THE MULTIDIMENSIONAL MARKOV-CHAIN WITH PRESPECIFIED ASYMPTOTIC MEANS, AND (AUTO-) COVARIANCES

In this paper a method is presented to construct random time series which, starting from their present values, converge to stationary time series with a priori specified mean values, standard deviations, correlations and autocorrelations. The method is applied to simulate time series of price-inflation, wage-inflation, and interest rates, whose mean values, standard deviations, correlations and autocorrelations converge to the values which are estimated from historical data. This application is a crucial part of a Decision Support System which assists management of pension funds in analysing new methods of calculating pension premiums.