Structural Time Series Modelling of Capacity Utilisation

In this paper we introduce a structural non-linear time series model for joint estimation of capacity and its utilisation, thereby providing the statistical underpinnings to a measurement problem that has received ad hoc solutions, often underlying arbitrary assumptions. The model we propose is a particular growth model subject to a saturation level which varies over time according to a stochastic process specified a priori. A bivariate extension is discussed which is relevant when survey based estimates of utilization rates are available. Illustrations are provided with respect to the US and the Italian industrial production.