Appraising elasticities in small samples. A bayesian approach

This paper presents the bayesian approach to analyze small sample elasticity distributions with locally regular flexible functional forms. It is known that for a translog cost function these are t student conditional on factor shares, otherwise they are a nonlinear combination of parameters and predictive, whose distributional properties cannot be analytically derived. However a simple Monte Carlo composition method can be used to approximate moments even with inequality constraints such as monotonicity and concavity. This approach is applied to the well known Berndt-Wood data set.