Invited Comment on: "The Uncertain Unit Root in Real GNP: A Re-examination"

With the objective to settle the often conflicting and inconclusive extant debate on whether the observed secular growth can be characterized by a deterministic or stochastic trend, Darné comes up with an alternative method – the outlier detection methodology and studies the persistence properties of US real GNP for the period 1869–1993. Given that the presence of outliers in a time series – unless detected and adjusted – often render wrong impression on the persistence properties and their implications for long-run growth, Darné's contribution in this regard is significant. Outliers of varied types assume varying persistence profiles, therefore detection of them in the GNP growth is a pre-requisite for characterizing the series in a stochastic or deterministic setting. To further intuition on the exact nature of persistence, I applied fractional integration test of the four real GNP series and found that outlier adjusted GNP series are mean-reverting and are more persistent than original. The rate of convergence of stochastic shocks to long-run mean have important implications for growth dynamics.