THE BAYESIAN COMPARISON OF CONDITIONAL INPUT DEMAND SYSTEMS IN THE PRESENCE OF INEFFICIENCY
Languages of publication
The main goal of the paper is to discriminate among stochastic conditional demand systems. In this paper the authoress presents the Bayesian comparison of conditional input demand functions derived from stochastic cost frontiers approximated by different flexible functional forms. The advantage of applying input demand (or cost share) systems over the production of cost frontiers relays on the fact that they allow for modelling both technical and allocative efficiency. The technical efficiency is related to the excessive inputs usage in a production process while the allocative inefficiency is related to the incorrect proportions of inputs. The composed error structure of each of the equations of the conditional demand systems is defined by the Bayesian random effects model, so-called common efficiency distribution model. In the empirical analysis there are considered three demand systems derived from the microeconomic short-run variable cost function approximated by three locally flexible functional forms: the translog, the Generalised Leontief and the Generalised McFadden. Details of prior specification are discussed on the basis of the short-run Generalised Leontief input demand system. The comparison of models is conducted by marginal likelihood that is calculated by implemetation of Markov Chain Monte Carlo terchniques. An example of input demand systems estimated using data obtained from 31 Polish electric power stations illustrates the methodology. The results show insignificant differences in the description of the underlying production process by competing systems but the authoress founds our significantly different explanatory power of various conditional demand models.
Publication order reference
CEJSH db identifier