The stochastic behavior of stock indices: a test of long memory in the Kuala Lumpur Stock Exchange

Abstract

Characterization of the stochastic process of stock market indices is vital in understanding the behavior of stock prices. Conventional share valuation models are subject to the nature of inter-temporal dependence between current and past movements. This paper studies the stochastic process of four KLsE indices (Composite Index,Industrial Index, Finatice Index, and Property Index). In addition to the standard unit root tests, the ARFlMA (Autoregressive Fractionally Integrated Moving Average) model which belongs to the class of long memory process is applied in the empirical analysis. The findings indicate that while the level of the indices is nonstationary, its growth rate exhibits stationary properties. Long memory is not supported for the level of the indices but is evidenced in its monthly growth rate. The growth rate of the indices can therefore be characterized as a long memory process that is mean reverting.