Capacity utilization and technology shocks in the US manufacturing sector

The ability of real business cycle models to generate reasonable aggregate fluctuations depends on the time series properties of technology shocks measured by the change of total factor productivity. Three specifications of a non-parametric productivity analysis which correct to different degrees for variations of capacity utilization are compared in this article using data for three- and four-digit US manufacturing industries during the years 1958-1996. The results show that correcting for utilization generally leads to substantially smaller technology shocks that are less strongly correlated with growth of output and hours. Moreover, the probability of technological regress is considerably lower after the correction.