a b s t r a c t
Beveridge and Nelson [Beveridge, Stephen, Nelson, Charles R., 1981. A new approach to decomposition
of economic time series into permanent and transitory components with particular attention to
measurement of the `business cycle'. Journal of Monetary Economics 7, 151174] proposed that the
long-run forecast is a measure of trend for time series such as GDP that do not follow a deterministic
path in the long run. They showed that if the series is stationary in first differences, then the estimated
trend is a random walk with drift that accounts for growth, and the cycle is stationary. In contrast to
linear de-trending, the smoother of Hodrick and Prescott (1981) and Hodrick and Prescott [Hodrick,
Robert, Prescott, Edward C., 1997. Post-war US business cycles: An empirical investigation. Journal of
Money Credit and Banking 29 (1), 116] and the unobserved components model of Harvey, [Harvey,
A.C., 1985. Trends and cycles in macroeconomic time series. Journal of Business and Economic Statistics
3, 216227]. Watson [Watson, Mark W., 1986. Univariate detrending methods with stochastic trends
Journal of Monetary Economics 18, 4975] and Clark [Clark, Peter K., 1987. The cyclical component of
US economic activity. The Quarterly Journal of Economics 102 (4), 797814], the BN decomposition
attributes most variation in GDP to trend shocks while the cycles are short and brief. Since each is an
estimate of the transitory part of GDP that will die out, it seems natural to compare cycle measures by
their ability to forecast future growth. The results presented here suggest that cycle measures contain
little if any information beyond the short-term momentum captured by BN.