Pay, Technology and the Cost of Worker Absence

Conventional studies of absenteeism concentrate on labour supply aspects. In this paper we analyse records of worker behaviour which enable us to investigate whether or not demand side effects exist. Are there differences in contract structure and behaviour across firms that can help explain the observed pattern of absence rates? The theoretical background to the study is the model in Coles and Treble (1996). Data are from the Enquête sur le coût de la main-d’œuvre et la structure des salaires, 1992, and Réponse, two linked cross-section surveys of French establishments, in which managers are asked about their firms as well as a sample of workers in each firm. We investigate the relationship between rates of pay and rates of absence, and how this relationship differs between firms with different technologies. It is to be expected that firms generally would pay no higher wages for a less reliable workforce than for a more reliable one, and we confirm this expectation. More subtle is the rate at which remuneration should fall with increased unreliability. We claim that just-intime technology implies that absence will be more expensive for firms adopting it. The loss of productivity when absence occurs will be greater for such firms than for others, and the wage premium for reliability should thus be higher for such firms. Our results indicate support for the main hypothesis of the model. The estimated costs of absence in firms using just-in-time are higher than in firms who do not. However, two separate ways of identifying the cost yield quite different quantitative estimates ACKNOWLEDGEMENTS: We are grateful to the British Council and the French Ministry of Foreign Affairs for financial support under the Alliance programme, and to Thomas Coutrot (DARES) and INSEE for permitting us access to the data. Coles and Treble are also grateful to ERMES at Université Panthéon-Assas, Paris II for their hospitality and assistance. Earlier drafts of the paper have been presented at seminars at Oxford, Manchester, CREST (Paris), the 2001 EEEG Conference (Leicester) and the 2001 CAED Conference (Aarhus). Comments from participants are gratefully acknowledged, especially those from Martyn Andrews, Francis Kramarz and JeanMarc Robin.