Convexity, quality and efficiency in education
本文指出数据包络分析（DEA）在教育效率评估中假设凸性可能不成立，尤其当包含教育产出质量指标时，会高估技术效率改进空间、低估配置效率重要性，并提出了替代的非参数方法和产出估值函数。
While Data Envelopment Analysis (DEA) has many attractions as a technique for analysing the efficiency of educational organisations, such as schools and universities, care must be taken in its use whenever its assumption of convexity of the prevailing technology and associated production possibility set may not hold. In particular, if the convexity assumption does not hold, DEA may overstate the scope for improvements in technical efficiency through proportional increases in all educational outputs and understate the importance of improvements in allocative efficiency from changing the educational output mix. The paper therefore examines conditions under which the convexity assumption is not guaranteed, particularly when the performance evaluation includes measures related to the assessed quality of the educational outputs. Under such conditions, there is a need to deploy other educational efficiency assessment tools, including an alternative non-parametric output-orientated technique and a more explicit valuation function for educational outputs, in order to estimate the shape of the efficiency frontier and both technical and allocative efficiency.