Models and Inference for Duration Data: A Survey

J.P. Florens
Toulouse Univeristy, Duke University

Nov 30 1989

This paper presents the main classes of models used for the treatment of duration data. We cover both the elementary (one spell, one state) case and the general (multispell, multistate) case. This last kind of specification is analysed through the univariate and multivariate counting processes models. Parametric, non parametric, or semi parametric inferences in these models are explored.


Duration data, couting processes, event history data, hazard and stochastic intensity, heterogeneity, non parametric estimation, cox partial likelihood


PDF icon 1989-05.pdf