Advances of matrix–analytic methods in risk modelling

Oscar Peralta Gutierrez

Research output: Book/ReportPh.D. thesis

101 Downloads (Pure)

Abstract

This work is concerned with the study of matrix–analytic methods with novel applications to the area of risk theory.
First, we review some topics of Applied Probability such as phase–type distributions, matrix–exponential distributions, Markovian arrival processes, Rational arrival processes, fluid flow processes and risk models. With these tools in hand, we propose a method to approximate the probability of ruin of any Cramér–Lundberg process using the theory of phase–type distributions, providing an error bound for such an approximation. With the goal of studying risk models with dependencies, we construct a class of bivariate distributions with given phase–type–distributed marginals and given Pearson’s correlation coefficient, which are later used to define different kinds of dependent Sparre–Andersen processes. Later on, we give an explicit formula for the probability of Parisian and cumulative Parisian ruin for a class of risk processes which are based on the theory of fluid flow processes. Next, we study some excursion properties of spectrally negative Lévy processes whenever they are inspected at an independent matrix–exponential time. Finally, inspired by the generalisation of the Markovian arrival process to the Rational arrival process, we construct a novel generalisation of the fluid flow process without Brownian components and study its first passage probabilities.
Original languageEnglish
PublisherDTU Compute
Number of pages215
Publication statusPublished - 2019

Cite this

Peralta Gutierrez, O. (2019). Advances of matrix–analytic methods in risk modelling. DTU Compute.
Peralta Gutierrez, Oscar. / Advances of matrix–analytic methods in risk modelling. DTU Compute, 2019. 215 p.
@phdthesis{bd13a6f893a74c7d9fe01fcc95deee66,
title = "Advances of matrix–analytic methods in risk modelling",
abstract = "This work is concerned with the study of matrix–analytic methods with novel applications to the area of risk theory. First, we review some topics of Applied Probability such as phase–type distributions, matrix–exponential distributions, Markovian arrival processes, Rational arrival processes, fluid flow processes and risk models. With these tools in hand, we propose a method to approximate the probability of ruin of any Cram{\'e}r–Lundberg process using the theory of phase–type distributions, providing an error bound for such an approximation. With the goal of studying risk models with dependencies, we construct a class of bivariate distributions with given phase–type–distributed marginals and given Pearson’s correlation coefficient, which are later used to define different kinds of dependent Sparre–Andersen processes. Later on, we give an explicit formula for the probability of Parisian and cumulative Parisian ruin for a class of risk processes which are based on the theory of fluid flow processes. Next, we study some excursion properties of spectrally negative L{\'e}vy processes whenever they are inspected at an independent matrix–exponential time. Finally, inspired by the generalisation of the Markovian arrival process to the Rational arrival process, we construct a novel generalisation of the fluid flow process without Brownian components and study its first passage probabilities.",
author = "{Peralta Gutierrez}, Oscar",
year = "2019",
language = "English",
publisher = "DTU Compute",

}

Peralta Gutierrez, O 2019, Advances of matrix–analytic methods in risk modelling. DTU Compute.

Advances of matrix–analytic methods in risk modelling. / Peralta Gutierrez, Oscar.

DTU Compute, 2019. 215 p.

Research output: Book/ReportPh.D. thesis

TY - BOOK

T1 - Advances of matrix–analytic methods in risk modelling

AU - Peralta Gutierrez, Oscar

PY - 2019

Y1 - 2019

N2 - This work is concerned with the study of matrix–analytic methods with novel applications to the area of risk theory. First, we review some topics of Applied Probability such as phase–type distributions, matrix–exponential distributions, Markovian arrival processes, Rational arrival processes, fluid flow processes and risk models. With these tools in hand, we propose a method to approximate the probability of ruin of any Cramér–Lundberg process using the theory of phase–type distributions, providing an error bound for such an approximation. With the goal of studying risk models with dependencies, we construct a class of bivariate distributions with given phase–type–distributed marginals and given Pearson’s correlation coefficient, which are later used to define different kinds of dependent Sparre–Andersen processes. Later on, we give an explicit formula for the probability of Parisian and cumulative Parisian ruin for a class of risk processes which are based on the theory of fluid flow processes. Next, we study some excursion properties of spectrally negative Lévy processes whenever they are inspected at an independent matrix–exponential time. Finally, inspired by the generalisation of the Markovian arrival process to the Rational arrival process, we construct a novel generalisation of the fluid flow process without Brownian components and study its first passage probabilities.

AB - This work is concerned with the study of matrix–analytic methods with novel applications to the area of risk theory. First, we review some topics of Applied Probability such as phase–type distributions, matrix–exponential distributions, Markovian arrival processes, Rational arrival processes, fluid flow processes and risk models. With these tools in hand, we propose a method to approximate the probability of ruin of any Cramér–Lundberg process using the theory of phase–type distributions, providing an error bound for such an approximation. With the goal of studying risk models with dependencies, we construct a class of bivariate distributions with given phase–type–distributed marginals and given Pearson’s correlation coefficient, which are later used to define different kinds of dependent Sparre–Andersen processes. Later on, we give an explicit formula for the probability of Parisian and cumulative Parisian ruin for a class of risk processes which are based on the theory of fluid flow processes. Next, we study some excursion properties of spectrally negative Lévy processes whenever they are inspected at an independent matrix–exponential time. Finally, inspired by the generalisation of the Markovian arrival process to the Rational arrival process, we construct a novel generalisation of the fluid flow process without Brownian components and study its first passage probabilities.

M3 - Ph.D. thesis

BT - Advances of matrix–analytic methods in risk modelling

PB - DTU Compute

ER -

Peralta Gutierrez O. Advances of matrix–analytic methods in risk modelling. DTU Compute, 2019. 215 p.