Unit root tests in finance

Abstract

The knowledge of whether a time series contains a unit root or not provides guidance to determine whether the series is stationary or not. This topic is one that covers vast amount of research given to its importance in the analysis of economic and other time series data. To understand the behavior, the properties of the series and the influence of any shock that occur to the series, stationary and unit root tests were constructed. In this thesis, we first present the Box and Jenkins ARMA models, discuss the conditions for station-arity. Then, we display different method to test autocorrelation. And finally, we examine several unit root tests and discuss their power.