On robustness

We view Sowells comments as threefold: (i) our priors were not exhaustive, (ii) our procedure does not deliver Classically standard 'size' in repeated samples, and (iii) we did not place prior point mass on the unit root. In response, we note that (i) our priors were general enough to permit researchers with disparate views to draw their own inferences, (ii) adopting a prior size adjustment does little to change our inferences, and (iii) assigning point mass to the unit root indicates clearly that strong priors are necessary to sustain this inference. We conclude that for many macroeconomic times series, the trend-stationarity inference is quite robust.