基于函数型数据的中国经济区划分

The Division of China’s Economic Zones Based on the Functional Data

国家通常按地域来划分中国经济区域,即将同一地区经济发展水平大致相同的省份划为一类,但常常忽略了同一地区不同省份间经济支柱和产业结构的差异性.例如随着改革开放的深入,东北三省的产业结构和经济发展逐渐分化：辽宁省第二产业占比下降,核心技术创新能力不强,工业呈负增长态势;吉林省工业增加值稳步提升,但产业结构单一;黑龙江省工业化进程深度不够,仍处于低速发展状态.可见,基于地域的粗犷型划分,容易因省情的差异而导致政策不能达到预期效果,甚至可能会导致某些省份发展停滞或者倒退.其次,为了避免单一指标不能很好的刻画各省真实的经济发展状态,本文采用了对五个指标（GDP、财政收入、进出口总额、最终消费、全社会固定资产投资）加权生成一个新的综合指标来对各省经济进行分区.再者,由于数据的结构特征及数据间的高度相关性使得传统统计方法结果表现较差.基于函数型数据在高度相关数据中的优异表现,本文建议采用函数型数据聚类分析方法对中国31个省市的综合指标进行聚类分析.研究结果表明,基于函数型数据聚类分析进行的经济区域的划分,能够打破地域的限制,且能充分考虑各省份的产业结构,即把产业结构相近的省份划分在同一经济区城.最后,基于本文提出的区域划分方法能制定更加有效的经济发展政策和促进全国经济的平衡发展.

China usually divides the economic zones according to region, that is, divide provinces with similar economic development levels in the neighboring area into one category, but often ignores the differences of economic pillar and industrial structure between different provinces in the same region. For example, with the deepening of reform and opening up, industrial structure and economic development in northeast China gradually differentiate： the proportion of the second industry in Liaoning province is declining, and the core technology innovation ability is not strong, which leads to the negative growth of industry; the industrial added value of Jilin province increases steadily, but the industrial structure is single; the industrialization process in Heilongjiang province is not deep enough, and it is still in a low speed state. Therefore, based on the geographical rough division, it is easy to cause the policy cannot achieve the desired effect due to the difference in the provincial situation, and may even lead to stagnation or retrogression of some provinces. In order to avoid that a single indicator can not characterize the true economic development of each province, this paper uses a weighted average of five indicators（i.e.,GDP,Fiscal Revenue, Total Imports and Exports, Final Consumption, and Total Investment in Fixed Assets）to generate a new comprehensive index to divide the economy zones of the provinces. At the same time,the traditional statistical methods＇ performance is poor because of the structural characteristics and the high correlation of data. Based on the excellent performance of functional data in highly correlated data, this paper proposes to use functional data clustering analysis method to cluster the 31 provinces in China by the comprehensive index. The results show that the division of economic regions based on functional data clustering analysis can break the geographical constraints, and can take full account of the industrial structure of each province, that is, the provinces with similar industrial structure are divided into the same economic region. Finally, based on the classification method proposed in this paper, we can formulate some more effective economic development policies and promote the balanced development of the national economy.