Title: Regular Variable Returns to Scale Production Frontier andEfficiency Measurement
定常变量回归规模化生产前沿与效率测量
李崇高
香港树仁大学、合肥工业大学
曾俊基
香港树仁大学、香港浸会大学
李树甘
香港树仁大学
何心巨
广西大学
The most frequently used empirical production frontier in data envelopment analysis, the variable returns to scale frontier, has a convex technology set and displays a special structure in economics, called the regular variable returns to scale in this paper; the production technology exhibits increasing returns to scale at the beginning of the production process followed by constant returns to scale and decreasing returns to scale. When the assumption of convexity is relaxed, modeling regular variable returns to scale becomes difficult, and currently, no satisfactory solution is available in multioutput production. Overcoming these difficulties, this paper adopts a suggestion in literature to incorporate regular variable returns to scale into the free disposal hull frontier under multiple outputs. We establish a framework for analyzing regular variable returns to scale and recommend an empirical production frontier for measuring technical efficiency with such pattern and multiple outputs. In the presence of regular variable returns to scale without convexity, the value of the technical efficiency measure computed from this new frontier is closer to the “true” value than that from the free disposal hull frontier, and the conventional variable returns to scale frontier may cause misleading implications.
在数据包络分析（DEA）中，最常用的经验生产前沿是具有规模报酬变化的前沿，即变量规模报酬前沿。这个前沿具有凸技术集，并在经济学中展现出一种特殊结构，本文称之为规则变量规模报酬；生产技术在生产过程的初期表现出规模报酬递增，随后是规模报酬不变和规模报酬递减。当放松凸性假设时，对规则变量规模报酬进行建模变得困难，目前在多产出生产中尚无满意的解决方案。为了克服这些困难，本文采纳了文献中的一个建议，将规则变量规模报酬纳入多产出情况下的自由处置壳前沿。我们建立了一个分析规则变量规模报酬的框架，并推荐了一个经验生产前沿，用于在这种模式和多产出情况下测量技术效率。在没有凸性的规则变量规模报酬存在的情况下，从这个新前沿计算出的技术效率值比从自由处置壳前沿得出的值更接近“真实”值，而传统的变量规模报酬前沿可能会导致误导性的结论。
Tips：“社科人工智能与人工智能经济学” 冬季学术研讨会即将举办，欢迎对人工智能经济学方法及其应用感兴趣的学者和学生报名！
疯狂暗示↓↓↓↓↓↓↓↓↓↓↓
