Filed under definitions

On-call hours and the question of what counts as a worked hour

Time spent available but not working sits awkwardly in every statistical system that tries to record it, and national estimates disagree for reasons that are largely definitional.

Statistical systems are built to count hours that are worked. Arrangements in which a person is available for work that may or may not materialise do not fit that frame, and the resulting estimates vary depending on who is asked and what they are asked.

Two ways of counting, two answers

The United Kingdom offers an unusually clear illustration. The Office for National Statistics produces estimates of contracts that do not specify a minimum number of hours from two separate sources. One is the Labour Force Survey, a household survey in which people describe their own employment arrangement. The other is a survey of employers, which counts contracts on the books during a reference period.

The two produce different totals, and ONS has published on why. Household respondents may not know what their contract specifies, or may not describe it in the terms the survey uses. The employer survey counts contracts rather than people, so a person holding two such arrangements appears twice, and a contract that exists but generated no work during the reference period may still be counted. Neither figure is wrong. They measure different objects, and the difference between them is a definitional artefact rather than a discrepancy to be resolved.

The international picture

Work of this kind appears in the International Labour Organization's treatment of non-standard employment, in research led by Janine Berg and colleagues, which groups on-call and casual arrangements alongside temporary and part-time work. The comparative difficulty is that national legal categories do not align. An arrangement classified as casual employment in one jurisdiction may fall under a fixed-term contract or self-employment in another, with different reporting obligations attached. Cross-country tables in this area rest on mappings between legal regimes, and those mappings are documented choices rather than neutral facts.

Availability as unmeasured time

Susan Lambert's research on what is often called just-in-time scheduling describes a related gap. When staffing is adjusted close to the point of need, some of the burden shifts to time that is neither worked nor free: a period held open in case a call comes. Standard instruments have no field for it. It is not working time, so it does not enter hours worked. It is not leisure in the ordinary sense either, but a diary asking for a primary activity will usually record whatever else was happening.

The consequence is systematic rather than random. Where availability is unpaid and unrecorded, the statistical record understates the extent to which a schedule constrains a week, while recording hours and earnings accurately. Two arrangements with identical hours and identical pay can differ substantially in how much of a week they hold open, and nothing in the published aggregates distinguishes them.

Reading the estimates

Three habits help. First, check whether an estimate counts people or contracts, since the two are not interchangeable. Second, check whether the source is a household survey or an employer survey, because the direction of likely error differs. Third, treat cross-national comparisons as comparisons of harmonised categories rather than of underlying arrangements.

Researchers working in this area are generally explicit about these limits in their methodological notes. The compression tends to happen later, when a single number is extracted from a technical report and repeated without the definition that produced it.