Modelling stochastic and cyclical components of technical change. An application of the Kalman filter

Abstract

In this paper, the Kalman filter is applied to the task of estimating the rate and direction of change in the technology of production at a micro level. The framework is the familiar system of factor-demand equations derived from a cost function. The state of technology, a latent variable, is modeled as a stochastic trend. In addition, estimates of total-factor productivity are corrected for measurement error that induces a procyclibal bias. As a result of decoupling trend and cyclical components and using state-space estimation techniques, significant cost changes are uncovered that fail to be detected when more traditional methods are employed. The application is to the U.S. primary-metals industry and the estimates appear to be consistent with the stylized facts in this sector. © 1989.