A multinomial logistic model for ranking technical efficiency of public project

Ranking efficiency of large projects based on Data Envelopment Analysis (DEA) could be used for grouping decision-making units. Utilizing the expert score tool on super efficiency results, we propose a multinomial regression model which aims to achieve the selection purpose of new public constructed projects. The efficiency of decision making units represented is estimated using the nonparametric frontier model of DEA and statistical inference for estimators is based on bootstrap. Conclusion is conducted that it ought to have at least one influential environmental factor that plays significant roles in terms of ranking public projects performance. And the method would be beneficial to the government decision-making process in public resource allocation and operation problems.