Further application of Narayan and Liu (2015) unit root model for trending time series

In this paper, we further subject the new GARCH-based unit root test for trending time series proposed by Narayan and Liu (NL) (2015) to empirical scrutiny. We utilize daily, weekly, and monthly data of 10-year bond yield for seventeen countries across the regions of America, Asia, and Europe. We find that the unit root test for sovereign bond yield data is better modeled in the presence of structural breaks, conditional heteroscedasticity, and time trend. More importantly, it may be necessary to pre-test for the existence of these statistical features when modeling with the bond yield data.