陕西省人口分布影响因素的空间计量分析

人口分布影响因素研究有利于揭示人口分布规律，预判人口分布趋势。基于陕西省区县级人口、经济社会、自然地理等数据，通过因子分析方法和空间计量建模解析人口分布的影响因素。研究发现，人口地域别比率不仅取决于一个特定区县内可观测的经济社会、历史基础、自然地理等外在特征，还取决于该区县不可观测的、模型遗漏的其他共有特征，其中经济与公共服务因子、人口基底因子对人口分布具有最显著的正向解释力，其他因素影响较弱或统计不显著；城市等级可显著强化产业结构、人均收入和地形因素对人口分布的影响。研究认为，经济与公共服务因素是优化人口分布的关键，同时需考虑自然地理因素的限制作用。研究对人口分布优化政策的制定具有参考价值。

Factors influencing population distribution in ShaanxiProvince using spatial econometric analysis

When conducting empirical research on the factors influencing population distribution,it is helpful to understand population distribution and its evolutionary trends.This study considers all the 107 counties in Shaanxi Province,northwest China as research objects and attempts to use demographic,socioeconomic,and natural geographic statistical data for 2015 to fit an ordinary least squares (OLS) regression and spatial econometric model to analyze the factors influencing the Shaanxi Province’s population distribution.The demographic and socioseconomic data were extracted from the 2016 Shaanxi Regional Statistical Yearbook,by the Statistics Bureau of Shaanxi Province in December 2016.The Shaanxi Province county level administrative map was drawn using the National Earth System Science Data Sharing Infrastructure,National Science & Technology Infrastructure of China (http://www.geodata.cn).Natural geographical data were extracted,calculated,and analyzed from one kilometer (km) Resolution Digital Elevation Model Data Set of China from the Scientific Data Center of Cold and Arid Regions in China (http://westdc.westgis.ac.cn).The population spatial database was established using the ArcGIS 10.0 software program populated with the aforementioned data.The explanatory variable in our models is the Regional Proportion of Population (RPP),which is a proportion calculated by dividing the population of one county by the total population of the region.The RPP can effectively avoid the heteroscedasticity that may be caused by large differentials between the areas of the Shaanxi Province counties.The independent variables(economic and public service,population base,industrial structure,per capita Income,topography,and average elevation)were obtained through factor analysis to avoid overlooking variables and issues of multicollinearity.The OLS regression model demonstrates the fact that there is a significant relationship between RPP and the explanatory variables,as well as the dummy variables,defined in the model.The administration rank variable might play an important role in influencing the coefficient.However,the dependent variable and its residuals cannot satisfy the no spatial autocorrelation assumption,although they can satisfy the other Gauss Markov assumptions.Therefore,we also attempt to fit the spatial lag model and the spatial error model (SEM).The SEM is the best model based on Lagrange multiplier (LM),Robust LM,and Akaike information criterion tests.The SEM reveals that the RPP of a particular county in Shaanxi Province depends on not only the characteristics of observable variables but also other characteristics that may be unobservable or omitted from the model.Among the model〖JP8〗’〖JP〗s independent variables,the economic and public service factor exhibits the most significant positive explanatory power on the RPP.The population base factor,which represents basic agricultural conditions,and the population in the year 2000 demonstrate positive explanatory power.The industrial structure factor is negatively correlated with RPP,which indicates that the second industry has limited absorptive capacity in terms of increasing employment compared with the third industry.Average elevation has negative explanatory influence on RPP.The effects of per capita income and topography are insignificant,possibly because some counties with higher per capita incomes have economies based on natural resources or minerals and are often located in remote mountain areas.Nonetheless,topography exhibits a complex spatial coupling relationship with climate,precipitation,temperature,and humidity.The administration rank of a county influences population distribution significantly.Our main conclusion is that the key,controllable determinative factors for optimizing population distribution are socioeconomic factors,although the restrictive role of natural geographical factors should not be overlooked.By considering the spatial interactions between the explanatory variables and error terms,this study corrects the biased estimations of the OLS model and provides a scientific analysis of the factors influencing population distribution,which is of great reference value in terms of projecting as well as optimizing population distribution trends.