Ethnicity pay gap reporting is about to stop being voluntary. In its consultation response of 25 March 2026, the government confirmed it "will introduce mandatory ethnicity and disability pay gap reporting for employers with 250 or more employees" — the same threshold, the same six calculations and the same service as gender pay gap reporting.
What was confirmed
The response settles several questions that had been open.
The threshold is 250 employees. The government took the view that this aligns with gender pay gap reporting and "avoids unnecessary burdens on smaller organisations, where the costs of reporting will be more impactful".
The calculations mirror the gender regime. Employers will report the same six measures:
- Mean hourly pay difference;
- Median hourly pay difference;
- Pay quartiles — the percentage in each of four equally sized pay bands;
- Mean bonus pay difference;
- Median bonus pay difference;
- Percentage receiving bonus pay.
Two further data points are added. Beyond the six, employers will report workforce composition and declaration rates — the percentage of employees who have not disclosed their ethnicity or disability status. That second one is the quiet centre of the whole regime, and the next section explains why.
The comparison is binary first. For ethnicity, the minimum required comparison is White, including White Other, against all other ethnic groups combined. A secondary requirement compares across five broad ethnic groups where minimum thresholds are met, with data collected using GSS harmonised standards. For disability, the comparison is binary only — disabled against non-disabled employees, using the Equality Act 2010 definition.
Timing follows the gender regime. The same reporting dates and the same online reporting service, with snapshot dates of 5 April for private and voluntary sector employers and 31 March for public bodies, and a 12-month deadline to publish.
The declaration rate problem
Gender pay gap reporting works on data the employer already holds for payroll. Ethnicity and disability do not work that way. Both depend on employees choosing to tell you, and nobody can be compelled to.
That creates a compliance problem that is not solved by better systems. A minimum group size of at least 10 employees in each group has been proposed, though the response is explicit that "policy work is ongoing to determine the most appropriate threshold". Below whatever threshold is settled, a group cannot be reported without risking identification of individuals.
So an employer with 300 staff and a 45% declaration rate may find that after removing non-disclosures and suppressing small groups, the reportable population is a fraction of the workforce — and a gap calculated on that fraction is not a reliable statement about the organisation. Publishing it anyway, without saying so, is the mistake to avoid.
What to do now
The legislation is not final. The response includes draft clauses for primary legislation with supporting regulations to follow, and says the government will continue to develop legislation and publish more information in due course. But the lead time on the underlying data is long, and that work does not depend on the final drafting.
Start collecting, and measure the declaration rate. This is the only task with a genuine lead time. Declaration rates improve slowly, through repeated asking, visible explanation of why the data is wanted, and demonstrated confidentiality. An employer starting at 40% will not reach a defensible rate in one campaign.
Use the harmonised standards from the outset. Data collected against a bespoke category list will have to be recoded, and recoding historic responses is unreliable. Collect once, in the form the regime will require.
Do a dry run on last year's data. Calculate the six measures now. The point is not the number — it is to discover whether your systems can produce it at all, and how much of the workforce drops out through non-disclosure and suppression.
Get the lawful basis and the privacy notice right. Ethnicity data is special category data and disability data concerns health. Both need a condition for processing and a clear privacy notice, and getting this wrong undermines the trust the declaration rate depends on.
Plan the narrative before the number. A pay gap figure without explanation invites the worst interpretation available. The organisations that handle gender pay gap reporting well publish the analysis and the actions alongside the statistic.
What the number means
One point worth being clear about internally, because it is misunderstood every year in the gender context. A pay gap is not a measure of unequal pay for equal work — that is a separate legal question under the Equality Act. A pay gap measures the difference in average pay across a whole population, and it is usually driven by the distribution of groups across roles and seniority levels.
That distinction is not a defence, and it does not make the number unimportant. It does mean the remedy usually lies in progression, recruitment and retention rather than in the payroll file. An employer that treats its gap as a pay problem will spend money without moving the figure.
Acumon supports employers on workforce data and reporting through outsourced payroll, payroll management and payroll audit work, with corporate governance support where the reporting reaches the board. If you employ 250 or more people, the declaration rate is the number to start improving now.