Unraveling Trend and Stationary Components of Total Factor Productivity

We propose a methodology to analyze the dynamic features of total factor productivity (TFP). Factor efficiency is assumed to evolve according to an unobserved component model which has the form of a dynamic version of factor analysis and which nests most of the specifications for technology shocks used in the literature. We show under which conditions the processes generating the components can be identified from measurements of TFP only. We also discuss how additional information on the technology shocks can be obtained from the factor demand equations associated with the production technology. The methodology is applied to quarterly data for the manufacturing industry in the Netherlands 1971.1-1990.IV. Technology is found to be driven by a stochastic trend with a first order moving average part. This finding is similar to results obtained by Lippi and Reichlin (1994) among others; The findings for TFP corroborate earlier findings for the dynamic features of technology in labor and cpaital demand equations for the manufacturing industry in the Netherlands.