Third-Order Translog Utility Functions
从理论和实证角度比较了二阶与三阶超越对数效用函数，发现三阶近似能更准确刻画偏好结构，避免参数估计不一致，对政策决策更有帮助。
Abstract This article examines the advantages of estimating a third-order rather than a second-order translog utility function in a theoretical and empirical context. It is demonstrated that the rigor of tests for appropriate functional form is increased by increasing the order of approximation. An empirical example demonstrates that a second-order approximation can lead to inconsistent parameter estimates, whereas the third-order translog allows for better modeling of the preference structure and more consistent estimates for policy decisions.