Testing and modeling nonlinearity of the stocks returns

Abstract

Most of the earlier work on time series analysis was based on the assumption of linearity of the data generating processes. However, there are occasions when the theory and data suggest that linear models are unsatisfactory. In this paper we have applied a few nonlinearity tests to the world stocks prices. It is found that all of the six stock prices considered in this paper are nonstationary. The stock's returns, however, are all stationary. Nonlinearity tests revel that not all of the returns series are linear. Hence modeling nonlinearity is eminent.