Modelling Markovian Migration in Finance and Medicine

For a time-continuous finite-state Markov process, constant transition intensities simplify the calculation of migration matrices. For two applications I investigate the adequacy of the model, first, rating migrations in finance, and second, cancer survival in medicine. In finance, ratings have usually more than one transient state. In contrast, in medicine “alive” as single transient state is common. Still, nonparametric methodology, like the NelsonAalen estimate, or likelihood ratio testing is very similar for both situations. The re-analysis of several data shows that the constant intensity is usually to be rejected, however alternative models need arguments of the specific field. Nonparametric smoothing may an intermediate step towards such models.