Semi-Markov models for human-resource modelling

For human-resource modelling, models of wastage – such as the lognormal model – ignore the variations in leaving rates between grades, while models of the grade structure – such as the Markov model – ignore duration dependence in the rates of movement to other grades or leaving. The semi-Markov approach allows us to combine a flexible means of modelling duration-dependent stay in a grade with a transition matrix of movements between grades. We may therefore describe the whole system and predict its future development. We discuss the use of semi-Markov models for human resource management and provide a means of estimating the parameters. The use of the model to assess promotion patterns is also considered.