A Monte Carlo Experiment Demonstrating Specifications Bias in Translog Cost Systems
Driscoll, Paul; Boisvert, Richard N.
A Monte Carlo experiment comparing translog cost and production systems indicates that the cost systems' biased estimates of returns to scale and input substitution elasticity's result from an error in variables problem and a disturbance with non-zero mean. Guidelines for when the cost or production system is more appropriate are suggested
Charles H. Dyson School of Applied Economics and Management, Cornell University