EXPLAINING BUSINESS CYCLES - A MULTIPLE-SHOCK APPROACH

Aiyagari (1992) and Prescott (1986, 1991) claim that a large fraction of the variance of United States quarterly detrended real GNP is attributable to an unobservable shock to total factor productivity. This paper argues that the importance of a productivity shock in explaining the variance of output is fundamentally indeterminate. Any model that is in accord with the several time series that make up United States macroeconomic data must feature multiple shocks that are correlated at all leads and lags. Sorting out the separate effects of these various shocks on a single variable such as real GNP is impossible. We illustrate this argument using a multiple-shock version of the King-Plosser-Rebelo (1988) benchmark real business cycle model.