The pay gap

The disability pay gap is real, persistent — and shaped by the fact that who is employed at all is not a random subset of disabled people.

What the measure captures

The disability pay gap is conventionally expressed as the difference between the median hourly pay of disabled employees and the median hourly pay of non-disabled employees. The Office for National Statistics publishes this figure from the Annual Population Survey, which provides large enough samples to support pay analysis broken down by disability status. The most recent releases show a gap in the region of 15 per cent, with disabled employees earning less per hour at the median — though readers should check the specific bulletin and reference year, since the APS-derived figure is revised and updated on a rolling basis.

Using the median rather than the mean is deliberate and standard: pay distributions are skewed, and the median is resistant to the distortion caused by a small number of very high earners. But the choice of measure still has consequences. The median gap at the top of the pay distribution and the median gap at the bottom can move in opposite directions in the same year, and a single headline figure obscures both.

The composition problem

The pay gap is not a pure measure of what disabled workers are paid relative to non-disabled workers doing the same jobs. It is a measure across all employed people in each group, which means it reflects the kinds of jobs the two groups are in, the hours they work, and their levels of qualification and experience — as well as any within-job pay differential. Because the employment gap is large, the population of disabled people who are in work is a selected group, and there is no straightforward way to know in which direction that selection biases the pay figure.

Disabled people in work are, on average, older and more experienced than the disabled working-age population as a whole — because younger disabled people face higher barriers to entering employment at all. That tends to compress the observed pay gap. At the same time, disabled workers are overrepresented in part-time work, which is associated with lower hourly pay, and in certain sectors. Decomposing the gap into these components requires regression methods, and published decompositions are infrequent.

The pay gap and the employment gap therefore need to be read together: a narrowing pay gap achieved by a narrowing employment gap requires different interpretation than either figure read alone.

Threshold is an independent publication about disability statistics and governance. It is not an advice, advocacy or support service, and nothing here should be read as guidance for any individual's circumstances.