Few studies of absenteeism have been undertaken by economists, despite the availability of data on absenteeism which have been used in a flourishing literature in applied psychology and of a well-understood model of time allocation and labour supply.

In Discussion Paper No. 434, Research Fellow John Treble, T A Barmby and C D Orme apply techniques commonly employed in the analysis of labour supply to a study of worker absenteeism, in which they seek first to identify the various factors that influence the rate of absence for individual workers and second to measure the size of the effects on absenteeism of the factors they identify as important. Candidates for inclusion are measurable factors relating to the terms and conditions of work (including the sick-pay scheme and disciplinary system) and personal characteristics of the workers themselves. These methods are designed to handle a wide variety of institutional arrangements and are thus well suited to the exercise carried out here.

The data are drawn from the payroll and personnel records of a manufacturing company, which employs about 5,000 production workers in four separate factories, over the 18 months from July 1987. The terms and conditions of work are for the most part centralized, including the sick-pay scheme which was redesigned about five years ago with the specific aims of improving its incentive properties and reducing the firm's sick-pay bill.

The results of this analysis should assist the firm in computing the costs and benefits of amending its personnel policies. For instance, using the sick-pay rate as a bargaining counter requires a careful assessment of the possible effects on absence and turnover for the firm to know the trade-offs involved.

The firm studied operates an experience-rated sick-pay scheme, in which workers' entitlement to sick pay in the current calendar year is determined by their record of absence over the previous two years. The analysis in the paper explores the effects on absence rates of changes in the various parameters of the scheme. In deciding whether or not to attend work, each worker must therefore take into account not only the immediate financial consequences, but also the implications of absence on their employment record in the future.

In the empirical part of the paper Barmby, Treble and Orme divide the problem into two parts, by analysing the incidence of absence using a sequential logit specification and its duration using a Weibull hazard model. They find that the structure of firm's sick-pay scheme has a significant impact on the duration of absence, but not on its incidence, which appears to be determined mostly by personal characteristics. This suggests that workers do not consider their entitlement to sick pay when commencing an absence, but that they do consider it in deciding when to return to work.

Worker Absenteeism: An Analysis Using Microdata
T A Barmby, C D Orme and John G Treble

Discussion Paper No. 434, August 1990 (HR)