Why Benefit Claims Do Not Mirror the Inactivity Signal

The UK study finds its clearest relative change in sickness-related inactivity, not in benefit receipt. This chapter diagnoses that mismatch by comparing the disability-benefit and incapacity-benefit estimates with the labour-market result, then tracing the institutional steps between a health limitation and a recorded award. The key judgment is careful rather than dismissive: no robust relative benefit increase was detected, but that is not evidence that no one experienced hardship, made a claim, or faced worsening health.

Disability benefits answer a functional question

For disability benefits, the study examines reported receipt of Personal Independence Payment or Disability Living Allowance. The largest estimate appears in the pooled Model 1 specification: 2.2 percentage points. Its confidence interval includes zero, however, and the reported p-value is 0.11. The other principal estimates are also not statistically robust: 1.4 percentage points in Model 1 with fixed effects, 1.6 points in the pooled aligned-period model, and 0.4 points in its fixed-effects version.

The first conclusion should therefore be stated precisely. The study does not find a statistically robust disproportionate increase in PIP/DLA receipt among clinically vulnerable respondents relative to low-risk respondents. It does not find that the estimate is exactly zero, that no vulnerable respondent received support, or that the aggregate rise in claims was imaginary. The result says that, under this outcome definition, sample, comparison, and uncertainty, the evidence is not strong enough to establish a relative concentration comparable to the inactivity finding.

The programme’s purpose helps explain why. PIP and DLA are based on functional limitations and the additional costs associated with disability. They are not conditioned on being outside employment. The outcome can therefore rise among people who remain in work, while a person who leaves the labour force because available work no longer fits their condition may not receive an award. Employment status is relevant context, but it is not the decision rule.

That difference changes what a “matching” result would even mean. If clinically vulnerable respondents became more inactive, a PIP/DLA effect would require the same relative difference to pass through a functional-limitation test and the process of applying for and receiving that support. A person’s inability to sustain a particular job is not automatically equivalent to meeting the conditions for help with disability-related costs. Nor does receiving such help establish that the person has withdrawn from work. The two outcomes may be connected in individual cases without moving in parallel at population level.

The distinction also prevents a common error in interpreting non-significant estimates. A confidence interval that includes zero means the data are compatible with no relative increase and with some positive effects that the study cannot estimate precisely. The smaller benefit samples make that uncertainty more consequential. The appropriate phrase is “no statistically robust relative increase detected,” not “there was no increase.” The absence of a clear group difference is an evidential boundary, not a statement about every person’s experience.

The 2.2-point estimate illustrates the difference between a point estimate and a conclusion. It is the largest of the principal disability-benefit estimates, so it cannot be described as evidence that the outcome moved in the opposite direction. But its p-value of 0.11 means that the uncertainty is too large for the study to treat the relative increase as established at the conventional levels used in the paper.