Abstract

The authors report on an investigation into Markov-models of demographic changes. Panel data are given on the number of six-year transitions between household statuses. The problem being dealt with is how to transfer these data to single year transition probabilities. First, a short introduction is given to stochastic and intensity matrices, and to necessary and sufficient conditions for the embeddability of stochastic matrices as the esponential of an intensity matrix and the identification (uniqueness) of this intensituy matrix. Second, the way the problem has been solved is described. The procedures used are: the determination of the 'nearest' intensity matrix. The determination of the maximum likelihood age-independent resp. age-dependent single year transition probabilities. the results are compared and discussed.