There is a trend in employment law towards transparency being used as a method of driving change. We see it in the “name and shame” regime for National Minimum Wage contraventions, the new CEO pay ratio reporting regime, the publication of age demographic statistics suggested by the Women and Equalities Committee, and in gender pay gap reporting.
And now, with the Government’s consultation, it looks like ethnicity pay reporting will be next.
What is in the consultation?
Ethnicity pay reporting featured in manifestos in last year’s snap election. The Conservative manifesto said that, if elected, the party would ask large employers to publish ethnic pay gaps and Labour’s manifesto also expressed concern over the pay gap suffered by black and Asian workers.
The consultation seeks feedback on the sort of information employers should be required to publish. It explains different ways in which this could be done, including having a single pay gap figure of “white vs non-white”, multiple pay gaps for different ethnicities, or publication of pay information by £20,000 pay band or quartile (as suggested by Baroness McGregor-Smith in her 2017 report “Race in the Workplace”).
The Government’s position is that employers with 250 or fewer employees should not be required to publish, but other views are sought.
Gender pay gap reporting and the small groups problem
The consultation also seeks comments on the extent to which it would be helpful to mirror gender pay gap reporting requirements: for example with the same definitions of “pay”, “bonus” and “relevant employee”, and the same six reportable statistics.
There would, however, be difficulties in such a “copy and paste” approach to legislating, because of how gender pay gaps are calculated.
The gender pay regulations require comparison of the average woman to the average man – if a company hires no/few women in either the lowest paid shop floor manufacturing jobs, or the highest paid board roles, and only a few women in decently-paid mid-level roles, it could end up with a very “good” looking gap. But that doesn’t necessarily reflect realities.
For example, imagine a workforce with 300 people, but which is 90% male. Because the average female pay is determined by just 30 people, the arrival/departure of one more woman in the lowest paid job, or one woman on the board, can have a big effect on the average female hourly rate, and therefore the pay gap. As this report shows, the gender pay gap for an employer with no equal pay issue which hires and promotes with no regard for gender could range massively: from -35% (i.e. in favour of women) to +45% just because of the laws of chance. When comparing a small group against another much larger one, this margin of error becomes very large.
The above applies by analogy to ethnicity pay gap reporting, but to a much greater extent because the size of groups will be even smaller. If, for example, that same employer with 300 people employs black individuals in proportion to the wider population (3% of the UK according to the ONS), it means their average pay would be calculated from just nine individuals.
Ethnic minorities are exactly that – a minority. Many large employers will only have a small proportion of non-white employees. The consultation recognises this issue. It notes that a headline pay gap figure does not reflect overall representation – a company with a highly paid non-white CEO might have good pay gap figures, but low minority representation in the wider workforce.
Classification of ethnicities
The consultation paper also identifies the classification of ethnic groups as a potential difficulty.
Firstly, many employers do not hold ethnicity data on all their staff. For example, when EY voluntarily reported its ethnicity pay gap, it found that it only had data for 80.3% of its workforce. Because ethnic origin is a special category of personal data, employers will need to be careful about how they collect, store and process it.
Secondly, the granularity of ethnicity could have a big impact on the complexity of reporting obligations. In the last census, the ONS grouped individuals into 18 specific ethnic classifications and five broader ones.
To avoid the complexity, one option suggested in the consultation is simply to calculate a pay gap of “non-white” against “white or white British”. By pooling all BAME individuals together, as sample sizes get bigger more reliable statistics may be produced.
But if a pay gap is to be used as proxy for the workplace chances afforded to someone, then quite apart from the possible divisiveness of binary distinctions between “white” or “white British” and everyone else, conflating ethnicities might do little to show this. The issues faced by one minority group are unlikely to be identical to those faced by another. For example, ONS data shows an overrepresentation of Asian workers in financial services, but an underrepresentation of black workers.
Another approach suggested in the consultation is a more granular reporting regime, with several pay gaps being reported across either five or 18 different ethnicity categories. However, this would exacerbate the “small groups” problem and add to the complexity of compliance. While the consultation proposes a comparison of ethnic minorities against white people, if the legislation went further and required comparison between ethnicities, employers would be required to calculate and report no fewer than 171 different pay gaps if all 18 categories were used.
A simple copy and paste of the gender pay gap regulations for ethnicity is likely to result in dubious statistics of limited value. It is clear that a different approach is needed.
An alternative approach which does not require the publication or calculation of pay gaps is more likely to yield reliable statistics. Pay figures reported on a quartile or centile basis, or the banded approach suggested by Baroness McGregor-Smith, would still give an indication of ethnic minority representation at an organisation, especially if published alongside official statistics showing the breakdown of ethnicity for the local and national population.