Stochastic unit roots modelling of stock price indices

Abstract

Recently developed methodology to allow the possibility of a stochastic unit root process as an alternative to a fixed parameter unit root model is applied to six national indices of stock market prices. Evidence supporting the stochastic unit root hypothesis is found. However, the implementation of this model generally leads to only very minuscule gains in the prediction of daily prices, except in the case of the Hang Seng index, where the predictive gain is somewhat larger.