Confidence Intervals for Elasticity Estimators in Translog Models.

This paper examines the distribution fu nctions of elasticity estimators in translog demand models. The authors consider the normaland ratio-of-normals distributions and present confidence intervals for the elasticity estimators. The results suggest that only elasticity estimato rs based on the means of the actual cost shares are likely to follow either the normal or ratio-of-normals distribution function. Examination of three published empirical studies demonstrates that inferences regarding the values of elasticities cannot be made from point estimates alone and suggests a trade- off between the level of aggregation and the width of confidence intervals for the elasticit y estimators. Copyright 1986 by MIT Press.