This study aims to measure the differentiations between actual government expenditures policies and experts' consensuses in Taiwan, and explore the determinants affecting these differentiations. Based mainly on the opinions of senior and experienced experts and scholars in different backgrounds, the conclusions are therefore developed. Scientific methods are adopted to transform these opinions into quantitative indicators, and the objective consensuses are then obtained and assessed the efficiency further. After qualitative analysis, the conclusions and suggestions could be the decision making basis. Since the efficiency of nonprofit organizations is usually difficult to assess objectively, the focal point of this research is combining quality and quantity information. The structure of the ideas accumulated from experiences will be found out by logical methods, and made these contents well arranged. Based on the results from the AHP, we employ Grey Association Analysis, by analyzing and comparing the associations and the differentiations of each policy-making determinant, to confirm uniformity of the analysis results. And further the Data Envelopment Analysis (DEA) is utilized to evaluate the appropriateness of size and structure of government expenditures and analyze the deviations from reasonable expenditures and their reasons. Therefore, this research integrates the utilization of Grey Association Analysis and DEA, so as to establish more objective pattern and carries on the case analysis, and to provide reference for government expenditure decision-making.