中文摘要：

Almon算法进行改造扩展可解决基于固定产业销售结构假定编制产业部门类型投入产出表出现负值的问题．根据Almon算法在迭代过程中避免负值的思想，基于固定产业销售结构假定下的推导公式，构造迭代公式，并针对迭代公式中可能出现负值的部分，引入比例因子，使在每一步的迭代中都能够避免负值，并且给出了迭代公式和比例因子的直观解释．为验证该方法的有效性，将扩展的Almon算法和传统负值处理方法进行实证比较，结果表明，扩展的Almon算法要明显优于传统负值处理方法，基于其编制出的投入产出表与官方公布投入产出表的误差更小．

英文摘要：

In this paper, the widely accepted Almon procedure was extended to estimate industry-by- industry IO tables based on fixed industry sales structure assumption to avoid negative flows. Based on the Almon procedure＇s point of avoiding negative flows during iteration and derivation formula of the fixed industry sales structure assumption, we constructed iterative formula, and adopted scale factors in the part where negative entries may arise to avoid negative flows in every iterative step. Also, we gave intuitive interpretations for iterative formula and scale factors. To test the validity of the method, empirical comparison between the extended Almon procedure and the traditional approach is employed. The extend Almon procedure is proved to have better performance than the traditional method, since it yields more accurate estimation than that of traditional method, against the official survey-based IO table.