Testing for a Unit Root and Structural Break Using SAS Software
This chapter employs SAS software to test for a unit root and structural break in a time series. We demonstrate, using the U.S. consumer price index (CPI) as a case study, how to test for a unit root using the Dickey-Fuller test; the Phillips-Perron test; and the Kwiatkowski, Phillips, Schmidt, and Shin test. Testing for a unit root determines whether the data series is stationary and if the series has a constant mean and variance over time. A basic ordinary least squares (OLS) analysis assumes that data is stationary. If the data is nonstationary, the OLS results are not reliable.
Another important feature of applied econometric analysis is to identify whether a time series contains a structural break. Furthermore, if an economic, financial, or business data series contains a structural break, then the series acts differently during different periods. Therefore, any estimated relationship in one time period does not work in another time period. Three different approaches are utilized to identify a structural break in the U.S. home prices: (1) a dummy-variable method, (2) the Chow test, and (3) the state-space approach. Last, the application of the Hodrick-Prescott (HP) filter is shown using the U.S. corporate profits series.
TESTING A UNIT ROOT IN A TIME SERIES: A CASE STUDY OF THE U.S. CPI
Is there an underlying pattern to a time series that could mislead analysts when they attempt to identify the behavior ...