Basic Properties of Stationary First-Order Autoregressive Processes and Random Walks

Abstract:     
This paper compares the basic properties of stationary autoregressive processes and random walks with special regard to their implications to unit root testing. In particular, it aims to answer three basic but important questions. Firstly: 'What do a constant term (drift) and a linear time trend mean in stationary first-order autoregressive equations and in random walks?'. Secondly: 'Can bounded series be modelled by random walk processes?'. Thirdly: 'Is there any difference between having a unit root in the level or in the logarithm of an economic time series?'.

Keywords: First-order autoregressive process, unit root

JEL Classifications: C22

Working Paper Series