Abstract
The study investigates the predictability of interest rates in South Africa with a focus on 14 stochastic modes of three stochastic process classes. Several research studies have focused on the efficiency of stochastic interest rate models to fit and forecast interest rates so as to gauge their predictive yields, but none of these researchers have, to the best of my knowledge, applied and compared three stochastic processes in the South African economy. With this modelling gap in mind, the aim of this research paper is to provide empirical analysis in this research area, where the performance of the 14 stochastic models, driven by the Wiener, Fractional and compound Poisson processes in terms of modelling the South African short-term interest rates are compared. The three processes can capture thick heavy tails, memory dependency and jumps. To do this, the paper considers each process style facts concerning its model: firstly, that the data follows a geometric Brownian motion (GBM) and secondly, that it is normally distributed.
M.Com. (Financial Economics)