Reconsidering ‘trends and random walks in macroeconomic time series’