产业结构数据的等距logratio变换与应用

Isometric logratio Transformation of Industrial Structure Data and Its Application

产业结构数据是典型的成分数据,其在经济问题研究中具有重要的地位,但定和约束却严重制约着其在统计分析中的应用。文章通过构建适合产业结构变迁特征的正交向量基,使用产业结构数据的等距logratio变换,将具有定和约束的产业结构数据转换为无约束的向量数据,研究结果表明:该变换不仅解除了定和约束,而且确保了变换前后数据的拓扑结构不变性,特别是变换后的数据还具有很好的经济解释力。运用该变换不仅可以直观呈现一国(或地区)产业结构的演变轨迹,而且可以与不带限制条件的传统分析方法有机结合。通过对中国和部分发达国家产业结构的实例分析,以及与传统计量方法的比较,验证了所提方法的合理性和优越性。

Industrial structure data is a typical compositional data,which plays an important role in the study of economic problems,but its application in statistics and econometric analysis is restricted by the fixed-sum constraint.By constructing the orthogonal vector basis suitable for the characteristics of industrial structure change,this paper uses the isometric logratio transformation of industrial structure data to transform the industrial structure data with fixed-sum constraint into unconstrained vector data.The results show that the transformation not only removes the fixed sum constraint,but also ensures the topological structure invariance of the data before and after the transformation,especially when the transformed data also has a good economic explanatory power,and that using this transformation can not only present the evolution track of a country,s(or region,s)industrial structure directly,but also combine with the traditional analysis methods without restrictions.The rationality and superiority of the proposed method are verified by analyzing the industrial structure of China and some developed countries and comparing with the traditional measurement methods.