Guidance

Making your calculations

Updated 26 May 2023

The ethnicity pay figures we recommend

If an employer chooses to calculate their ethnicity pay figures, the following measures can provide a well-rounded understanding of any disparities.

These measures are recommended to help an employer understand aspects of differences between ethnic groups including:

• ethnicity representation at different pay levels in their organisation, and across their organisation as a whole
• differences in average pay between ethnic groups
• levels of engagement with ethnicity questions

We don’t recommend looking at any one of these measures in isolation (such as the mean or median ethnicity pay gap).

The measures we recommend are:

• percentage of each ethnic group in each hourly pay quarter
• mean (average) ethnicity pay gap using hourly pay
• median ethnicity pay gap using hourly pay
• percentages of employees in different ethnic groups in your organisation
• percentage of employees who did not disclose their ethnicity – they either answered ‘prefer not to say’ or gave no answer when you attempted to collect their ethnicity.

Details of how to make each of these calculations can be found below.

Bonus pay gaps

Employers may additionally choose to calculate:

• the percentage of each ethnic group receiving bonus pay
• the mean (average) pay gap for bonus pay
• median ethnicity pay gap for bonus pay.

We would strongly encourage organisations to do this where bonuses make up a large proportion of employee pay.

Choose which ethnic groups to analyse

Before you begin to calculate your ethnicity pay measures, you need to decide which ethnic groups you will include in your calculations. The decision will be based on the number of employees in different ethnic groups. We do this so that:

• individual employees cannot be identified
• the data is statistically robust

Ensuring individual employees cannot be identified

To comply with the General Data Protection Regulation (GDPR), it must not be possible to identify an individual from information put into any report. A common way of doing this is to not publish statistics for an ethnic group with less than a certain number of employees.

Ensuring statistical robustness

Having a certain number of employees in each ethnic group for your analysis can also help you be confident that the data is robust. This can help prevent the following scenarios from occurring:

• you appear to have pay disparities which are simply the result of the time that you collected the data and the make up of your workforce at that particular time – for example, you have temporarily opened or closed a certain site and this has had a short term impact on pay
• you appear to have no pay gaps and disparities now but in reality there are issues which have been temporarily hidden because of random, unusual circumstances

It might be difficult for you to know if these situations have occurred. You will need to think about for example:

• whether the make-up of your workforce has changed in an unusual way over the year – for example whether you have recruited more lower or higher paid employees than usual
• if you have opened new sites in different areas of the country with different proportions of different ethnic groups

Producing ethnicity pay reporting analyses for more than one year can help see whether results for any given year are unusual.

A minimum category size

To guard against both of these issues, we recommend setting a level for the minimum number of employees in each group (a ‘minimum category size’) that you will analyse. This minimum category size will depend on whether your analysis is:

• for internal use only
• to be published externally

You will probably be able to have a lower minimum category size for each ethnic group for internal use. If you are publishing your analysis, we recommend a higher minimum category size.

If you are only using your analysis internally

Usually your main concern here will be not disclosing information about individual employees. Current good practice from publishers of statistics (such as the ONS) suggests the minimum the ‘minimum category size’) to avoid this should be somewhere between 5 and 20 employees.

Exceptions to this requirement are employees who chose ‘prefer not to say’ and who gave no answer. Since the employees have not disclosed their ethnicity, it should be much harder to identify these individuals from any statistic published. Therefore these categories do not have to comply with the minimum category size you decide upon and if the number of employees in these categories is really small, they can be excluded from analysis.

If your organisation has a dedicated data rights team, it may be worth consulting them to make sure you are compliant with the GDPR, and not at risk of disclosing information about individual employees.

If you are publishing your analysis

Your concerns here might be about not disclosing information about employees and ensuring your analysis is robust.

We recommend that a minimum category size of 50 employees should be specified to ensure statistical robustness when publishing data.

This will also guard against information about individuals being disclosed.

If you intend to publish data for specific groups of your employees, for example separate analyses for 2 different sites, the minimum category size applies to each analysis published, not just an analysis with all of your employees combined. Therefore, to ensure you have sufficient numbers of employees per category for each group of employees, you may need to specify a larger minimum category size for all employees combined.

If you are unsure what value to use, we recommend you take advice from an analyst and your data rights team.

Once you have decided your minimum category size, you can count the number of employees within each ethnic group you have collected. This will tell you which groups are below the minimum category size you have collected. If some groups are below that number, you may need to aggregate some ethnicities into larger groups.

Deciding how to aggregate

There are many ways to aggregate the ethnic categories that are below the minimum category size for either internal analysis or publication. It requires judgement and understanding of the ethnicity of your workforce. However, you can bear in mind the following considerations when you aggregate data for ethnic groups.

Try and show as many ethnic groups as possible

You should try and show as many ethnicity categories as possible in your analysis. Some ethnic groups may be earning much more than others and breaking down the different categories will give a much richer picture and better inform your action plans.

Aggregate to 5 larger ethnic groups

You could use the 5 aggregated groups below and the ‘prefer not to say’ option:

• Asian
• black
• mixed
• white
• other
• prefer not to say

You might also be able to show white British and ‘other white groups’ separately.

The risk of aggregating in this way is that differences between ethnic minority groups could be hidden. For example Bangladeshi and Pakistani employees generally earn less than those from the Indian ethnic group, but this might not be clear for analysis of a broader ‘Asian’ group.

Aggregate to 2 groups (‘binary’ reporting)

Sometimes you might only be able to report pay gaps between either:

• white and all other ethnic minorities combined
• white British and ethnic minorities

We strongly discourage you from doing this in isolation. Only reporting 2 ethnic groups will mask detail and nuance which might be vital for understanding ethnicity pay gaps and identifying relevant actions. However, you may have to do this should employees’ confidentiality be at risk if any further details were released.

If binary reporting is the only available first step for you, particularly if you are a smaller employer or have small numbers of employees in certain ethnic groups, you can keep this under review and aim towards reporting on a more granular level in future years.

Where you can analyse data for more than 2 groups, you might also want to do a binary analysis. This could help you compare data consistently over time in your organisation, or your data with that of employers who are only able to produce a binary analysis.

Recording aggregated groups

If you have aggregated some of your ethnic groups together, you need to record the aggregated group for each employee. You will be using these aggregated groups for each of the ethnicity pay calculations described later.

Employees who ‘prefer not to say’ or who did not respond

You should collect data on these 2 groups of employees and record this information in any published data. The ‘prefer not to say’ category can indicate if there are employees who have responded to the data collection but do not feel comfortable disclosing their ethnicity. For employees who have not responded, this will help you consider how many employees are not engaging with activities to collect diversity data. Employers might need to do further work to boost engagement in this case.

Both categories can provide useful information for employers and add useful context to their published data. We recommend keeping these 2 categories separate. It is also very important that they are not aggregated with any other ethnic group.

Presenting and writing about ethnic groups

In their harmonisation webpage, the GSS provide information on how to present data for different ethnic groups. The RDU also provides guidance about how to write about ethnic groups. Employers might find these resources useful when preparing their data and writing their action plans.

Illustrative examples

For some of the examples in the calculations below, for simplicity we have assumed that the calculations are using a split between 2 ethnic groups. In some cases, you might need to perform the same calculations for each ethnic category that you are using.

Calculation 1: percentage of employees in different ethnic groups in each hourly pay quarter

To produce your quarterly pay bands by ethnic category data, you will need to:

1. Define your pay quarters and assign each employee to a pay quarter.

2. For each pay quarter, count the number of employees within each ethnic category, ‘prefer not to say’ and those who did not respond.

These steps are the same as those used for gender pay gap reporting except you replace the gender categories with your ethnic categories. If you have aggregated your ethnic groups together, then you will use these aggregated groups for the analysis (along with ‘prefer not to say’ and those who did not respond).

You could present your data using:

• the number of employees in each combination of ethnic category and pay quarter
• the percentage of employees from each pay quarter that fall in each ethnic category
• both of these

Divide hourly pay into quarters

• refer to your list of hourly pay for all full-pay relevant employees in the relevant pay period)
• sort your full-pay relevant employees from highest to lowest based on their hourly pay
• divide this list into 4 quarters, with an equal number of employees in each section

These quarters will be the:

• upper hourly pay quarter
• upper middle hourly pay quarter
• lower middle hourly pay quarter
• lower hourly pay quarter

Example: When the number of employees isn’t divisible by 4

If the number of employees is not divisible by 4, distribute them as evenly as possible.

For example, Acme Ltd has 4,445 full-pay relevant employees. To distribute them equally into quarters would mean 1,111 employees in each quarter, with 1 employee left over.

In this case, you should add the 1 employee left over to the lower hourly pay quarter.

This means there are 1,112 employees in the lower hourly pay quarter in this example, 1,111 employees in the lower middle hourly pay quarter, 1,111 employees in the upper middle hourly pay quarter, and 1,111 employees in the upper hourly pay quarter.

If you have 2 employees left over, add one employee to the lower hourly pay quarter and one employee to the upper middle hourly pay quarter.

If you have 3 left over, you can distribute them between lower, lower middle and upper middle pay quarters.

Check the ethnicity distribution of matching hourly pay

If there are employees on the same hourly pay that overlap between hourly pay quarters, adjust the categories to ensure different ethnic groups are split as evenly as possible across the hourly pay quarters, either side of the overlap.

Example: distributing employees across pay quarters when they have the same hourly pay

Acme Ltd has 4,445 full-pay relevant employees, has sorted them by highest hourly pay to the lowest hourly pay, and has then divided the list into 4 hourly pay quarters.

However, 40 employees all have the same hourly pay:

• 36 are are from an ethnic minority group and 4 are white British
• of these, 10 have fallen into the lower hourly pay quarter
• 30 have fallen into the lower middle hourly pay quarter

To evenly distribute these employees by gender, Acme Ltd should list 1 white British employee for for every 9 ethnic minority employees listed:

• of the ethnic minority employees, the employer lists 9 of these, and 1 employee who is white British, in the lower hourly pay quarter
• 27 ethnic minority employees, and 3 employees who are white British in the lower middle hourly pay quarter

Optional: Use ‘pay halves’ instead if you have too few employees

It could be the case that although an ethnic category has enough employees for your minimum category size, when you split this category by pay quarters, you end up with some combinations of pay quarter and ethnic category with only 1 or 2 employees. If this were to happen, it might be possible to identify those people – for example, the 2 Asian employees in the upper pay quarter might be senior people known to many employees.

There are 2 options for you to mitigate this outcome. The first is to perform a further aggregation of your ethnic categories but this may not be possible or desirable. The second is to combine your pay quarters into pay halves instead as follows:

• upper pay half = upper pay quarter + upper middle pay quarter
• lower pay half = lower middle pay quarter + lower pay quarter

You can then present and interpret a pay half breakdown by ethnic category

Work out the percentage of employees in each ethnic group in each hourly pay quarter

For each hourly pay quarter and for each ethnic group you are using in your analysis, you need to:

• divide the number of full-pay relevant employees in the quarter who are in each ethnic group, ‘prefer not to say’ or who did not respond, by the total number of full-pay relevant employees in the quarter and multiply by 100

This calculation gives you the percentage of employees in the hourly pay quarter in each ethnic group (as well as for those who preferred not to give their ethnicity and those who did not respond).

The number of calculations you need to do will be the same as the number of ethnic group categories you are using, plus 2 calculations for ‘prefer not to say’ and employees who did not respond. This will be 4 ethnicity calculations for each pay quarter if you are using the binary split, for example:

• white British and ethnic minority groups
• white and ethnic minority groups other than white minorities
• ‘prefer not to say’
• did not respond

If you are using the 5 aggregated groups (white, Asian, black, mixed and other), you will need to do 7 calculations for each pay quarter.

If you are using all the England and Wales Census 2021 categories, you will need to do 21 calculations for each pay quarter.

Example: Calculating the percentage of white British and ethnic minority employees in each hourly pay quarter

Acme Ltd has 4,445 full-pay relevant employees and has:

• sorted them by highest hourly pay to the lowest hourly pay
• divided the list into 4 hourly pay quarters

• checked that employees on the same hourly pay are distributed as evenly as possible by ethnicity where they cross the quarter boundaries

• of the 1,112 employees in the lower hourly pay quarter, 900 are white British and 187 are ethnic minority employees. 25 answered ‘prefer not to say’

This means 80.9% are white British and 16.8% are ethnic minority employees. 2.2% were ‘prefer not to say’.

• of the 1,111 employees in the lower middle hourly pay quarter, 1,001 are white British and 100 are ethnic minority employees. 10 answered ‘prefer not to say’.

This means 90.1% are white British and 9.0% are ethnic minority employees. 0.9% were ‘prefer not to say’.

• of the 1,111 employees in the upper middle hourly pay quarter, 813 are white British and 283 are ethnic minority employees. 15 answered ‘prefer not to say’

This means 73.2% are white British and 25.5% are ethnic minority employees. 1.4% were ‘prefer not to say’.

• of the 1,111 employees in the upper hourly pay quarter, 361 are white British and 735 are ethnic minority employees. 15 answered ‘prefer not to say’

This means 32.5% are white British and 66.2% are ethnic minority employees. 1.4% were ‘prefer not to say’.

Calculations 2 and 3: mean (average) and median ethnicity pay gap for hourly pay

We have included advice here on mean (average) and median calculations as they can reveal different dimensions of pay differences. Many employers are already familiar with the processes involved due to the mandatory requirements for gender pay gap reporting. The mean calculation shows the average earnings taking account of the earnings of all employees in an organisation. The median calculation shows the average middle earner in an organisation and is less likely to be skewed by a few high earners in an organisation. Both calculations can provide useful information for an employer.

In gender pay gap reporting, the pay of women is compared with that of men. Men are the ‘comparator group’. In ethnicity pay reporting, choosing a comparator group is more difficult because you might want to see disparities between a number of different ethnic groups. Analysing pay gaps between each combination of ethnic groups you are using might show more nuance and help you develop a stronger action plan to tackle any pay disparities. We suggest calculating mean (average) and median gaps between each combination of ethnic groups you are using.

If you are using the binary split (white British employees and ethnic minority employees, for example), this will be 1 calculation each for the mean and median pay gap.

We recommend caution in drawing conclusions about your ethnic pay results if using a binary split. In such cases more attention should be placed on the other calculations outlined in this guidance (hourly pay quarters, ethnic minority representation and ethnicity disclosure rates) as these measures will provide more useful information about any actual disparities.

If you are using the 5 aggregated groups, you will need to do 10 calculations each for the mean and median pay gap.

If you are using the full Census breakdown, you will need to do 171 calculations each for the mean and median pay gaps.

Calculation 2: the mean (average) ethnicity pay gap for hourly pay

The mean (average) ethnicity pay gap figure you might report uses hourly pay of all full-pay employees to calculate the difference between the mean (average) hourly pay of each combination of ethnic groups you are using. For example, the difference in mean (average) hourly pay between Indian employees and black African employees, or between white British employees and ethnic minority employees.

A mean involves adding up all of the numbers and dividing the result by how many numbers were in the list.

Means are useful because they place the same value on every number they use, giving a good overall indication of the ethnicity pay gap. But very high or very low hourly pay can distort the figure.

The calculations of these figures are based on the payroll data of your full-pay relevant employees.

Start by referring to your list of hourly pay for full-pay relevant employees created in Task 5: Hourly pay for each ethnic group. Choose the first 2 ethnic groups you will analyse (we will call them group 1 and group 2):

Step 1. Calculate the mean (average) hourly pay for employees in group 1

• add together the hourly pay of all full-pay relevant employees who are in group 1
• divide this figure by the number of full-pay relevant employees who are in group 1

This gives you the mean (average) hourly pay for employees in group 1.

Step 2. Calculate the mean (average) hourly pay for group 2

• add together the hourly pay of all full-pay relevant employees who are in group 2
• divide this figure by the number of full-pay relevant employees who are in group 2

This gives you the mean (average) hourly pay for group 2.

Step 3. The mean (average) pay gap using hourly pay figure:

• take the mean (average) hourly pay for your group 1 employees and subtract the mean (average) hourly pay for group 2 employees
• divide the result by the mean (average) hourly pay for group 1 employees
• multiply the result by 100

This gives you the mean (average) ethnicity pay gap in hourly pay for group 2 as a percentage of the hourly pay of employees in group 1.

If you have aggregated your data into more than 2 ethnic groups (for example, Asian, black, mixed, white and other) you should perform the same calculation for each combination of ethnic groups you are using and present each mean (average) pay gap separately. Calculating the pay gaps for all combinations of groups can provide crucial information to inform your action plan to tackle pay disparities.

You might present the mean hourly pay gaps between each of the ethnic groups in tabular form. For example if you were analysing 5 ethnic groups, this might look like:

Ethnic group	Mixed	Asian	Black	Other
White	Gap between white and mixed ethnic groups	Gap between white and Asian ethnic groups	Gap between white and black ethnic groups	Gap between white and mixed ethnic groups
Mixed		Gap between mixed and Asian ethnic groups	Gap between mixed and black ethnic groups	Gap between mixed and other ethnic groups
Asian			Gap between Asian and black ethnic groups	Gap between Asian and other ethnic groups
Black				Gap between black and other ethnic groups

A worked example for 3 ethnic groups is given below.

Example: Calculating the mean (average) ethnicity pay gap using hourly pay for 3 groups

Acme Ltd has 4,445 relevant full-pay employees. Of these, 1,345 are white British employees, and 1,500 are Pakistani employees and 1,600 are Bangladeshi employees.

We call white British employees group 1, Pakistani employees group 2 and Bangladeshi employees group 3.

White British employees hourly pay amounts to:

• (1,000 employees x £15) + (300 employees x £21)+ (45 employees x £50) = £23,550
• divided by 1,345, the mean (average) white British employee earns £17.51 in hourly pay

Pakistani employees hourly pay amounts to:

• (200 employees x £10) + (500 employees x £13) + (500 employees x £15) + (200 employees x £21) + (100 employees x £40) = £24,200
• divided by 1,500, the mean (average) Pakistani employee earns £16.13 in hourly pay

Bangladeshi employees hourly pay amounts to:

• (250 employees x £10) + (600 employees x £13) + (400 employees x £15) + (300 employees x £21) + (50 employees x £50) = £25,100
• divided by 1,600, the mean (average) Bangladeshi employee earns £15.69 in hourly pay

Acme Ltd’s mean (average) ethnicity pay gaps (using hourly pay rounded to pence) are:

• Between white British employees and Pakistani employees: £17.51 minus £16.13; divided by £17.51, and multiplied by 100 give a figure of 7.88%
• Between white British employees and Bangladeshi employees: £17.51 minus £15.69; divided by £17.51, and multiplied by 100 give a figure of 10.39%
• Between Pakistani employees and Bangladeshi employees: £16.13 minus £15.69; divided by £16.13, and multiplied by 100 give a figure of 2.73%

This means that on average:

• white British employees at Acme Ltd are paid 7.88% more than Pakistani employees
• white British employees at Acme Ltd are paid 10.39% more than Bangladeshi employees
• Pakistani employees at Acme Ltd are paid 2.73% more than Bangladeshi employees

This means that on average for every £1 a Pakistani employee earns at Acme Ltd a Bangladeshi employee will earn 97p.

Calculation 3: the median ethnicity pay gap using hourly pay

The median ethnicity pay gap figure is the difference between the hourly pay of the median full-pay relevant employee for an ethnic group and the hourly pay of the median full-pay for another ethnic group. The median for each is the employee who is in the middle of a list of hourly pay ordered from highest to lowest paid.

A median involves listing all of the numbers in numerical order. If there is an odd number of results, the median is the middle number. If there is an even number of results, the median will be the mean (average) of the 2 central numbers.

Medians are useful to indicate what the ‘typical’ situation is. They are not distorted by very high or low hourly pay (or bonuses). However, this means that not all ethnicity pay gap issues will be picked up. They could also fail to pick up as effectively where the ethnicity pay gap issues are most pronounced in the lowest paid or highest paid employees. For example, if a Chief Executive Officer of a company has a substantially higher salary than the rest of the company, the median pay gap is not really affected by this.

The calculations of these figures are based on the payroll data of your full-pay relevant employees.

Start by referring to your list of hourly pay for your full-pay relevant employees created in Task 5: Hourly pay. Choose the first 2 groups you will analyse (we will call them group 1 and group 2) and then:

Step 1. Calculate group 1 employee’s median hourly pay:

• identify all full-pay relevant employees who are in group 1
• sort these employees in a list, in order of their hourly pay, with the lowest paid first and the highest paid last
• identify the employee (and their hourly pay) who is in the middle of this list

This gives you the median hourly pay for employees in group 1.

Step 2. Calculate group 2 employee’s median hourly pay rate:

• identify all full-pay relevant employees who are in group 2
• sort these employees in a list, in order of their hourly pay, with the lowest paid first and the highest paid last
• identify the group 2 employee (and their hourly pay) who is in the middle of this list

This gives you the median hourly pay for group 2.

Step 3. The median ethnicity pay gap using hourly pay figure:

• take the median hourly pay for group 1 employees and minus the median hourly pay for group 2
• divide the result by the median hourly pay for group 1 employees
• multiply the result by 100

This gives you the median ethnicity pay gap in hourly pay for group 2 as a percentage of group 1 pay.

If there is an even number of employees in your list of full-pay relevant employees for an ethnic group

You may find there is an even number of employees in your list of full-pay relevant employees for different ethnic groups. In this instance, to identify the median employee (the person in the middle of each list), use the average of these 2 people’s hourly pay to identify the median hourly pay.

For example, if you have 80 full-pay employees who are Indian employees, the 40th and 41st Indian employees would be the middle of this list (the median). To find the median hourly pay for Indian employees, take the mean (average) of these 2 ethnic minority employee’s hourly pay.

Example: Calculating the median ethnicity pay gap using hourly pay for 2 groups

Acme Ltd has a headcount of 4,500, of whom 4,445 are full-pay relevant employees. Of these, 1,345 are white British, and 3,100 are ethnic minority employees

Identify the median white British employee using hourly pay of full-pay relevant employees only:

• if all full-pay relevant employees who are white British stood in a line in order of their hourly pay there will be 1,345 white employees in this line
• the 673rd white British employee is the employee in the middle of this line and earns £14 an hour. They will have 672 employees to their left whose hourly pay is the same or less than theirs, and 672 employees to their right whose hourly pay is the same or more than theirs

In this example, the median hourly pay for white British employees is the hourly pay of the 673rd white employee in this line, which is £14

Identify the median ethnic minority employee using hourly pay of full-pay relevant employees only:

• if all full-pay relevant employees who are ethnic minority employees stood in a line in order of their hourly pay there will be 3,100 ethnic minority employees in this line
• because there is an even number of ethnic minority employees, when they are sorted based on their hourly pay, there would be 2 ethnic minority employees standing in the middle of this line. They will be the 1,550th and the 1,551st ethnic minority employees in the line
• the 1,550th ethnic minority employee will have 1,549 ethnic minority employees to their left earning the same or less than them, and the 1,551st ethnic minority employee will have 1,549 ethnic minority employees to their right earning the same or more than them
• the 1,550th ethnic minority employee earns £14 an hour, and the 1,551st ethnic minority employee earns £16 an hour

In this example, the median hourly pay for ethnic minority employees is the mean (average) of the 2 ethnic minority employees in the middle of the line (1,550th and 1,551st).

This is the average of £14 and £16, which is £15 median hourly pay.

Acme Ltd’s median ethnicity pay gap using hourly pay:

• take the white British employee’s median hourly pay and subtract the median ethnic minority employee’s hourly pay
• divide this figure by the median white British employee’s hourly rate of pay
• multiply by 100 to get the percentage figure
• £14 minus £15, divided by £14, multiplied by 100

This results in a median ethnicity pay gap based on hourly pay of -7.14%.

This means that when using the median, white British employees at Acme Ltd are paid 7.14% less than ethnic minority employees. Which means for every £1 an ethnic minority employee earns at Acme Ltd, a white British employee will earn 93p.

Like calculation 2, you might perform the median pay gap calculation for each combination of ethnic groups you are analysing. You might present this in tabular form.

Calculation 4: representation of ethnic groups in your organisation

The representation of ethnic groups is simply the percentage of employees falling in each ethnic category you have decided to analyse. This additional data can help contextualise your result for the other calculations, as well as provide more information for decision makers in an organisation about where there may be disparities or barriers in recruitment or progression within the organisation.

Calculation 5: percentage of employees whose ethnicity is ‘unknown’ or ‘prefer not to say’

Publishing, or at least internally reviewing, the number or percentage of employees who choose not to state their ethnicity, or who you were unable to gather this information from, is a useful baseline. You should try to analyse these 2 groups of employees separately.

It can be used to contextualise the other ethnicity pay calculations, as well as highlighting whether further work is needed to improve data collection. It might identify whether further work is needed to reassure and explain to your employees why collecting ethnicity data is important.

Calculation 6: percentage of employees in different ethnic groups receiving bonus pay

This calculation shows the percentage of people in different ethnic groups who received bonus pay in the 12 months ending on your snapshot date.

Use the list of all relevant employees (not only full-pay relevant employees), including the ethnic group they are in (after aggregation if necessary) and bonus pay. You might have aggregated your employees into larger ethnic groups.

For each ethnic group you are analysing, add together the number of employees in your list in the group who received bonus pay in the 12 months ending on your snapshot date. Divide this figure by the total number of employees in the ethnic group. This gives you the percentage of employees in the ethnic group who received bonus pay.

Example: Calculating the percentage of different ethnic groups who received a bonus payment

Acme Ltd has 4,500 relevant employees. 1,375 are from an ethnic minority group, and 1,300 of them received bonus pay. 3,125 are white British, and 2,000 of them received a bonus.

To calculate the percentage who received a bonus payment, Acme Ltd:

• divides 1,300 (ethnic minority employees receiving bonus pay) by 1,375 (total ethnic minority employees) – this equals 0.945
• multiplies 0.945 by 100
• divides 2,000 (white British employees receiving bonus pay) by 3,125 (total number of white British employees) – this equals 0.64
• multiplies 0.64 by 100

The results show that 94.5% of ethnic minority employees and 64.0% of white British employees received a bonus.

Calculation 7: mean (average) ethnicity pay gap for bonus pay

This calculation shows the difference in the mean (average) bonus pay paid to different ethnic groups.

Use the list of all relevant employees (not only full-pay relevant employees), including their ethnicity and bonus pay. You might have aggregated your employees into larger ethnic groups. Choose the first 2 groups you will analyse (we will call them group 1 and group 2) and then do the following.

1. Add together the bonus payments made to all employees in group 1 in your list in the 12 months to your snapshot date. Divide this figure by the number of group 1 employees who received bonus pay. This gives you the mean (average) bonus pay for group 1.

2. Repeat this for all group 2 in your list. This gives you the mean (average) bonus pay for group 2.

3. Take the mean (average) bonus pay for group 1 and subtract the mean (average) bonus pay for group 2.

4. Divide the result by the mean (average) bonus pay for group 1.

5. Multiply the result by 100.

This gives you the mean (average) ethnicity pay gap in bonus pay as a percentage of group 1’s pay.

Like calculation 2, you might perform the mean (average) pay gap for bonus pay calculation for each combination of ethnic groups you are analysing. You might present this in tabular form.

Example: Calculating the mean (average) ethnicity pay gap for bonus pay

Acme Ltd has a mean (average) bonus pay of £1,650 for black African employees, and £1,490 for black Caribbean employees.

To calculate this mean (average) ethnicity pay gap for bonus pay, Acme Ltd:

• takes the mean (average) bonus pay for black African employees (£1,650), and subtracts the mean (average) bonus pay for black Caribbean employees (£1,490) – this equals £160
• divides £160 by the mean (average) bonus pay for black African employees (£1,650) – this equals 0.097
• multiplies 0.097 by 100 to find the mean (average) ethnicity pay gap for bonus pay as a percentage
• Acme Ltd has a 9.7% mean (average) ethnicity pay gap using bonus pay between these 2 groups.

This means that, using the mean (average), black Caribbean employees at Acme Ltd are paid 9.7% less in bonus pay than black African employees. This means for every £1 a black African employee receives in bonus pay at Acme Ltd, a black Caribbean employee receives 90p.

Calculation 8: median ethnicity pay gap for bonus pay

This calculation shows the difference in the median bonus pay paid to employees in different ethnic groups.

Use the list of all relevant employees (not only full-pay relevant employees), including their ethnicity and bonus pay. You might have aggregated your employees into larger ethnic groups. Choose the first 2 groups you will analyse (we will call them group 1 and group 2) and then:

1. Create a list of all employees in group 1 who received bonus pay in the 12 months ending on your snapshot date. Sort them in order of highest to lowest bonus pay amounts. Identify the employee in the middle of the list, and write down their bonus pay. This figure is the median bonus pay for group 1.

2. Repeat this for group 2. This figure is the median bonus pay for group 2.

3. Take the median bonus pay for group 1 and subtract the median bonus pay for group 2.

4. Divide the result by the median bonus pay for group 1.

5. Multiply the result by 100.

This gives you the median ethnicity pay gap in bonus pay as a percentage of group 1’s bonus pay.

Like calculation 7, you might perform the median pay gap for bonus pay calculation for each combination of ethnic groups you are analysing. You might present this in tabular form.

If there is an even number of people in an ethnic group

Identify the 2 employees in the middle of the list. Use the mean (average) of their bonus pay to identify the median bonus pay figure.

For example, if you have 80 Bangladeshi employees, numbers 40 and 41 are in the middle of the list. To find the median bonus pay for Bangladeshi employees, take the mean (average) of these 2 employees bonus pay.

Example: Calculating the median ethnicity pay gap for bonus pay

Acme Ltd has a median bonus pay of £2,300 for all relevant employees who are Chinese, and £2,225 for all relevant employees who are Indian.

To calculate the median ethnicity pay gap using bonus pay for these 2 groups, Acme Ltd does the following:

• takes the median bonus pay for Chinese employees (£2,300) and subtracts the median bonus pay for Indian employees (£2,225) – this equals £75
• divides £75 result by the median bonus pay for Chinese employees (£2,300) – this equals 0.033
• multiplies 0.033 by 100 to find the median ethnicity pay gap for bonus pay for these 2 groups as a percentage
• Acme Ltd has a 3.3% median bonus ethnicity pay gap for these 2 groups

This means that when using the median, Indian employees at Acme Ltd receive 3.3% less bonus pay than Chinese employees. This means for every £1 a Chinese employee receives in bonus pay, an Indian employee receives 97p.

Understanding what a positive or negative percentage figure means

In Calculation 2, Calculation 3, Calculation 7 and Calculation 8:

• a positive percentage figure reveals that typically, or overall, employees in the second ethnic group have lower pay or bonuses than employees who are in the first group
• a negative percentage figure reveals that typically, or overall, employees who are in the first group have lower pay or bonuses than employees in the second group
• a zero percentage figure would reveal no gap between the pay or bonuses of employees in the 2 groups