Poisson compounding of dependent random variables: A stochastic model for total claim costs

  • Authors:
  • J. E. Angus

  • Affiliations:
  • Department of Mathematics The Claremont Graduate School, 143 E. Tenth Street Claremont, CA 91711, U.S.A.

  • Venue:
  • Mathematical and Computer Modelling: An International Journal
  • Year:
  • 1993

Quantified Score

Hi-index 0.98

Visualization

Abstract

The Compound Poisson process is a useful model for describing total claim costs in the insurance industry. In this investigation, the compound Poisson model is modified to allow dependence among the compounded variables in an effort to more accurately model situations in which successive claim awards are correlated and/or form a nonstationary process. Specifically, the sequence is assumed to follow an ARIMA(p,d,q) model. This type of model is illustrated with the use of actual asbestosis claim cost data collected from Naval shipyards. Some asymptotic theory (namely a central limit theorem) is developed for the case in which the Poisson process mean approaches infinity.