4.1 Introduction

In this chapter we will discuss reasons that a time series may be non-stationary and methods to convert it into a stationary series. We will cover three primary reasons a time series may be non-stationary:

1. Presence of a trend

2. Presence of a unit root

3. Presence of seasonal fluctuations

We will also cover three methods to convert a non-stationary time series into a stationary one, loosely paralleling the three causes above:

1. Detrending

2. Differencing

3. Seasonal adjustment

In the context of differencing we will also introduce a useful convention in the form of the backshift operator. We will conclude the chapter with a brief discussion of non-stationarity stemming from increasing variance and methods to correct it.