The cross-sectional and cross-temporal universality of nonlinear serial dependencies: evidence from world stock indices and the Taiwan stock exchange

Abstract

Numerous papers have examined the existence of nonlinearity in financial time series. However, most of this work has revolved around a relative few well-studied time series. This leaves the question of whether nonlinearity is an inherent, universal feature of financial time series or whether it is an anomaly of the time series that have been studied. This paper seeks to answer this question first by studying a variety of world financial market indices in order to determine how widespread the phenomenon of nonlinear serial dependency is, and then by studying a relatively financially isolated market, the Taiwan Stock Exchange of the l 980's, which was relatively isolated from the various economic feed back effects faced by other markets, in order to examine more closely whether non linearity appears to be an inherent feature of financial trading behaviour or a consequence of feedbacks across different markets. Our results indicate that the phenomenon of nonlinearity is universal, existing within all the markets studied and within the vast majority of individual stocks traded on the Taiwan Stock Exchange. However, upon closer examination of the extant nonlinearity, it is found that such dependencies do not exist universally across all time periods; rather, they tend to appear and disappear fairly rapidly, rendering difficult any exploitation of their existence. In summary, nonlinearity appears to be a universal phenomenon cross- sectional but not cross-temporally.