Usefulness of Pretests for Estimating Underlying Technologies Using Dual Profit Functions
通过模拟数据检验了超越对数与广义线性近似形式在对偶利润函数估计中的表现，发现凸性假设在部分模型中被拒绝，而单调性从未被拒绝，两种形式在估计替代弹性上无显著优劣。
Translog and generalized-linear approximating forms are used to estimate dual profit functions from stochastically simulated data generated from a known, relatively complex technology under profit-maximizing conditions. The hypotheses of monotonicity, convexity, and equality of parameters common to the profit function and demand equations are tested. This last hypothesis is always rejected and monotonicity is never rejected. Convexity is rejected for about a third of the translog models and fewer than 5 percent of generalized-linear models. Neither approximating form strongly dominates the other for accuracy in estimating the underlying elasticities of substitution. Copyright 1987 by Economics Department of the University of Pennsylvania and the Osaka University Institute of Social and Economic Research Association.