正则化方法在生产函数识别中的应用

AN APPLICATION OF REGULARIZATION METHOD FOR IDENTIFYING THE PRODUCTIVE FUNCTION

本文讨论了Tikhnov的正则化方法在生产函数识别中的应用。它无需对技术效率的变动模式作先验的假定,可同时估计每个生产周期的技术效率和投入要素的产出弹性;同时,它对于因投入要素之间可能存在的强相关性而带来的“病态”有较强的处理能力。对于十多组实际投入产出数据的分析表明,当用普通的线性二乘法估计参数失效时,正则化方法在大多数情况都能给出比较合理的参数估计。

This paper discusses an application of Regularization Method for identifyins CobbDouglas Productive Function .In the application,there is no need to make the prior hypothesis on the pattern of technical progess, the rate of technieal progress for each period and the elasticity coefficients of all the input elements can be obtaianed simultaneously by this method. Moreover it is powerful to treat the ill-conditioned circumstances which may appear in the case where strong colinerity between the input elements exists . The comparative studies on more than-ten groups of real input-output data indicate that the rational parameter estimates can be obtained mostly by this method, but rarely by Ordinary Least Squares(OLS)