Evaluating non-structural measures of the business cycle

This paper evaluates a number of non-structural measures of the business cycle. It adopts a structural definition of the cycle, interprets non-structural measures as noisy approximations, and seeks a proxy that is reliable across a variety of plausible trend-cycle structures. The results favor a consumption-based measure proposed by Cochrane (1994). Across a variety of structures, it has the highest correlation and coherence with structural cycles, and best matches their dynamic properties. When applied to U.S. data, consumption-based measures conform closely to the dates of NBER recessions. They also yield a strong negative correlation between the cyclical components of productivity and hours, a fact that deepens the challenge to models which emphasize technology shocks as the primary source of business cycles.