中文摘要：

建立基于DEA模型的城市效率投入产出指标体系，以2005-2009年甘肃省12个地市截面数据为样本，采用超效率DEA模型对其城市效率进行测算，以克服传统数据包络分析模型对有效的决策单元无法做进一步区分的困惑．同时，运用盯收敛甘肃省城市效率变化趋势进行收敛性检验．结果发现：甘肃省城市效率值一般，城市之间存在着一定的差异性，并且BCC模型下效率值为1的城市数量要大于CCR模型下效率值为1的城市数量；超效率DEA值整体变化比较稳定，但少数城市的超效率值变化剧烈，超效率值大于1的城市数目基本保持不变；陇东地区城市效率值最高，陇中地区城市效率值最低，河西地区和陇南地区介于两者之间，甘肃省城市效率呈现出“两边高中间低”的格局；全省城市效率差异呈现出先扩大后缩小的趋势，河西地区、陇中地区、陇东地区三大区域内部的城市效率差异均呈现出扩大的趋势，而陇南地区则正好相反．

英文摘要：

Based on the data drawn from 12 cities in Gansu province during 2005-2009, an input-output index system was built to evaluate the urban efficiency thereof. The super efficiency model was employed to avoid the disadvantage that the results from a traditional DEA evaluation might have some DMUs with efficiency equaling 1; meanwhile, the convergence of the changing trend of urban efficiency in Gansu was examined by a convergence and the results showed that the efficiency of 12 cities in Gansu province was just at a general level. The number of DMUs with efficiency equaling 1 under BCC model was greater than the CCR model. The number of cities whose super-efficient DEA value was greater than 1 remained stable. The overall super-efficient DEA values generally kept stable except for some dramatic changes in a few cities. The urban efficiency of Longdong was the highest, and the lowest was Longzhong, with Hexi and Longnan in between. The spatial pattern of urban efficiency of Gansu province was that ＂both sides are higher than the middle＂. The differences between 12 cities showed a narrowing trend in the first expansion. The differences between cities in Longnan showed a trend of expansion, while the other areas were in an opposite direction.