Estimating capacity utilization from survey data

In this paper, we review the history and concepts behind the Federal Reserve's measures of capacity and capacity utilization, summarize the methods used to construct the measures, and describe the principal source data for these measures--the Census Bureau's Survey of Plant Capacity. We show that the aggregate manufacturing utilization rate from the Survey of Plant Capacity does not exhibit the \"cyclical bias\" possessed by utilization rates from the less statistically rigorous utilization rate surveys previously used to estimate the Federal Reserve's measures. At the detailed industry level, utilization rates from the Survey of Plant Capacity for several industries do appear to possess a cyclical bias, but we demonstrate that this bias is removed in the construction of the Federal Reserve capacity measures. We further show that the Federal Reserve measures, by combining the Census survey utilization rates with other indicators of capacity, do not discard significant information contained in the Census rates. In fact, the Federal Reserve procedures add to the predictive content of the Census utilization rates in models of capital spending, capacity expansion, and changes in price inflation.