Comment on 'Adaptive estimation in time series regression models' by D.G. Steigerwald