Talk by Taehan Bae
Room: 248
SPEAKER: Taehan Bae, University of Regina, Saskatchewan, Canada
TITLE: Backward Simulation of Correlated Point Processes for Quantitative Risk Modelling
ABSTRACT: The incorporation of extremal dependence structure is one of the main concerns in quantitative risk modelling especially under abnormal situations such as market failures. The occurrences of these unexpected shocks can be naturally modelled with a multivariate point process. Some recent studies on backward construction and simulation of correlated Poisson processes show that backward methods are computationally efficient, and they allow for flexible and extremal correlation structures. In this talk, the backward approach for correlated Poisson processes will be reviewed, and an extension to the correlated negative binomial Lévy process, an appealing model for over-dispersed count data such as operational losses, will be discussed. The attainable correlation boundaries under the forward and backward approaches will also be discussed.