TIME SERIES SIMULATION WITH QUASI-MONTE CARLO METHODS

The purpose of this paper is to compare the use of quasi-Monte Carlo methods, in particular the so--called $(t,m,s)-nets$ technique, versus classical Monte Carlo approaches for the simulation of econometric time series models. Some theoretic results indicate the superiority of quasi-Monte Carlo methods. Successful applications already exist in image processing, physics, and the evaluation of finance derivatives. However, so far, quasi--Monte Carlo methods are rarely used in the field of econometrics. In this paper, we apply both traditional Monte Carlo and quasi--Monte Carlo simulation methods to time series models as they typically arise in macroeconometrics. The numerical evidence demonstrates that quasi--Monte Carlo methods outperform the traditional Monte Carlo for many time series models including non-linear and multivariate models.