Further Topics in Time Series Analysis
Usually macroeconomic time series show a tendency of growing over time and thus can be characterized as trending. This can be clearly seen in the real GDP performance, which is the primary indicator for tracking a domestic economy. However, if we were to take snapshots of an economy at different points in time, no two photos would look alike. The reason for this has concerned economists for some time. It appears that there exist seemingly random fluctuations around the growth trend. Indeed as Nelson wrote:
Moreover, there seems to be a connection between this phenomenon and the existence of the business cycles. Economists refer to this “noisiness” or movements about the trend as business cycles or business fluctuations. The existence of cyclical movements in macroeconomic data has always been an important issue in the business cycle research since Burns and Mitchell (1946) and different methodologies have been derived for its identification. One of the most popular approaches is to define the cyclical fluctuations as the difference from the trend line which is a deterministic function of time. The problem with time deterministic approach according to Zarnowitz and Ozyildirim (2001: 16-17) is that:
This chapter will review some fundamental contributions, based on the papers of Beveridge and Nelson (BN) (1981), Nelson and Plosser (NP) (1982), and Hodrick and Prescott (HP) (1980, 1997).