Abstract: Inferences based on time series heavily depend on the integration of variables of interest. This paper aims to examine the integrated property of gross domestic products (GDP) in China. Data were quarterly time series and covered the period from 1993 to 2016. Data were seasonally adjusted and in logarithms. Nominal GDP comprised two types: one was measured in current values. The other was measured in accumulated statistics. Alternative unit root techniques have the dissimilar power, size, and finite-sample property; hence, differing inferences often arise. The Phillips-Perron test indicated one unit root for the two series. However, the augmented Dickey-Fuller test and the Elliott-Rothenberg-Stock point-optimal test suggested two unit roots. The Perron test suggested that these two series variables were trend-stationary, with a break date occurred around Quarter 1, 2008, which might have increased the order of series variables suggested by two conventional tests. Overall, two quarterly GDP series in nominal terms contained a unit root. Tests indicated consistent results for GDP in current values and in accumulated statistics, which implies that they could substitute each other in future analyses. The paper suggests that at least two conventional unit root tests along with break-date tests might lead to right unit root inferences.
Keywords: Unit root, structural break, trend, time series stationarity.
Title: The Integrated Properties of Gross Domestic Products in Current Values and in Accumulated Statistics in China
Author: Gaolu Zou
International Journal of Social Science and Humanities Research
ISSN 2348-3156 (Print), ISSN 2348-3164 (online)
Research Publish Journals
