Trend Interpolation and the Persistence of Fluctuations in U.S. GNP

Abstract

Two recent papers by Cochrane [1988] and Cogley [1990] examined the persistence of fluctuations in real GNP. Both papers find little persistence in real U.S. per capita GNP series covering the periods 1869–1986 (Cochrane) and 1871–1985 (Cogley). Campbell and Mankiw [1987], however, argue that the persistence of U.S. post World War II GNP is substantial. Furthermore, comparing the persistence of real output series for various countries in the prewar and postwar era, DeLong and Summers [1988] report that first, prewar output series are in general less persistent than postwar series, and second, U.S. prewar GNP exhibits much lower persistence than comparable prewar output series for Sweden and the United Kingdom. These puzzling findings raise a natural question: Why is the persistence of U.S. prewar GNP fluctuations so unusually low?

We would like to thank our discussant John Chipman for helpful remarks on an earlier version of this note.