Determining the Impact of Different Forms of Stationarity on Financial Time Series Analysis

This chapter aims to demonstrate how an experimental research strategy was followed to prove that the differencing technique that econometricians and financial data analysts use to render data stationary does matter. The first difference approach has become popular mainly because of the work of (Nelson and Plosser, Journal of Monetary Economics 10:139–162, 1982), who argued that many macroeconomic time series are difference stationary and not trend stationary. However, since most time series data are in fact non-stationary in its level form (considerable amount of time series data are actually fractionally integrated), the econometrician and data analyst are required to make the data stationary before embarking on any econometric analysis in order to avoid spurious results. Although there are several different ways to render a non-stationary time series stationary, few econometricians and data analysts look past the first differencing and log-differencing methods when rendering data stationary through a data transformation process. The end result of this careless approach is that they often over difference the data. When over-differenced, the data is stripped of its original underlying statistical properties resulting in a loss of important information. In doing so, many research studies do in fact arrive at inferior results because of over-differencing the data. To investigate the implications of this practice, this research had to test existing methodologies for their ability to properly render time series data stationary. The research approach was therefore deductive in nature, following a positivist research philosophy.