SEPARATING TECHNICAL CHANGE FROM TIME-VARYING TECHNICAL INEFFICIENCY IN THE ABSENCE OF DISTRIBUTIONAL ASSUMPTIONS

This paper proposes an alternative approach for separating technical change from time-varying technical inefficiency. The approach uses the general index, developed by Baltagi and Griffin (1988), to model technical change along the production function, and a quadratic function of time trend, as in Cornwell, Schmidt and Sickles (1990), to capture the temporal pattern of technical inefficiency. In such a setting, all parameters associated with the rate of technical change and the temporal pattern of technical inefficiency are identified separately. Moreover, the proposed approach is independent of any distributional assumption concerning the one-sided error term associated with technical inefficiency, and can easily be estimated using FGLS. Comparative empirical results based on a translog production frontier function, and estimates of technical inefficiency and technical change are presented for the UK dairy sector over the period 1982-1992.