Saturday, November 13, 2010

Bang goes the theory: menstruation and absenteeism

This paper casts doubt on the well known paper, recently published in the AEA Applied Economics that attributed much of the absenteeism of female workers to menstruation basically as it followed a 28 day cycle and was absent in older (post-menopausal) women. Its important not to be too persuaded by one's own theory, no matter how clever.

Does menstruation explain gender gaps in worker absenteeism
Jonah Rockoff, Mariesa Herrman
Ichino and Moretti (2009) find that menstruation may contribute to gender gaps in absenteeism and earnings, based on evidence that absences of young female Italian bank employees follow a 28-day cycle. We analyze absenteeism of teachers and find no evidence of increased female absenteeism on a 28-day cycle. We also show that the evidence of 28-day cycles in the Italian data is not robust to the correction of coding errors or small changes in specification. We show that five day workweeks can cause misleading group differences in absence hazards at multiples of 7, including 28 days.
NBER 16523

No comments: