An Empirical Model of Work Attendance

Abstract

T WO recent studies have indicated on any given day about 3% to 4% of all workers do not report to their jobs. The Bureau of National Affairs (BNA) has reported absence data for a sample of several hundred establishments since January 1974. The median absence rate for the entire sample was 3.4% in 1974, 3.0% in 1975 and 1976, 2.8% in 1977, and 2.9% in 1978.1 Absence rates vary widely across establishments; in December 1978 they ranged from 0 to 23.0%. Recent changes in the Current Population Survey permit the computation of work time lost to absenteeism with household data. In May 1976 Hedges (1977) reported full time workers missed about 3.5% of scheduled hours for health and personal reasons. It is interesting to note in comparison in the same month (1) less than one-third of 1% of scheduled hours were lost to strikes, and (2) 3.4% of the labor force were out of work because they had lost their previous job. Economists have paid little attention to absenteeism despite the sizable number of man-hours involved and the probable impacts on productivity and income distribution. Most of the previous research on work attendance has been done by applied psychologists, who generally argue, according to a recent survey article by Steers and Rhodes (1978), that job dissatisfaction represents the primary cause of absenteeism.2 Another widely held view is absenteeism results from inadequate managerial concern for the problem. Countless authors of articles in personnel and business journals have argued any firm can control absenteeism by keeping adequate records, establishing guidelines for permissible absences, or rewarding workers who attend regularly. Given efficient markets for entrepreneurial expertise, it is unclear why such measures have not already been taken. The purpose of this paper is to suggest an alternate interpretation of absenteeism and to develop an empirical model to test various hypotheses about its incidence. Absences result when an individual decides to engage in nonwork activity throughout a scheduled work period. It will be argued below utility increments obtained by not reporting are likely to vary across individuals and the cost of absenteeism will not be identical across employers. Employers can reduce absenteeism in three ways. One option is to make it more costly to employees, e.g., through decisions regarding promotions, merit wage increases, dismissals, and the availability of sick leave and attendance bonuses. Another is to reduce the worker's demand for absences by making schedules more flexible. In cases where substitute workers are available at no extra cost, absenteeism is costly only when it is unexpected. Both the firm and the worker will then be better off if they agree to adjust the worker's schedule in advance. A final