Coming to work ill, a cost rarely counted

Presenteeism denotes working while a health condition would justify absence. The term is sometimes used in another sense, that of prolonged presence without necessity, which creates lasting confusion in the literature.

It is the first sense that is the object of empirical work, notably in occupational health, and that produces the most cited cost estimates. Gary Johns proposed a review of it in 2010 that carefully distinguishes the two usages and lists the instruments used.

Measurement rests on questionnaires asking for the number of days worked despite a health problem, and sometimes an estimate of the associated drop in output. The questions vary from one survey to another in the reference period and in the severity of the problem mentioned.

That second estimate is the most fragile: it asks a person to assess their own relative productivity, an exercise for which no external validation exists. It also depends on what the person compares themselves with, an ordinary day or an ideal one.

Several studies conclude that the cost associated with presenteeism exceeds that of absenteeism, a result regularly cited in the management literature.

Those estimates depend entirely on the assumed drop in output, and the most solid publications present wide ranges rather than a single value.

They also leave aside an argument advanced in the opposite direction: for some minor conditions, staying active is not necessarily harmful, and research cannot say what share of presenteeism days constitutes a loss rather than a reasonable trade-off.

Studies find an association with difficulty in being replaced, with job insecurity, with the cumulative nature of undone work, and with arrangements for compensating the first days of sick leave. The question arises in different terms for self-employed workers, whose sick leave cover follows its own rules and for whom absence translates directly into lost income.

The waiting period is one of the few elements whose effect has been evaluated through quasi-experimental designs, with results showing a fall in short absences and more uncertain effects on total duration. In France, the unpaid waiting day introduced in the civil service in 2012, abolished in 2014 and reinstated in 2018, provided such a field of observation.

An organisation tracking only its absence rate may record an improvement while the situation worsens, if absences are replaced by work in a degraded state.

That case illustrates the general limit of single indicators: they direct attention towards what they count, and make invisible what substitutes for it.