Robust Inference for Near-Unit Root Processes with Time-Varying Error Variances

Heteroskedasticity-Robust Unit Root Testing for Trending Panels

“…Results in the upper panel of this table show that, under homoskedasticity, the recursive detrending scheme in Demetrescu and Hanck (2014) leaves the pivotalness of heteroskedsticity-robust tests unaffected. With respect to rejection frequencies under the alternative hypothesis, it can be seen that using estimated However, this power loss vanishes with increasing T .…”

“…Demetrescu and Hanck (2014) show that, under homoskedasticity, ∆y t could also be centered by means of forward demeaning instead of (8). In the presence of heteroskedasticity, both full sample centering and forward demeaning affect the pivotalness of even the heteroskedasticityrobust tests t HS and t DH and, hence, invoke marked size distortions (see Demetrescu and Hanck (2014) for rigorous arguments on this issue). As forward demeaning additionally leads to relatively large power losses in comparison with full sample centering, the test proposed in this work relies on full sample demeaning.…”

“…Unlike in Herwartz and Siedenburg (2008) and Demetrescu and Hanck (2014), where the White-type covariance estimator is applied, the more complicated form ofỹ ′ t−1 ∆y * t invokes the following representation of the variance of the numerator in (12):…”

“…Moreover, the bias-correction becomes highly complicated with the presence of variance breaks. The detrending procedures in Breitung and Das (2005) and Demetrescu and Hanck (2014) …”

“…The construction of the test is simple. We begin by detrending the data according to the method suggested in Demetrescu and Hanck (2014), and trace the effects of the detrending scheme on the (detrended) integrated level data. The drift term is estimated as the unconditional mean of first-differenced series.…”