No label defined (Q4428607)

The aim of this paper is to define and examine a new, generalized Waring process (GWP) associated to a discrete distribution, and with a large spectrum of applications in areas such as accident statistics, income analysis, environmental statistics etc. Section 2 introduces the generalized Waring process as a Markov process, analogously to the case of Poisson and Pólya processes. The starting point in GWP is a process of negative binomial form, but different from a Pólya process. This process is then mixed with a BetaII distribution of the second type. This scheme is shown to lead to the generalized Waring process, having the property of being a stationary but non-homogeneous Markov process. An alternative genesis scheme referring the so-called spells model (W. L. Cresswell, P. Froggatt; 1963) is also proposed, in the framework of E. Xekalaki (1983). This scheme is shown to allow for stationary increments and to yield the validity of the Markovian property. Section 3 describes the framework for the definition of the GWP as a stationary but non-homogeneous Markov process. In this section there are derived: (a) expressions for the first two moments of the involved Markov process; (b) results on the (individual) intensity of the process; (c) transition probabilities of the process; (d) the associated forward and backward Chapman-Kolmogorov differential equations. Section 4 provides another perspective: showing that Poisson and Pólya processes are special cases of the GWP, the authors prove several known theoretical results and properties concerning these processes, their transition probabilities and the associated Chapman-Kolmogorov equations. The final Section 5 considers two further genesis schemes where the underlying mechanism of the generalized Waring distribution is interpreted within the contagion model of accident theory, studied also by the authors (2001). The new involved Waring process is proved not to be a Markov process (since it does not satisfy the Chapman-Kolmogorov equations), but the concepts and terminology used can be applied within several other application fields, ranging from economics, inventory control and insurance, to demometry, biometry and psychometry. For the entire collection see [Zbl 1012.00044].