Evaluation of Tweedie exponential dispersion model densities by Fourier inversion
- 416 Downloads
The Tweedie family of distributions is a family of exponential dispersion models with power variance functions V(μ)=μ p for \(p\not\in(0,1)\) . These distributions do not generally have density functions that can be written in closed form. However, they have simple moment generating functions, so the densities can be evaluated numerically by Fourier inversion of the characteristic functions. This paper develops numerical methods to make this inversion fast and accurate. Acceleration techniques are used to handle oscillating integrands. A range of analytic results are used to ensure convergent computations and to reduce the complexity of the parameter space. The Fourier inversion method is compared to a series evaluation method and the two methods are found to be complementary in that they perform well in different regions of the parameter space.
KeywordsCompound Poisson distribution Generalized linear models Numerical integration Numerical acceleration Power variance function
Unable to display preview. Download preview PDF.
- Abramowitz, M., Stegun, I.A. (eds.): A Handbook of Mathematical Functions. Dover, New York (1965) Google Scholar
- Dunn, P.K.: Likelihood-based inference for Tweedie exponential dispersion models. Unpublished Ph.D. Thesis, University of Queensland (2001) Google Scholar
- Johnson, N.L., Kotz, S.: Continuous Univariate Distributions—I. Houghton Mifflin, Boston (1970) Google Scholar
- Jørgensen, B., Paes de Souza, M.C.: Fitting Tweedie’s compound Poisson model to insurance claims data. Scand. Actuar. J. 1, 69–93 (1994) Google Scholar
- R Development Core Team: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna (2005). http://www.R-project.org Google Scholar
- Seigel, A.F.: Modelling data containing exact zeros using zero degrees of freedom. J. R. Stat. Soc. B 47, 267–271 (1985) Google Scholar