A Wiener-Hopf Monte Carlo simulation technique for Lévy processes (Q657695): Difference between revisions

Lévy processes play a central role in the theory of financial mathematics and statistics, and they are extensively used in modern insurance risk theory. In both financial and insurance settings, a key quantity of generic interest is the joint law of the current position and the running maximum of a Lévy process at a fixed time. In the present paper the authors develop a completely new and straightforward method for simulating the joint law of the position and running maximum at a fixed time of a general Lévy process with a view of an application in insurance and financial mathematics. The Wiener-Hopf decomposition for the Lévy process as well as taking advantage of recent developments in factorization techniques is given. The Wiener-Hopf Monte Carlo method for different processes, including a new family of Lévy processes called hypergeometric Lévy processes, are illustrated. In the Introduction the idea of the general Lévy process is recalled. The technique of Wiener-Hopf factorization is developed and the advantages of this technique are discussed. Also the Monte Carlo simulations of expectations are considered. The advantages of the Wiener-Hopf techniques are presented. In Section 2 the Wiener-Hopf Monte Carlo simulation technique is developed. A sequence \(e_{1}, e_{2}, \dots\) of i.i.d. exponentially distributed random variables which are equal in law to a gamma random variable with parameters \(n\) and \(\frac{n}{t}\) is considered. The process of Wiener-Hopf factorization is given. In Theorem 1 the results of the Wiener-Hopf factorization are shown. In Section 3 two large families of two-sided jump Lévy processes are considered and implemented to the theory of mathematical finance. The first type is the \(\beta\)-class of Lévy processes. The theoretical base of these processes is given. In Theorem 2 the roots of the equation \(\lambda + \psi(\theta) = 0\) are analytically identified. They are simple and occur on the imaginary axis. The second type consists of general hypergeometric Lévy processes. The theoretical base of these processes is given. The philantropy of this type of processes is discussed. In Theorem 3 the roots of the equation \(\lambda + \psi(\theta) = 0\) are analytically described. In Section 4 another alternative for extending the application of the Wiener-Hopf Monte Carlo technique to a very large class of Lévy processes is presented. In Subsection 4.1 it is shown how to build an arbitrarily large jump. In Theorem 4 Lévy processes, whose Lévy measures are written as a sum of a Lévy measure from the \(\beta\)-family or hypergeometric family plus any other measure with finite mass, is studied. In this theorem a very general class is considered. Here, the considered process is a sum of a Lévy process and a compound Poisson process. The distributional Wiener-Hopf decomposition of this process is described. In Subsection 4.2 an approximative simulation of the law of the triple \((X_{t}, \overline{X}_{t}, \underline{X}_{t})\) is considered. A slight modification of the Wiener-Hopf Monte Carlo technique is developed. In Theorem 5 it is shown how the two triples of random variables \((V(n, \lambda), J(n, \lambda), K(n, \lambda))\) and \((V(n, \lambda), \widetilde{J}(n, \lambda), \widetilde{K}(n, \lambda))\) can be considered as estimates for \((X_{g(n, \lambda)}, \overline{X}_{g(n, \lambda)}, \underline{X}_{g(n, \lambda)}).\) In Section 5 some numerical results are presented. A concrete process in the \(\beta-\)family with given parameters is considered. The results of the computations are graphically illustrated. The problem of pricing up-and-out barrier is considered. The results, obtained by an application of the Wiener-Hopf Monte Carlo method and the classical Monte Carlo method, are compared. The Wiener-Hopf method gives a better accuracy.