中国跨区域人口迁移流空间结构研究——描述、模型表达与预测

Research on the Spatial Structure of China’s Interregional Migration Flow: Description, Model Expression and Prediction

关于人口跨区域流动的空间结构研究与数理统计明显不足,尤其是鲜见对人口迁移流量数据自身运行规律和数据内生化过程的统计研究。描述、模型化表达与预测是人口迁移流空间结构研究的三个不应分割的系统过程。鉴于此,将人口迁移流空间结构的描述、模型化表达和预测的“基本程式”整合,采用乘法分量模型、对数线性模型和双边比例调整法分别来描述、模型化表达和预测我国人口迁移流空间结构特征。实证过程着眼于人口统计学的视角,应用2011-2017年中国流动人口动态监测调查数据和2000-2020年三次人口普查数据等资料,以长江经济带11省份和中国东部、中部、西部和东北四大区域之间人口迁移流为例,基于人口迁移流矩阵的内生规律和数据自我生成过程来全景式刻画研究我国跨区域人口迁移流空间结构。主要结果显示,列联表及其内生的乘法分量模型可以很好地描述具有二维结构的人口迁移流;[ODAOA]四参数的非饱和对数线性模型可以刻画中国跨区域人口流动空间结构的主要特征,拟合程度达到82%;人口迁移流预测意义重大但探索不足,双边比例调整法可作为人口迁移流预测的良好工具。研究主要目的是尝试引入和搭建国内人口迁移流空间结构研究的体系框架或基本程式,推进我国人口空间统计或空间人口学的研究与应用。

The research on the spatial structure and mathematical statistic of the cross-regional population flow is obviously insufficient, especially the statistical research on the operation law of the population migration data itself and the internalization process of the data is very rare. In this article, description, model expression and prediction are three separate systematic processes which should not be separated in the study of spatial flow structure of population migration. In view of this, this paper integrates the “basic formula” of descripting, modelling and predicting the spatial structure of migration flows: multiplicative component model, log-linear model and bi-proportional adjustment method are used to describe, model and predict the spatial structure characteristics of migration flow in China. From the perspective of demographics, the empirical process applies the data of China Migration Dynamics Survey from 2011 to 2017 and the three census data from 2000 to 2020,and takes the population migration between the 11 provincial areas in the Yangtze River Economic Belt and the four major regions of China’s east, central, west and northeast flows as an example. Based on the endogenous law of the migration flow matrix and the data self-generation process the spatial structure of cross-regional population migration flow in China is studied comprehensively. The main results show that the contingency table and its endogenous multiplicative component model can well describe the population migration flow with a two-dimensional structure. The unsaturated log-linear model with four-parameter [O D A OA] can describe the cross-regional main characteristics of the population flow structure in China, with a fitting degree reaches 82%. The prediction of migration flow is of great significance, but the exploration is insufficient. The two sided proportional adjustment method can be used as a good tool for the population migration flow prediction. The main purpose of the article is to introduce and build a system framework or basic program for the study of the spatial structure of population migration, and to promote the research and application of spatial statistics or spatial population in China.