中文摘要：

传统DEA—Malmquist生产率指数的测量与分析都建立在精确的投入产出数据基础之上，缺乏对非精确数据的分析与应用研究。针对投入为确定性而产出为区间型数据的情形，构建了基于有效前沿面的区间DEA—Malmquist指数，探讨了区间Malmquist指数及其分解部分的性质，并提出了一种综合可能度所有位次重要性的区间数排序方法，将构建的理论应用于2006年至2009年间全国11个主要沿海省市工业行业的全要素生产率分析，结果表明：天津和海南的工业行业全要素生产率呈现增长态势，其余省市则落在降低与增长的区间内，其中技术效率对全要素生产率的贡献起主导作用的有海南，辽宁，河北，广西，山东，广东，福建7个省份，而技术进步对全要素生产率的贡献起主导作用的地区有浙江，天津，上海，江苏省的技术进步和技术效率的作用相当．

英文摘要：

Typical Malmquist productivity index （MPI） research under the frame- work of data envelopment analysis is based on the precise data of input and output rather than the imprecise data. This paper investigates the interval MPI with pre- cise input and imprecise output based on the efficient frontier. Furthermore, we decompose the interval MPI into efficiency change （EC） and technical change （TC） to discern how deeply EC and TC will influence the MPI. In order to rank the interval MPI, interval EC and interval TC, we propose an approach which integrates all the ranking places for every decision making unit based upon the concerning theory of possibility. To validate the rationality and effectiveness of our approach, we apply it to the empirical analysis of industrial productivity of 11 coastal cities in China from the year 2006 to 2009. It turns out that EC and TC impact the rank of MPI, the leading factor of MPI differs from every DMU. It is necessary to distribute industrial recourse rationally and efficiently according to the leading factors and hints.