Title: Heteroskedasticity-robust unit root testing for panels with linear trends
Authors: Yabibal Walle - Georg-August-University Goettingen (Germany) [presenting]
Helmut Herwartz - Georg-August-University Goettingen (Germany)
Abstract: Volatilities of most economic time series display significant fluctuations over time. Standard (panel) unit root tests are not robust in such cases and, hence, could lead to misleading conclusions. Consequently, recent studies have proposed heteroskedasticity-robust panel unit root tests. A major limitation of these tests, however, is that they work well only as long as the data is not trending. We propose a modification of the detrending procedure previously suggested that enables heteroskedasticity-robust panel unit root tests to work well in the presence of linear trends and innovation variance breaks. Evidence from Monte Carlo simulations shows satisfactory small sample performance of the tests under the new detrending strategy. As an empirical illustration, the evidence on the stationarity of GDP per capita is re-examined. Results show that GDP per capita is best characterized as a unit root process.