Multiple time series modelling and tests of market efficiency

Levy, Edmond (1987) Multiple time series modelling and tests of market efficiency. University of Southampton, Doctoral Thesis.

Abstract

This thesis is concerned with the statistical examination of the market efficiency hypothesis. It is argued that when the information set considered is confined to the current and past history of market forecasts and the underlying variable, multiple time series modelling properly conducted offers an intuitively appealing framework which results in test procedures which are the most efficient amongst procedures in the same class.Chapter 1 presents an overview of the thesis. Chapter 2 examines market efficiency in the context of expectations being formed in accordance with rational expectations. The use of regression methods is examined in Chapter 3. There the relationship between these methods and variance bounds procedures is considered. Observations on shortcomings of the regression approach leads to the advancement of a bivariate time series methodology in Chapter 4. The commonly applied first difference filter for nonstationary series is shown to be inappropriate. Admissible structures of nonstationarity is present in Chapter 5. There it is argued that the relation between spot and forward prices, under the efficiency hypothesis, form a co-integrated pair. Various identification procedures for multivariate ARMA model-building are discussed and a Wald-test procedure is supplied for testing efficiency under co-integration. Chapter 6 focuses on the structure of nonstationarity of multiple time series through results relating to the theory of lambda-matrices. An alternative description of the restrictions general exact expectations relations impose on bivariate AR parameters is also presented. Chapter 7 illustrates the issues raised in the thesis by examining the efficiency of the foreign exchange market. (D80996)

This record has no associated files available for download.

Contributors

Download statistics