour to £50 an hour needs to be divided into £10-19ph, £20-29ph, £30-39ph, £40-50ph with the result that the majority of employees are in the lowest quarter and only a few in the upper quarter. If an employer has an odd number of employees, say 801, then one pay quarter would have 201 employees and the others 200. So the GEO will know that the totals for each pay quarter should not vary by more than 1 and if any employer does submit data violating this rule, the GEO can automatically flag the error to the employer and tell them to recalculate and resubmit.
It is important to note though that any numbers submitted by the employer to the GEO are never made public. They are retained by the GEO who will then publish the % breakdown by category as they do today. This point is covered by recommendation #5.
Recommendation 4 – Employers to submit average pay per Pay Quarter
For each of the 4 pay quarters, I am calling for employers to tell the GEO what the average hourly pay is. This is across all employees within the pay quarter, not separately for men & women (though a case can be made for doing this as explained in point 6 of “10 recommendations for improving pay gap reporting” published by the Royal Statistical Society in 2019). So for example, an employer might say the mean hourly pay in the lower pay quarter is £10 per hour, in the lower middle pay quarter it’s £15 per hour, etc.
Again as for recommendation #3, this data is never made public and is retained by the GEO. What is made public as per recommendation #5 are four Pay Ratios where the lower pay quarter is set to 1 and the other 3 quarters are expressed as a ratio of this. For example, if an employer submits £10ph, £15ph, £25ph, £50ph as the averages for the 4 pay quarters, the 4 published ratios will respectively be 1, 1.5, 2.5 & 5.
The motivation for doing this is explained more in recommendation #5 but I want to point out that pay ratios have been discussed by other people in relation to executive pay. There have been calls for companies to be required to publish the ratio between CEO and lowest paid employee for example. If my recommendation is acted upon, it would mean that the gender pay gap reporting system could also be used to illuminate pay ratios in general which may be more efficient than requiring a separate system for executive pay.
As mentioned in recommendation #1, if recommendation #4 is not enacted then I would support the retention of the mean gender pay gap statistic. The swap number covered in recommendation #5 is intended to be the replacement for the median gender pay gap statistic.
Recommendation 5 – GEO to calculate & publish Fingerprints, Swap Numbers & Pay Ratios for employers
Once the GEO has received and verified the data submitted as per recommendations #2, #3 & #4, they can then make public the following statistics for the employer.
- The overall % breakdown by category (Male, Female, Other, N/A) subject to any minimum category size as discussed in recommendation #2.
- The Fingerprint i.e. the % breakdown by category for each of the 4 pay quarters.
- The Pay Ratios for each of the 4 pay quarters where the lower pay quarter is set to 1 and the other 3 pay quarters are expressed as a ratio of the lower pay quarter.
- The Swap Number for each pair of categories. For gender pay gap reporting, in almost all cases, only the Male-Female Swap Number will be published.
Fingerprints & Swap Numbers are explained in the articles linked in the list above (click on Fingerprint & Swap Number). My call for Pay Ratios is new but is inspired by this graphic which compares two employers, Hitachi Capital & Staffordshire Police who have essentially the same fingerprints & swap numbers but very different median gender pay gaps. If the 4 pay ratios were also published, it should show that the pay ratio in the upper pay quarter for Hitachi is probably double that of Staffordshire Police
Of course there is nothing to stop an employer doing all of these calculations themselves as they are simple to do. My article “Abolish the gender pay gap!” gives you the step by step process for the calculations but anyone who has the fingerprint in front of them like those in the graphic above can do the following calculation:-
- Treat the percentages in the pay quarter charts as numbers instead e.g. 73% is 73. Thus each employer is assumed to have 400 employees in total.
- Female swap number for Hitachi is FSN(h) = (65+62-46-27)/2 = +27 & for Staffordshire Police is FSN(s) = (59+60-38-29)/2 = +26
- Male swap number for Hitachi is MSN(h) = (38+35-54-73)/2 = -27 & for Staffordshire Police is MSN(s) = (41+40-62-71)/2 = -26
- Gender Swap Number for Hitachi is GSN(h) = [ FSN(h)-MSN(h) ]/2 = +27 & for Staffordshire Police is GSN(s) = [ FSN(s)-MSN(s) ]/2 = +26
- Pro-rata the GSNs so that the assumed number of employers is 1000 rather than 400 as per point 1 so this means multiply GSN by 2.5 to give the GSN per 1K of +67.5 for Hitachi Capital & +65 for Staffordshire Police.
The reason why swap numbers are superior to median pay gaps is that they directly tell you how many men & women have to swap places on the pay scale for the median gender pay gap to be zero. In Staffs Police case, assuming they have 1,000 employees, then if 65 women from the two lower pay quarters swap places with 65 men from the two upper pay quarters, the number of men in the top two and bottom two pay quarters will be equal. Similarly the number of women in the top two pay quarters will be the same as the bottom two pay quarters. If this situation is reached, mathematically the median gender pay gap must be zero as explained in my article “How to spot an incorrect median gender pay gap” which uses the logic of a swap number to identify why Cleveland Police’s pay gap was incorrect. Note, back then I was using the term gender ratio differential rather than gender swap number but they both follow the same logic.
Once an employer know how many men and women have to swap places they can then answer the question of how long this is likely to take. I’ve explored this already in my “Eliminate your gender pay gap by playing Blackjack!” where again I use the concept of a swap number though without explicitly calling it that. The key point to take away from that article is that closing a pay gap can take a very long time, perhaps 25 years or more, even if an employer does not discriminate in recruitment & promotions. An option the GEO could explore is that when they publish a swap number for an employer, they could also publish (using the model that can be downloaded from that article) an estimate of how long it will take the employer to close their pay gap given certain assumptions. I think such an estimate would prompt an employer to think more deeply about what their goals are in respect of diversity & inclusion and how to get there.
This article is focusing on gender but as I said earlier, fingerprints, pay ratios & swap numbers can also be published for ethnicity if ethnicity pay gap reporting is mandated. Suppose an employer has enough employees to provide fingerprints breaking down their pay quarters by White, Asian, Black, Other & N/A, then swap numbers can be published for each possible pair of these categories. There are 10 possible pairs in all but probably only the 4 pairs involving white employees i.e. White:Asian, White:Black, White:Other & White:N/A would actually be published.
Recommendation 6 – Employers to include a 2nd breakdown by category in their pay gap report
In my article “Why Ryanair’s pay gap report is my favourite“, I showed how they had inadvertently stumbled upon the swap number concept. This was because in addition to publishing the statutory pay gap information (using horrible graphics), they also broke their workforce down by job role, specifically Pilots, Cabin Crew & Others. Because pilots earn considerably more than cabin crew and that Ryanair employ similar numbers of pilots & cabin crew in the UK, this meant their swap number & fingerprint was almost completely explained by this secondary breakdown that Ryanair had voluntarily disclosed.
The top half of the chart shows that Ryanair’s gender swap number is +196 for a workforce of 1450 in order to reach a point in the future where the median gender pay gap is zero. For a pro-rated workforce of 1,000 this gives Ryanair a GSN per 1k of +129. The bottom half of the chart is however more insightful since that shows that the swap number is more like +225 out of 1450 since the median gender pay gap will close if 227 male pilots swap places with 224 female cabin crew. When you see it laid out like that, you realise that Ryanair may never close their pay gap due to the paucity of female pilots worldwide.
I am calling for employers to be required to publish a second breakdown by gender (& ethnicity) in their pay gap report using a categorisation of their choice. This 2nd categorisation would not be submitted to the GEO and would only appear in the narrative that employers are required to publish. Exactly how this could be made mandatory needs further discussion so it may end up as a strong recommendation instead but the point is to leave it to the employer to choose something that best reflects how they operate today. Some possible options are –
- Job Role – this what Ryanair did and is completely the right choice since these roles are the heart of an airline.
- Sites or Divisions – some employers might have multiple sites e.g. my article “What is the gender pay gap at Novartis UK?“ noted that they had a head office in London and a factory in Hull.
- Perm/Temp – some industries like hospitality or agriculture rely heavily on a casual or seasonal workforce that contracts and expands. Splitting by Permanent & Temporary roles would make sense here.
- Full Time/Part Time – retailers are known for having a large workforce of part time workers in their stores and depots. This might be an appropriate categorisation for them to use.
- Pay Bands – the public sector uses pay bands to pay their employees and a number of such employers disclose gender breakdowns by pay bands in their reports e.g. the Foreign Office in 2020.
As I say, this secondary data is not submitted to the GEO but the GEO could issue guidance to employers as to the most appropriate breakdown to use.
Recommendation 7 – Repeat steps 2 to 6 including & excluding Variable Pay
In recommendation D later on in this article, I call for Bonus Pay to be reclassed as Variable Pay. I believe it is better for employers to report data based on total remuneration including bonuses rather than the current separation between ordinary pay and bonus pay. For some employers though it may make more sense to exclude variable pay. So this recommendation is to require employers to submit two sets of data, one including and one excluding variable pay. The GEO would then publish two sets of fingerprints, swap numbers and pay ratios accordingly.
An option to consider is for the employer to tell the GEO which set of data they consider more meaningful. This employer preference could then be made public to assist those interrogating the GEO database.
This recommendation replaces the requirement to report bonuses separately. I don’t consider that requirement insightful and I’ve yet to see such data taken seriously.
That concludes my list of recommendations aimed at improving what is reported by employers. I believe taken together, recommendations 1 to 7 would greatly enhance interpretation of pay gap data and make it easier to prepare and report in the first place.
I now go through recommendations A to E which aim to improve the quality of the data used in the calculations and reduce potential distortions in the data.
Recommendation A – All calculations based on a 12 months/52 weeks period
The current regulations require the pay gap to be calculated based on the pay period which includes the snapshot date (31st March public sector, 5th April private sector). For most employers that is usually a month or a week’s worth of data. Bonus gaps on the other hand are required to be based on the 12 months to the snapshot date.
This major difference in required data causes employers a lot of difficulty today as it requires them to manipulate their payroll data in two completely different ways. Why did Parliament mandate this? I think the answer is because when Parliament passed the Equality Act in 2010 which mandated pay gap reporting, they didn’t know how to do it and they made the decision to copy what the Office of National Statistics (ONS) was already doing for the gender pay gap data they were publishing using their ASHE survey. I explain more about this is my article “No employer has a pay gap! Let’s celebrate!” but the ONS method largely excludes bonuses and is based on a sample of HMRC payroll data taken in April of each year. Parliament may have thought that they needed to use the same method to ensure comparability between ONS and employer figures but as I explain in the article, the ONS method can never be comparable with employer pay gap data.
Since it is now clear that comparability is no longer possible or indeed desirable, I am calling for the legislation to be changed so that employer pay gaps are calculated on a full year’s worth of data. I think the most common full year for payroll is 12 months to the end of March or 52 weeks to last Friday of March. If this is adopted, this is basically the time period used for bonus pay except that there is no longer separate snapshots for public and private sector and all employers use 31st March as the snapshot date.
One issue that will be more prominent for employers when collecting their data under this recommendation is that more employees will have started or left an employer during the year. The regulations will need to state that all such employees should have their earnings pro-rated to the full year e.g. someone who has only worked 3 months of a year and already earned £10,000 would have that multiplied by 4 to arrive at a pro-rated £40,000.
Recommendation B – Reconsider current employee exclusions
In my article “10 quick and easy ways to eliminate your pay gap” I introduce 10 tools of Creative Pay Gapping. Tools 3 to 6 all exploit provisions in the current legislation that allow certain employees to be excluded from the calculations. These include partners, being on reduced pay leave, non-UK employees, those identifying as non-binary, etc.
Such exclusions have the potential to distort pay gap figures. Some make sense because the current pay gap is based on one pay period (reduced pay leave), others because of the current split between ordinary pay and bonus pay (partners). Since recommendation A calls for 12 months of data for all calculations and recommendation #7 calls for two sets of figures (including & excluding Variable Pay), I think many of these exclusions can now be overturned and the employees included. For example, someone taking a month of reduced pay leave (e.g. furlough) is going to heavily influenced the existing pay gap but over a 12 month period, that impact will be reduced.
An important point that needs to be debated is that I think all employees who continue to be excluded from the new set of regulations would now have to be entered in the N/A category as per recommendation #2. That way, we will know how many employees and from which pay quarters they are in. If an employer says they are excluding a large number of employees, I think they need to explain why.
Recommendation C – Use Total Remuneration Package as much as possible
I am sure most employees evaluate what they are paid based on the total package rather than the headline pay figure. I am calling for the regulations to be changed so employers try to include as many benefits as is reasonable within the 12 month period as called for in recommendation A. That way the overall pay for each employee takes into elements that might have a gendered impact. A simple way may be to simply add together what is recorded in an employees P60 & P11D forms each year.
One element that will be affected by this is salary sacrifice for pensions. If you’ve not heard this before, this involves an employee giving up part of their salary and the employer contributing that amount to their pension. At first sight, that appears to be no different from an employee deciding to contribute the same amount to their pension pot from their gross pay. However, the current pay gap regulations treat these differently with employees choosing salary sacrifice having their net pay after sacrifice used in the calculations and employees choosing pension contributions having their gross pay before pension contribution used in the calculations. This is weird but it is a direct result of the fact the ONS in their pay gap method take a sample of HMRC records and in HMRC eyes, the former is a different situation to the latter. This is something that should be changed.
Recommendation D – Change Bonus Pay to Variable Pay
I am calling for employers to make a distinction between remuneration that is considered “steady” and remuneration that is considered “variable” year on year. As per recommendation #7, employers would report two sets of data, the first for total remuneration including variable pay and the second for total remuneration excluding variable pay.
The current definition of bonuses is probably close enough to a future definition of variable pay but I would suggest some consultation over this. I think employers should allowed more discretion so that the key principle becomes a split what is reliable pay and what is variable pay. I do think employers should be explaining in their pay gap report their definitions of these two concepts.
Together with recommendation #7, this replaces the current requirement to report both bonus pay gaps and the percentage of men and women receiving bonuses.
Recommendation E – Conglomerates to report group level data
Tool number 10 of my article “10 quick and easy ways to close your pay gap” is the best tool for any creative pay gapper. It allows you to slice and dice your workforce into pieces that flatter your pay gap when in fact if you were required to report the conglomerate as well, your pay gap would not look so rosy. I think it is time to mandate that any holding or conglomerate be required to report for the whole group as well as the separate legal entities that make up the group.
I appreciate that I am not an expert in company law and the definition of what makes a conglomerate or group will have to be defined. One possibility is to the use the “common terms” clause of the Equal Pay Act 1970. This has been in the news recently with the Supreme Court ruling that Asda stores and logistic depots are considered to have “common terms” for the purposes of the Act. The key principle is does Walmart, who own Asda, have the power to change the terms and conditions in both stores and depots? If they do then depots and stores are part of the same establishment. Whilst gender pay gaps is not the same as unequal pay, there is no harm making the link for this point.
I do hope you will give me some feedback on these 7 + 5 recommendations using the links given below. If I receive feedback that suggests additional recommendations should be made, I will add them here and use my newsletter to inform you on the update.
— 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 & diversity? —
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 & employment disparities? —
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 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.