Spell Durations and the Impact of Censoring
Posted: 11 Apr 2001
A grouped hazard approach for analysing multiple-spell durations subject to censoring is applied to spells of absence from the workplace. We follow Barmby, Orme and Treble's (1991) procedure for dealing with unobserved heterogeneity, but argue that their treatment of the observed discrete data, and the inherent censoring, is inappropriate and could lead to significant overestimation of duration dependence.
Keywords: spell durations, censoring, Monte Carlo analysis
JEL Classification: C15, C41, J29
Suggested Citation: Suggested Citation